أثر راجيف بانكر الدائم على تحليل كفاءة البيانات Rajiv Banker’s lasting impact on data envelopment analysis

المجلة: Annals of Operations Research، المجلد: 351، العدد: 2
DOI: https://doi.org/10.1007/s10479-025-06473-3
تاريخ النشر: 2025-02-24

أثر راجيف بانكر الدائم على تحليل كفاءة البيانات

علي إمروزنجاد (د) ⋅ فيكتور بودينوفكسي (د) ⋅ فنسنت تشارلز (د) ⋅ تشيكسياو لو (د) ⋅ أمير مرادي مطلق (د)

تاريخ الاستلام: 6 سبتمبر 2024 / تاريخ القبول: 1 يناير 2025 / تاريخ النشر على الإنترنت: 24 فبراير 2025
© المؤلفون 2025

الملخص

تقدم هذه الورقة تحليلًا شاملاً للأثر الكبير للبروفيسور راجيف بانكر في مجال تحليل تطوير البيانات (DEA). من خلال مراجعة شاملة لمساهماته الأكاديمية، نستكشف ثلاث مجموعات رئيسية ضمن أبحاث DEA: (1) العوائد على المقياس (RTS) وحجم المقياس الأكثر إنتاجية (MPSS)، (2) الاستدلال الإحصائي في DEA، و(3) التحليل السياقي. لقد تقدم البحث الرائد لبانكر بشكل كبير منهجيات DEA، معالجًا التحديات الأساسية المتعلقة بكفاءة المقياس، والصلابة الإحصائية، وتأثير المتغيرات السياقية على الأداء. لقد جسر عمله بين التطورات النظرية والتطبيقات العملية، مؤثرًا في مجالات متنوعة مثل الاقتصاد، والمالية، وعلوم الإدارة. من خلال دراسة اتجاهات الاقتباس والبيانات الببليومترية، نتتبع تطور وأهمية مساهماته المستمرة، مع تسليط الضوء على الأوراق الرئيسية التي شكلت مسار أبحاث DEA. تناقش هذه الورقة أيضًا تطور نماذج وأساليب DEA، بما في ذلك دمج العناصر العشوائية والتحليلات من المرحلة الثانية. في الاعتراف بتفاني بانكر مدى الحياة في DEA، نحتفل بإرثه الدائم وتأثيره التحويلي على كل من المجتمع الأكاديمي والتطبيقات العملية لـ DEA في جميع أنحاء العالم.

الكلمات الرئيسية: راجيف بانكر • تحليل تطوير البيانات • العوائد على المقياس • حجم المقياس الأكثر إنتاجية • الاستدلال الإحصائي في DEA • التحليل السياقي

1 المقدمة

يعد تحليل تطوير البيانات (DEA) منهجية أساسية في مجال تقييم الكفاءة والأداء عبر مجالات متنوعة، بما في ذلك البحث التشغيلي، وعلوم الإدارة، والاقتصاد والمالية، والرعاية الصحية، والتعليم. تم تصور DEA في البداية من قبل تشارنز وآخرين (1978)، ويقدم إطارًا قويًا لتقييم الكفاءة النسبية لوحدات اتخاذ القرار (DMUs) التي تحول مدخلات متعددة إلى مخرجات متعددة. في جوهره، يعمل DEA على مبدأ مقارنة كفاءة DMUs بناءً على علاقات المدخلات والمخرجات الخاصة بها، دون الحاجة إلى افتراضات صريحة حول وظيفة الإنتاج الأساسية. تميز هذه الطريقة غير المعلمية DEA عن الطرق الاقتصادية التقليدية، مما يجعلها مناسبة بشكل خاص للحالات التي تكون فيها الشكل الوظيفي لعملية الإنتاج غير معروفة أو يصعب تحديدها. تكمن مرونة DEA في قدرته على استيعاب مدخلات ومخرجات متعددة في وقت واحد، مما يمكّن من تقييمات كفاءة شاملة عبر أبعاد مختلفة. من خلال تحديد DMUs الفعالة التي تعمل كنقاط مرجعية، يسهل DEA تحسين الأداء من خلال تسليط الضوء على المجالات التي يمكن أن تعزز فيها الوحدات غير الفعالة إنتاجيتها.
على مر السنين، تطور DEA من صيغته الأولية ليشمل مجموعة واسعة من التقدمات المنهجية والتطبيقات. تستكشف هذه الورقة الإرث الدائم لبانكر في مجال DEA، مع تسليط الضوء على مساهماته مدى الحياة وتأثيرها على تشكيل المنهجية وتطبيقاتها. من خلال التعمق في أبحاث بانكر الرائدة وتفانيه الثابت في تعزيز DEA، نسعى لتقديم نظرة شاملة على تأثيره الكبير في هذا المجال المؤثر من الدراسة.
تتجلى مساهمات بانكر الأساسية في منهجية DEA في منشوراته العديدة، التي حظيت باعتراف وتأثير واسع. ورقته لعام 1984، “بعض النماذج لتقدير الكفاءات الفنية وكفاءة المقياس في تحليل تطوير البيانات”، التي تم الاقتباس منها أكثر من 27000 مرة وفقًا لجوجل سكولار (في وقت كتابة هذه الورقة)، قدمت نماذج قوية لتقدير الكفاءات الفنية وكفاءة المقياس، مما صقل الأسس النظرية لـ DEA. ورقته لعام 1989، “الحساسية، والدقة، والتجميع الخطي للإشارات لتقييم الأداء”، التي تم الاقتباس منها أكثر من 1300 مرة وفقًا لجوجل سكولار (في وقت كتابة هذه الورقة)، عززت بشكل أكبر صرامة DEA وقابليتها للتطبيق. استمر في البناء على هذا الأساس من خلال ورقته لعام 1993، “الاحتمالية القصوى، والاتساق، وتحليل تطوير البيانات: أساس إحصائي”، التي قدمت أسسًا إحصائية للتقنية. بالإضافة إلى ذلك، عمله على تقنيات DEA المتقدمة لتقييم مقاييس الأداء غير المالية، كما يتضح في ورقته “تحقيق تجريبي لخطة حوافز تتضمن مقاييس أداء غير مالية”، وسع نطاق DEA إلى ما هو أبعد من المقاييس المالية التقليدية. كانت هذه التقدمات حاسمة في تعزيز التطبيقات العملية لـ DEA عبر صناعات مختلفة، بما في ذلك البنوك، والرعاية الصحية، والتعليم، والزراعة. تظهر هذه الأوراق المؤثرة للغاية التأثير الكبير لعمل بانكر في مجال DEA.
بعيدًا عن تقدمه المنهجي، لعب بانكر دورًا محوريًا في نشر معرفة DEA وتعزيز مجتمع DEA النابض. لا يزال كتابه المدرسي المشترك، “المحاسبة الإدارية”، حجر الزاوية في تعليم DEA، حيث يقدم التقنية لأجيال من المحللين والباحثين الطموحين.
تقدم هذه الورقة تكريمًا شاملاً لمساهمات بانكر مدى الحياة في DEA، مستكشفة أبحاثه الرائدة والتزامه الثابت بنشر الموضوع. تبدأ بمراجعة عملية تغيير النموذج في سياق العوائد على المقياس (RTS) وحجم المقياس الأكثر إنتاجية (MPSS) في DEA (القسم 2).
يركز القسم 3 على مساهمات بانكر في الأساس الإحصائي لـ DEA، وبناء إطار اختبار الفرضيات وإطار محاكاة مونت كارلو في DEA. بالإضافة إلى ذلك، يتم مناقشة التطورات النظرية الحديثة والتحولات المحتملة في النموذج. في القسم 4، يتم تقديم مساهمة بانكر في التحليل السياقي القائم على DEA. يتم إنشاء شبكة التأليف بناءً على مسيرة بانكر الأكاديمية في القسم 5. ثم، يتم تقديم تحليل ببليومتري شامل لمساهمات بانكر الأكاديمية على المستويين الكلي والجزئي في القسم 6. أخيرًا، يختتم القسم 7 الورقة.

2 العوائد على المقياس وحجم المقياس الأكثر إنتاجية

بعد نموذج CCR-DEA الأولي الذي طوره تشارنز وآخرون (1978)، ناقشت العديد من الدراسات المفهوم الاقتصادي للعوائد على المقياس (RTS) من خلال نهج مختلفة. يقتصر الجزء الأكبر من المناقشة على التوصيفات النوعية للعوائد على المقياس. على سبيل المثال، ما إذا كانت العوائد على المقياس تُعرف بأنها متزايدة، أو متناقصة، أو ثابتة. دع تمثل النسبة، الزيادة النسبية في المدخلات والزيادة النسبية في المخرجات ممثلة بـ ، على التوالي. وبالتالي، . في نظرية الإنتاج التقليدية، يتم تعريف أنواع العوائد على المقياس بناءً على قيمة النسبة بين الزيادة النسبية في المخرجات والزيادة النسبية في المدخلات في سيناريو مخرج واحد. على وجه التحديد، عندما تكون الزيادة النسبية في المدخلات أكبر من الزيادة النسبية في المخرجات ( )، تكون العوائد على المقياس متناقصة. على العكس، إذا كانت الزيادة النسبية في المخرجات أكبر من الزيادة النسبية في المدخلات ( )، تكون العوائد على المقياس متزايدة. تعتبر العوائد على المقياس ثابتة إذا كانت . تم توسيع مفهوم العوائد على المقياس إلى سيناريوهات متعددة المخرجات من خلال عمل بانكر (1984)، بانكر وآخرين (1984)، وبانكر وثرال (1992). من خلال التحقيق في حالة العوائد على المقياس ونقطة الحجم الأكثر إنتاجية لوحدة اتخاذ القرار (DMU)، يدعم نموذج DEA عملية اتخاذ القرار في صياغة السياسات، وتطوير استراتيجيات الأعمال، وتخصيص الموارد. بشكل عام، تحدد نتائج التحليل استراتيجيات على المستوى الكلي لتعظيم إنتاجية وحدة اتخاذ القرار (DMU). على سبيل المثال، سويوشي وغوتو قيمت شروط RTS للأداء التشغيلي لمحطات الطاقة التي تعمل بالفحم ومحطات الطاقة التي تعمل بالوقود الأحفوري في السوق الأمريكية. بالنسبة لوحدات القياس التي تظهر زيادة في RTS، يُوصى بزيادة الحجم التشغيلي للوصول إلى نقطة MPSS وزيادة الإنتاجية. على العكس من ذلك، يمكن لوحدات القياس التي تعاني من انخفاض RTS تقليل حجمها التشغيلي لتعزيز الإنتاجية. في تطوير نظرية تحديد RTS، اقترحت الدراسات طرقًا لتحديد شروط RTS في عمليات الإنتاج الأكثر تعقيدًا والسيناريوهات متعددة الفترات (انظر القسم 2.3.1 لمزيد من التفاصيل). كما تم توسيع مفهوم RTS ليشمل الضرر على المقياس (DTS)، مما يوفر دعمًا إضافيًا لعمليات اتخاذ القرار المتعلقة بالاستدامة (انظر القسم 2.3.2 لمزيد من التفاصيل).
نظرًا لأن تحديد RTS مرتبط ارتباطًا وثيقًا بتحديد MPSS، نستعرض أولاً الأوراق الأصلية حول RTS وMPSS من قبل بانكر في القسمين 2.1 و2.2. يتم مناقشة التقدم الأخير في تحديد RTS وMPSS في القسم 2.3. يتم توضيح تطوير النظرية في المرحلة المبكرة من قبل بانكر في الشكل 1.

2.1 تحديد RTS وMPSS من قبل بانكر

أشار بانكر وآخرون (1984) إلى أن كفاءة CCR يمكن تفكيكها إلى كفاءة تقنية وكفاءة مقياس. للمرة الأولى في الأدبيات، قام بانكر (1984) بتوسيع مفهوم MPSS، الذي اقترحه فريش (1964)، إلى نظرية DEA. لتمييز المفهوم
الشكل 1 مراحل تطوير تحديد RTS وMPSS في DEA من قبل بانكر
من MPSS عن مفهوم مزيج التكلفة الأدنى، عرف بانكر (1984) MPSS بموجب نظرية DEA كنقطة وحدة على حدود الكفاءة التي تعظم الإنتاجية المتوسطة لمزيج المدخلات والمخرجات المعطاة، وبعدها يبدأ انخفاض RTS. كما تم اقتراح نموذج للتحقيق في نقطة MPSS في هذه الورقة. بالإضافة إلى ذلك، اقترح بانكر (1984) نهجًا جديدًا لتحديد النوع الفريد من RTS بناءً على نموذج CCR. من خلال توسيع عمل بانكر (1984)، يمكن تطبيق النموذج المقترح من قبل بانكر وآخرون (1984) بموجب فرضية VRS. كلا النموذجين المقترحين من قبل بانكر (1984) وبانكر وآخرون (1984) يستندان إلى فرضية أن مجموعة إمكانيات الإنتاج (PPS) المقدرة بواسطة نموذج DEA موجودة فقط كطائرة دعم فريدة. ومع ذلك، تشير الأدبيات إلى أنه في التطبيقات التجريبية، تسبب النماذج المقترحة من قبل بانكر (1984) وبانكر وآخرون (1984) مشكلة حيث قد تحدث طائرات دعم متعددة (على سبيل المثال، في الشكل 2، كلا النقطتين B وC هما نقاط MPSS)، كما هو موضح في تشارنز وآخرون (1986) وسيفورد وثرال (1990). نعتبر هذه الفترة كمرحلة ما قبل النموذج (كون، 1970)، حيث أن النهج المقترح له عيوب واضحة.
طور بانكر وثرال (1992) إطارًا صارمًا لمعالجة العيوب المحددة في بانكر (1984) وبانكر وآخرون (1984). يسمح النموذج المقترح بإمكانية وجود طائرات دعم متعددة. جادل بانكر وثرال (1992) بأن القضايا يمكن حلها من خلال تحديد حدود ميل الطائرات الداعمة المتعددة للنقاط الموجودة على حدود مجموعة PPS المحدبة. كما اقترحوا نموذجًا مطابقًا لتحديد حدود الميل ونهجًا للتحقيق في RTS وMPSS في حالة وجود طائرة دعم فريدة. بالإضافة إلى ذلك، اقترح بانكر وثرال (1992) أيضًا معيارين عامين للتحقيق في RTS وMPSS تم تطبيقهما على نطاق واسع في الأبحاث اللاحقة:
(1) مفهوم RTS محدد جيدًا فقط للنقاط الموجودة على حدود PPS. بالنسبة للنقاط الداخلية، لا يمكن التحقيق في RTS حيث أن التغيرات في الإنتاجية بسبب RTS تتداخل مع التغيرات في الإنتاجية بسبب القضاء على عدم الكفاءة.
(2) يرتبط مفهوم MPSS ارتباطًا مباشرًا بـ RTS. على وجه التحديد، الشعاع من نقطة الأصل (المعروف باسم في الشكل 2) إلى نقطة الحدود يتقاطع مع PPS في نقطتين. إذا كانت نقطة الحدود هي الأولى من هاتين النقطتين، فإنها تشير إلى زيادة RTS. على العكس من ذلك، إذا كانت نقطة الحدود أبعد من الأصل، فإنها تشير إلى انخفاض RTS.
الشكل 2 PPS مع نقطتين MPSS (بانكر وآخرون، 2011)

2.2 تطوير النظرية في تحديد RTS وMPSS من قبل بانكر

تم تطوير العديد من الدراسات الموسعة بناءً على المعايير التي اقترحها بانكر وثرال (1992). تركز هذه الدراسات بشكل أساسي على تحسين أداء الحساب للنموذج المقترح. على سبيل المثال، اقترح فاري وآخرون (1985) نهجًا جديدًا لتحديد RTS بناءً على مجموعة من النسب من القياسات الشعاعية الناتجة عن أزواج النماذج التي تختلف في تلبية شروط المحدبة والمحدبة الفرعية. جادل تشانغ وغوه (1991) بأن نهج بانكر (1984) وبانكر وآخرون (1984) غير صالح بسبب الافتراض الضمني لوظائف الإنتاج الخطية. على وجه التحديد، أصروا على أن النتيجة 1 (التي تنص على أنه إذا كان وزن الكثافة الإجمالية أكبر من 1، فإن وحدة القياس قيد التقييم تشير إلى زيادة RTS) التي اقترحها بانكر (1984) غير صالحة في نموذج CCR. تم تقديم مثال مضاد في ملحق تشانغ وغوه (1991). ردًا على ذلك، أشار بانكر وآخرون (1996ب) إلى أن النموذج المقترح من قبل فاري وآخرون (1985) يعادل نموذج بانكر وثرال (1992). من خلال إثبات العلاقة المعادلة بين النهجين، أزال بانكر وآخرون (1996ب) بعض الافتراضات الأساسية للنظريات في فاري وآخرون (1985). على وجه التحديد، جادلوا بأن الافتراض المتعلق بالقدرة على التخلص يمكن إزالته من خلال تعيين معاملات صفرية للمتغيرات الزائدة في أهداف نموذج BCC وCCR. عزز بانكر وآخرون (1996أ) سهولة الحساب لنهج بانكر وثرال (1992) من خلال توفير أشكال أبسط لتنفيذ نظريات بانكر-ثرال. تنفي النهج المقترح الحاجة إلى التحقيق في جميع الأمثل البديلة عند استخدام نموذج BCC.
تم توسيع النموذج المبسط المقترح بشكل أكبر إلى نموذج CCR من قبل بانكر وآخرون (1996ب).
يحدد الإطار المقترح من قبل بانكر (1984) وبانكر وآخرون (1984) وبانكر وثرال (1992) RTS وMPSS في سياق نظرية DEA. أسس بانكر وآخرون (1996أ، ب) الصلة بين المسارين لحل مشكلة تحديد RTS وسهلوا عملية حساب النموذج المقترح. تشكل هذه الأوراق الأساس لتحديد RTS وMPSS في نظرية DEA. بالإضافة إلى ذلك، ناقش بانكر وآخرون (2004أ، ب) تحديد RTS وMPSS من خلال نماذج مختلفة، بما في ذلك نماذج CCR وBCC الموجهة للمدخلات، بالإضافة إلى النموذج المضاعف الموجه للمخرجات.

2.3 الثورة والتقدم

تم نشر العديد من الدراسات بناءً على أعمال بانكر وآخرون (1984، 1996أ، ب، 2004أ، ب). مؤخرًا، يمكن تقسيم الأدبيات حول RTS وMPSS إلى مجموعتين. تهدف المجموعة الأولى إلى معالجة الفجوة البحثية في عمل بانكر، حيث يقتصر نهج تحديد RTS وMPSS على تكنولوجيا CRS وVRS التقليدية. تمتد الأدبيات إلى النماذج غير الشعاعية وتقترح هياكل إنتاج أكثر تعقيدًا مثل الشبكات والهياكل الديناميكية. المجموعة الثانية توسع تعريف RTS وMPSS إلى مفاهيم اقتصادية ذات صلة تتعلق بتوصيفات المقياس. بناءً على عمل بانكر، تعرف هذه المجموعة من الأدبيات المفاهيم الاقتصادية مثل معدلات الهامش المختلفة ومرونة المقياس وتحددها في سياق نظرية DEA. نستعرض الأدبيات المتعلقة بهاتين المجموعتين في هذا القسم.

2.3.1 الثورة والتقدم في النمذجة

نصنف المنشورات المتعلقة بتطوير النمذجة إلى ثلاث مجموعات كما هو موضح في الشكل 3.
أحد النماذج الموسعة لتحديد العوائد المتناقصة في حالة غير موجهة هو نهج قياس الفجوة القائم على Slack الذي اقترحه وو وآن (2013). اقترح فورسوند وآخرون (2007، 2009) التعامل مع قياس العوائد المتناقصة في نماذج DEA غير الشعاعية من خلال شروط الفجوة التكميلية القوية من نظرية التحسين. اقترح كريفونوجكو وآخرون (2014) وقدموا دعماً لنهج لتقليل التعقيد الحسابي في فورسوند وآخرون (2007، 2009). قام فوكوياما (2000، 2003) بتوسيع أدبيات بانكر إلى نموذج دالة المسافة الاتجاهية (DDF) لتحديد العوائد المتناقصة ونقطة MPSS. حدد كيرستنز وإيكوت (1999) العوائد المتناقصة من خلال نموذج الفجوة الحرة (FDH)، الذي تم خطيته بواسطة بودينوفيسكي (2004) للحصول على ميزة حسابية. اقترح سليماني-دامانه وآخرون (2006) وسليماني-دامانه ورشادي (2007) أيضًا خوارزمية زمنية متعددة الحدود لتحسين الحسابات
الشكل 3 تقسيم تطوير النموذج بناءً على هيكل النموذج وأنواع البيانات
الميزة. تشمل الأدبيات لتحديد MPSS وRTS في نماذج DEA المقيدة بالوزن الدراسات التي أجراها لطفي وآخرون (2007)، كورهونين وآخرون (2011)، وتون (2001).
اقترح غولاني ويو (1997) نهجًا جديدًا لتقدير RTS في نموذج BCC بدقة من خلال مقارنة نتائج النموذج الموجه نحو المدخلات والنموذج الموجه نحو المخرجات. على وجه التحديد، اختبروا وجود حلول في المناطق الأربعة المحددة بواسطة جيران DMU قيد التقييم. تمثل المناطق الأربعة حالة العوائد المتناقصة (DRS)، والعوائد المتزايدة (IRS) CRS وعدم الكفاءة التقنية، على التوالي. يحل هذا النهج المشكلة في بانكر وثرال (1992) بأن النموذج قابل للتطبيق فقط على DMUs الفعالة تقنيًا ويفشل في توفير اكتشاف غير غامض للعوائد المتناقصة. تم توسيع هذا النهج بواسطة طالب وآخرون (2022) كنهج غير موجه لتحديد RTS من خلال نموذج تكامل البيانات ثنائي الهدف مع مراعاة المخرجات غير المرغوب فيها. تقترح بعض الدراسات أيضًا نهجًا للتعامل مع البيانات السلبية (إيمروزنجاد وآخرون 2010؛ ساهو وآخرون، 2016)، وبيانات النسبة (أولسن وآخرون، 2022؛ إيمروزنجاد وأمين، 2009)، وبيانات الفترات (هتامي-مربيني وآخرون، 2014).
النماذج المذكورة أعلاه تعتمد على هياكل الصندوق الأسود التقليدية. تم تقديم امتدادات نظرية لأخذ العمليات الإنتاجية الأكثر تعقيدًا في الاعتبار. اقترح لوزانو وفيلا (2010) وسينغ ورانجان (2018) نموذجًا لتحديد حالة RTS ونقطة MPSS في أنظمة الإنتاج المتوازية. نظرًا لعدم وجود وسائط في النظام المتوازي، فإن النهج المقترح مشابه للنماذج المعتمدة على نظام الصندوق الأسود. لأخذ العملية الإنتاجية الداخلية في الاعتبار، اقترح لوزانو (2011) الأساس النظري لتحديد الكفاءة التقنية والتكلفة والتخصيص، بالإضافة إلى RTS المحلي، في نظام من مرحلتين. اقترح زانغ ويانغ (2015) نهجًا لتحديد RTS لعملية إنتاج من مرحلتين من خلال نموذج احتواء معدّل من مرحلتين. حدد أساني وآخرون (2018) نقطة MPSS لنموذج DEA من مرحلتين. ومع ذلك، لا يزال هناك فجوة بحثية كبيرة في تحديد RTS للهياكل الإنتاجية الشبكية الأكثر تعقيدًا، مثل الأنظمة المتوازية والسلسلة والعامة. تعتمد معظم نهج تحديد RTS الشبكية على نموذج من مرحلتين، والذي يصف العملية الإنتاجية حيث يتم تحويل جميع المدخلات إلى وسائط ثم تستخدم لاحقًا لإنتاج المخرجات في المرحلة التالية. ومع ذلك، كما أشار خذريموطلاغ وزو (2023)، يتم توليد PPS المدخلات-المخرجات في نظام من مرحلتين من خلال دمج PPS المدخلات-الوسائط وPPS الوسائط-المخرجات. وبالتالي، قد تتضمن PPS المدخلات-المخرجات في نظام من مرحلتين نقاط إسقاط خارجية، وتتطلب النماذج المذكورة أعلاه مزيدًا من التحقيق لمعالجة هذه القضية.
تم تطوير نظرية تحديد RTS وMPSS تحت التحليل العرضي أيضًا بناءً على عمل بانكر. اقترح سوييوشي وسيكيتاني (2005) نموذجًا لمعالجة مشكلة تحديد RTS في نظام إنتاج ديناميكي. لفترة زمنية معينة يتم التعامل مع الحمل من الفترة كمدخل بينما يتم التعامل مع الحمل إلى الفترة كمخرج. بناءً على المعيار أعلاه، يمكن تحديد MPSS وRTS للفترة باستخدام نهج مشابه اقترحه سوييوشي (1999). يعتمد نهج سوييوشي (1999) المقترح على نموذج مخطط الكفاءة التكلفة لبانكر وآخرون (1984). وبالتالي، هناك حاجة إلى عملية حسابية متعددة المراحل لتحديد جميع الطائرات الفائقة المدعومة. استجابةً لذلك، اقترح سليماني-دامانه (2013) نهجًا جديدًا لتحديد RTS في الأنظمة الديناميكية باستخدام الشكل الاحتوائي لنموذج CRS. يمكن العثور على المناقشة المتعلقة بتحديد نقطة MPSS في تحليل مؤشر مالموكويست في بودينوفيسكي وآخرون (2017). النموذج الذي اقترحه أساني وآخرون (2018) قادر على تحديد MPSS لعمليات الإنتاج المتوازية متعددة الفترات. عالج كونتولايمو وتسيكورا (2010) قضايا القياس الناجمة عن تحديد MPSS في نماذج الحدود الفوقية. لقد عرفوا نظريًا نقطة MPSS في نموذج الحدود الفوقية كنقطة فعالة تقنيًا عبر جميع التقنيات.
مدد بودينوفيسكي (2022) مفاهيم CRS وVRS إلى CRS متعددة المكونات (MCRS) وVRS متعددة المكونات (MVRS) واقترح نهجًا مطابقًا لتحديد مرونة المقياس وRTS. على وجه التحديد، عرّف بودينوفيسكي (2022) التقنيات متعددة المكونات على أنها تقنيات إنتاج حيث يستخدم كل عملية مدخلاتها المحددة ونسبة غير معروفة من المدخلات المشتركة لإنتاج مخرجاتها المحددة ونسبة غير معروفة من المخرجات المشتركة. يكمن الاختلاف الرئيسي بين تقنيتين متعددتي المكونات في افتراضاتهما: يعامل نموذج MVRS كل عملية مكونة كتقنية محدبة منفصلة، بينما يفترض نموذج MCRS بالإضافة إلى ذلك أن كل عملية قابلة للتوسع. بناءً على عمل بانكر وثرال (1992) وبودينوفيسكي (2017)، ذكر بودينوفيسكي (2022) أن حالات RTS يمكن أن تُميز بناءً على مرونات المقياس أحادية الجانب المقابلة. يتميز DMU قيد التقييم بزيادة RTS إذا كانت مرونات المقياس اليمنى المقابلة له أقل من 1. وبالمثل، يظهر DMU قيد التقييم انخفاض RTS إذا كانت مرونات المقياس اليسرى أقل من 1. يظهر DMU CRS إذا وفقط إذا كان يقع عند نقطة MPPS للتكنولوجيا الحالية. قام بابايواننو وبودينوفيسكي (2023) بتوسيع عمل بودينوفيسكي (2022) إلى سيناريوهات مع تخصيصات مقيدة للمدخلات والمخرجات المشتركة.

2.3.2 الثورة والتقدم في توصيفات المقياس

بدلاً من التركيز فقط على تقييم الأداء، يمكن توسيع منهجية DEA التقليدية لمعالجة العديد من المهام الأخرى المهمة لصنع القرار والسياسة. تمتد معظم الأدبيات في هذه المجموعة إلى مفاهيم RTS وMPSS ضمن نموذج DEA. في هذا القسم، نستعرض الدراسات الحديثة التي تقترح وتحدد توصيفات جديدة للمقياس. يوضح الشكل 4 تقسيم المفاهيم الجديدة.
استهدفت نظرية DEA المبكرة بشكل أساسي تحديد الأنواع المحددة من RTS لـ DMU معين. ومع ذلك، في نظرية اقتصاديات الإنتاج الفعال، يتم قياس مصطلح RTS كقيمة محددة لـ RTS، والمعروفة باسم مرونة المقياس (فوكوياما، 2000). اقترح بانكر وآخرون (1984) وبانكر وثرال (1992) نهجًا بسيطًا غير معلمي لحساب مرونة المقياس من حيث الحلول المثلى المزدوجة لنماذج CCR وBCC. اقترح فورسوند وآخرون ذكر أن نقطة البداية في نهج تحليل البيانات الفعالة (DEA) هي تحديد تكنولوجيا الإنتاج الخطية القطعية، بدلاً من تكنولوجيا الإنتاج العامة. على وجه التحديد، فإن افتراض السلاسة والتفاضل المستمر لحدود الإنتاج لا ينطبق في نظرية DEA لأن حدود الإنتاج الناتجة عن نموذج DEA هي خطية قطعياً. مع أخذ هذه الاختلافات في الاعتبار، قاموا بتصنيف الأساليب لحساب مرونة الحجم إلى مجموعتين: (1) النهج غير المباشر و (2) النهج المباشر.
تقدّر الطريقة غير المباشرة خصائص المقياس والاستبدال من خلال الدوال الحدودية غير المعلمية في تحليل البيانات الاقتصادية (DEA)، مما ينتج عنه صيغة بسيطة لمرونة المقياس من حيث الحلول المثلى المزدوجة لنموذج VRS (بودينوسكي وآخرون، 2009). على سبيل المثال،
الشكل 4 تقسيم خصائص المقياس
حسب Førsund و Hjalmarsson (2004)، تم حساب قيمة مرونة المقياس للإسقاطات الشعاعية للنقاط غير الفعالة إلى الحدود بناءً على درجات الكفاءة والقيم الظلية على قيود التقعر. تم تطبيق نهج مشابه من قبل Førsund وآخرون (2007) لتقدير مرونة المقياس للوحدات غير الفعالة وإسقاطاتها المقابلة على الحدود. على وجه التحديد، قاموا بحساب القيمة القصوى والدنيا لسعر الظل على قيد التقعر وحددوا حدود قيم مرونة المقياس وفقًا لذلك. قام Podinovski وآخرون (2009) بتبسيط الصيغة التي اقترحها Førsund و Hjalmarsson (2004) وذكروا أن المرونات أحادية الجانب يمكن اعتبارها مشتقات اتجاهية لدالة القيمة المثلى في نموذج DEA المعدل. يوسع هذا النهج تحقيق مرونة المقياس إلى الحالة الأكثر عمومية، حيث يتعامل مع حلول مثلى متعددة، والانحلال، ووجوه غير كاملة الأبعاد.
من ناحية أخرى، يستكشف النهج المباشر خصائص المقياس والاستبدال من خلال دوال الإنتاج النيوكلاسيكية المعلمية. على سبيل المثال، جادل كريفونوجكو وآخرون (2004) بأن نهج المصرفي محدود في التحقيق في المشتقات الجزئية (معدل الاستبدال الهامشي، معدل التحول الهامشي، المنتج الهامشي) لتلك وحدات القياس التي تقع داخل حدود الإنتاج. واقترحوا نهجًا معلميًا لحساب المشتقات الجزئية في أي نقطة على الحدود. فورسوند وآخرون. تم تقييم مرونة المقياس العددية مباشرة على سطح DEA على طول التقاطعات مع المستويات، باستخدام نهج التحسين البارامتري لبناء حدود تكنولوجيا الإنتاج VRS. ثم يتم حساب قيمة مرونة المقياس بناءً على حدود تكنولوجيا الإنتاج المقابلة. بناءً على عمل بودينوفكسي وآخرون (2009)، قام بودينوفكسي وفورسون (2010) أولاً بحساب فئة من مقاييس المرونة الجزئية المختلطة في كل من تكنولوجيا VRS وCRS. اقترحوا نهجين لقياس المرونة: أحدهما يعالج مشكلة عدم الاستقرار التي تم مناقشتها في كوبر وآخرون (2000) من خلال برامج خطية مزدوجة مصممة خصيصًا، والآخر يعتمد على نموذج المضاعف القياسي، الذي يزيد من عبء الحساب ولكنه يقدم مزايا تفسيرية. قام زيلينيك (2013) بحساب مرونة المقياس لتكنولوجيا المدخلات والمخرجات المتعددة بناءً على نموذج DDF، موضحًا التكافؤ مع مقاييس مرونة المقياس الموجهة نحو المدخلات والمخرجات. كما أشار مقياس المرونة القائم على DDF ومقياس المرونة القائم على دالة الربح إلى علاقة مزدوجة. قام كاو وهوانغ (2011) وساهو وآخرون (2014) بتفكيك مرونة المقياس في نموذج DEA بمضاعف من مرحلتين.
لقد درست الدراسات أيضًا الإمكانات الإنتاجية للوحدات التي تعمل بكفاءة استجابةً لتغييرات معينة في ملفات مدخلاتها ومخرجاتها من خلال تقديم توصيفات هامشية. قام بانكر ومايندراتا (1986) أولاً بحساب معدلات تحويل الاستبدال. جادل بيسنت وآخرون (1988) بأن نهج بانكر ومايندراتا (1986) فشل في معالجة المشكلة الناتجة عن واجهات الجوانب. بالنسبة لوحدة القياس التي تحتوي على مزيج من الموارد أو المخرجات التي تختلف عن أي نقطة حدودية، اقترحوا نهج تحليل الجوانب المقيدة للنظر في الحد الأدنى من الكفاءة للوحدات التنظيمية. تهدف النهج المذكورة أعلاه إلى تقدير معدلات الاستبدال الهامشية في سيناريو مع تغييرات غير محسوسة، أو تغييرات صغيرة محدودة، في متغير واحد أو أكثر. جادل أسميلد وآخرون (2006) بأن التغيير قد يكون أكبر في التطبيقات الواقعية. لذلك، اقترحوا نماذج لتقدير تأثير التغييرات غير الهامشية الأكبر، والتغييرات القياسية، والتغييرات المضافة في تحليل الكفاءة باستخدام نهج روزن وآخرون (1998). طور تشارنز وآخرون (1996) دالة إنتاج فعالة تجريبية قوية وقاسوا التبادلات الهامشية لدالة الإنتاج الفعالة. قام كوبر وآخرون (2000) بتحليل التبادلات الهامشية من خلال المتغيرات الوسيطة. تقديرات الدراسات الأحدث، مثل بودينوفكسي وفورسون (2010)، لمعدلات الاستجابة الهامشية التي تظهرها مدخل واحد (مخرج) بالنسبة للمخرج (المدخل) تحت تقنيات VRS وCRS القياسية. ثم قام بودينوفكسي وآخرون (2016) بـ
اقترح إطارًا أكثر عمومية لحساب معدلات الاستبدال الحدية في الحدود الإنتاجية غير المعلمية.
تمت مناقشة مفهوم سعر الظل في العديد من الدراسات. من منظور التحليل الحسابي والكفاءة، ينبع سعر الظل من استخدام تقنية لاغرانج لحل مشاكل التحسين ذات القيود (Førsund، 2018). استهدفت الأدبيات الحديثة توسيع مفهوم سعر الظل للإنتاج غير المرغوب فيه من منظور اقتصادي، بهدف حساب تكلفة التخفيف الهامشية لإنتاج غير مرغوب فيه معين. على سبيل المثال، أشار Leleu (2013) إلى عمل Färe وآخرين. وحسبوا تسعير الظل للمخرجات غير المرغوب فيها مع الأخذ في الاعتبار آثار القابلية الضعيفة للتخلص. النموذج الهجين المقترح يأخذ في الاعتبار التوازن بين نمذجة التداخل بين المخرجات الجيدة والسيئة وتوقع الحصول على سعر الظل للمخرجات السيئة. يمكن العثور على مراجعة أدبية حول حساب سعر المخرجات غير المرغوب فيها في Zhou et al. (2014). قام Kao و Hwang (2019) بالتحقيق في سعر الظل للمخرجات غير المرغوب فيها في كل من نموذج مخطط المضاعف ونموذج مخطط الإحاطة، محققين أعلى درجة كفاءة مقاسة ممكنة. قدّر Podinovski (2019) الخصائص الهامشية للحدود الإنتاجية غير المعلمية في وجود مخرجات غير مرغوب فيها. حسب Wang et al. (2016) المكاسب المحتملة من توفير تكاليف تقليل انبعاثات الكربون من خلال نموذج DEA الديناميكي الخاص. قام Lee و Wang (2019) بقياس تقدير تكلفة التخفيف الهامشية للملوثات الهوائية تحت توازن ناش. قام Wu et al. (2023) بالتحقيق في خيارات التخفيف المختلفة لحساب تكلفة التخفيف الهامشية لـ الانبعاثات وقدم مفهوم الاشتقاق التفسيري لحساب المكاسب الهامشية الناتجة عن التحول في الطاقة.
بالإضافة إلى وحدات القياس الفردية، ساهمت الأدبيات أيضًا في قياس خصائص النطاق على مستوى المجموعة. بانكر اقترح فرضية للتحقيق في نوع RTS لتكنولوجيا معينة. كما تم مناقشته في Banker (1984) و Banker وآخرون (1984)، تحدد الانحرافات عن نقطة MPSS عدم الكفاءة في النطاق ضمن مدى العمليات. وبالتالي، فإن Banker اقترح استخدام عدم كفاءة عدم المقياس كفرضية صفر، مما يشير إلى أن تكنولوجيا الإنتاج تلبي شروط الكفاءة النسبية الثابتة. كانت نقطة البداية في بانكر هو حساب درجات عدم الكفاءة والكفاءة المقابلة. تفترض الفرضية الصفرية عدم وجود عدم كفاءة في المقياس، مما يعني أن بيانات العينة يمكن تبريرها من خلال مجموعة إنتاجية تظهر RTS ثابتة. بناءً على توزيع عدم الكفاءة الحقيقي (الموزع بشكل أسي، الموزع بشكل نصف طبيعي، توزيع غير معروف)، تم حساب إحصائيات الاختبار باستخدام صيغ مختلفة. كما طور بانكر (1996) إطارًا لاختبار الفرضية الصفرية لعدم وجود DRS مقابل البديل من DRS والفرضية الصفرية لعدم وجود IRS مقابل البديل من IRS. هذا الإطار مشابه لاختبار CRS. الخطوة الأولى هي تقدير عدم الكفاءة تحت IRS (DRS) عن طريق ضبط مجموع وزن الشدة ليكون أقل (أو أكبر) من 1. ثم تتم مقارنة عدم الكفاءة المقدرة بالكفاءة الحقيقية باستخدام إحصائية الاختبار المقابلة. سيمار وويلسون وسعت عمل بانكر من خلال تقديم إجراءات تقدير bootstrap، وتوفير قيم حرجة مناسبة لإحصائيات الاختبار. ومع ذلك، جادل أليرازائي وآخرون (2018) بأن الافتراضات الإحصائية التي تستند إليها بانكر صعبة التحقيق، وقد يغفل صانعو القرار بعض العوامل أو يحكمون مسبقًا على سلوك التكنولوجيا في التطبيقات الواقعية. استجابةً لذلك، قدم أليرازائي وآخرون (2018) مفهوم RTS التكنولوجي واقترحوا نهجًا غير إحصائي لتحديد RTS التكنولوجي (TRTS) بناءً على طريقة الزوايا. على وجه التحديد، تم تقديم مقياس لحساب الفجوة بين المستوى الثابت لـ TRTS والمستوى المتغير لـ TRTS في كل من الأقسام المتزايدة والمتناقصة من الحدود.

3 التقدم في الأسس الإحصائية لـ DEA: المساهمات، الأطر، والتطورات الحديثة

يمكن تصنيف التقدم في الأسس الإحصائية لـ DEA، التي رائدها بانكر، إلى مجموعتين. تركز الأولى على اختبار الفرضيات بناءً على نموذج DEA، بينما تتناول الثانية أساس وهيكل تصميم اختبار المحاكاة في DEA. باختصار، يؤسس العمل الأصلي لبانكر حول الأسس الإحصائية لـ DEA نموذج البحث لاختبارات الفرضيات واختبارات المحاكاة تحت إطار DEA. توفر اختبارات الفرضيات لبانكر الأساس لبناء مجموعة واسعة من الاختبارات الإحصائية الرسمية لنماذج DEA، بما في ذلك التغيير الفني، تأثير المتغيرات السياقية، أنواع RTS، وقابلية فصل المدخلات. من ناحية أخرى، تقترح اختبارات المحاكاة تحت إطار DEA نهجًا عامًا لاختيار نماذج التحليل المناسبة. كما يتم تطبيق هذه الاختبارات على نطاق واسع لتقييم أداء نماذج التحليل غير المعلمية. وقد وسعت الدراسات الحديثة نهج بانكر لاختبار الأسس الإحصائية لنماذج التحليل غير المعلمية الأخرى. كما توفر الاستدلالات الإحصائية التي طورها بانكر الأساس النظري لنظرية الحد المركزي (تحت نموذج DEA)، وتحليل DEA العشوائي، وتقدير الحدود الكمية غير المعلمية. تقدم هذه النماذج تقديرات أكثر دقة لأسعار الظل ومرونة أكبر في اختيار المتجهات الاتجاهية. علاوة على ذلك، تقترح الدراسات نهج الحدود الجزئية بناءً على الاستدلال الإحصائي لنموذج DEA. تقترح مجموعة أخرى من الدراسات نهج الانحدار الكمي، الذي يعتبر امتدادًا لعمل بانكر وآخرين (1991). توفر هذه النماذج نهجًا أكثر ملاءمة للتعامل مع الضوضاء العشوائية، والتباين غير المتجانس، والقيم الشاذة.
نستعرض الأدبيات الأصلية لبانكر في القسم 3.1. في القسم 3.2، نقدم الامتداد النظري لبانكر بناءً على المناقشة في القسم 3.1. يتم مراجعة التقدم الأخير في اختبارات المحاكاة، واختبارات الفرضيات، والاستدلال الإحصائي لـ DEA وامتداداته في القسم 3.3. نقدم أيضًا بعض النماذج الموسعة الأكثر حداثة، بناءً على النظرية الإحصائية، لمعالجة ضوضاء البيانات في القسم 3.4.

3.1 أصل الأسس الإحصائية لـ DEA بواسطة بانكر

يمكن تقسيم عملية الثورة في الأساس الإحصائي لـ DEA إلى ثلاث مراحل: ما قبل النموذج، الأصل، والتطوير. تستعرض هذه القسم الأدبيات التي تمثل هذه المراحل الثلاث بواسطة بانكر. يتم تصور العملية في الشكل 5.
الشكل 5 مراحل الثورة للإحصائيات في DEA بواسطة بانكر
نموذج DEA هو نهج غير معلمي يستند إلى المسلمات العامة لنظرية الإنتاج، مثل التزايد، والتحدب، والتجانس (بانكر وآخرون، 1984). في الثمانينيات، كانت الأبحاث حول الإحصائيات في DEA في ما وصفه كوهين (1970) بمرحلة ما قبل النموذج. تجاهلت الأدبيات المبكرة لـ DEA إلى حد كبير المناقشات حول الاستدلال الإحصائي للحدود الفعالة الناتجة. حتى عمل بانكر وماينديراتا (1988) لم يؤسس سوى رابط بين نموذج DEA واختبار فارين (1984) الجبري للتوافق مع المسلمة الضعيفة لتقليل التكلفة، مما يقيس ما إذا كانت الكفاءات الفنية أو التخصيصية أو المقياسية لوحدات اتخاذ القرار منخفضة. وبالتالي، خلال هذه الفترة، اعتُبر DEA نهجًا غير إحصائي (بانكر، 1996؛ غونغ وسيكلز، 1992؛ شميت، 1985). ومع ذلك، انتقد شميت (1985) هذا النهج القياسي غير الإحصائي، معبرًا عن شكوكه بشأن عدم وجود أساس قياسي مطابق. جادل بأنه، لأن الطريقة غير الإحصائية وغير المعلمية لا تأخذ في الاعتبار الخطأ، يجب التعامل مع الأدبيات بحذر، خاصة عندما يُفترض أن البيانات قيد التقييم خالية من الضوضاء الإحصائية.
استجابةً لانتقادات شميت، ناقش بانكر (1993) وأسس الأساس الإحصائي لـ DEA. حدد بانكر (1993) أربعة مسلمات تتعلق بمجموعة الإنتاج ودالة كثافة الاحتمال لـ “عدم الكفاءة الحقيقي”. تشمل هذه المسلمات التزايد، والتحدب، والتغليف، واحتمالية الأداء الفعال. تشكل هذه المسلمات الأربعة الأساس لدراسات الاختبار المعتمدة على المحاكاة (بانكر، 1993، 1996؛ بانكر وتشانغ، 1995؛ بانكر وآخرون، 1994).
من خلال حل نماذج BCC لتقدير درجات عدم الكفاءة، تتم مقارنة درجات الكفاءة المقدرة بـ “عدم الكفاءة الحقيقي” الناتج بناءً على هذه المسلمات. تشير النتائج إلى أن حدود الإنتاج BCC تتماشى مع دالة كثافة الاحتمال المتزايدة، الناتجة بناءً على انحراف المخرجات قيد التقييم عن مستويات المخرجات الفعالة. بالإضافة إلى ذلك، فإن عدم الكفاءة المقدرة تعظم دالة الاحتمال إذا كانت دالة كثافة الاحتمال لعدم الكفاءة تظهر اتجاهًا متزايدًا.
اقترح بانكر (1993) أيضًا ثلاث فرضيات لاختبار الفروق في عدم الكفاءة بين نوعين مختلفين من وحدات اتخاذ القرار، لتقييم ما إذا كانت مجموعة واحدة أكثر كفاءة من الأخرى. يمكن أن تستوعب هذه الاختبارات الثلاثة أنواعًا مختلفة من التوزيعات (الموزعة بشكل أسي، الموزعة بشكل نصف طبيعي، توزيع غير معروف) من عدم الكفاءة. يضع النهج المقترح الأساس النظري للأدبيات المستقبلية، مؤسسًا أساسًا إحصائيًا رسميًا لمقدرات عدم كفاءة DEA واقتراح إطار عام لاختبار الفرضيات الإحصائية بشأن الفروق في توزيعات عدم الكفاءة في DEA.

3.2 تطوير نهج التمديد القائم على الإحصائيات في DEA بواسطة بانكر

استنادًا إلى عمل بانكر (1993)، تم تطوير نهجين رئيسيين قائمين على الإحصائيات في مجال DEA. يركز النهج الأول على تطوير الأساس الإحصائي لـ DEA بشكل أكبر واقتراح نهج اختبار فرضيات معدلة لمعالجة القضايا المحتملة الناجمة عن التحيز الإحصائي. يبني النهج الثاني على نهج المحاكاة الذي استخدمه بانكر (1993) ويهدف إلى اختبار أداء نماذج معينة. تستعرض هذه القسم هذين النهجين والأدبيات ذات الصلة بواسطة بانكر.

3.2.1 أسس المحاكاة في DEA بواسطة بانكر

قام بانكر (1993) بوضع الأساس لإطار اختبار قائم على المحاكاة في تحليل الكفاءة. تتضمن إجراءات اختبار المحاكاة مقارنة درجة الكفاءة المقدرة بواسطة تحليل الكفاءة مع “الكفاءة الحقيقية” الناتجة عن تكنولوجيا الإنتاج التي تلبي الافتراضات الأربعة المحددة بواسطة بانكر (1993). العوامل الحاسمة في هذه المحاكاة هي الخصائص
لتكنولوجيا الإنتاج، وتوزيع عدم الكفاءة، وحجم العينة (بانكر، 1996).
في جميع دراسات المحاكاة تقريبًا التي أجراها بانكر، تهدف التصاميم إلى تقييم أداء نموذج أو اختبار من خلال مقارنة درجات الكفاءة المقدرة مع “الكفاءة الحقيقية” من خلال محاكاة مونت كارلو. عادةً ما يتم توليد “عدم الكفاءة الحقيقي” باستخدام دالة إنتاج كوب-دوغلاس في سيناريو إنتاج أحادي، مع توزيع قيم “الكفاءات الحقيقية” في نمط محدد. على سبيل المثال، قام بانكر وتشانغ (1995) بتقييم أداء اختبارات تحليل الكفاءة غير المتناهية لاختلافات عدم الكفاءة، كما هو موضح في بانكر (1993). تم مقارنة أداء هذه الاختبارات مع اختبارات المربعات الصغرى العادية المصححة (COLS) واختبارات ويلش ومان-ويتني. أظهرت النتائج أن اختبارات تحليل الكفاءة غير المتناهية تفوقت على اختبارات COLS واختبارات مان-ويتني. كما تم استخدام إطار المحاكاة المقترح لتقييم دقة تقدير الكفاءة في نموذج تحليل الكفاءة الفائق (بانكر وتشانغ، 2006). تظهر النتائج أن نموذج الكفاءة الفائقة يتفوق عليه نموذج تحليل الكفاءة التقليدي في تحديد البيانات الملوثة بالقيم الشاذة وإزالة القيم الشاذة المقابلة.

3.2.2 أسس اختبار الفرضيات في تحليل الكفاءة بواسطة بانكر

منذ نشر بانكر (1993)، تطورت الأدبيات حول الأساس الإحصائي لتحليل الكفاءة بشكل كبير. إحدى الخصائص الحاسمة لحدود الإنتاج هي وجود عوائد متزايدة أو متناقصة. كما تم استعراضه في القسم 2.3، طور بانكر (1996) إطارًا لاختبار عوائد التكنولوجيا على مستوى التكنولوجيا.
فرضية أخرى اقترحها بانكر وآخرون (1994) كانت تهدف إلى اختبار استقلال التقديرات والمدخلات. على وجه التحديد، فإن قابلية فصل المدخلات – الافتراض بأن المدخلات يمكن تحسينها بشكل مستقل دون أن تكون قابلة للاستبدال بأخرى – هو افتراض شائع في نموذج تحليل الكفاءة. ومع ذلك، في التطبيقات العملية، قد تكون مدخلتان (مثل العمل ورأس المال) بدائل، ولم يتم اختبار الافتراض بشأن القابلية للاستبدال أو الفصل في نماذج تحليل الكفاءة التقليدية. لاختبار هذه الفرضية على مجموعة بيانات كاملة، قدم بانكر وآخرون (1994) ارتباطًا بين نموذج تحليل الكفاءة التقليدي ودالة المسافة لشيبرد (1970). يتم تقدير عدم الكفاءة القابل للفصل بشكل مثالي كمعكوس لمقياس المسافة لشيبرد (1970). ذكروا أنه إذا كانت المدخلات قابلة للفصل، في مجموعة من وحدات اتخاذ القرار مع المدخلات، فإن عدم الكفاءة المدخلية المقدرة بواسطة مدخل معين فريدة. على وجه التحديد، دع يكون عدم الكفاءة المدخلي لـ الناتج عن المدخل فقط ; لدينا . لذلك، يمكن اعتبار فرضية عدم الفصل المدخلي كالحالة التي يكون فيها . مشابهًا لبانكر (1993)، اقترح بانكر (1994) أيضًا استخدام ثلاثة إحصائيات اختبار مختلفة بناءً على توزيع عدم الكفاءة الحقيقي.
كانت دراسة بانكر وتشانغ (1995) الأولى التي اقترحت نهج اختبار التخصيص (المشار إليه باسم “اختيار المتغيرات” في الدراسات الحديثة) لاختبار أهمية المتغيرات عند الهامش في توصيف عملية الإنتاج. تفترض الفرضية الصفرية أن المتغيرات الإضافية لا تؤثر على تطابق الإنتاج. على وجه التحديد، لمعالجة مشكلة الأبعاد، يجب أن تكون درجة عدم الكفاءة بدون المتغيرات الإضافية أكبر من درجة عدم الكفاءة الناتجة الناتجة عن النموذج الذي يتضمن المتغيرات الإضافية. كما تم اقتراح ثلاثة إحصائيات اختبار مختلفة بناءً على توزيع الكفاءة الحقيقية.

3.3 التقدم في الأسس الإحصائية لتحليل الكفاءة وتوسعاته

أثبت عمل بانكر أنه نموذج البحث لاختبار الإحصاء واختبار المحاكاة في تحليل الكفاءة. يستعرض القسم 3.3.1 التقدم في تصميم المحاكاة ضمن تحليل الكفاءة، مع تسليط الضوء على كيفية تطبيق إطار المحاكاة الخاص ببانكر لتقييم أداء النماذج المصممة لتقييم عمليات الإنتاج الأكثر تعقيدًا. يفحص القسم 3.3.2 التوسعات في الاختبار الإحصائي ضمن إطار تحليل الكفاءة. بالإضافة إلى توسيع نطاق الاختبار الإحصائي، تقترح الدراسات طرقًا لمعالجة ضوضاء البيانات خلال كل من مراحل الاختبار وتوليد البيانات. يناقش القسم 3.3.3 التطورات الأخيرة في إثبات الاستدلال الإحصائي لنماذج تحليل الكفاءة الموسعة. يقدم القسم أيضًا بعض نماذج تحليل الكفاءة العشوائية الملحوظة، والتي تستند إلى المسلمات الإحصائية التي وضعها بانكر.

3.3.1 التقدم في تصميم المحاكاة في تحليل الكفاءة

تم تصميم العديد من اختبارات المحاكاة منذ بانكر (1993)، وتشارك معظمها نفس الدافع وتهدف إلى فحص رضا نموذج تحليل الكفاءة. تصميم المحاكاة مشابه لعمل بانكر (1993)، الذي يقارن عدم الكفاءة المقدرة في تحليل الكفاءة مع “عدم الكفاءة الحقيقية” الناتجة. كما هو موضح في الشكل 6، تستهدف معظم التوسعات النظرية عملية الإنتاج من خلال إدخال المزيد من العوامل، مثل المخرجات المتعددة، والمخرجات غير المرغوب فيها، والمتغيرات غير القابلة للتقدير، أو اقتراح بيئة إنتاج شبكية أكثر تعقيدًا.
كتجربة محاكاة ذات مخرج واحد، قارن باردهان وآخرون (1998) وكوبر وتون (1997) أداء تحليل الكفاءة، وتركيبات تحليل الكفاءة والانحدار، ونهج الانحدار على الحدود العشوائية تحت أحجام عينات مختلفة. صمم روجيرو التجارب المحاكاة لاختبار أداء نماذج تحليل الكفاءة التي تأخذ في الاعتبار المتغيرات غير القابلة للتقدير. قارن جيرالياس وآخرون (2012) أداء تحليل الكفاءة، والمربعات الصغرى العادية، وتحليل الحدود العشوائية (SFA) عند توليد الحدود بناءً على بيانات اللوحة. كما تمت مقارنة أداء تحليل الكفاءة التقليدي، وSFA، ونموذج تحليل الكفاءة العشوائي (ريستي، 2000؛ سيمار، 2007)، ودقة الكفاءة التقنية وكفاءة الحجم التي يقيسها نموذج تحليل الكفاءة (بيرلمان وسانتين، 2009) باستخدام إطار المحاكاة الخاص ببانكر (1993). يمكن العثور على المزيد من الأمثلة في خذريموطلغ (2022)، الذي يستعرض الأدبيات الحالية حول التجارب المحاكاة الموجودة.
اختبار المحاكاة الخاص ببانكر (1993) يعتمد على دالة إنتاج كوب-دوغلاس التقليدية في حالة ذات مخرج واحد. ومع ذلك، في التطبيقات الواقعية، من المحتمل أن تتعامل نماذج تحليل الكفاءة مع سيناريوهات متعددة المخرجات. لذلك، تهدف بعض الدراسات إلى بناء تجارب محاكاة باستخدام دالة الإنتاج مع مدخلات متعددة ومخرجات متعددة. لتوليد “عدم الكفاءة الحقيقية” في تكنولوجيا الإنتاج متعددة المخرجات، يتم تطبيق دالة إنتاج كوب-دوغلاس الموسعة (وانغ وآخرون، 2020). تم اقتراح تكنولوجيا إنتاج أخرى تعتمد على الحدود التربيعية بواسطة سيمار (2007). تم اقتراح وتطبيق دالة الإنتاج الترانسلوغ ذات المدخلين والمخرجات من قبل بيرلمان وسانتين (2009). فيرنانديز
الشكل 6 التقدم الأخير في تصميم المحاكاة في تحليل الكفاءة
وآخرون (2002) استخدموا دالتين مختلفتين للإنتاج لتقدير المخرجات المرغوبة وغير المرغوبة بشكل منفصل. تم تطوير هذا النهج بشكل أكبر بواسطة تشين وديلماس (2012) لتوليد دالة إنتاج متعددة المخرجات مع مراعاة القابلية الضعيفة للتخلص من المخرجات غير المرغوب فيها.
لأخذ العملية الإنتاجية الداخلية في الاعتبار، يتم توليد “عدم الكفاءة الحقيقية” بواسطة حدود الإنتاج الشبكي (ميشالي وآخرون، 2023). كما اقترح خزريموتلاخ (2022) نهجًا عامًا لبناء حدود الإنتاج الشبكي لتقييم أداء نموذج DEA الشبكي. جوهر هذا النهج هو توليد “عدم الكفاءة الحقيقية” ومستوى الإنتاج الأصلي، مما يضمن أن قيمة “عدم الكفاءة الحقيقية” ومستوى الإنتاج الأصلي تلبي توزيعًا محددًا. يعمل مستوى الإنتاج الأصلي كإسقاط للإنتاج أو معيار للإنتاج لوحدة القياس المعنية. لذلك، يتم حساب حجم إنتاج وحدة القياس من خلال ضرب مستوى الإنتاج الأصلي في “الكفاءة الحقيقية” المقابلة، والتي هي معكوس “عدم الكفاءة الحقيقية”. في معظم الحالات، تكون النماذج المختبرة هي نماذج DEA شعاعية موجهة نحو المدخلات (الإنتاج). تم اقتراح نهج عام للتعامل مع DEA غير الشعاعي من قبل خزريموتلاخ (2022). ومع ذلك، فإن قابليته للتطبيق تحتاج إلى مزيد من المناقشة لأنه يفترض عدم كفاءة المدخلات (الإنتاج) كجمع بدلاً من مجموعة من الفجوات غير الشعاعية. “عدم الكفاءة الحقيقية” غير الشعاعية المتولدة هي نفسها “عدم الكفاءة الحقيقية” الشعاعية حيث تمثل الانكماش الشعاعي النسبي للمدخلات أو التوسع في المخرجات بدلاً من الفائض في المدخلات و/أو النقص في المخرجات بين المعايير (هاليكا وترنوفسكا، 2021). على سبيل المثال، في الحالات التي تحتاج فيها أداء SBM إلى الاختبار، يفشل الإطار الذي اقترحه خزريموتلاخ (2022) في تقديم معلومات حول “عدم الكفاءة الحقيقية” لمتغير معين ولكنه بدلاً من ذلك يعطي “عدم الكفاءة الحقيقية” الإجمالية. علاوة على ذلك، قد يؤدي افتراض نفس مستوى عدم الكفاءة للمدخلات والمخرجات إلى تناقضات في التطبيقات الواقعية. علاوة على ذلك، هناك فجوة بحثية في اختبار أداء نموذج DEA الشبكي غير الشعاعي (أي، نموذج SBM الشبكي) باستخدام النهج المحاكي الذي اقترحه بانكر.

3.3.2 التقدم والثورة في اختبارات الفرضيات في DEA

قدم بانكر (1993) أساسًا إحصائيًا رسميًا لتقنيات تقدير DEA، مما مهد الطريق لتطوير اختبارات فرضيات متنوعة. كما هو موضح في الشكل 7، بالإضافة إلى توسيع نطاق البحث لاختبار أداء مجموعة من وحدات القياس أو توسيع النظرية التي اقترحها بانكر وآخرون (1996أ، ب)، ركزت العديد من الدراسات أيضًا على دمج مصطلحات الضوضاء المستقلة ومعالجة قضايا مثل البيانات المفقودة أو بيانات اللوحات (اختبارات الفرضيات لتغيير الإنتاجية والكفاءة التخصيصية).
في الأبحاث الحديثة، تم تطوير اختبارات تقدير وإحصائية جديدة للتعامل مع البيانات الطولية والمقطعية. الغرض من هذه الدراسات هو التحقيق في تحول فترة الزمن لحدود الإنتاج التي تم إنشاؤها بواسطة نموذج DEA من منظور إحصائي. طور بانكر وآخرون (2004أ، ب) اختبارات إحصائية لفرضية العدم لعدم الكفاءة التخصيصية من خلال تقديم متغير تكلفة إجمالية واحد لتمثيل الإجمالي.
الشكل 7 التقدم الحديث في اختبارات الفرضيات في DEA
عدم الكفاءة الفنية والتخصيصية. يسمح هذا النهج بتحليل اتجاهات الكفاءة بمرور الوقت ويأخذ في الاعتبار الاختلافات عبر المجموعات الفرعية داخل مجموعة بيانات اللوحة. في السيناريوهات التي تتضمن معلومات غير متكافئة، قد لا تكون بيانات كمية الإنتاج متاحة، ولكن قد تكون القيمة النقدية للمخرجات الفردية ومعلومات كمية المدخلات متاحة. قام بانكر وآخرون (2007) بتوسيع عمل بانكر وآخرون (2004أ، ب) لإنشاء إطار لاختبار عدم الكفاءة التخصيصية المعتمدة على الإنتاج مع الأخذ في الاعتبار بيانات كمية الإنتاج المفقودة. اقترح بانكر وآخرون (2005) ثلاث فرضيات عدم لاختبار تغييرات الكفاءة وطبقوا هذه الاختبارات على صناعة المحاسبة العامة كحالة تجريبية. بالمقارنة مع نهج مؤشر مالموكويست التقليدي، يوفر نهجهم خصائص إحصائية لمكونات تغيير الإنتاجية ويستخرج اختبارات إحصائية بناءً على هذه التقديرات.
تم تطوير وتطبيق اختبارات الفرضيات التي اقترحها بانكر وآخرون (1996أ، ب) في مجالات متنوعة. على سبيل المثال، طبق ألبروفيتش وآخرون (2015) اختبار الفرضية الذي اقترحه بانكر وآخرون (1996أ، ب) لاختبار فرضية RTS المتعلقة بأداء الشركات. استخدم بانكر وآخرون (2010أ، ب) نهج بانكر وآخرون (1993) لفحص تغييرات الإنتاجية داخل النظام المصرفي الكوري بعد أزمة. قام سيمار وويلسون (2002) بتوسيع عمل بانكر وآخرون (1996أ، ب) لاختبار الفرضيات المتعلقة بـ RTS في نماذج الكفاءة الفنية غير المعلمية. اقترح باستور وآخرون (2002) اختبارًا لنماذج DEA الشعاعية المتداخلة من خلال توسيع اختبارات فرضيات اختيار المتغيرات التي اقترحها بانكر وآخرون (1996أ، ب). يهدف اختبار الفرضية هذا إلى تحليل الدور الهامشي لمتغير معين فيما يتعلق بالكفاءة المقاسة بواسطة نموذج DEA. أثبتت محاكاة مونت كارلو أن أداء الاختبار المقترح كان متفوقًا على أداء بانكر وآخرون (1996أ، ب). اقترح سويوشي وآوكي (2001) نهجًا إحصائيًا غير معلمي لاختبار تغييرات الإنتاجية بناءً على إنتاج تحليل نافذة DEA واقترحوا اختبار رتبة كروسكال-واليس المقابل. قام باستور وآخرون (1999) بتوسيع إطار فرضيات بانكر (1993) واقترحوا اختبارًا إحصائيًا لاكتشاف الملاحظات المؤثرة. اقترح كوسمانين وآخرون (2007) اختبار فرضية لاكتشاف الأخطاء في المتغيرات، كما ذُكر في فاريان (1985).
في الأدبيات المذكورة أعلاه، عادةً ما تعالج عملية توليد البيانات عدم الكفاءة العشوائية كسبب وحيد لانحراف الإنتاج الفعلي عن حدود الإنتاج. ومع ذلك، اقترحت العديد من الدراسات أيضًا اعتبار تأثير مصطلح ضوضاء مستقل في عملية توليد البيانات. كان بانكر وآخرون (2002) أول من اقترح تقدير دالة متطرفة تتضمن كل من مصطلح كفاءة أحادي الجانب ومصطلح ضوضاء ثنائي الجانب. يتم تطبيق هذه الدالة لتقييم ملاءمة الأشكال الوظيفية المعلمية من خلال أربعة اختبارات إحصائية: اختبار كولموغوروف-سميرنوف، اختبار الانحدار المرتب، اختبار ويلكوكسون لمجموع الرتب، واختبار ثايل غير المعلمي. عالج بانكر وناتاراجان (2008) مصطلح الخطأ كتكامل لثلاثة مكونات متميزة: دالة خطية لعدة متغيرات سياقية، قد تكون مرتبطة. كما أن ورقتهم أنشأت إطارًا عامًا باستخدام DEA في المرحلة الأولى وOLS أو ML في المرحلة الثانية لتقدير عدم الكفاءة الفردية مشروطة بقيمة مصطلح الخطأ المركب. قدم بانكر وآخرون (2010أ، ب) مصطلح ضوضاء عشوائي ثنائي الجانب في عملية توليد البيانات واقترحوا خمس فرضيات لمقارنة أداء مجموعات مختلفة من وحدات القياس. أظهر النهج المقترح أداءً أفضل من بانكر (1993) عندما كانت مستويات الضوضاء كبيرة، كما تم توضيحه من خلال محاكاة مونت كارلو. علاوة على ذلك، جادل سيمار وويلسون (2007) بأن الأدبيات السابقة فشلت في وصف عملية توليد بيانات متماسكة. استجابةً لذلك، اقترحوا إجراءات bootstrap التي تسمح باستنتاج صالح وتحسن الكفاءة الإحصائية في الانحدار في المرحلة الثانية.

3.3.3 التطورات في الاستدلال الإحصائي لـ DEA وتطبيقه

منذ أن قدم بانكر وآخرون (1993)، استكشفت الأدبيات الاستدلال الإحصائي في DEA وتوسعاته إلى حالات أكثر عمومية. استعرض غروسكوب (1996) الاستدلال الإحصائي لنموذج DEA ونموذج FDH. كما استعرضت هذه الورقة اختبارات الانتظام غير المعلمية، وتحليل الحساسية، والتحليل ذو المرحلتين مع الانحدار، والاختبارات الإحصائية غير المعلمية. يمكن العثور على مسوحات مماثلة في سيمار وويلسون (2000ب، 2007). في الأعمال الأكثر حداثة، عرّف داويا وجيبيلز (2011) وأثبتوا الاستدلال الإحصائي للحدود الجزئية غير المعلمية (DEA، FDH).
لقد تم توسيع نهج اختبار الاستدلال الإحصائي المقترح من قبل بانكر وآخرون (1993) ليشمل التحليل متعدد الفترات. طبق أوديك (2009) طريقة البوتستراب لضمان الاستدلال الإحصائي لمقياس مؤشر مالموكويست. قام تورطوسا-أوسينا وآخرون (2008) بدراسة حساسية مقياس مؤشر مالموكويست. اختبر كنيب وآخرون (2021) الاستدلال الإحصائي لتغيير الإنتاجية (كلا من التغييرات الفردية والمتوسطة في الإنتاجية لتقنية معينة) التي تم قياسها بواسطة مؤشرات مالموكويست. قام فام وآخرون (2024) بتوسيع عمل كنيب وآخرون (2021) من خلال اقتراح مؤشر إنتاجية مالموكويست جديد. يعكس هذا المؤشر الجديد الأهمية النسبية للأفراد، وقد تم إثبات استدلاله الإحصائي أيضًا.
توسيعًا لعمل بانكر وآخرون (1993)، ناقش كورستيلف وآخرون (1995أ، ب) اتساق وسرعة تقارب مقدرات DEA وFDH. تم توسيع هذا النهج أكثر ليشمل نموذج DEA المضاعف لبانكر وماينديراتا (1988) بواسطة كنيب وآخرون (1998). طبق تسيوناس وفيليبس (2023) تقنيات بايزيانية لاختبار الحساسية العالمية لنموذج DEA. إلى جانب تطوير نموذج DEA، تم تطبيق الإطار الذي وضعه بانكر وآخرون (1993) أيضًا لاختبار الاستدلال الإحصائي لنماذج DEA الموسعة، بما في ذلك نموذج DDF (سيمار وآخرون، 2012)، مقياس راسل (بادونينكو وموزهاروفسكي، 2020)، التقنية العامة ومقاييس التخصيص (سيمار وويلسون، 2020؛ سيمار وآخرون، 2024أ)، ومؤشرات إنتاجية هيكس-مورستين (سيمار وآخرون، 2024ب).
بعيدًا عن التوسعات المباشرة لعمل بانكر وآخرون (1993)، توفر الأسس الإحصائية لـ DEA من قبل بانكر وآخرون (1993) الأسس النظرية لنظرية الحد المركزي (كنيب وآخرون، 2015، 2016؛ نغوين وآخرون، 2022؛ سيمار وزيلينيك، 2020). علاوة على ذلك، تم استخدام البديهيات الإحصائية التي وضعها بانكر (1993) للاقتراح بنماذج عشوائية لمعالجة ضوضاء البيانات. من خلال أخذ ضوضاء البيانات في الاعتبار أثناء بناء حدود الإنتاج، تكون هذه النماذج العشوائية قادرة على تقليل خطر التقدير الناقص وتحسين دقة تقدير الكفاءة (تسيوناس، 2021). وفقًا لبانكر وآخرون (1996أ، ب)، تتجاهل نماذج DEA التقليدية ضوضاء البيانات والبديهيات المتعلقة بتوزيع الانحرافات عن حدود الممارسة الأفضل. من ناحية أخرى، تم انتقاد نهج SFA بسبب افتراضاته الوظيفية التقييدية، خاصة في سياق الإنتاج المشترك (كوسمانين وجونسون، 2017). تم تطوير عدة نماذج لتحسين القوة ضد أخطاء البيانات والنقاط الشاذة ولإدماج الإعدادات الاحتمالية في نظرية SFA، مستندة إلى البديهيات الإحصائية التي وضعها بانكر (1993) وسيمار وويلسون (2000أ، ب). قام أولسن وبيترسن (2016) بتصنيف المنهجيات العشوائية إلى ثلاثة اتجاهات:
  1. معاملة الكفاءات المقدرة كانحرافات عشوائية.
  2. تطوير نماذج تأخذ في الاعتبار إما أخطاء القياس أو أخطاء التحديد.
  3. تطوير نماذج لتوليد مجموعات إمكانيات الإنتاج العشوائية بناءً على التغيرات العشوائية في مجموعات البيانات.
يمكن تقسيم نماذج DEA العشوائية الحالية إلى ثلاث مجموعات بناءً على هذه الاتجاهات:
  • المجموعة 1: توسع في الاتجاه الأول ويشمل نموذج DEA البوتستراب (أجيليوبولوس وجيورجوبولوس، 2017؛ بوبدي وتاناكا، 2018؛ بوبكر وآخرون، 2023؛ ديا وآخرون، 2022؛ دو وآخرون، 2018؛ كانغ وآخرون، 2024؛ ميشالي وآخرون، 2023؛ ستات، 2002؛ سيمار وويلسون، 1998، 1999، 2000أ، ب؛ مورادي-مطلغ وآخرون، 2022).
  • المجموعة 2: توسع في كلا الاتجاهين الأول والثاني ويشمل نموذج DEA شبه المعلمي (عساف وجيلين، 2012؛ جونسون وماكغينيس، 2008؛ جيرادي وراجيرو، 2019؛ كوسمانين وكورتيلين، 2012؛ سيمار وويلسون، 2007).
  • المجموعة 3: توسع في الاتجاهين الثاني والثالث ويشمل نماذج DEA المقيدة بالفرص (أميرتييموري وآخرون، 2023؛ كوبر وآخرون، 1996، 1998، 2002؛ لين ولو، 2023؛ ميتروبولوس وآخرون، 2015؛ شيراز وآخرون، 2020؛ تالوري وآخرون، 2006).
يمكن العثور على مراجعة أكثر تفصيلًا لنموذج DEA العشوائي في أولسن وبيترسن (2016).

3.4 التقدم والتطورات في تقدير حدود الكوانتيل غير المعلمي

تعالج النماذج المذكورة أعلاه في القسم 3.3.3 عدم اليقين الكامن في البيانات وتعزز القوة. من المهم ملاحظة أن هذه النماذج تستند إلى البديهيات الإحصائية بدلاً من معالجة الضوضاء العشوائية مباشرة في البيانات الملاحظة. ومع ذلك، بسبب تطبيق مبدأ الاستقراء الأدنى الذي يقلل من الانحراف عن دالة الإنتاج في مجموعة البيانات، تجادل عدة دراسات (إيستيف وآخرون، 2020؛ تسيوناس، 2022) بأن DEA تعاني من مشكلة الإفراط في التخصيص. لمعالجة هذه العيوب، تركز الأساليب المقترحة حديثًا على تقدير حدود الكوانتيل بدلاً من الحدود الكاملة التي تشمل جميع الملاحظات. توفر هذه النماذج، المعروفة بتقدير حدود الكوانتيل غير المعلمي، تقديرات أكثر دقة للأسعار الظلية وتكون أكثر قوة في اختيار المتجهات الاتجاهية. بالإضافة إلى ذلك، يمكن لهذه النماذج معالجة أنواع مختلفة من ضوضاء البيانات، مثل الضوضاء العشوائية، والتباين غير المتجانس، والنقاط الشاذة (داي وآخرون، 2022؛ كوسمانين وزو، 2021؛ لياو وآخرون، 2024). وفقًا لداي وآخرون (2022)، يمكن تصنيف طرق تقدير حدود الكوانتيل غير المعلمي الحالية إلى مجموعتين: (1) الحدود الجزئية و(2) الانحدار الكوانتي. تولد طريقة الحدود الجزئية مقدر حدود يتناسب مع مجموعة فرعية من الملاحظات (داي وآخرون، 2022). في المقابل، تقوم طريقة الانحدار الكوانتي ببناء الحدود من خلال تطبيق معيار غير متماثل على العينة الكاملة من الملاحظات (داي وآخرون، 2022؛ كوينكر وباسيت، 1978). في هذا القسم، نستعرض عدة دراسات تمثيلية حول تقدير حدود الكوانتيل غير المعلمي.

3.4.1 طريقة الحدود الجزئية

تولد طريقة الحدود الجزئية حدودًا تتناسب مع مجموعة فرعية من الملاحظات. يتبع الخوارزمية لتوليد مثل هذه الحدود نموذج تحليل DEA التقليدي، الذي يزيد من المسافة بين وحدة القرار قيد التقييم ونقطة الإسقاط المرتبطة بها. بالإضافة إلى ذلك، مستوحاة من نموذج البحث الذي وضعه بانكر (1993)، تم إثبات النظرية الإحصائية الكامنة وراء طريقة الحدود الجزئية بشكل صارم.
اقترح كازالز وآخرون (2002) في البداية طريقة الترتيب- لتوليد حدود FDH مع معالجة النقاط الشاذة. بناءً على عمل كازالز وآخرون (2002)، اقترح أراجون وآخرون (2005) طريقة حدود الكوانتيل FDH من الترتيب-m. في الدراسات اللاحقة، فحص داويا وسيمار (2005) الخصائص الحدية لهذه الطرق وأثبتوا تقارب الترتيب-m و الترتيب- الطرق. من خلال توسيع هذا الخط من البحث، دمج داويا وسيمار (2007) تأثيرات المتغيرات البيئية على الكفاءة. قدم ويلوك وويلسون (2008) مفهوم الجذر-n لضمان اتساق النموذج وتقاربه، مع الفائدة الإضافية في التخفيف من لعنة الأبعاد. مؤخرًا، اقترح أتوود وشايك (2020) نموذج DEA كوانتيل في كل من الشكل المحيط
وشكل المضاعف. تم التحقق من الخصائص الإحصائية لهذا النموذج باستخدام محاكاة مونت كارلو وعينات nCm. تم تطبيق نهج مشابه على DEA المعتمد على السياق بواسطة سيفورد وزو (2003)، والذي يتضمن خوارزمية تحسب مستويات مختلفة من الحدود الفعالة من خلال استبعاد وحدات القرار الفعالة من مجموعة المرجع بشكل تكراري. ومع ذلك، لم يتم إثبات الاستدلال الإحصائي لـ DEA المعتمد على السياق بشكل مباشر. استكشف كارفالو وماركيز (2014) اقتصادات النطاق والمقياس لحساب اقتصادات التكامل الرأسي من خلال طريقة الحدود الجزئية.
بشكل عام، يمكن اعتبار الأساليب المذكورة أعلاه امتدادات لنماذج تحليل الكفاءة النسبية التقليدية. إنها تعالج القضايا الناتجة عن ضوضاء البيانات من خلال استبعاد القيم الشاذة، مع تحديد القيم الشاذة المستندة إلى النظرية الإحصائية. مستلهمًا من بانكر (1993)، تم تأسيس الاستدلال الإحصائي لمعظم الأساليب المقدمة للحدود الجزئية. ومع ذلك، لا تزال هذه الطرق تعتمد على هيكل الإنتاج التقليدي المغلق. علاوة على ذلك، مثل أساليب تحليل الكفاءة النسبية التقليدية، تظل الأسعار الظلية للمتغيرات غير فريدة. يمكن أن تركز الأبحاث المستقبلية على توسيع هذه الأساليب لتشمل نماذج الشبكات أو اقتراح مقاييس ذات أوزان مشتركة.

3.4.2 نهج الانحدار الكمي

الفرق الرئيسي بين نهج الانحدار الكمي ونهج الحدود الجزئية يكمن في استراتيجياتهما لتقدير عدم الكفاءة (الضوضاء). بينما يركز نهج الحدود الجزئية على استراتيجيات تقدير عدم الكفاءة في تحليل الكفاءة النسبية (DEA)، فإن نهج الانحدار الكمي عادةً ما يناسب مقدر الحدود لجميع الملاحظات، مع الأخذ في الاعتبار كل من عدم الكفاءة والضوضاء. علاوة على ذلك، أشار داي وآخرون (2022) إلى أن نهج الانحدار الكمي الحالي يمكن اعتباره امتدادًا لنموذج DEA العشوائي الذي اقترحه بانكر وآخرون (1991). تستعرض هذه القسم عدة نهج بارزة في الانحدار الكمي.
استلهمت من بانكر وماينديراتا (1992)، الذين طبقوا عدم المساواة لأفريات في تقدير الاحتمالات القصوى من خلال إدخال مصطلحات خطأ غير غاوسي، اقترح كوسمانين (2008) نموذج المربعات غير المعلمية المحدبة (CNLS). لاحقًا، أسس كوسمانين وجونسون (2010) ارتباطًا أوثق بين نماذج CNLS وDEA. على وجه التحديد، أظهر كوسمانين وجونسون (2010) أنه يمكن صياغة DEA كنموذج انحدار للمربعات الأقل حيث تكون مصطلح الخطأ دائمًا سالبًا. استنادًا إلى هذا الإثبات، اقترحوا نموذج المربعات غير المعلمية المحدبة المصححة ( تم اختبار نموذج NLS) ) ووجد أنه يتفوق على نموذج DEA التقليدي ونموذج COLS من خلال محاكاة مونت كارلو. علاوة على ذلك، تم تقديم نهج enveloping البيانات غير المعلمية العشوائية (StoNED) كإطار عمل أكثر عمومية يجمع بين DEA التقليدي وSFA، كما طوره كوسمانين (2006) وكوسمانين وكورتيلين (2012). قام كوسمانين وجونسون (2017) بتوسيع نموذج CNLS (كوسمانين، 2008) من خلال دمج القيود لقياس DDF.
بينما تعتبر الأساليب المذكورة سابقًا في الانحدار الكمي غير خطية، قدم وانغ وآخرون (2014) في أبحاثهم الحديثة نموذج الانحدار الكمي المحدب (CQR) لتقدير حدود الإنتاج. يستخدم هذا النموذج البرمجة الخطية ويمكن حله بسهولة باستخدام الخوارزميات القياسية (Gurobi، GAMS، CPLEX، MOSEK). ومع ذلك، يجب ملاحظة أن نموذج CQR الذي قدمه وانغ وآخرون (2014) لا ينتج حدود إنتاج فريدة. لمعالجة ذلك، اقترح كوسمانين وآخرون (2015) صياغة تربيعية كدالة هدف معدلة لنموذج CQR. دمج جيرادي وراجيرو (2019) نهج الانحدار الكمي في نموذج DEA كقياس عشوائي. جادل كوسمانين وزو (2021) بأن النهج التقليدي لـ DEA قد يبالغ في تقدير تكلفة التخفيف الهامشية. لمعالجة ذلك، جمعوا بين نموذج CQR الخاص بوانغ وآخرون (2014) مع قياس DDF الذي اقترحه كوسمانين وجونسون (2017) لتطوير نهج الانحدار الكمي المحدب.
تأخذ الطريقة المقترحة في الاعتبار بشكل صريح كل من الضوضاء وعدم الكفاءة للتخفيف من المشكلات الناجمة عن الحساسية العالية لبيانات الضوضاء في تحليل الكفاءة النسبية (DEA). وبالتالي، تحسب الطريقة المقترحة تكلفة التخفيف الهامشية مع خلوها من مشكلات ضوضاء البيانات في نماذج DEA التقليدية. في أبحاث أكثر حداثة، قام داي وآخرون (2023) بتلخيص الأساليب الحالية للانحدار الكمي واقترحوا نموذجًا عامًا للدوال غير المعلمية المقيدة بالشكل. متجاوزين النماذج المحدبة المستخدمة في الانحدار الكمي الشرطي (CQR) (برمجة خطية) والانحدار المحدب المتوقع (CER) (برمجة رباعية)، قدم داي وآخرون (2023) معلمة لتمثيل الترتيب الجزئي، مما أدى إلى توليد نماذج CQR وCER غير المحدبة العامة. بالإضافة إلى ذلك، استخدم إسبانيا وآخرون (2024) نموذج DEA لتشكيل القيود وتقدير دوال الإنتاج من خلال نماذج إضافية تعتمد على انحدار الأشرطة. اقترح لياو وآخرون (2024) نهج انحدار غير معلمي محدب لمعالجة مشكلة الإفراط في التكيف والقيم الشاذة.
بشكل عام، عند توليد دالة الإنتاج، يستفيد بعض الباحثين من النهج القائم على الخطية في نموذج DEA لاقتراح قيود شكلية تضمن أن الحدود الإنتاجية الناتجة تلبي خصائص التزايد والتقعر في نظرية الإنتاج. ومع ذلك، فإن الحدود الناتجة هي حدود كوانتيلية بدلاً من أن تكون واحدة تغلف جميع الملاحظات. تعالج النماذج المقترحة الضوضاء العشوائية، وعدم التجانس، ومشكلات القيم الشاذة في تقنيات DEA التقليدية. وتعتبر الأساليب المذكورة أعلاه جميعها امتدادات لعمل بانكر وآخرين (1991)، حيث تعتبر حدود DEA حالة خاصة من هذه النماذج.

4 التحليل السياقي في تحليل البيانات

اعتمدت معظم الأبحاث المبكرة حول تقييم المتغيرات السياقية على إطار الحدود العشوائية، كما هو موضح في دراسات مثل روجيرو وفيتاليانو (1999) وروسكو (2001). قدّرت هذه الأساليب درجات كفاءة DEA في المرحلة الأولى ثم اعتبرت هذه الدرجات كمتغيرات تابعة في تحليل قائم على الانحدار في المرحلة الثانية. ومع ذلك، تفتقر الطريقة ذات المرحلتين إلى أساس إحصائي قوي وفشلت في توضيح عملية توليد البيانات المتوافقة مع DEA ذي المرحلتين (بانكر وآخرون، 2019). ونتيجة لذلك، تتساءل عدة دراسات عن قدرة الطريقة ذات المرحلتين على التحقيق بفعالية في تأثير المتغيرات السياقية (فورسوند، 1999؛ غروسكوف، 1996).
ردًا على هذه الانتقادات، قام بانكر وناتاراجان (2008) بتطبيق محاكاة مونت كارلو لأول مرة لمقارنة أداء نهج المرحلة الثانية القائم على تحليل الكفاءة (باستخدام تقدير المربعات الصغرى، والاحتمالية القصوى، وتقدير توبيط في المرحلة الثانية) مع النهج البارامترية ذات المرحلة الواحدة والمرحلتين. تشير نتائج المحاكاة إلى أن الإجراءات القائمة على تحليل الكفاءة تتفوق على الطرق البارامترية. كما أسست هذه الدراسة الأساس الإحصائي للنهج القائم على تحليل الكفاءة في مرحلتين واقترحت إطارًا لتقدير المتغيرات السياقية مع الأخذ في الاعتبار الضوضاء العشوائية ذات الجانبين. قام بانكر وآخرون (2019) بمقارنة أداء DEA+OLS مع الانحدار DEA+Tobit ونهج سيمار وويلسون (2007) ضمن بيئة إنتاجية مليئة بالضوضاء. في بيئة خالية من الضوضاء، تظهر نتائج المحاكاة أن نهج DEA+OLS هو الأفضل، خاصة عند تقييم بيانات عينات كبيرة (تعرف بأنها 400 وحدة قياس كفاءة في اختبارات المحاكاة). يتم ملاحظة نفس النتيجة في البيئات المليئة بالضوضاء، حيث يتفوق نهج DEA+OLS على الآخرين. يجب ملاحظة أنه بينما يكون أداء نهج DEA+OLS مشابهًا لأداء الانحدار DEA+Tobit، لم يتم تأسيس الأساس النظري للانحدار DEA+Tobit. علاوة على ذلك، يظهر نهج DEA+OLS أداءً أفضل من نهج سيمار وويلسون (2007) عندما تؤثر المتغيرات السياقية بشكل كبير على الناتج. حتى في الحالات التي لا تؤثر فيها المتغيرات السياقية على الناتج، فإن نهج DEA+OLS
يؤدي بشكل مشابه أو أحيانًا أفضل من نهج سيمار وويلسون (2007). وبالتالي، يُوصى باستخدام نهج DEA+OLS لتقييم تأثير المتغيرات السياقية.
بالإضافة إلى نموذج تحليل البيانات التقليدي، يتم استخدام عدة نماذج موسعة لتقييم الأداء في المرحلة الأولى من التحليل. تشمل الأمثلة نموذج تحليل البيانات مع المخرجات غير المرغوب فيها (بانديوبادياي، 2011) ونموذج تحليل البيانات الشبكي ذو المرحلتين (تان وآخرون، 2021). قام جيمينيز وآخرون (2024) بتطبيق نموذج الاتجاه لفائدة الشك. في المرحلة الأولى من التحليل لتقييم إدارة الدول للجائحة.
لقد وجدنا أيضًا عدة دراسات توسع تحليل المتغيرات السياقية ضمن إطار DEA. اقترح راماليو وآخرون (2010) استخدام نماذج الانحدار الكسري كقياس في المرحلة الثانية. من خلال دمج نماذج الانحدار الكسري مع الطريقة الديناميكية العامة للحظات، اقترح دي سوزا وغوميز (2015) نهجًا مقطعيًا لقياس الآثار الديناميكية الناتجة عن المتغيرات السياقية. قام كاراجيانيس (2015) بتوسيع عمل بانكر وناتاراجان (2008) لتفكيك تأثير الكفاءة إلى تأثيرات متزامنة على الكفاءة الفنية واستخدام القدرة. ذكر جونسون وكوسمانين (2012) أن DEA يمكن صياغته كحالة خاصة مقيدة من نموذج CNLS واقترحوا نهجًا شبه غير بارامتري لتقدير معاملات المتغيرات السياقية من خلال حساب من مرحلة واحدة. اقترح يو وآخرون (2024) نموذج تخصيص موارد مركزي لتقييم تأثير المتغيرات السياقية من خلال حساب من مرحلة واحدة. طبق شي وآخرون (2025) تفسيرات شابلي الإضافية لفحص تأثيرات المتغيرات السياقية. على حد علمنا، فقط جونسون وكوسمانين (2012) بنوا إطار محاكاة. لمقارنة أداء النموذج المقترح مع نهج DEA+OLS. نقترح تطوير إطار محاكاة أكثر شمولاً يتضمن مستويات متغيرة من حجم العينة والضوضاء لمقارنة أداء النهجين المذكورين بشكل أفضل.

5 شبكة المؤلفين والمؤتمرات المنظمة

لتقديم نظرة شاملة على مساهمات الباحث بنك، تم إجراء تحليل ببليومتري استنادًا إلى 203 من منشوراته. كانت قاعدة بيانات سكوبس هي المصدر الرئيسي. لاستخراج البيانات ذات الصلة التي تم جمعها في ديسمبر 2024. تقدم الجدول 1 تاريخًا زمنيًا لارتباطات بنك مع مختلف الجامعات ومدارس الأعمال عبر ولايات مختلفة في الولايات المتحدة من 1980 إلى 2023. بدأت رحلته الأكاديمية في عام 1980 في كلية هارفارد للأعمال في ماساتشوستس. على مر السنين، قدم مساهمات كبيرة للعديد من الجامعات، بينما قضى معظم مسيرته الأكاديمية اللاحقة من 2005 إلى 2023 في كلية فوكس للأعمال بجامعة تمبل في فيلادلفيا. يعرض هذا التاريخ الواسع تفاني بنك في الأكاديميا ومساهماته عبر مؤسسات تعليمية متعددة في الولايات المتحدة.
الجدول 2 يوضح منشورات بنكير حسب النوع، بإجمالي 203. ويبرز أن المقالات تمثل الغالبية بواقع 133. )، تليها أوراق المؤتمر في 36 ( ). المراجعات،
الجدول 1 انتماءات المصرفيين الجامعية (1980-2023)
سنة الانتماء دولة
2005-2023 جامعة تمبل/كلية فوكس للأعمال فيلادلفيا
2006 جامعة ولاية أريزونا / كلية و. ب. كاري للأعمال أريزونا
2003-2005 جامعة كاليفورنيا، ريفرسايد كاليفورنيا
2005 جامعة كاليفورنيا/مدرسة أندرسون للإدارة كاليفورنيا
1992-2004 جامعة تكساس في دالاس تكساس
1989-1997 جامعة مينيسوتا/كلية كارلسون للإدارة مينابوليس
1984-1996 جامعة كارنيجي ميلون بيتسبرغ
1994 جامعة مينيسوتا/كلية كارلسون للإدارة مينيسوتا
1980 جامعة هارفارد / كلية هارفارد للأعمال ماساتشوستس
الجدول 2 منشورات المصرفيين حسب النوع
نوع عدد المنشورات نسبة مئوية
مقالة ١٣٣ 65.5
ورقة مؤتمر ٣٦ 17.7
مراجعة 11 ٥.٤
كتاب محرر من وقائع المؤتمر ٧ 3.4
فصل الكتاب ٨ 3.9
تحرير ٧ 3.4
ملاحظة 1 0.5
إجمالي ٢٠٣ 100
تشكل وقائع المؤتمرات، فصول الكتب، الافتتاحيات، والملاحظات أجزاءً أصغر من أعماله، مما يبرز مساهماته الأكاديمية المتنوعة عبر منصات مختلفة.
تقدم الجدول 3 توزيع بنك المؤلفين في منشوراته. ويظهر أنه كان المؤلف الأول في 159 ورقة بحثية. )، مما يدل على مساهمته الأساسية كباحث رئيسي. بالإضافة إلى ذلك، تم إدراجه كالمؤلف الثاني في 28 ورقة ( ) والمؤلف الثالث في 14 ورقة ( في المنشورات التي تم إدراجه فيها كمؤلف رابع أو أكثر، كانت هناك حالتان فقط، تمثلان 1% من إجمالي منشوراته. توضح هذه التفاصيل بروز دوره كمساهم رئيسي أو مؤلف رئيسي في الغالبية العظمى من أعماله الأكاديمية.
يوضح الجدول 4 أفضل 10 مؤلفين مشاركين لبانكر بناءً على عدد المشاركات في التأليف التي شاركوها في الأعمال الأكاديمية. ويقدم أسماء المؤلفين المشاركين المعنيين، وعدد الأوراق التعاونية مع بانكر، وعناوينها، والجامعات أو المؤسسات التابعة لها، والمدارس المحددة.
الجدول 3 ترتيب مؤلفي البنك في المنشورات
المؤلف الأول المؤلف الثاني المؤلف الثالث الرابع مؤلف إجمالي
عدد المنشورات 159 ٢٨ 14 2 ٢٠٣
نسبة مئوية 78.3 13.8 6.9 1.0 100
الجدول 4 أفضل 10 علماء تعاونوا مع بانكر
مؤلف مشارك عدد المشاركات في التأليف عنوان الجامعة/الانتماء مدرسة/ كلية بلد
تشانغ، شيهوي ٢٤ أستاذ المحاسبة جامعة دريكسل جامعة دريكسل الولايات المتحدة
ناتاراجان، راماشاندران نات ١٣ أستاذ إدارة جامعة تكساس في دالاس مدرسة نافين جيندال لإدارة الأعمال الولايات المتحدة
إمرونجيد، أ 11 أستاذ تحليل الأعمال جامعة ساري كلية الأعمال في ساري المملكة المتحدة
كوفمان، روبرت ج 11 أستاذ نظم المعلومات جامعة سنغافورة للإدارة جامعة سنغافورة للإدارة سنغافورة
ويليام و. كوبر 11 أستاذ بحوث العمليات جامعة تكساس في أوستن مدرسة مكومبز للأعمال الولايات المتحدة
داتار، سريكانت م 10 أستاذ الإدارة جامعة هارفارد مدرسة هارفارد للأعمال الولايات المتحدة
أندرسون، مارك سي 10 أستاذ مشارك في المحاسبة جامعة كالجاري كلية هاسكين للأعمال كندا
ريدل، رينيه 9 أستاذ الأعمال الرقمية والابتكار جامعة العلوم التطبيقية العليا كلية الإدارة النمسا
باول أ. بافلو 9 أستاذ علوم المعلومات جامعة هيوستن كلية باوئر للأعمال الولايات المتحدة
ديفيس، فريد د.د 9 أستاذ تكنولوجيا المعلومات جامعة تكساس تك كلية راولز للأعمال الولايات المتحدة
الشكل 8 شبكة المؤلفين المشاركين للبنك
أو الكليات داخل تلك المؤسسات، والدول التي تقع فيها. توضح هذه الجدول العلاقات التعاونية التي أنشأها بانكر مع هؤلاء الأفراد عبر مختلف الجامعات، مما يبرز شراكاته داخل المجتمع الأكاديمي، لا سيما في مجالات المحاسبة، والإدارة، وتحليل الأعمال، ونظم المعلومات، وبحوث العمليات.
الشكل 8 يظهر شبكة من التعاون المؤلف مع بانكر، حيث تمثل العقد المؤلفين المشاركين، والألوان المختلفة تشير إلى مجموعات مختلفة من المؤلفين. استخدمنا VOSviewer، وهو أداة برمجية تم تطويرها بواسطة فان إيك ووالتمان (2010)، لإنشاء وتصوير خرائط استنادًا إلى بيانات الشبكة. حجم العقد يمثل عدد المنشورات المشتركة. بانكر هو العقدة المركزية في الشبكة، وقد شارك في تأليف أوراق مع العديد من الباحثين في عدة مجموعات. تمثل مجموعات المؤلفين، المعروضة بألوان مختلفة، باحثين ذوي اهتمامات ومواقع و/أو خبرات مشتركة. على سبيل المثال، تحتوي المجموعة ذات اللون الأزرق الداكن على علماء مثل كوبر وتشيرنز، الذين قدموا مساهمات كبيرة في تقديم وتطوير DEA، بينما تشمل المجموعة ذات اللون الأزرق الفوسفوري باحثين في الغالب في مجال المحاسبة. إن ارتباط بانكر بالباحثين في مجموعات متعددة يشير إلى أن لديه مجموعة واسعة من التعاونات العالمية واهتمامات البحث. وهذا ليس مفاجئًا، نظرًا لأنه عالم محترم للغاية قدم مساهمات كبيرة في مجالات متعددة، بما في ذلك الرياضيات التطبيقية، والمحاسبة، والاقتصاد.
الجدول 5 الافتتاحيات التحريرية للمجلة التي شارك في تأليفها بانكر
عنوان مجلة مرجع
التقدم في تحليل كفاءة البيانات: الاحتفال بالذكرى الأربعين لتحليل كفاءة البيانات والذكرى المئوية لميلاد البروفيسور أبراهام تشارنس المجلة الأوروبية للبحوث التشغيلية إمرونج نيجاد وآخرون (2019)
التطورات الأخيرة في استخدام DEA في القطاع العام علوم التخطيط الاجتماعي والاقتصادي أهن وآخرون (2018)
إدارة أداء الأعمال في ظل بيئات غير مؤكدة – II مجلة مركز الكاتدرائية: مجلة أبحاث الأعمال والاقتصاد تشارلز وبانكر (2017)
إدارة أداء الأعمال في ظل بيئات غير مؤكدة – الجزء الأول مجلة مركز الكاتدرائية: مجلة أبحاث الأعمال والاقتصاد تشارلز وبانكر (2016)
تحليل كفاءة البيانات في القطاع العام علوم التخطيط الاجتماعي والاقتصادي إمرونج نيجاد وآخرون، (2014أ، ب)
الكفاءة والإنتاجية: النظرية والتطبيقات سجلات بحوث العمليات إمرونجاد وبانكر (2010)
تقدم الجدول 5 تجميعًا لافتتاحيات المجلات التي شارك في تأليفها بانكر، بما في ذلك عناوينها، والمجلات المعنية، ومراجعها. لقد قدم بانكر مساهمات كبيرة في القطع التحريرية في مجلات مرموقة، مع التركيز بشكل أساسي على تحليل البيانات (DEA) وتطبيقاته عبر مختلف القطاعات. بالتعاون مع مؤلفين مشاركين مثل إمروزنجاد، تشارلز، آhn، وآخرين، كتب افتتاحيات تحتفل بالمعالم في تاريخ DEA، وتستكشف التطورات الأخيرة، لا سيما في القطاع العام، وتفحص الجوانب النظرية للكفاءة والإنتاجية.
يوفر الجدول 6 قائمة بمؤتمرات الأبحاث التي تم تحريرها بالتعاون مع بانكر عبر مؤتمرات دولية مختلفة حول تحليل البيانات (DEA)، بما في ذلك عناوين الأبحاث، وعناوين المؤتمرات المعنية والمراجع. تمتد هذه الأبحاث من عام 2012 إلى 2017، وتبرز التطبيقات المتنوعة والتطورات في منهجيات DEA التي تم مناقشتها خلال التجمعات الأكاديمية الهامة في جميع أنحاء العالم. تعكس العناوين مجموعة واسعة من المواضيع، بما في ذلك التطبيقات الحديثة لـ DEA، ودورها في التنمية المستدامة، وقياس الأداء، والاستكشافات النظرية.
الجدول 7 يمثل سجل نشر بانكر عبر المجلات الأكاديمية حيث ساهم بثلاثة أو أكثر من المنشورات. يُظهر هذا الجدول مساهماته الكبيرة في مجالات دراسية متنوعة. يبدأ بمجلة الأبحاث التشغيلية الأوروبية، التي تحتوي على أعلى عدد من المنشورات بواقع 17، تليها مجلة علوم الإدارة بـ 15 منشورًا. كما يبرز الجدول 7 المجلات التي نشر فيها بانكر أكثر من نصف أعماله الأكاديمية. مظهرًا تأثيره الكبير من خلال هذه المجلات المتميزة.
توضح هذه البيانات الإنتاج الأكاديمي المتنوع والواسع لبانكر، مما يبرز تأثيره عبر مجالات أكاديمية متعددة ومشاركته النشطة في تعزيز المعرفة وتطبيق تحليل البيانات (DEA) والمنهجيات ذات الصلة.
الجدول 6: وقائع المؤتمر التي تم تحريرها بالتعاون مع بانكر*
عنوان مؤتمر المؤلفون/المحررون رقم الكتاب الدولي الموحد
التطبيقات الحديثة لتحليل كفاءة البيانات المؤتمر الدولي السادس عشر للـ DEA، يونيو 2017، جامعة الاقتصاد، براغ، جمهورية التشيك (DEA2017) إمرونج نيجاد، بانكر وآخرون (2017) 978185449 4337
التطبيقات الحديثة لتحليل كفاءة البيانات المؤتمر الدولي الرابع عشر للـ DEA، مايو 2016، جامعة جيانغهان، ووهان، الصين (DEA2016) إمرونج نيجاد، بانكر وآخرون (2016) 978185449 4139
تحليل كفاءة البيانات وتطبيقاته المؤتمر الدولي الثالث عشر للـ DEA، أغسطس 2015، براونشفايغ، ألمانيا (DEA2015) إمرونجيد، بانكر وآخرون (2015) 978185449 4979
التنمية المستدامة وقياس الأداء ورشة العمل الدولية لمكافحة المخدرات، 17-19 سبتمبر 2014، هيرموسيلو، سونورا، المكسيك بانكر وآخرون (2014) 978185449 4825
نظرية وتطبيقات تحليل كفاءة البيانات المؤتمر الدولي الثاني عشر للـ DEA، أبريل 2014، جامعة مالايا، كوالالمبور، ماليزيا (DEA2014) إمرونجاد وآخرون (2014أ، ب) 978185449 4870
تحليل كفاءة البيانات وقياس الأداء المؤتمر الدولي الحادي عشر للـ DEA، يونيو 2013، سامسون، تركيا (DEA2013) بانكر، إمروزنجاد وآخرون (2013) 978185449 4771
تحليل كفاءة البيانات: النظرية والتطبيقات المؤتمر الدولي العاشر حول تحليل البيانات، ناتال، البرازيل (DEA2012) بانكر، إمروزنجاد وآخرون (2012) 978185449 4375
الجدول 7 المجلات التي تحتوي على ثلاث منشورات أو أكثر كتبها بانكر
المجلة/المصدر عدد المنشورات % المنشورات % التراكمي
المجلة الأوروبية للبحوث التشغيلية 17 ٨.٧ ٨.٧
علوم الإدارة 15 ٧.٧ 16.3
سجلات بحوث العمليات 9 ٤.٦ ٢٠.٩
مراجعة المحاسبة ٧ 3.6 ٢٤.٥
البحث المحاسبي المعاصر ٧ 3.6 ٢٨.١
مجلة المحاسبة والاقتصاد ٧ 3.6 31.6
أبحاث نظم المعلومات ٦ 3.1 ٣٤.٧
ربع سنوي نظم المعلومات الإدارية ٦ 3.1 37.8
ملاحظات المحاضرات في نظم المعلومات والتنظيم ٥ 2.6 ٤٠.٣
مجلة بحوث المحاسبة ٤ 2.0 42.3
مجلة المحاسبة والتدقيق والمالية ٤ 2.0 ٤٤.٤
مجلة نظم إدارة المعلومات ٤ 2.0 ٤٦.٤
مجلة تحليل الإنتاجية ٤ 2.0 ٤٨.٥
أعمال المؤتمر الدولي السنوي في هاواي لعلوم الأنظمة ٤ 2.0 50.5
اتصالات جمعية الحاسبات الآلية ٣ 1.5 ٥٢.٠
معاملات IEEE في هندسة البرمجيات ٣ 1.5 53.6
مجلة أبحاث المحاسبة الإدارية ٣ 1.5 ٥٥.١
الجدول 8 أعلى عشرة أوراق بحثية تم الاستشهاد بها من تأليف بانكر
نشر مجلة عدد الاقتباسات
بانكر وآخرون (1984) علوم الإدارة 11,547
بانكر وموري (1986) بحوث العمليات 863
المصرفي (1984) المجلة الأوروبية للبحوث التشغيلية 696
أندرسون وآخرون (2003) مجلة أبحاث المحاسبة 639
المصرفي (1993) علوم الإدارة ٦١٦
بانكر و ناتاراجان (2008) بحوث العمليات ٥٥٩
بانكر وثرال (1992) المجلة الأوروبية للبحوث التشغيلية ٥١٧
بانكر وآخرون (2000) مراجعة المحاسبة ٤٩٠
بانكر وآخرون (1998) علوم الإدارة ٤٥١
بانكر وموري (1986) علوم الإدارة 418

6 التحليل البيبليومتري

6.1 المستوى الكلي

توضح الشكل 9 شبكة الكلمات الرئيسية المستخدمة في منشورات بانكر، حيث تظهر كل منها ثلاث مرات على الأقل، مما يبرز الموضوعات المترابطة داخل أبحاثه. تشير العقد الأكبر داخل الشبكة إلى الكلمات الرئيسية المستخدمة بشكل أكثر تكرارًا، مما يدل على أهميتها في أبحاثه. تمثل تجمعات الكلمات الرئيسية موضوعات مترابطة وحقول فرعية داخل عمله، مثل تحليل البيانات، البرمجة الخطية، والإنتاجية في السنوات السابقة، والابتكار، والتمويل، والتجارة في الأوقات الأكثر حداثة. تعكس هذه التحولات في الكلمات الرئيسية على مر الزمن
الشكل 9 شبكة الكلمات الرئيسية المستخدمة من قبل المصرفي
تطور تركيز أبحاثه، من المواضيع التقليدية إلى مواضيع أكثر حداثة ضمن هذا المجال.
تقدم الجدول 8 أكثر عشرة أوراق بحثية تم الاستشهاد بها والتي كتبها أو شارك في كتابتها بانكر. يتضمن عناوين المنشورات، والمجلات المعنية التي نُشرت فيها، وعدد الاستشهادات التي تلقتها كل ورقة. هذه الأوراق، التي نُشرت في مجموعة من المجلات الأكاديمية المرموقة مثل علوم الإدارة، أبحاث العمليات، المجلة الأوروبية لأبحاث العمليات، مجلة أبحاث المحاسبة، ومراجعة المحاسبة، تبرز التأثير الكبير لبانكر في تطوير نماذج DEA وتطبيقاتها.

6.2 المستوى الدقيق

في هذا القسم، نقدم نتيجة تحليل بيبليومتري مفصل للثلاثة مجموعات البحث الرئيسية التي تم تحديدها في عمل بنك.

6.2.1 تحليل موضوع العوائد على المقياس

توضح الشكل 10 اتجاهات الاقتباس للأوراق الرئيسية لبانكر حول RTS و MPSS. تبرز ورقة بانكر وآخرون (1984) كأكثر ورقة تم الاقتباس منها، حيث تظهر اتجاهًا تصاعديًا مستمرًا على مدى ما يقرب من ثلاثة عقود، مما يبرز دورها الأساسي في أدبيات DEA. تُعتبر ورقة بانكر (1984) ثاني أكثر ورقة تم الاقتباس منها، حيث تظهر اتجاهًا تصاعديًا، على الرغم من أن عدد الاقتباسات سنويًا أقل بكثير مقارنةً بورقة بانكر وآخرون (1984). بينما تحتوي الأوراق الأخرى في الشكل 10 (بخلاف تلك المنشورة في 1984) على عدد أقل من الاقتباسات، لا تزال مواضيع RTS و MPSS تمثل جزءًا كبيرًا من أدبيات DEA.
الشكل 10 اتجاهات الاقتباس للأوراق المتعلقة بـ RTS و MPSS
الشكل 11 اتجاهات الاقتباس للأوراق المتعلقة بالتحليل الإحصائي

6.2.2 تحليل التحليل الإحصائي في موضوع تحليل البيانات

تظهر الشكل 11 اتجاهات الاقتباس لأوراق بانكر الرئيسية حول التحليل الإحصائي في تحليل البيانات. الورقة الأكثر اقتباسًا في هذا التجمع هي ورقة بانكر (1993)، التي أظهرت اتجاه اقتباس مستقر نسبيًا على مدار العقد الماضي. تليها ورقة بانكر وتشانغ (2006) عن كثب، مع اتجاه تصاعدي في سنواتها الأولى ولكنها توقفت في السنوات الأخيرة. بشكل عام، يبدو أن مساهمات بانكر في التحليل الإحصائي في تحليل البيانات لم تحقق نفس مستوى البروز مثل أوراقه حول RTS.
تسلط هذه التحليلات الضوء على تأثير وأهمية أبحاث بانكر المستمرة داخل المجتمع الأكاديمي، لا سيما في مجالات RTS وMPSS والمنهجيات الإحصائية في DEA. لا يزال عمله يؤثر ويشكل اتجاه البحث في هذه المجالات، كما يتضح من اتجاهات الاقتباس المستمرة وتركيز البحث المتطور على مر الزمن.

6.2.3 تحليل موضوع التحليل السياقي

توضح الشكل 12 المسارات المتناقضة في الاقتباسات بين ورقتين من أوراق بانكر حول التحليل في المرحلة الثانية في تحليل البيانات. شهدت ورقة بانكر ونتراجان، التي نُشرت في عام 2008، في البداية زيادة في الاقتباسات خلال سنواتها الأولى ولكنها أظهرت تراجعًا في الزخم في السنوات الأخيرة. على العكس من ذلك، استمتعت ورقة بانكر لعام 2019 بزيادة مستمرة في الاقتباسات، مما يدل على اهتمام مستمر وأهمية في هذا المجال. ومع ذلك، عند مقارنة عدد الاقتباسات بشكل عام، تظهر ورقة بانكر حول التحليل في المرحلة الثانية عددًا أقل نسبيًا من الاقتباسات مقارنة بالنهج البديل الذي اقترحه سيمار وويلسون (2007). لمناقشة شاملة حول الطريقة البديلة وتحليلها الببليومتري، يرجى الرجوع إلى مرادي-مطلغ وإيمروزنجاد (2022).
الشكل 12 اتجاهات الاقتباس للأوراق المتعلقة بتحليل الكفاءة في المرحلة الثانية

7 الخاتمة

استكشفت هذه الورقة التأثير الكبير والدائم للبروفيسور راجيف بانكر على مجال تحليل البيانات (DEA). على مدار مسيرته المرموقة، كان بانكر له دور حاسم في تشكيل مسار DEA، تاركًا إرثًا يستمر في إلهام العلماء والباحثين والممارسين على حد سواء.
كشفت تحيتنا الشاملة لمساهمات بانكر مدى الحياة عن نهج متميز ورائد في DEA، يركز على ثلاثة تجمعات رئيسية: العوائد على المقياس (RTS) وحجم المقياس الأكثر إنتاجية (MPSS) في DEA، الاستدلال الإحصائي لـ DEA، والتحليل السياقي. تمثل كل من هذه التجمعات تحولًا في نموذج DEA، مدفوعًا بأبحاث بانكر الرائدة والتزامه الثابت بتطوير هذا المجال.
في تكريم تفاني بانكر مدى الحياة لـ DEA، نعترف ليس فقط بالعمق الفكري لمساهماته ولكن أيضًا بالإلهام الذي قدمه للعديد من العلماء. بينما يتقدم مجتمع DEA، فإنه يفعل ذلك على الأساس المتين الذي وضعه بانكر. من خلال هذه الورقة، نحتفل بإرثه الدائم ونعبر عن امتناننا لتأثيره التحويلي على تحليل البيانات.
الشكر يود المؤلفون أن يشكروا المحرر والمراجعين المجهولين على ملاحظاتهم القيمة حول النسخة السابقة من هذه المخطوطة.

الإعلانات

تعارض المصالح يعلن المؤلفون أنهم ليس لديهم مصالح مالية متنافسة معروفة أو علاقات شخصية قد تبدو أنها تؤثر على العمل المبلغ عنه في هذه الورقة.
الوصول المفتوح هذه المقالة مرخصة بموجب رخصة المشاع الإبداعي للاستخدام والمشاركة والتكيف والتوزيع وإعادة الإنتاج في أي وسيلة أو صيغة، طالما أنك تعطي الائتمان المناسب للمؤلفين الأصليين والمصدر، وتوفر رابطًا لرخصة المشاع الإبداعي، وتوضح ما إذا كانت هناك تغييرات قد أُجريت. الصور أو المواد الأخرى من طرف ثالث في هذه المقالة مشمولة في رخصة المشاع الإبداعي للمقالة، ما لم يُشار إلى خلاف ذلك في سطر ائتمان للمادة. إذا كانت المادة
غير مشمولة في رخصة المشاع الإبداعي للمقالة واستخدامك المقصود غير مسموح به بموجب اللوائح القانونية أو يتجاوز الاستخدام المسموح به، ستحتاج إلى الحصول على إذن مباشرة من صاحب حقوق الطبع والنشر. لعرض نسخة من هذه الرخصة، قم بزيارةhttp://creativecommons.org/licenses/by/4.0/.

References

Aggelopoulos, E., & Georgopoulos, A. (2017). Bank branch efficiency under environmental change: A bootstrap DEA on monthly profit and loss accounting statements of Greek retail branches. European Journal of Operational Research, 261(3), 1170-1188.
Ahn, H., Afsharian, M., Emrouznejad, A., & Banker, R. (2018). Recent developments on the use of DEA in the public sector. Socio-Economic Planning Science, 61, 1-3.
Alirezaee, M., Hajinezhad, E., & Paradi, J. C. (2018). Objective identification of technological returns to scale for data envelopment analysis models. European Journal of Operational Research, 266(2), 678-688.
Alperovych, Y., Hübner, G., & Lobet, F. (2015). How does governmental versus private venture capital backing affect a firm’s efficiency? Evidence from Belgium. Journal of Business Venturing, 30(4), 508-525.
Amirteimoori, A., Charles, V., & Mehdizadeh, S. (2023). Stochastic data envelopment analysis in the presence of undesirable outputs. Journal of the Operational Research Society, 74(12), 2619-2632.
Anderson, M. C., Banker, R. D. & Janakiraman, S. N. (2003). Are selling, general, and administrative costs “sticky”?. Journal of Accounting Research, 41(1), 47-63.
Aragon, Y., Daouia, A., & Thomas-Agnan, C. (2005). Nonparametric frontier estimation: A conditional quantile-based approach. Econometric Theory, 21(2), 358-389.
Asmild, M., Paradi, J. C., & Reese, D. N. (2006). Theoretical perspectives of trade-off analysis using DEA. Omega, 34(4), 337-343.
Assaf, A. G., & Gillen, D. (2012). Measuring the joint impact of governance form and economic regulation on airport efficiency. European Journal of Operational Research, 220(1), 187-198.
Assani, S., Jiang, J., Cao, R., & Yang, F. (2018). Most productive scale size decomposition for multi-stage systems in data envelopment analysis. Computers & Industrial Engineering, 120, 279-287.
Atwood, J., & Shaik, S. (2020). Theory and statistical properties of quantile data envelopment analysis. European Journal of Operational Research, 286(2), 649-661.
Badunenko, O., & Mozharovskyi, P. (2020). Statistical inference for the Russell measure of technical efficiency. Journal of the Operational Research Society, 71(3), 517-527.
Bandyopadhyay, S. (2011). In search of contextual variables in a stochastic DEA framework: Effect of regulation on efficiency of Indian cement industry. Journal of the Operational Research Society, 62(9), 1621-1637.
Banker, R. D. (1984). Estimating most productive scale size using data envelopment analysis. European Journal of Operational Research, 17(1), 35-44.
Banker, R. D. (1993). Maximum likelihood, consistency and data envelopment analysis: A statistical foundation. Management Science, 39(10), 1265-1273.
Banker, R. D. (1996). Hypothesis tests using data envelopment analysis. Journal of Productivity Analysis, 7, 139-159.
Banker, R. D., & Chang, H. (1995). A simulation study of hypothesis tests for differences in efficiencies. International Journal of Production Economics, 39(1-2), 37-54.
Banker, R.D ., & Chang, H. (2006). The super-efficiency procedure for outlier identification, not for ranking efficient units. European Journal of Operational Research, 175(2), 1311-1320.
Banker, R. D., & Maindiratta, A. (1986). Piecewise loglinear estimation of efficient production surfaces. Management Science, 32(1), 126-135.
Banker, R. D., & Maindiratta, A. (1988). Nonparametric analysis of technical and allocative efficiencies in production. Econometrica: Journal of the Econometric Society, 56, 1315-1332.
Banker, R. D., & Natarajan, R. (2008). Evaluating contextual variables affecting productivity using data envelopment analysis. Operations Research, 56(1), 48-58.
Banker, R. D., & Thrall, R. M. (1992). Estimation of returns to scale using data envelopment analysis. European Journal of Operational Research, 62(1), 74-84.
Banker, R. D., Potter, G., & Srinivasan, D. (2000). An empirical investigation of an incentive plan that includes nonfinancial performance measures. The Accounting Review, 75(1), 65-92.
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science, 30(9), 1078-1092.
Banker, R. D., Datar, S. M., & Kemerer, C. F. (1991). A model to evaluate variables impacting the productivity of software maintenance projects. Management Science, 37, 1-18.
Banker, R. D., Chang, H., & Sinha, K. K. (1994). Tests to evaluate the separability or substitutability of inputs to a production system.
Banker, R. D., Bardhan, I., & Cooper, W. W. (1996a). A note on returns to scale in DEA. European Journal of Operational Research, 88(3), 583-585.
Banker, R. D., Chang, H., & Cooper, W. W. (1996b). Equivalence and implementation of alternative methods for determining returns to scale in data envelopment analysis. European Journal of Operational Research, 89(3), 473-481.
Banker, R. D., Janakiraman, S., & Natarajan, R. (2002). Evaluating the adequacy of parametric functional forms in estimating monotone and concave production functions. Journal of Productivity Analysis, 17, 111-132.
Banker, R. D., Khosla, I., & Sinha, K. K. (1998). Quality and competition. Management science, 44(9), 1179-1192.
Banker, R. D., Cooper, W. W., Seiford, L. M., Thrall, R. M., & Zhu, J. (2004a). Returns to scale in different DEA models. European Journal of Operational Research, 154(2), 345-362.
Banker, R. D., Janakiraman, S., & Natarajan, R. (2004b). Analysis of trends in technical and allocative efficiency: An application to Texas public school districts. European Journal of Operational Research, 154(2), 477-491.
Banker, R. D., Chang, H., & Natarajan, R. (2005). Productivity change, technical progress, and relative efficiency change in the public accounting industry. Management Science, 51(2), 291-304.
Banker, R. D., Chang, H., & Natarajan, R. (2007). Estimating DEA technical and allocative inefficiency using aggregate cost or revenue data. Journal of Productivity Analysis, 27, 115-121.
Banker, R. D., Chang, H., & Lee, S. Y. (2010a). Differential impact of Korean banking system reforms on bank productivity. Journal of Banking & Finance, 34(7), 1450-1460.
Banker, R. D., Zheng, Z. E., & Natarajan, R. (2010b). DEA-based hypothesis tests for comparing two groups of decision making units. European Journal of Operational Research, 206(1), 231-238.
Banker, R. D., Cooper, W. W., Seiford, L. M. & Zhu, J. (2011). Returns to scale in DEA. Handbook on data envelopment analysis, pp. 41-70.
Banker, R.D., Emrouznejad, A., Lopes, A. L. M. & de Almeida, M. R. (2012). Data envelopment analysis: Theory and applications. In Proceedings of the 10th International Conference on DEA, Natal, Brazil.
Banker, R. D., Emrouznejad, A., Bal, H., Alp, I. & Cengiz, M. A. (2013). Data envelopment analysis and performance measurement. In Proceedings of the 11th International Conference of DEA, June 2013, Samsun, Turkey.
Banker, R.D., Emrouznejad, A., Vargas, F. & Flores, P. (2014). Sustainable development and performance measurement. In Proceedings of the International DEA Workshop, 2014, Hermosillo, Sonora, Mexico.
Banker, R., Natarajan, R., & Zhang, D. (2019). Two-stage estimation of the impact of contextual variables in stochastic frontier production function models using data envelopment analysis: Second stage OLS versus bootstrap approaches. European Journal of Operational Research, 278(2), 368-384.
Bardhan, I. R., Cooper, W. W., & Kumbhakar, S. C. (1998). A simulation study of joint uses of data envelopment analysis and statistical regressions for production function estimation and efficiency evaluation. Journal of Productivity Analysis, 9, 249-278.
Bessent, A., Bessent, W., Elam, J., & Clark, T. (1988). Efficiency frontier determination by constrained facet analysis. Operations Research, 36(5), 785-796.
Bobde, S. M., & Tanaka, M. (2018). Efficiency evaluation of electricity distribution utilities in India: A twostage DEA with bootstrap estimation. Journal of the Operational Research Society, 69(9), 1423-1434.
Boubaker, S., Le, T. D., Manita, R., & Ngo, T. (2023). The trade-off frontier for ESG and Sharpe ratio: a bootstrapped double-frontier data envelopment analysis. Annals of Operations Research. https://doi.org/ 10.1007/s10479-023-05506-z
Carvalho, P., & Marques, R. C. (2014). Computing economies of vertical integration, economies of scope and economies of scale using partial frontier nonparametric methods. European Journal of Operational Research, 234(1), 292-307.
Cazals, C., Florens, J. P., & Simar, L. (2002). Nonparametric frontier estimation: A robust approach. Journal of Econometrics, 106(1), 1-25.
Chang, K. P., & Guh, Y. Y. (1991). Linear production functions and the data envelopment analysis. European Journal of Operational Research, 52(2), 215-223.
Charles, V., & Banker, R. D. (2016). Business performance management under uncertain environments-I. Editorial. Journal of Centrum Cathedra: The Business and Economics Research Journal, 9(2), 90-91. https://doi.org/10.1108/JCC-01-2017-0003
Charles, V., & Banker, R. D. (2017). Business performance management under uncertain environments-II Editorial. Journal of Centrum Cathedra: The Business and Economics Research Journal, 10(1), 2-3. https://doi.org/10.1108/JCC-07-2017-018
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research, 2(6), 429-444.
Charnes, A., Cooper, W. W., & Thrall, R. M. (1986). Classifying and characterizing efficiencies and inefficiencies in data development analysis. Operations Research Letters, 5(3), 105-110.
Charnes, A., Gallegos, A., & Li, H. (1996). Robustly efficient parametric frontiers via multiplicative DEA for domestic and international operations of the Latin American airline industry. European Journal of Operational Research, 88(3), 525-536.
Chen, C. M., & Delmas, M. A. (2012). Measuring eco-inefficiency: A new frontier approach. Operations Research, 60(5), 1064-1079.
Chen, K., & Zhu, J. (2019). Computational tractability of chance constrained data envelopment analysis. European Journal of Operational Research, 274(3), 1037-1046.
Cooper, W. W., & Tone, K. (1997). Measures of inefficiency in data envelopment analysis and stochastic frontier estimation. European Journal of Operational Research, 99(1), 72-88.
Cooper, W. W., Huang, Z., & Li, S. X. (1996). Chapter 13 satisficing DEA models under chance constraints. Annals of Operations Research, 66, 279-295.
Cooper, W. W., Huang, Z., Lelas, V., Li, S. X., & Olesen, O. B. (1998). Chance constrained programming formulations for stochastic characterizations of efficiency and dominance in DEA. Journal of Productivity Analysis, 9, 53-79.
Cooper, W. W., Park, D. K. S., & Ciurana, P. J. T. P. (2000). Marginal rates and elasticities of substitution with additive models in DEA. Journal of Productivity Analysis, 13, 105-123.
Cooper, W. W., Deng, H., Huang, Z., & Li, S. X. (2002). Chance constrained programming approaches to technical efficiencies and inefficiencies in stochastic data envelopment analysis. Journal of the Operational Research Society, 53, 1347-1356.
Dai, S., Kuosmanen, T. and Zhou, X. (2022) Partial frontiers are not quantiles. arXiv preprint arXiv:2205. 11885.
Dai, S., Kuosmanen, T., & Zhou, X. (2023). Generalized quantile and expectile properties for shape constrained nonparametric estimation. European Journal of Operational Research, 310(2), 914-927.
Daouia, A., & Gijbels, I. (2011). Robustness and inference in nonparametric partial frontier modeling. Journal of Econometrics, 161(2), 147-165.
Daouia, A., & Simar, L. (2005). Robust nonparametric estimators of monotone boundaries. Journal of Multivariate Analysis, 96(2), 311-331.
Daouia, A., & Simar, L. (2007). Nonparametric efficiency analysis: A multivariate conditional quantile approach. Journal of Econometrics, 140(2), 375-400.
Dia, M., Takouda, P.M. & Golmohammadi, A., (2022). Assessing the performance of Canadian credit unions using a three-stage network bootstrap DEA. Annals of Operations Research, 1-33.
Du, K., Worthington, A. C., & Zelenyuk, V. (2018). Data envelopment analysis, truncated regression and double-bootstrap for panel data with application to Chinese banking. European Journal of Operational Research, 265(2), 748-764.
e Souza, G. D. S., & Gomes, E. G. (2015). Management of agricultural research centers in Brazil: A DEA application using a dynamic GMM approach. European Journal of Operational Research, 240(3), 819-824.
Emrouznejad, A., & Amin, G. R. (2009). DEA models for ratio data: Convexity consideration. Applied Mathematical Modelling, 33(1), 486-498.
Emrouznejad, A., & Banker, R. D. (2010). Efficiency and productivity: theory and applications. Annals of Operations Research, 173(1), 1-10.
Emrouznejad, A., Anouze, A. L., & Thanassoulis, E. (2010). A semi-oriented radial measure for measuring the efficiency of decision making units with negative data, using DEA. European Journal of Operational Research, 200(1), 297-304.
Emrouznejad, A., Banker, R. D., Lopes, A. L. M., & de Almeida, M. R. (2014a). Data envelopment analysis in the public sector. Socio-Economic Planning Sciences, 48(1), 2-3.
Emrouznejad, A., Banker, R., Munisamy, S. & Arabi, B. (2014b). Theory and applications of data envelopment analysis. In Proceedings of the 12th International Conference of DEA, April 2014, University of Malaya, Kuala Lumpur, Malaysia.
Emrouznejad, A., Banker, R.D., Ahn, H. & Afsharian, M. (2015). Data envelopment analysis and its applications. In Proceedings of the 13th International Conference of DEA, August 2015, Braunschweig, Germany.
Emrouznejad, A., Banker, R., Ray, S. C. & Chen, L. (2016). Recent applications of data envelopment analysis. In Proceedings of the 14th International Conference of DEA, May 2016, Jianghan University, Wuhan, China.
Emrouznejad, A., Jablonský, J., Banker, R. & Toloo, M. (2017). Recent applications of data envelopment analysis. In Proceedings of the 15th International Conference of DEA, June 2017, University of Economics, Prague, Czech Republic.
Emrouznejad, A., Banker, R. D., & Neralic, L. (2019). Advances in data envelopment analysis: Celebrating the 40th anniversary of DEA and the 100th anniversary of Professor Abraham Charnes’ birthday. European Journal of Operational Research, 278(2), 365-367.
España, V. J., Aparicio, J., Barber, X., & Esteve, M. (2024). Estimating production functions through additive models based on regression splines. European Journal of Operational Research, 312(2), 684-699.
Esteve, M., Aparicio, J., Rabasa, A., & Rodriguez-Sala, J. J. (2020). Efficiency analysis trees: A new methodology for estimating production frontiers through decision trees. Expert Systems with Applications, 162, 113783.
Färe, R., Grosskopf, S., & Lovell, C. K. (1985). The measurement of efficiency of production (6th ed.). Springer Science & Business Media.
Färe, R., Grosskopf, S., Lovell, C. K., & Yaisawarng, S. (1993). Derivation of shadow prices for undesirable outputs: a distance function approach. The Review of Economics and Statistics, 75, 374-380.
Färe, R., Grosskopf, S., & Lovell, C. K. (1994). Production frontiers. Cambridge University Press.
Fernandez, C., Koop, G., & Steel, M. F. J. (2002). Multiple-output production with undesirable outputs: An application to nitrogen surplus in agriculture. Journal of the American Statistical Association, 97(458), 432-442.
Førsund, F. R. (1999). The evolution of DEA-The economics perspective. University of Oslo Working paper, Oslo, Norway.
Førsund, F. R. (2018). Economic interpretations of DEA. Socio-Economic Planning Sciences, 61, 9-15.
Førsund, F. R., & Hjalmarsson, L. (2004). Calculating scale elasticity in DEA models. Journal of the Operational Research Society, 55(10), 1023-1038.
Førsund, F. R., Hjalmarsson, L., Krivonozhko, V. E., & Utkin, O. B. (2007). Calculation of scale elasticities in DEA models: Direct and indirect approaches. Journal of Productivity Analysis, 28, 45-56.
Førsund, F. R., Kittelsen, S. A., & Krivonozhko, V. E. (2009). Farrell revisited-Visualizing properties of DEA production frontiers. Journal of the Operational Research Society, 60, 1535-1545.
Frisch, R. (1964). Theory of production. Springer Science & Business Media.
Fukuyama, H. (2000). Returns to scale and scale elasticity in data envelopment analysis. European Journal of Operational Research, 125(1), 93-112.
Fukuyama, H. (2003). Scale characterizations in a DEA directional technology distance function framework. European Journal of Operational Research, 144(1), 108-127.
Giménez, V., Prior, D., Thieme, C., & Tortosa-Ausina, E. (2024). International comparisons of COVID-19 pandemic management: What can be learned from activity analysis techniques? Omega, 122, 102966.
Giraleas, D., Emrouznejad, A., & Thanassoulis, E. (2012). Productivity change using growth accounting and frontier-based approaches-Evidence from a Monte Carlo analysis. European Journal of Operational Research, 222(3), 673-683.
Golany, B., & Yu, G. (1997). Estimating returns to scale in DEA. European Journal of Operational Research, 103(1), 28-37.
Gong, B. H., & Sickles, R. C. (1992). Finite sample evidence on the performance of stochastic frontiers and data envelopment analysis using panel data. Journal of Econometrics, 51(1-2), 259-284.
Grosskopf, S. (1996). Statistical inference and nonparametric efficiency: A selective survey. Journal of Productivity Analysis, 7, 161-176.
Halická, M., & Trnovská, M. (2021). A unified approach to non-radial graph models in data envelopment analysis: Common features, geometry, and duality. European Journal of Operational Research, 289(2), 611-627.
Hatami-Marbini, A., Emrouznejad, A., & Agrell, P. J. (2014). Interval data without sign restrictions in DEA. Applied Mathematical Modelling, 38, 2028-2036.
Johnson, A. L., & Kuosmanen, T. (2012). One-stage and two-stage DEA estimation of the effects of contextual variables. European Journal of Operational Research, 220(2), 559-570.
Johnson, A. L., & McGinnis, L. F. (2008). Outlier detection in two-stage semiparametric DEA models. European Journal of Operational Research, 187(2), 629-635.
Jradi, S., & Ruggiero, J. (2019). Stochastic data envelopment analysis: A quantile regression approach to estimate the production frontier. European Journal of Operational Research, 278(2), 385-393.
Kang, H. J., Kim, C., & Choi, K. (2024). Combining bootstrap data envelopment analysis with social networks for rank discrimination and suitable potential benchmarks. European Journal of Operational Research, 312(1), 283-297.
Kao, C., & Hwang, S. N. (2011). Decomposition of technical and scale efficiencies in two-stage production systems. European Journal of Operational Research, 211(3), 515-519.
Kao, C., & Hwang, S. N. (2019). Efficiency evaluation in the presence of undesirable outputs: The most favorable shadow price approach. Annals of Operations Research, 278, 5-16.
Karagiannis, R. (2015). A system-of-equations two-stage DEA approach for explaining capacity utilization and technical efficiency. Annals of Operations Research, 227, 25-43.
Kerstens, K., & Eeckaut, P. V. (1999). Estimating returns to scale using non-parametric deterministic technologies: A new method based on goodness-of-fit. European Journal of Operational Research, 113(1), 206-214.
Khezrimotlagh, D. (2022). Simulation designs for production frontiers. European Journal of Operational Research, 303(3), 1321-1334.
Khezrimotlagh, D., & Zhu, J. (2023). The role of unobserved units in two-stage network data envelopment analysis. Journal of the Operational Research Society, 74(5), 1275-1285.
Kneip, A., Park, B. U., & Simar, L. (1998). A note on the convergence of nonparametric DEA estimators for production efficiency scores. Econometric Theory, 14(6), 783-793.
Kneip, A., Simar, L., & Wilson, P. W. (2015). When bias kills the variance: Central limit theorems for DEA and FDH efficiency scores. Econometric Theory, 31(2), 394-422.
Kneip, A., Simar, L., & Wilson, P. W. (2016). Testing hypotheses in nonparametric models of production. Journal of Business & Economic Statistics, 34(3), 435-456.
Kneip, A., Simar, L., & Wilson, P. W. (2021). Inference in dynamic, nonparametric models of production: Central limit theorems for Malmquist indices. Econometric Theory, 37(3), 537-572.
Koenker, R., & Bassett, G., Jr. (1978). Regression quantiles. Journal of the Econometric Society., 23, 33-50.
Kontolaimou, A., & Tsekouras, K. (2010). Are cooperatives the weakest link in European banking? A nonparametric metafrontier approach. Journal of Banking & Finance, 34(8), 1946-1957.
Korhonen, P. J., Soleimani-Damaneh, M., & Wallenius, J. (2011). Ratio-based RTS determination in weightrestricted DEA models. European Journal of Operational Research, 215(2), 431-438.
Korostelëv, A. P., Simar, L., & Tsybakov, A. B. (1995a). Efficient estimation of monotone boundaries. The Annals of Statistics, 23, 476-489.
Korostelëv, A. P., Simar, L., & Tsybakov, A. (1995b). On estimation of monotone and convex boundaries. In Annales de l’ISUP, 39(1), 3-18.
Krivonozhko, V. E., Førsund, F. R., & Lychev, A. V. (2014). Measurement of returns to scale using non-radial DEA models. European Journal of Operational Research, 232(3), 664-670.
Krivonozhko, V. E., Utkin, O. B., Volodin, A. V., Sablin, I. A., & Patrin, M. (2004). Constructions of economic functions and calculations of marginal rates in DEA using parametric optimization methods. Journal of the Operational Research Society, 55, 1049-1058.
Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.). The University of Chicago Press.
Kuosmanen, T. (2006). Stochastic nonparametric envelopment of data: Combining virtues of SFA and DEA in a unified framework. MTT Discussion paper no. 3/2006, MT
Kuosmanen, T. (2008). Representation theorem for convex nonparametric least squares. The Econometrics Journal, 11(2), 308-325.
Kuosmanen, T., & Johnson, A. L. (2010). Data envelopment analysis as nonparametric least-squares regression. Operations Research, 58(1), 149-160.
Kuosmanen, T., & Johnson, A. (2017). Modeling joint production of multiple outputs in StoNED: Directional distance function approach. European Journal of Operational Research, 262(2), 792-801.
Kuosmanen, T., & Kortelainen, M. (2012). Stochastic non-smooth envelopment of data: Semi-parametric frontier estimation subject to shape constraints. Journal of Productivity Analysis, 38, 11-28.
Kuosmanen, T., Post, T., & Scholtes, S. (2007). Non-parametric tests of productive efficiency with errors-invariables. Journal of Econometrics, 136(1), 131-162.
Kuosmanen, T., Johnson, A. & Saastamoinen, A. (2015). Stochastic nonparametric approach to efficiency analysis: A unified framework. InData Envelopment Analysis: A Handbook of Models and Methods, pp. 191-244
Kuosmanen, T., & Zhou, X. (2021). Shadow prices and marginal abatement costs: Convex quantile regression approach. European Journal of Operational Research, 289(2), 666-675.
Lee, C. Y., & Wang, K. (2019). Nash marginal abatement cost estimation of air pollutant emissions using the stochastic semi-nonparametric frontier. European Journal of Operational Research, 273(1), 390-400.
Leleu, H. (2013). Shadow pricing of undesirable outputs in nonparametric analysis. European Journal of Operational Research, 231(2), 474-480.
Liao, Z., Dai, S., & Kuosmanen, T. (2024). Convex support vector regression. European Journal of Operational Research, 313(3), 858-870.
Lin, S. W., & Lu, W. M. (2023). Efficiency assessment of public sector management and culture-led urban regeneration using the enhanced Russell-based directional distance function with stochastic data. Journal of the Operational Research Society, 75, 1-19.
Lotfi, F. H., Jahanshahloo, G. R., & Esmaeili, M. (2007). An alternative approach in the estimation of returns to scale under weight restrictions. Applied Mathematics and Computation, 189(1), 719-724.
Lozano, S. (2011). Scale and cost efficiency analysis of networks of processes. Expert Systems with Applications, 38(6), 6612-6617.
Lozano, S., & Villa, G. (2010). Gradual technical and scale efficiency improvement in DEA. Annals of Operations Research, 173, 123-136.
Mergoni, A., Emrouznejad, A., & De Witte, K. (2025). Fifty years of data envelopment analysis. European Journal of Operational Research. https://doi.org/10.1016/j.ejor.2024.12.049
Michali, M., Emrouznejad, A., Dehnokhalaji, A., & Clegg, B. (2023). Subsampling bootstrap in network DEA. European Journal of Operational Research, 305(2), 766-780.
Mitropoulos, P., Talias, MA., & Mitropoulos, I. (2015). Combining stochastic DEA with Bayesian analysis to obtain statistical properties of the efficiency scores: An application to Greek public hospitals. European Journal of Operational Research, 243(1), 302-311.
Moradi-Motlagh, A., & Emrouznejad, A. (2022). The origins and development of statistical approaches in non-parametric frontier models: A survey of the first two decades of scholarly literature (1998-2020). Annals of Operations Research, 318(1), 713-741.
Nguyen, B. H., Simar, L., & Zelenyuk, V. (2022). Data sharpening for improving central limit theorem approximations for data envelopment analysis-type efficiency estimators. European Journal of Operational Research, 303(3), 1469-1480.
Odeck, J. (2009). Statistical precision of DEA and Malmquist indices: A bootstrap application to Norwegian grain producers. Omega, 37(5), 1007-1017.
Olesen, O. B., & Petersen, N. C. (2016). Stochastic data envelopment analysis-A review. European Journal of Operational Research, 251(1), 2-21.
Olesen, O. B., Petersen, N. C., & Podinovski, V. V. (2022). Scale characteristics of variable returns-to-scale production technologies with ratio inputs and outputs. Annals of Operations Research, 318(1), 383-423.
Papaioannou, G., & Podinovski, V. V. (2023). Multicomponent production technologies with restricted allocations of shared inputs and outputs. European Journal of Operational Research, 308(1), 274-289.
Pastor, J. T., Ruiz, J. L., & Sirvent, I. (1999). A statistical test for detecting influential observations in DEA. European Journal of Operational Research, 115(3), 542-554.
Pastor, J. T., Ruiz, J. L., & Sirvent, I. (2002). A statistical test for nested radial DEA models. Operations Research, 50(4), 728-735.
Perelman, S., & Santín, D. (2009). How to generate regularly behaved production data? A Monte Carlo experimentation on DEA scale efficiency measurement. European Journal of Operational Research, 199(1), 303-310.
Pham, M., Simar, L., & Zelenyuk, V. (2024). Statistical inference for aggregation of Malmquist productivity indices. Operations Research, 72(4), 1615-1629.
Podinovski, V. V. (2004). On the linearisation of reference technologies for testing returns to scale in FDH models. European Journal of Operational Research, 152(3), 800-802.
Podinovski, V. V. (2017). Returns to scale in convex production technologies. European Journal of Operational Research, 258(3), 970-982.
Podinovski, V. V. (2019). Direct estimation of marginal characteristics of nonparametric production frontiers in the presence of undesirable outputs. European Journal of Operational Research, 279(1), 258-276.
Podinovski, V. V. (2022). Variable and constant returns-to-scale production technologies with component processes. Operations Research, 70(2), 1238-1258.
Podinovski, V. V., Førsund, F. R., & Krivonozhko, V. E. (2009). A simple derivation of scale elasticity in data envelopment analysis. European Journal of Operational Research, 197(1), 149-153.
Podinovski, V. V., & Førsund, F. R. (2010). Differential characteristics of efficient frontiers in data envelopment analysis. Operations Research, 58(6), 1743-1754.
Podinovski, V. V., Ismail, I., Bouzdine-Chameeva, T., & Zhang, W. (2014). Combining the assumptions of variable and constant returns to scale in the efficiency evaluation of secondary schools. European Journal of Operational Research, 239(2), 504-513.
Podinovski, V. V., Chambers, R. G., Atici, K. B., & Deineko, I. D. (2016). Marginal values and returns to scale for nonparametric production frontiers. Operations Research, 64(1), 236-250.
Ramalho, E. A., Ramalho, J. J., & Henriques, P. D. (2010). Fractional regression models for second stage DEA efficiency analyses. Journal of Productivity Analysis, 34, 239-255.
Resti, A. (2000). Efficiency measurement for multi-product industries: A comparison of classic and recent techniques based on simulated data. European Journal of Operational Research, 121(3), 559-578.
Rosen, D., Schaffnit, C., & Paradi, J. C. (1998). Marginal rates and two-dimensional level curves in DEA. Journal of Productivity Analysis, 9, 205-232.
Rosko, M. D. (2001). Impact of HMO penetration and other environmental factors on hospital X-inefficiency. Medical Care Research and Review, 58, 430-454.
Ruggiero, J. (1996). On the measurement of technical efficiency in the public sector. European Journal of Operational Research, 90(3), 553-565.
Ruggiero, J. (1998). Non-discretionary inputs in data envelopment analysis. European Journal of Operational Research, 111(3), 461-469.
Ruggiero, J., & Vitaliano, D. F. (1999). Assessing the efficiency of public schools using data envelopment analysis and frontier regression. Contemporary Economic Policy, 17(3), 321-331.
Sahoo, B. K., Khoveyni, M., Eslami, R., & Chaudhury, P. (2016). Returns to scale and most productive scale size in DEA with negative data. European Journal of Operational Research, 255(2), 545-558.
Sahoo, B. K., Zhu, J., Tone, K., & Klemen, B. M. (2014). Decomposing technical efficiency and scale elasticity in two-stage network DEA. European Journal of Operational Research, 233(3), 584-594.
Schmidt, P. (1985). Frontier production functions. Econometric Reviews, 4(2), 289-328.
Seiford, L. M., & Thrall, R. M. (1990). Recent developments in DEA: The mathematical programming approach to frontier analysis. Journal of Econometrics, 46(1-2), 7-38.
Seiford, L. M., & Zhu, J. (2003). Context-dependent data envelopment analysis-Measuring attractiveness and progress. Omega, 31(5), 397-408.
Shephard, R. W. (1970). Theory of cost and production fictions. Princeton University Press.
Shi, Y., Charles, V., & Zhu, J. (2025). Bank financial sustainability evaluation: Data envelopment analysis with random forest and Shapley additive explanations. European Journal of Operational Research, 321(2), 614-630.
Shiraz, R. K., Hatami-Marbini, A., & Emrouznejad, H. F. (2020). Chance-constrained cost efficiency in data envelopment analysis model with random inputs and outputs. Operational Research-An International Journal, 20, 1863-1898.
Simar, L. (2007). How to improve the performances of DEA/FDH estimators in the presence of noise. Journal of Productivity Analysis, 28, 183-201.
Simar, L., Vanhems, A., & Wilson, P. W. (2012). Statistical inference for DEA estimators of directional distances. European Journal of Operational Research, 220(3), 853-864.
Simar, L., & Wilson, P. W. (1998). Productivity growth in industrialized countries (No. UCL-Université Catholique de Louvain). Université Catholique de Louvain.
Simar, L., & Wilson, P. W. (1999). Estimating and bootstrapping Malmquist indices. European Journal of Operational Research, 115(3), 459-471.
Simar, L., & Wilson, P. W. (2000a). A general methodology for bootstrapping in non-parametric frontier models. Journal of Applied Statistics, 27(6), 779-802.
Simar, L., & Wilson, P. W. (2000b). Statistical inference in nonparametric frontier models: The state of the art. Journal of Productivity Analysis, 13, 49-78.
Simar, L., & Wilson, P. W. (2002). Non-parametric tests of returns to scale. European Journal of Operational Research, 139(1), 115-132.
Simar, L., & Wilson, P. W. (2007). Estimation and inference in two-stage, semi-parametric models of production processes. Journal of Econometrics, 136(1), 31-64.
Simar, L., & Wilson, P. W. (2020). Technical, allocative and overall efficiency: Estimation and inference. European Journal of Operational Research, 282(3), 1164-1176.
Simar, L., & Zelenyuk, V. (2020). Improving finite sample approximation by central limit theorems for estimates from data envelopment analysis. European Journal of Operational Research, 284(3), 1002-1015.
Simar, L., Zelenyuk, V., & Zhao, S. (2024a). Inference for aggregate efficiency: Theory and guidelines for practitioners. European Journal of Operational Research, 316(1), 240-254.
Simar, L., Zelenyuk, V. & Zhao, S., (2024b). Statistical inference for Hicks-Moorsteen productivity indices. Annals of Operations Research, 1-29
Singh, S., & Ranjan, P. (2018). Efficiency analysis of non-homogeneous parallel sub-unit systems for the performance measurement of higher education. Annals of Operations Research, 269, 641-666.
Soleimani-Damaneh, M. (2013). Another approach for estimating RTS in dynamic DEA. Journal of Productivity Analysis, 39(1), 75-81.
Soleimani-Damaneh, M., Jahanshahloo, G. R., & Reshadi, M. (2006). On the estimation of returns-to-scale in FDH models. European Journal of Operational Research, 174(2), 1055-1059.
Soleimani-Damaneh, M., & Reshadi, M. (2007). A polynomial-time algorithm to estimate returns to scale in FDH models. Computers & Operations Research, 34(7), 2168-2176.
Staat, M. (2002). Bootstrapped efficiency estimates for a model for groups and hierarchies in DEA. European Journal of Operational Research, 138(1), 1-8.
Sueyoshi, T. (1999). DEA duality on returns to scale (RTS) in production and cost analyses: An occurrence of multiple solutions and differences between production-based and cost-based RTS estimates. Management Science, 45(11), 1593-1608.
Sueyoshi, T., & Aoki, S. (2001). A use of a nonparametric statistic for DEA frontier shift: The Kruskal and Wallis rank test. Omega, 29(1), 1-18.
Sueyoshi, T., & Goto, M. (2012). Returns to scale and damages to scale on US fossil fuel power plants: Radial and non-radial approaches for DEA environmental assessment. Energy Economics, 34(6), 2240-2259.
Sueyoshi, T., & Goto, M. (2013). Returns to scale vs. damages to scale in data envelopment analysis: An impact of US clean air act on coal-fired power plants. Omega, 41(2), 164-175.
Sueyoshi, T., & Sekitani, K. (2005). Returns to scale in dynamic DEA. European Journal of Operational Research, 161(2), 536-544.
Taleb, M., Khalid, R., Ramli, R., Ghasemi, M. R., & Ignatius, J. (2022). An integrated bi-objective data envelopment analysis model for measuring returns to scale. European Journal of Operational Research, 296(3), 967-979.
Talluri, S., Narasimhan, R., & Nair, A. (2006). Vendor performance with supply risk: A chance-constrained DEA approach. International Journal of Production Economics, 100(2), 212-222.
Tan, Y., Wanke, P., Antunes, J., & Emrouznejad, A. (2021). Unveiling endogeneity between competition and efficiency in Chinese banks: A two-stage network DEA and regression analysis. Annals of Operations Research, 306(1), 131-171.
Tone, K. (2001). On returns to scale under weight restrictions in data envelopment analysis. Journal of Productivity Analysis, 16, 31-47.
Tortosa-Ausina, E., Grifell-Tatjé, E., Armero, C., & Conesa, D. (2008). Sensitivity analysis of efficiency and Malmquist productivity indices: An application to Spanish savings banks. European Journal of Operational Research, 184(3), 1062-1084.
Tsionas, M. G. (2021). Optimal combinations of stochastic frontier and data envelopment analysis models. European Journal of Operational Research, 294(2), 790-800.
Tsionas, M. G. (2022). Convex non-parametric least squares, causal structures and productivity. European Journal of Operational Research, 303(1), 370-438.
Tsionas, M. G., & Philippas, D. (2023). Measures of global sensitivity in linear programming: Applications in banking sector. Annals of Operations Research, 330(1), 585-607.
Van Eck, N., & Waltman, L. (2010). Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, 84(2), 523-538.
Varian, H. R. (1984). The nonparametric approach to production analysis. Econometrica: Journal of the Econometric Society, 579-597.
Varian, H. R. (1985). Non-parametric analysis of optimizing behavior with measurement error. Journal of Econometrics, 30(1-2), 445-458.
Wang, H., Pan, C., Wang, Q., & Zhou, P. (2020). Assessing sustainability performance of global supply chains: An input-output modeling approach. European Journal of Operational Research, 285(1), 393-404.
Wang, Y., Wang, S., Dang, C., & Ge, W. (2014). Nonparametric quantile frontier estimation under shape restriction. European Journal of Operational Research, 232(3), 671-678.
Wang, K., Wei, Y. M., & Huang, Z. (2016). Potential gains from carbon emissions trading in China: A DEA based estimation on abatement cost savings. Omega, 63, 48-59.
Wheelock, D. C., & Wilson, P. W. (2008). Non-parametric, unconditional quantile estimation for efficiency analysis with an application to Federal Reserve check processing operations. Journal of Econometrics, 145(1-2), 209-225.
Wu, J., & An, Q. (2013). Slacks-based measurement models for estimating returns to scale. International Journal of Information and Decision Sciences, 5(1), 25-35.
Wu, F., Wang, S. Y., & Zhou, P. (2023). Marginal abatement cost of carbon dioxide emissions: The role of abatement options. European Journal of Operational Research, 310(2), 891-901.
Yu, A., Zhang, H., Liu, H. C., Shi, Y., & Bi, W. (2024). Dynamic centralized resource allocation approach with contextual impacts: Analyzing Chinese carbon allocation plans. Annals of Operations Research, 341(1), 451-483.
Zelenyuk, V. (2013). A scale elasticity measure for directional distance function and its dual: Theory and DEA estimation. European Journal of Operational Research, 228(3), 592-600.
Zelenyuk, V. (2020). Aggregation of inputs and outputs prior to data envelopment analysis under big data. European Journal of Operational Research, 282(1), 172-187.
Zhang, Q., & Yang, Z. (2015). Returns to scale of two-stage production process. Computers & Industrial Engineering, 90, 259-268.
Zhou, P., Zhou, X., & Fan, L. W. (2014). On estimating shadow prices of undesirable outputs with efficiency models: A literature review. Applied Energy, 130, 799-806.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

  1. Victor Podinovski
    V.Podinovski@lboro.ac.uk
    Vincent Charles
    c.vincent@qub.ac.uk
    Amir Moradi-Motlagh
    amoradi@swin.edu.au
    1 Surrey Business School, University of Surrey, Guildford GU2 7XH, UK
    2 Loughborough Business School, Loughborough University, Leicestershire LE11 3TU, UK
    3 Queen’s Business School, Queen’s University Belfast, Belfast BT9 5EE, UK
    4 Swinburne University of Technology, Melbourne, Australia
  2. From a definitional perspective, the applied directional Benefit-of-the-Doubt model can be regarded as a DDF model without explicit inputs. However, from a computational perspective, it can be treated as a conventional DDF model due to the presence of undesirable outputs and the absence of inputs.
    The sample size for the simulation tests in Johnson and Kuosmanen’s (2012) study is set at 100 , which is categorised as a ‘medium-sized’ sample in Banker et al. (2019). Their proposed model demonstrated better performance than the DEA + OLS approach in simulation testing.
    In addition to Scopus, Google Scholar is used to extract seven conference proceedings co-edited by Banker.
  3. *Banker also co-organised or served as an invited speaker at several other DEA conferences, including: DEA2002 (invited speaker; Moscow, Russia), DEA2004 (invited speaker; Birmingham, UK), DEA2007 (coorganiser; Hyderabad, India), DEA2009 (organiser; Philadelphia, USA), DEA2010 (invited speaker; Beirut, Lebanon), DEA2011 (invited speaker; Thessaloniki, Greece), DEA workshop 2013 (co-organiser; Hebrew University, Israel), DEA2018 (co-organiser; Wuhan, China), DEA2019 (co-organiser; Calgary, Canada)

Journal: Annals of Operations Research, Volume: 351, Issue: 2
DOI: https://doi.org/10.1007/s10479-025-06473-3
Publication Date: 2025-02-24

Rajiv Banker’s lasting impact on data envelopment analysis

Ali Emrouznejad (D) ⋅ Victor Podinovski (D) ⋅ Vincent Charles (D) ⋅ Chixiao Lu (D) ⋅ Amir Moradi-Motlagh (D)

Received: 6 September 2024 / Accepted: 1 January 2025 / Published online: 24 February 2025
© The Author(s) 2025

Abstract

This paper provides a comprehensive analysis of Professor Rajiv Banker’s significant impact on the field of Data Envelopment Analysis (DEA). Through an extensive review of his scholarly contributions, we explore three major clusters within DEA research: (1) Returns-to-Scale (RTS) and Most Productive Scale Size (MPSS), (2) Statistical Inference in DEA, and (3) Contextual Analysis. Banker’s pioneering research has significantly advanced DEA methodologies, addressing fundamental challenges related to scale efficiency, statistical robustness, and the influence of contextual variables on performance. His work has bridged theoretical developments and practical applications, influencing diverse fields such as economics, finance, and management science. By examining citation trends and bibliometric data, we trace the evolution and enduring relevance of his contributions, highlighting key papers that have shaped the trajectory of DEA research. This paper also discusses the evolution of DEA models and approaches, including the integration of stochastic elements and second-stage analyses. In recognising Banker’s lifetime dedication to DEA, we celebrate his lasting legacy and his transformative influence on both the academic community and practical implementations of DEA worldwide.

Keywords Rajiv Banker • Data envelopment analysis • Returns-to-scale • Most productive scale size • Statistical inference in DEA • Contextual analysis

1 Introduction

Data Envelopment Analysis (DEA) serves as a foundational methodology in the field of efficiency and performance evaluation across diverse fields, including operational research, management science, economics and finance, healthcare, and education. Initially conceptualised by Charnes et al. (1978), DEA offers a powerful framework for assessing the relative efficiency of decision-making units (DMUs) that convert multiple inputs into multiple outputs. At its core, DEA operates on the principle of comparing the efficiency of DMUs based on their input-output relationships, without requiring explicit assumptions about the underlying production function. This non-parametric approach differentiates DEA from traditional econometric methods, making it particularly well-suited for situations where the functional form of the production process is unknown or difficult to specify. The versatility of DEA lies in its ability to accommodate multiple inputs and outputs simultaneously, enabling comprehensive efficiency evaluations across various dimensions. By identifying efficient DMUs that serve as benchmarks, DEA facilitates performance improvement by highlighting areas where inefficient units can potentially enhance their productivity.
Over the years, DEA has evolved from its initial formulations to encompass a wide array of methodological advancements and applications. This paper explores the enduring legacy of Banker in the field of DEA, highlighting his lifetime contributions and their impact on shaping the methodology and its applications. By delving into Banker’s pioneering research and unwavering dedication to advancing DEA, we seek to provide a comprehensive overview of his significant influence on this influential field of study.
Banker’s seminal contributions to DEA methodology are exemplified in his numerous publications, which have garnered widespread recognition and impact. His 1984 paper, “Some models for estimating technical and scale inefficiencies in data envelopment analysis”, cited over 27,000 times according to Google Scholar (at the time of writing this paper), introduced robust models for estimating technical and scale inefficiencies, refining DEA’s theoretical underpinnings. His 1989 paper, “Sensitivity, precision, and linear aggregation of signals for performance evaluation”, cited over 1300 times according to Google Scholar (at the time of writing this paper), further enhanced DEA’s rigour and applicability. He continued to build on this foundation with his 1993 paper, “Maximum likelihood, consistency, and data envelopment analysis: a statistical foundation”, which introduced statistical foundations for the technique. Additionally, his work on advanced DEA techniques for evaluating non-financial performance measures, as reflected in his paper “An empirical investigation of an incentive plan that includes nonfinancial performance measures”, expanded the scope of DEA beyond traditional financial metrics. These advancements have been instrumental in enhancing DEA’s practical applications across various industries, including banking, healthcare, education, and agriculture. These highly influential papers demonstrate the significant impact of Banker’s work on the field of DEA.
Beyond his methodological advancements, Banker played a pivotal role in disseminating DEA knowledge and fostering a vibrant DEA community. His co-authored textbook, “Management Accounting”, remains a cornerstone of DEA education, introducing the technique to generations of aspiring analysts and researchers.
This paper provides a comprehensive tribute to Banker’s lifelong contributions to DEA, exploring his pioneering research and unwavering commitment to disseminating the subject. It begins with a review of the paradigm-shifting process in the context of returns-to-scale (RTS) and most productive scale size (MPSS) in DEA (Sect. 2).
Section 3 focuses on Banker’s contributions to the statistical foundation of DEA, building the hypothesis test framework and Monte Carlo simulation framework in DEA. Additionally, recent theoretical developments and potential paradigm shifts are discussed. In Sect. 4, Banker’s contribution to DEA-based contextual analysis is presented. The authorship network based on Banker’s academic career is generated in Sect. 5. Then, a comprehensive bibliometric analysis of Banker’s scholarly contributions at both macro and micro levels is provided in Sect. 6. Finally, Sect. 7 concludes the paper.

2 Returns-to-scale and most productive scale size

Following the initial CCR-DEA model developed by Charnes et al. (1978), numerous studies have discussed the economic concept of RTS through different approaches. The bulk of the discussion is confined to qualitative characterisations of the RTS. For instance, whether the RTS is identified as increasing, decreasing, or constant. Let denote the ratio, the proportional increase in input and proportional increase in output are represented by , respectively. Thus, . In the conventional production theory, the types of RTS are defined based on the value of the ratio between the proportional increase in output and the proportional increase in input in a single output scenario. Specifically, when the proportional increase in inputs is larger than the proportional increase in outputs ( ), RTS is decreasing. Conversely, if the proportional increase in outputs is larger than the proportional increase in inputs ( ), RTS is increasing. The RTS is considers to be constant if . The concept of RTS has been extended to multiple-output scenarios by the work of Banker (1984), Banker et al. (1984), and Banker and Thrall (1992). By investigating the RTS condition and the MPSS point of a DMU, the DEA model supports the decision-making process in policy formulation, business strategy development, and resource allocation. In general, the analysis outcomes determine macro-level strategies to maximise the productivity of a DMU. For instance, Sueyoshi and Goto evaluated the RTS conditions of the operational performance for coal-fired power plants and fossil fuel power plants in the US market. For DMUs exhibiting increasing RTS, an increase in operational size is recommended to reach the MPSS point and maximise productivity. Conversely, DMUs under decreasing RTS can further reduce their operational size to enhance productivity. In the development of RTS identification theory, studies have proposed approaches to identify RTS conditions in more complex production processes and multi-period scenarios (See Sect. 2.3.1. for details). The concept of RTS has also been extended to Damage to Scale (DTS), offering additional support for sustainability-related decision-making processes (See Sect. 2.3.2. for details).
Since the identification of RTS is highly related to the determination of MPSS, we first review the original papers on RTS and MPSS by Banker in Sects. 2.1 and 2.2. Recent progress in the identification of RTS and MPSS is discussed in Sect. 2.3. The development of the theory at the early stage by Banker is demonstrated in Fig. 1.

2.1 RTS and MPSS identification by Banker

Banker et al. (1984) indicated that CCR efficiency could be decomposed into technical efficiency and scale efficiency. For the first time in the literature, Banker (1984) extended the concept of MPSS, suggested by Frisch (1964), into DEA theory. To distinguish the concept
Fig. 1 Development phases of RTS and MPSS identification in DEA by Banker
of MPSS from the concept of minimum cost mix, Banker (1984) defined MPSS under DEA theory as a unit point on the efficiency frontier that maximises the average productivity for its given input-output mix and after which decreasing RTS set in. The model to investigate the MPSS point is also proposed in this paper. Additionally, Banker (1984) suggested a novel approach to identify the unique type of RTS based on the CCR model. By extending the work of Banker (1984), the model proposed by Banker et al. (1984) can be applied under the VRS assumption. Both models proposed by Banker (1984) and Banker et al. (1984) are based on the assumption that the Production Possibility Set (PPS) estimated by the DEA model only exists as a unique supporting hyperplane. However, the literature suggests that in empirical applications, the proposed models by Banker (1984) and Banker et al. (1984) cause an issue where multiple hyperplanes might occur (e.g., in Fig. 2, both points B and C are MPSS points), as seen in Charnes et al. (1986) and Seiford and Thrall (1990). We consider this period as the pre-paradigm stage (Kuhn, 1970), as the proposed approach has apparent pitfalls.
Banker and Thrall (1992) developed a rigorous framework to address the pitfalls identified in Banker (1984) and Banker et al. (1984). The proposed model allows for the possibility of multiple hyperplanes. Banker and Thrall (1992) argued that the issues can be solved by identifying the bounds of the slopes of the multiple supporting hyperplanes for points located at the boundary of the convex PPS. They further proposed a corresponding model to identify the bounds of the slope and an approach to investigate RTS and MPSS in the case of a unique hyperplane. Additionally, Banker and Thrall (1992) also suggested two general standards for RTS and MPSS investigation that have been widely applied in subsequent research:
(1) The concept of RTS is well-defined only for the points located on the boundary of the PPS. For interior points, RTS cannot be investigated since productivity changes due to RTS are confounded with productivity changes due to inefficiency elimination.
(2) The concept of MPSS is directly related to RTS. Specifically, the ray from the origin point (known as in Fig. 2) to the boundary point intersects the PPS at two points. If the boundary point is the first of these two points, it indicates increasing RTS. Conversely, if the boundary point is farther from the origin, it indicates decreasing RTS.
Fig. 2 PPS with two MPSS points (Banker et al., 2011)

2.2 Development of theory in RTS and MPSS identification by Banker

Numerous extended studies have been developed based on the two standards suggested by Banker and Thrall (1992). These studies primarily focus on enhancing the computation performance of the proposed model. For example, Färe et al. (1985) proposed a novel approach to identify RTS based on a group of ratios of radial measures generated from model pairs that differ in the satisfaction of convexity and sub-convexity conditions. Chang and Guh (1991) argued that Banker’s (1984) and Banker et al.’s (1984) approach is invalid due to the implicit assumption of linear production functions. Specifically, they contended that Corollary 1 (which states that if the aggregate intensity weight is larger than 1 , the DMU under evaluation indicates increasing RTS) suggested by Banker (1984) is invalid in the CCR model. A counterexample is presented in the Appendix of Chang and Guh (1991). In response, Banker et al. (1996b) pointed out that the model proposed by Fare et al. (1985) is equivalent to Banker and Thrall’s (1992) model. By proving the equivalent relationship between the two approaches, Banker et al. (1996b) eliminated some of the assumptions underlying the theorems in Fare et al. (1985). Specifically, they argued that the assumption regarding disposability could be eliminated by assigning zero coefficients to the slack variables in the objectives of the BCC and CCR model. Banker et al. (1996a) enhanced the computational convenience of Banker and Thrall’s (1992) approach by providing simpler forms for implementing the Banker-Thrall theorems. The suggested approach negates the need to investigate all alternative optima when using the BCC model.
The proposed simplified model is further extended to the CCR model by Banker et al. (1996b).
The framework proposed by Banker (1984), Banker et al. (1984), and Banker and Thrall (1992) defines RTS and MPSS within the context of DEA theory. Banker et al. (1996a, b) established the connection between the two paths to solve the RTS identification problem and simplified the computation process of the proposed model. These papers lay the foundation for RTS and MPSS identification in DEA theory. Additionally, Banker et al. (2004a, b) discussed the identification of RTS and MPSS through various models, including the input-oriented CCR and BCC models, as well as the output-oriented multiplicative model.

2.3 Revolution and progress

Several studies have been published based on the works of Banker et al. (1984, 1996a, b, 2004a, b). Recently, the literature on RTS and MPSS can be divided into two groups. The first group aims to address the research gap in Banker’s work, where the RTS and MPSS identification approach is limited to conventional black-box CRS and VRS technology. The literature extends to non-radial models and further proposes more complicated production structures such as network and dynamic structures. The second group extends the definition of RTS and MPSS to related economic concepts regarding scale characterisations. Building on Banker’s work, this group of literature defines economic concepts such as various marginal rates and scale elasticity and identifies them within the context of DEA theory. We review the literature regarding these two groups in this section.

2.3.1 Revolution and progress in modelling

We classify the publications regarding modelling development into three groups as shown in Fig. 3.
One of the extended models to identify RTS in a non-oriented case is Wu and An’s (2013) Slack-Based Measure (SBM)-based approach. Førsund et al. (2007, 2009) suggested dealing with RTS measurement in non-radial DEA models through Strong Complementary Slackness Conditions from optimisation theory. Krivonozhko et al. (2014) proposed and substantiated an approach to decrease computational complexity in Førsund et al. (2007, 2009). Fukuyama (2000, 2003) extended Banker’s literature into the Direction Distance Function (DDF) model to identify RTS and MPSS. Kerstens and Eeckaut (1999) identified RTS through the free disposal hull (FDH) model, which is linearised by Podinovski (2004) to gain a computational advantage. Soleimani-Damaneh et al. (2006) and Soleimani-Damaneh and Reshadi (2007) also suggested a polynomial-time algorithm to improve computational
Fig. 3 Division of model development based on model structure and data types
advantage. Literature to identify MPSS and RTS in weight-restricted DEA models includes the studies by Lotfi et al. (2007), Korhonen et al. (2011), and Tone (2001).
Golany and Yu (1997) suggested a new approach to estimate the RTS in BCC model precisely by comparing the outcome of the input-oriented model and the output-oriented model. Specifically, they tested the existence of solutions in the four projection regions defined by the neighbourhood of the DMU under evaluation. The four projection regions represent the case of Decreasing Returns to Scale (DRS), Increasing Returns to Scale (IRS) CRS and technical inefficiency, respectively. This approach solves the issue in Banker and Thrall (1992) that the model is applicable only for technically efficient DMUs and fails to provide unambiguous detection of RTS. This approach was extended by Taleb et al. (2022) as a non-oriented approach to identify RTS through an integrated bi-objective data envelopment model while considering undesirable outputs. Some studies also suggest approaches to deal with negative data (Emrouznejad et al. 2010; Sahoo et al., 2016), ratio data (Olesen et al., 2022; Emrouznejad and Amin, 2009), and interval data (Hatami-Marbini et al., 2014).
The aforementioned models are based on conventional black-box structures. Theoretical extensions have been introduced to account for more complex production processes. Lozano and Villa (2010) and Singh and Ranjan (2018) proposed a model to identify the RTS condition and MPSS point in parallel production systems. Since there are no intermediates in the parallel system, the proposed approach is similar to models based on the black-box system. To account for the internal production process, Lozano (2011) proposed the theoretical foundation for identifying technical, cost, and allocative efficiency, as well as local RTS, in a two-stage system. Zhang and Yang (2015) suggested an approach for identifying the RTS of a two-stage production process through a modified two-stage envelopment model. Assani et al. (2018) identified the MPSS point for a two-stage DEA model. However, there is still a significant research gap in RTS identification for more complex network production structures, such as parallel, series, and general systems. Most network RTS identification approaches are based on the two-stage model, which describes the production process where all inputs are transformed into intermediates and subsequently used to produce outputs at the next stage. However, as noted by Khezrimotlagh and Zhu (2023), the input-output PPS in a two-stage system is generated by integrating the input-intermediate PPS and intermediate-output PPS. Consequently, the input-output PPS in a two-stage system may include exterior projection points, and the aforementioned models require further investigation to address the issue.
The theory of the RTS and MPSS identification under the cross-sectional analysis has also been developed based on the work of Banker. Sueyoshi and Sekitani (2005) suggested a model to address the RTS identification problem in a dynamic production system. For a given time period , the carryover from period is treated as an input while the carryover to period is treated as an output. Based on the above standard, the MPSS and RTS for period can be identified using a similar approach suggested by Sueyoshi (1999). Sueyoshi’s (1999) proposed approach builds on the cost-efficiency scheme model of Banker et al. (1984). Consequently, a multi-stage computation process is needed to identify all the supported hyperplanes. In response, Soleimani-Damaneh (2013) suggested a novel approach to identify RTS in dynamic systems using the envelopment form of the CRS model. The discussion regarding the identification of the MPSS point in Malmquist index analysis can be found in Podinovski et al. (2017). The model suggested by Assani et al. (2018) is capable of identifying the MPSS of multi-period parallel production processes. Kontolaimou and Tsekouras (2010) addressed the measurement issues caused by the identification of MPSS in meta-frontier models. They theoretically defined the MPSS point in a meta-frontier model as the point that is technically efficient across all technologies.
Podinovski (2022) extended the concepts of CRS and VRS to Multiple Component CRS (MCRS) and Multiple Component VRS (MVRS) and suggested a corresponding approach to identify the Scale Elasticity and RTS. Specifically, Podinovski (2022) defined multiple component technologies as production technologies where each process uses its specific inputs and an unknown portion of shared inputs to produce its specific outputs and an unknown portion of shared outputs. The main difference between the two Multiple Component technologies lies in their assumptions: the MVRS model treats each component process as a separate convex technology, whereas the MCRS model additionally assumes that each process is scalable. Building on the work of Banker and Thrall (1992) and Podinovski (2017), Podinovski (2022) stated that RTS situations can be characterised based on the corresponding one-sided scale elasticities. A DMU under evaluation is characterised by increasing RTS if its corresponding right-hand scale elasticities are smaller than 1 . Similarly, the DMU under evaluation exhibits decreasing RTS if its left-hand scale elasticities are smaller than 1 . A DMU exhibits CRS if and only if it is located at the MPPS point of the current technology. Papaioannou and Podinovski (2023) further extended the work of Podinovski (2022) to scenarios with restricted allocations of shared inputs and outputs.

2.3.2 Revolution and progress in scale characterisations

Rather than focusing solely on performance evaluation, the conventional DEA methodology can be extended to address several other tasks important for decision and policymaking. Most of the literature in this group extends the concepts of RTS and MPSS within the DEA model. In this section, we review recent studies that propose and define new scale characterisations. Figure 4 illustrates the division of the new concepts.
Early DEA theory primarily targeted identifying the specific types of RTS for a given DMU. However, in the theory of economics of efficient production, the term RTS is quantified as the specific value of RTS, known as Scale Elasticity (Fukuyama, 2000). Banker et al. (1984) and Banker and Thrall (1992) suggested a simple non-parametric approach for calculating the scale elasticity in terms of the dual optimal solutions to the CCR and BCC models. Førsund et al. stated that the starting point of the DEA approach is the specification of a piecewise-linear production technology, rather than a general production technology. Specifically, the assumption of smoothness and continuous differentiability of the production frontier does not hold in DEA theory because the production frontier generated by the DEA model is piecewise linear. Taking this difference into account, they categorised approaches for calculating scale elasticity into two groups: (1) the indirect approach and (2) the direct approach.
The indirect approach estimates scale and substitution properties through the nonparametric frontier functions in DEA, producing a simple formula for scale elasticity in terms of the dual optimal solutions to the VRS model (Podinovski et al., 2009). For instance,
Fig. 4 Division of scale characterisations
Førsund and Hjalmarsson (2004) calculated the value of scale elasticity for radial projections of inefficient points to the frontier based on efficiency scores and shadow values on the convexity constraint. A similar approach was applied by Førsund et al. (2007) to estimate the scale elasticity of inefficient units and their corresponding projections on the frontier. Specifically, they calculated the maximum and minimum value of the shadow price on the convexity constraint and defined the bound of scale elasticity values accordingly. Podinovski et al. (2009) simplified the formula suggested by Førsund and Hjalmarsson (2004) and stated that one-sided elasticities could be viewed as directional derivatives of the optimal value function in a modified envelopment DEA model. This approach extends the scale elasticity investigation to the most general case, dealing with multiple optimal solutions, degeneracy, and non-full dimensional facets.
The direct approach, on the other hand, investigates scale and substitution properties through parametric neoclassical production functions. For instance, Krivonozhko et al. (2004) argued that Banker’s approach is limited to investigating partial derivatives (marginal rate of substitution, marginal rate of transformation, marginal product) of those DMUs that are interior to the production frontier. They suggested a parametric approach to calculate partial derivatives at any point on the frontier. Førsund et al. directly evaluated numerical scale elasticity on the DEA surface along intersections with planes, applying the parametric optimisation approach to construct the boundary of the VRS production technology. The scale elasticity value is then calculated based on the corresponding production technology boundary. Following the work of Podinovski et al. (2009), Podinovski and Førsund (2010) first calculated a class of mixed partial elasticity measures in both VRS and CRS technology. They proposed two approaches for elasticity measurement: one addressing the instability problem discussed in Cooper et al. (2000) through specially constructed dual linear programmes, and the other based on the standard multiplier model, which increases computational burden but offers interpretative advantages. Zelenyuk (2013) calculated the scale elasticity for multiple inputs and outputs technology based on the DDF model, demonstrating equivalence to input-oriented and output-oriented scale elasticity measures. The DDF-based elasticity measure and the profit function-based elasticity measure also indicated a dual relationship. Kao and Hwang (2011) and Sahoo et al. (2014) decomposed scaly elasticity in a two-stage multiplier DEA model.
Studies have also investigated the productive potential of efficiently operating units in response to particular changes in their input and output profiles by introducing marginal characterizations. Banker and Maindiratta (1986) first calculated the rates of substitution transformation. Bessent et al. (1988) argued that Banker and Maindiratta’s (1986) approach failed to address the issue caused by facet interfaces. For the DMU that has a mix of resources or outputs which are different from any frontier point, they suggested a constrained facet analysis approach to consider the lower bound of efficiency for organisational units. The aforementioned approaches aim to estimate the marginal rates of substitution in a scenario with infinitesimal, or small finite, changes in one or more variables. Asmild et al. (2006) argued that the change might be larger in realistic applications. Therefore, they proposed models to estimate the impact of larger non-marginal changes, scalar changes, and additive changes in DEA by extending Rosen et al.’s (1998) approach. Charnes et al. (1996) developed an empirical robust efficient production function and measured the marginal trade-offs of the efficient production function. Cooper et al. (2000) analysed marginal trade-offs through intermediate variables. More recent studies, such as Podinovski and Førsund (2010), estimated the marginal rates of response exhibited by a single input (output) with respect to output (input) under standard VRS and CRS technologies. Podinovski et al. (2016) subsequently
suggested a more general framework for calculating the marginal rates of substitution in non-parametric production frontiers.
The concept of shadow price has been discussed in numerous studies. From a computation and efficiency analysis perspective, shadow price stems from using the Lagrangian technique to solve optimisation problems with constraints (Førsund, 2018). Recent literature has targeted extending the concept of shadow price of the undesirable output from an economic perspective, aiming to calculate the marginal abatement cost of a particular undesirable output. For instance, Leleu (2013) referred to the work of Färe et al. and calculated the shadow pricing of undesirable outputs while considering the implications of weak disposability. The proposed hybrid model considers the trade-off between modelling the jointness of good and bad output and the expectation of getting the shadow price of bad output. A literature review on calculating the price of undesirable output can be found in Zhou et al. (2014). Kao and Hwang (2019) investigated the shadow price of undesirable output in both the multiplier scheme model and the envelopment scheme model, achieving the highest possible measured efficiency score. Podinovski (2019) estimated the marginal characteristics of non-parametric production frontiers in the presence of undesirable outputs. Wang et al. (2016) calculated the potential gains from carbon emissions abatement cost savings through a special dynamic DEA model. Lee and Wang (2019) measured the marginal abatement cost estimation of air pollutants under Nash equilibrium. Wu et al. (2023) investigated different abatement options to calculate the marginal abatement cost of emissions and introduced the concept of expository derivation to calculate the marginal gains caused by energy switching.
Beyond individual DMUs, the literature has also contributed to measuring scale characterisations at the group level. Banker suggested a hypothesis to investigate the type of RTS for a given technology. As discussed in Banker (1984) and Banker et al. (1984), deviations from the MPSS point determine the scale inefficiency within the extent of operations. Consequently, Banker suggested using no scale inefficiency as the null hypothesis, indicating that the production technology satisfies CRS. The starting point in Banker is to calculate inefficiency scores and the corresponding scale efficiency. The null hypothesis assumes no scale inefficiency, meaning the sample data could be rationalised by a production set exhibiting constant RTS. Based on the distribution of true inefficiency (exponentially distributed, half-normally distributed, unknown distribution), the test statistics were calculated using different formulas. Banker (1996) also developed a framework to test the null hypothesis of non-DRS against the alternative of DRS and the null hypothesis of nonIRS against the alternative of IRS. This framework is similar to the CRS test. The first step is to estimate inefficiency under IRS (DRS) by setting the sum of intensity weight to be less (or greater) than 1 . The estimated inefficiency is then compared to true efficiency using the corresponding test statistic. Simar and Wilson extended Banker’s work by introducing bootstrap estimation procedures, providing appropriate critical values for the test statistics. However, Alirezaee et al. (2018) argued that the statistical assumptions underlying Banker are difficult to satisfy, and decision-makers may overlook certain factors or pre-judge the behaviour of the technology in realistic applications. In response, Alirezaee et al. (2018) introduced the concept of technological RTS and suggested a non-statistical approach to identify technological RTS (TRTS) based on the Angles method. Specifically, a measure was introduced to calculate the gap between the constant TRTS hyperplane and the variable TRTS in both the increasing and the decreasing sections of the frontier.

3 Advancements in the statistical foundations of DEA: contributions, frameworks, and recent developments

Advancements in the statistical foundations of DEA, pioneered by Banker, can be categorised into two groups. The first focuses on hypothesis testing based on the DEA model, while the second addresses the foundation and structure of simulation test design in DEA. In summary, the original work by Banker on the statistical foundations of DEA establishes the research paradigm for hypothesis tests and simulations tests under the DEA framework. Banker’s hypothesis tests provide the foundation for constructing a wide range of formal statistical tests for DEA models, including technical change, the impact of contextual variables, types of RTS, and input separability. On the other hand, simulation tests under the DEA framework suggest a general approach for selecting suitable analysis models. These simulation tests are also widely applied to assess the performance of nonparametric analysis models. More recent studies have extended Banker’s approach to testing the statistical foundations of other nonparametric analysis models. The statistical inference developed by Banker also provides the theoretical foundation for central limit theory (under the DEA model), stochastic DEA analysis, and nonparametric quantile frontier estimation. These models offer more accurate estimations of shadow prices and greater robustness in selecting directional vectors. Furthermore, studies propose the partial frontier approach based on the DEA model’s statistical inference. Another group of studies suggests the quantile regression approach, which is considered an extension of Banker et al. (1991). These models provide a more suitable approach for addressing random noise, heteroscedasticity and outliers.
We review the original literature by Banker in Sect. 3.1. In Sect. 3.2, we present the theoretical extension by Banker based on the discussion in Sect. 3.1. Recent progress in simulation tests, hypothesis tests, and statistical inference of DEA and its extensions are reviewed in Sect. 3.3. We also present some more recent extended models, based on statistical theory, to address data noise in Sect. 3.4.

3.1 The origin of statistical foundations of DEA by Banker

The revolution process of the statistical foundation of DEA can be divided into three phases: Pre-Paradigm, Origin and Development. This section reviews the literature that represents these three phases by Banker. The process is visualised in Fig. 5.
Fig. 5 Revolution phases of Statistics in DEA by Banker
The DEA model is a non-parametric approach grounded in the general axioms of production theory, such as monotonicity, convexity, and homogeneity (Banker et al., 1984). In the 1980s, research on statistics in DEA was in what Kuhn (1970) described as the pre-paradigm stage. Early DEA literature largely ignored discussions of the statistical inference of the generated efficient frontier. Even the work of Banker and Maindiratta (1988) merely established a link between the DEA model and Varian’s (1984) algebraic test for consistency with the weak axiom of cost minimisation, measuring whether the technical, allocative, or scale efficiencies of DMUs were low. Thus, during this period, DEA was considered a non-statistical approach (Banker, 1996; Gong and Sickles, 1992; Schmidt, 1985). However, Schmidt (1985) criticised this non-statistical measurement approach, expressing skepticism about its lack of a corresponding measurement foundation. He argued that, because the non-statistical, nonparametric method does not account for error, the literature should be approached cautiously, particularly when the data under evaluation are assumed to be free from statistical noise.
In response to Schmidt’s critique, Banker (1993) discussed and established the statistical basis for DEA. Banker (1993) identified four postulates regarding the production set and the probability density function of the “true inefficiency”. These postulates include monotonicity, convexity, envelopment, and the likelihood of efficient performance. These four postulates form the foundation for simulation-based testing studies (Banker, 1993, 1996; Banker and Chang, 1995; Banker et al., 1994).
By solving the BCC models to estimate the inefficiency scores, the estimated efficiency scores are compared to the “true inefficiency” generated based on these postulates. The findings suggest that the BCC production frontier aligns with a monotone decreasing probability density function, generated based on the deviation of the outputs under evaluation from the efficient output levels. Additionally, the estimated inefficiency maximises the likelihood function if the inefficiency probability density function exhibits a monotone decreasing trend.
Banker (1993) also proposed three hypotheses to test differences in inefficiency between two different types of DMUs, assessing whether one group is more efficient than another. These three tests can accommodate different kinds of distributions (exponentially distributed, half-normally distributed, unknown distribution) of inefficiency. The proposed approach lays the theoretical foundation for future literature, establishing a formal statistical foundation for DEA inefficiency estimators and suggesting a general framework for statistical hypothesis testing regarding differences in inefficiency distributions in DEA.

3.2 Development of statistical-based extension approaches in DEA by Banker

Based on the work of Banker (1993), two main statistical-based extension approaches have been developed in the field of DEA. The first approach focuses on further developing the statistical foundation of DEA and proposing adjusted hypothesis testing approaches to address potential issues caused by statistical bias. The second approach builds on the simulation approach used by Banker (1993) and aims to test the performance of specific models. This section reviews these two approaches and the related literature by Banker.

3.2.1 Foundations of simulation in DEA by Banker

Banker (1993) laid the groundwork for the simulation-based testing framework in DEA. The simulation testing procedure involves comparing the efficiency score estimated by DEA with the “true efficiency” generated by a production technology that satisfies the four postulates defined by Banker (1993). The critical factors in these simulations are the characteristics
of the production technology, the distribution of inefficiency, and the sample size (Banker, 1996).
In almost all of Banker’s simulation studies, the designs are aimed at evaluating the performance of a model or test by comparing the estimated efficiency scores to the “true efficiency” through Monte Carlo simulations. The “true inefficiency” is typically generated using a Cobb-Douglas production function in a single-output scenario, with the values of “true efficiencies” distributed in a specific pattern. For instance, Banker and Chang (1995) assessed the performance of asymptotic DEA tests for differences in inefficiency, as presented in Banker (1993). These tests’ performance was compared with conventional parametric corrected ordinary least squares (COLS) tests and the Welch and Mann-Whitney tests. The results indicated that the asymptotic DEA tests outperformed COLS-based tests and Mann-Whitney tests. The proposed simulation framework is also utilised to evaluate the accuracy of efficiency estimation in the sup-efficiency DEA model (Banker and Chang, 2006). The findings show that the super-efficiency model is outperformed by the conventional DEA model in identifying data contaminated with outliers and removing the corresponding outliers.

3.2.2 Foundations of hypothesis testing in DEA by Banker

Since the publication of Banker (1993), the literature on the statistical foundation of DEA has evolved significantly. A critical characteristic of production frontiers is the presence of increasing or decreasing RTS. As reviewed in Sect. 2.3, Banker (1996) developed a framework to test RTS at the technology level.
Another hypothesis proposed by Banker et al. (1994) aimed to test the independence of estimations and inputs. Specifically, the separability of inputs-the assumption that inputs can be optimised independently without being substitutable for others-is a common assumption in the DEA model. However, in practical applications, two inputs (such as labour and capital) might be substitutes, and the assumption regarding substitutability or separability is not tested in conventional DEA models. To test this hypothesis over an entire dataset, Banker et al. (1994) provided a connection between the conventional DEA model and Shephard’s (1970) distance function. The perfect separable inefficiency is then estimated as the reciprocal of Shephard’s (1970) distance measure. They stated that if inputs are separable, in a group of DMUs with inputs, the input inefficiency estimated by a particular input is unique. Specifically, let be the input inefficiency of generated by only input ; we have . Therefore, the null hypothesis of input separability can be treated as the situation where . Similar to Banker (1993), Banker (1994) also suggested using three different test statistics based on the distribution of true inefficiency.
Banker and Chang’s (1995) study was the first to propose a specification testing approach (referred to as “variable selection” in recent studies) to test the significance of variables at the margin in characterising the production process. The null hypothesis posits that extra variables do not affect production correspondence. Specifically, to address the dimensionality issue, the inefficiency score without extra variables should be larger than the generated inefficiency score generated by the model that includes the extra variables. Three different test statistics were also suggested based on the true efficiency distribution.

3.3 Progress in the statistical foundations of DEA and its extensions

Banker’s work has established the research paradigm for statistical testing and simulation testing in DEA. Section 3.3.1 reviews progress in simulation design within DEA, highlighting how Banker’s simulation framework has been applied to evaluate the performance of models designed to assess more complex production processes. Section 3.3.2 examines extensions in the statistical testing under the DEA framework. In addition to broadening the scope of statistical testing, studies propose methods to address data noise during both the testing and data-generating stages. Section 3.3.3 discusses recent developments in the proof of statistical inference for extended DEA models. The section also presents some notable stochastic DEA models, which are grounded in the statistical axioms established by Banker.

3.3.1 Progress in simulation design in DEA

Several simulation tests have been designed since Banker (1993), and most of them share the same motivation and aim to examine the satisfaction of a DEA model. The simulation design is similar to Banker’s (1993) work, which compares estimated DEA inefficiency to the generated “true inefficiency”. As illustrated in Fig. 6, most theoretical extensions target the production process by introducing more factors, such as multiple outputs, undesirable outputs, nondiscretionary variables, or proposing a more complex network production environment.
As a single-output simulation experiment, Bardhan et al. (1998) and Cooper and Tone (1997) compare the performance of DEA, DEA-regression combinations, and stochastic frontier regression approaches under different sample sizes. Ruggiero designed the simulation experiments to test the performance of DEA models that consider nondiscretionary variables. Giraleas et al. (2012) compared the performance of DEA, COLS, and Stochastic Frontier Analysis (SFA) when generating the frontier based on panel data. The performance of conventional DEA, SFA, and the stochastic DEA model (Resti, 2000; Simar, 2007), and the accuracy of technical efficiency and scale efficiency measured by the DEA model (Perelman and Santín, 2009) are also compared by using Banker’s (1993) simulation framework. More examples can be found in Khezrimotlagh (2022), which reviews the current literature on existing simulation experiments.
Banker’s (1993) simulation test is based on a conventional Cobb-Douglas production function in a single-output case. However, in realistic applications, DEA models are likely to deal with multiple-output scenarios. Therefore, some studies aim to build simulation experiments by using the production function with multiple inputs and multiple outputs. To generate the “true inefficiency” in multiple-output production technology, the extended Cobb-Douglas production function is applied (Wang et al., 2020). Another production technology based on the quadratic frontier was proposed by Simar (2007). The two-input and two-output translog production function was proposed and applied by Perelman and Santín (2009). Fernandez
Fig. 6 Recent progress in simulation design in DEA
et al. (2002) used two different production functions to estimate desirable and undesirable outputs separately. This approach was further developed by Chen and Delmas (2012) to generate a multiple-output production function while considering the weak disposability of the undesirable outputs.
To account for the internal production process, the “true inefficiency” is generated by a network production frontier (Michali et al., 2023). Khezrimotlagh (2022) also suggested a general approach for constructing the network production frontier to evaluate the performance of the network DEA model. The core of this approach is to generate the “true inefficiency” and original output level, ensuring that the value of “true inefficiency” and the original output level satisfy a specific distribution. The original output level serves as the output projection or output benchmark for the corresponding DMU. Therefore, the output volume of the DMU is calculated by multiplying the original output level by the corresponding “true efficiency”, which is the reciprocal of “true inefficiency”. In most cases, the tested models are radial DEA models that are input (output) oriented. A general approach for handling non-radial DEA was proposed by Khezrimotlagh (2022). However, its applicability needs further discussion since it assumes input (output) inefficiency as an aggregation rather than a group of non-radial slacks. The generated non-radial “true inefficiency” is the same as radial “true inefficiency” as it represents the proportional, radial contraction of inputs or expansion of outputs rather than input excesses and/or output shortfalls between benchmarks (Halická and Trnovská, 2021). For instance, in cases where the performance of SBM needs to be tested, the framework suggested by Khezrimotlagh (2022) fails to provide information regarding the “true inefficiency” of a specific variable but instead gives the aggregate “true inefficiency”. Moreover, assuming the same level of inefficiency for inputs and outputs might lead to contradictions in realistic applications. Furthermore, there is a research gap in testing the performance of the network non-radial DEA model (i.e., network SBM model) by using the simulation approach suggested by Banker.

3.3.2 Progress and revolution in hypothesis tests in DEA

Banker (1993) provided a formal statistical basis for DEA estimation techniques, which paved the way for the development of various hypothesis tests. As illustrated in Fig. 7, beyond expanding the research scope to test group performance of DMUs or extending the theory proposed by Banker et al. (1996a, b), several studies have also focused on incorporating independent noise terms and addressing issues like missing data or panel data (hypothesis tests for productivity change and allocative efficiency).
In recent research, new estimation and statistical tests have been developed to handle longitudinal and cross-sectional data. The purpose of these studies is to investigate the timeperiod shifting of the production frontier generated by the DEA model from a statistical perspective. Banker et al. (2004a, b) developed statistical tests for the null hypothesis of no allocative inefficiency by introducing a single aggregate cost variable to represent aggregate
Fig. 7 Recent progress in hypothesis tests in DEA
technical and allocative inefficiency. This approach allows for the analysis of efficiency trends over time and consider the differences across subgroups within a panel dataset. In scenarios involving asymmetric information, output quantity data may not be available, but the monetary value of individual outputs and input quantity information might be. Banker et al. (2007) extended the work of Banker et al. (2004a, b) to create a framework for testing outputbased allocative inefficiency while accounting for missing output quantity data. Banker et al. (2005) proposed three null hypotheses to test efficiency changes and applied these tests to the public accounting industry as an empirical case. Compared with the conventional Malmquist index approach, their proposed approach provides statistical properties of the components of productivity change and derives statistical tests based on these estimators.
The hypothesis tests proposed by Banker et al. (1996a, b) have been developed and applied in various areas. For instance, Alperovych et al. (2015) applied the hypothesis test suggested by Banker et al. (1996a, b) to test the RTS assumption related to firm performance. Banker et al. (2010a, b) used Banker et al.’s (1993) approach to examine changes in productivity within the Korean banking system following a crisis. Simar and Wilson (2002) extended Banker et al. (1996a, b)’s work to test hypotheses regarding RTS in non-parametric models of technical efficiency. Pastor et al. (2002) proposed a test for Nested Radial DEA Models by extending the variable selection hypothesis tests suggested by Banker et al. (1996a, b). This hypothesis test aims to analyse the marginal role of a given variable concerning efficiency measured by a DEA model. Monte Carlo simulation proved the performance of the proposed test to be superior to that of Banker et al. (1996a, b). Sueyoshi and Aoki (2001) proposed a non-parametric statistical approach to test productivity changes based on the output of DEA window analysis and suggested the corresponding Kruskal-Wallis rank test. Pastor et al. (1999) extended Banker’s (1993) hypothesis framework and proposed a statistical test for detecting influential observations. Kuosmanen et al. (2007) suggested a hypothesis test to detect errors-in-variables, as mentioned in Varian (1985).
In the literature mentioned above, the data-generating process typically treats stochastic inefficiency as the sole cause of deviations of actual output from the production frontier. However, several studies have also suggested considering the influence of an independent noise term in the data-generating process. Banker et al. (2002) is the first to propose the estimation of an extremal function incorporating both a one-sided efficiency term and a twosided noise term. This function is applied to evaluate the adequacy of parametric functional forms through four statistical tests: Kolmogorov-Smirnov test, the Rank-regression test, the Wilcoxon rank-sum test, and Theil’s distribution-free test. Banker and Natarajan (2008) treated the error term as an integration of three distinct components: a linear function of multiple, possibly correlated, contextual variables. Their paper also established a general framework using DEA at the first stage and OLS or ML at the second stage to estimate individual inefficiency conditional on the value of the composed error term. Banker et al. (2010a, b) introduced a two-sided random noise term into the data-generating process and proposed five hypotheses to compare the performance of different groups of DMUs. The proposed approach was shown to perform better than Banker (1993) when noise levels were significant, as demonstrated through Monte Carlo simulation. Further, Simar and Wilson (2007) argued that previous literature failed to describe a coherent data-generating process. In response, they proposed bootstrap procedures that permit valid inference and improve statistical efficiency in second-stage regression.

3.3.3 Developments in statistical inference of DEA and its application

Ever since Banker et al. (1993), the literature has explored statistical inference in DEA and its extensions to more general cases. Grosskopf (1996) surveyed the statistical inference of the DEA model and the FDH model. This paper also reviewed nonparametric regularity tests, sensitivity analysis, two-stage analysis with regression, and nonparametric statistical tests. Similar surveys can be found in Simar and Wilson (2000b, 2007). In more recent work, Daouia and Gijbels (2011) defined and proved the statistical inference of nonparametric (DEA, FDH) partial frontiers.
The statistical inference testing approach proposed by Banker et al. (1993) has also been extended to multi-period analysis. Odeck (2009) applied a bootstrapping method to ensure the statistical inference of the Malmquist index measure. Tortosa-Ausina et al. (2008) examined the sensitivity of the Malmquist index measure measured. Kneip et al. (2021) tested the statistical inference of productivity change (both individual and mean productivity changes for a given technology) measured by Malmquist indices. Pham et al. (2024) extended Kneip et al.’s (2021) work by proposing a novel Malmquist Productivity Index. This new index reflects the relative importance of individuals, and its statistical inference is also proven.
Extending the work of Banker et al. (1993), Korostelev et al. (1995a, b) discussed the consistency and convergence speed of DEA and FDH estimators. This approach is further extended to Banker and Maindiratta’s (1988) multiplicative DEA model by Kneip et al. (1998). Tsionas and Philippas (2023) applied Bayesian techniques to test the global sensitivity of the DEA model. Alongside the development of the DEA model, the framework by Banker et al. (1993) has also been applied to test the statistical inference of extended DEA models, including the DDF model (Simar et al., 2012), the Russell measure (Badunenko and Mozharovskyi, 2020), the general technique and allocative measures (Simar and Wilson, 2020; Simar et al., 2024a), and the Hicks-Moorsteen productivity indices (Simar et al., 2024b).
Beyond the direct extensions of Banker et al. (1993), the statistical underpinnings of DEA by Banker et al. (1993) provide the theoretical foundations for the central limit theory (Kneip et al., 2015, 2016; Nguyen et al., 2022; Simar and Zelenyuk, 2020). Moreover, the statistical axioms established by Banker (1993) have been used to suggest stochastic models for addressing data noise. By considering the data noise while building the production frontier, these stochastic models are able to minimise the risk of underestimation and improve the precision of efficiency estimation (Tsionas, 2021). According to Banker et al. (1996a, b), conventional DEA models ignore the data noise and the axioms on the distribution of deviations from a best practice frontier. On the other hand, the SFA approach has been criticised for its restrictive functional form assumptions, particularly in the context of joint production (Kuosmanen and Johnson, 2017). Several models have been developed to improve robustness against data errors and outliers and to incorporate probabilistic settings in SFA theory, drawing on the statistical axioms established by Banker (1993) and Simar and Wilson (2000a, b). Olesen and Petersen (2016) categorised stochastic methodologies into three directions:
  1. Treating estimated inefficiencies as random deviations.
  2. Developing models to account for either measurement errors or specification errors.
  3. Developing models to generate random Production Possibility Sets based on random variations in datasets.
The existing Stochastic DEA models can be divided into three groups based on these directions:
  • Group 1: Extends the first direction and includes the bootstrapping DEA model (Aggelopoulos and Georgopoulos, 2017; Bobde and Tanaka, 2018; Boubaker et al., 2023; Dia et al., 2022; Du et al., 2018; Kang et al., 2024; Michali et al., 2023; Staat, 2002; Simar and Wilson, 1998, 1999, 2000a, b; Moradi-Motlagh and Emrouznejad, 2022).
  • Group 2: Extends both the first and second directions and includes the semi-parametric DEA model (Assaf and Gillen, 2012; Johnson and McGinnis, 2008; Jradi and Ruggiero, 2019; Kuosmanen and Kortelainen, 2012; Simar and Wilson, 2007).
  • Group 3: Extends the second and third directions and includes Chance Constrained DEA models (Amirteimoori et al., 2023; Cooper et al., 1996, 1998, 2002; Lin and Lu, 2023; Mitropoulos et al., 2015; Shiraz et al., 2020; Talluri et al., 2006).
A more detailed review of stochastic DEA can be found in Olesen and Petersen (2016).

3.4 Progress and advances in nonparametric quantile frontier estimation

The aforementioned models in Sect. 3.3.3 address the inherent uncertainty in data and enhance robustness. It is important to note that these models are based on statistical axioms rather than directly addressing stochastic noise in the observed data. However, due to the application of the minimal extrapolation principle that minimises the deviation to the production function in the dataset, several studies (Esteve et al., 2020; Tsionas, 2022) argue that DEA suffers from an overfitting problem. To address these deficiencies, newly proposed approaches focus on estimating quantile frontiers instead of full frontiers that envelop all observations. These models, known as nonparametric quantile frontier estimation, provide more accurate estimations of shadow prices and are more robust in selecting directional vectors. Additionally, these models can address various types of data noise, such as random noise, heteroscedasticity, and outliers (Dai et al., 2022; Kuosmanen and Zhou, 2021; Liao et al., 2024). According to Dai et al. (2022), existing nonparametric quantile frontier estimation methods can be categorised into two groups: (1) Partial frontier and (2) Quantile regression. The partial frontier approach generates a frontier estimator that fits a subset of observations (Dai et al., 2022). In contrast, the quantile regression approach constructs the frontier by applying an asymmetric norm to the full sample of observations (Dai et al., 2022; Koenker and Bassett, 1978). In this section, we review several representative studies on nonparametric quantile frontier estimation.

3.4.1 Partial frontier approach

The partial frontier approach generates a frontier that fits a subset of the observations. The algorithm for generating such a frontier follows the conventional DEA analysis paradigm, which maximises the distance between the DMU under evaluation and its associated projection point. Additionally, inspired by the research paradigm established by Banker (1993), the statistical theory underlying the partial frontier approach has been rigorously proven.
Cazals et al. (2002) initially suggested the order- approach to generate the FDH frontier while addressing outliers. Building on the work of Cazals et al. (2002), Aragon et al. (2005) proposed the FDH quantile frontier approach of order-m. In subsequent studies, Daouia and Simar (2005) examined the asymptotic properties of these methods and proved the convergence of the order-m and order- approaches. Further extending this line of research, Daouia and Simar (2007) incorporated the effects of environmental variables on efficiency. Wheelock and Wilson (2008) introduced the concept of root-n to ensure model consistency and convergence, with the added benefit of mitigating the curse of dimensionality. More recently, Atwood and Shaik (2020) proposed a quantile DEA model in both envelopment
and multiplier form. The statistical properties of this model are validated using Monto Carlo simulation and nCm subsampling. A similar approach has been applied to context-dependent DEA by Seiford and Zhu (2003), which involves an algorithm that calculates different levels of efficient frontiers by iteratively excluding efficient DMUs from the reference set. However, the statistical inference for context-dependent DEA has not been directly proven. Carvalho and Marques (2014) explored the economies of scope and scale for computing economies of vertical integration through the partial frontier approach.
In general, the aforementioned approaches can be considered extensions of traditional DEA models. They address issues caused by data noise by excluding outliers, with the determination of outliers grounded in statistical theory. Inspired by Banker (1993), the statistical inference for most of the presented partial frontier approaches has been established. However, these methods are still based on the conventional black-box production structure. Furthermore, like conventional DEA approaches, the shadow prices of variables remain non-unique. Future research could focus on extending these approaches to network models or proposing common-weight measures.

3.4.2 Quantile regression approach

The main difference between the quantile regression approach and the partial frontier approach lies in their strategies for inefficiency (noise) estimation. While the partial frontier approach focuses on inefficiency estimation strategies in DEA, the quantile regression approach typically fits a frontier estimator to all observations, considering both inefficiency and noise. Further, Dai et al. (2022) pointed out that the existing quantile regression approaches can be viewed as extensions of the stochastic DEA model proposed by Banker et al. (1991). This section reviews several notable quantile regression approaches.
Inspired by Banker and Maindiratta (1992), who applied Afriat inequalities in maximumlikelihood estimation by introducing non-Gaussian error terms, Kuosmanen (2008) proposed the convex nonparametric least squares (CNLS) model. Later, Kuosmanen and Johnson (2010) established a closer connection between the CNLS and DEA models. Specifically, Kuosmanen and Johnson (2010) demonstrated that DEA could be formulated as a least squares regression where the error term is always negative. Based on this proof, they suggested the Corrected Concave Nonparametric Least Squares ( NLS) model, which was tested and found to outperform the conventional DEA model and the COLS model through Monte Carlo simulation. Further, the Stochastic Nonparametric Envelopment of Data (StoNED) approach was introduced as a more general framework that combines conventional DEA and SFA, as developed by Kuosmanen (2006) and Kuosmanen and Kortelainen (2012). Kuosmanen and Johnson (2017) extended the CNLS model (Kuosmanen, 2008) by incorporating constraints for the DDF measure.
While the aforementioned quantile regression approaches are nonlinear, in more recent research, Wang et al. (2014) introduced a convex quantile regression (CQR) model to estimate the production frontier. This model uses linear programming and can be easily solved with standard algorithms (Gurobi, GAMS, CPLEX, MOSEK). However, it should be noted that the CQR model by Wang et al. (2014) fails to generate a unique production frontier. To address this, Kuosmanen et al. (2015) proposed a quadratic formulation as a modified objective function for the CQR model. Jradi and Ruggiero (2019) integrated the quantile regression approach into the DEA model as a stochastic measure. Kuosmanen and Zhou (2021) argued that the conventional DEA approach may overestimate the marginal abatement cost. To address this, they combined Wang et al.’s (2014) CQR model with the DDF measure proposed by Kuosmanen and Johnson (2017) to develop a convex quantile regression approach. This
approach explicitly accounts for both noise and inefficiency to mitigate issues caused by high sensitivity to noise data in DEA. Consequently, the proposed approach calculates the marginal abatement cost while free from the data noise issues in conventional DEA models. In more recent research, Dai et al. (2023) summarised existing Quantile regression approaches and proposed a general model of shape-constrained nonparametric functions. Moving beyond the convex models used in CQR (linear programming) and convex expectile regression (quadric programming), Dai et al. (2023) introduced a parameter to represent partial order, generating general nonconvex CQR and CER models. In addition, España et al. (2024) used the DEA model to shape constraints and estimate production functions through additive models based on regression splines. Liao et al. (2024) proposed a nonparametric convex regression approach to address overfitting and outliers.
In general, when generating the production function, some researchers leverage the DEA model’s linear-based approach to propose shape constraints that ensure the generated production frontier satisfies the monotonicity and concavity properties of production theory. However, the generated frontier is a quantile frontier rather than one that envelops all observations. The proposed models address random noise, heteroscedasticity, and outlier issues in conventional DEA techniques. The aforementioned approaches are all extensions of Banker et al. (1991), with the DEA frontier serving as a special case of these models.

4 Contextual analysis in DEA

Most early research on evaluating contextual variables relied on the stochastic frontier framework, as seen in studies like Ruggiero and Vitaliano (1999) and Rosko (2001). These approaches estimated DEA efficiency scores in the first stage and then treated these scores as dependent variables in a regression-based analysis in the second stage. However, the twostage approach lacks a robust statistical foundation and failed to articulate a data-generating process consistent with the two-stage DEA (Banker et al., 2019). Consequently, several studies question the ability of the two-stage approach to effectively investigate the influence of contextual variables (Førsund, 1999; Grosskopf, 1996).
In response to these critiques, Banker and Natarajan (2008) first applied Monto Carlo simulation to compare the performance of the DEA-based two-stage approach (using OLS, maximum likelihood, and Tobit estimation in the second stage) with one-stage and twostage parametric approaches. The simulation results indicate that DEA-based procedures outperform the parametric methods. This study also established the statistical foundation of the two-stage DEA-based approach and proposed a framework to estimate contextual variables while considering two-sided random noise. Banker et al. (2019) further compared the performance of DEA+OLS with DEA+Tobit regression and Simar and Wilson’s (2007) approach within a noisy production environment. In a noise-free environment, simulation results demonstrate that the DEA+OLS approach performs best, particularly when evaluating large sample data (defined as 400 DMUs in the simulation tests). The same result is observed in noisy environments, where the DEA+OLS approach outperforms others. It should be noted that while the performance of the DEA+OLS approach is similar to that of DEA+Tobit regression, the theoretical foundation for DEA+Tobit regression has not been established. Moreover, the DEA+OLS approach shows better performance than Simar and Wilson’s (2007) approach when contextual variables significantly affect the output. Even in cases where contextual variables have no impact on the output, the DEA+OLS approach
performs similarly or occasionally better than Simar and Wilson’s (2007) approach. Consequently, the DEA+OLS approach is recommended for evaluating the influence of contextual variables.
In addition to the conventional DEA model, several extended models are used for performance evaluation in the first stage of analysis. Examples include the DEA model with undesirable outputs (Bandyopadhyay, 2011) and two-stage network DEA (Tan et al., 2021). Giménez et al. (2024) applied the directional Benefit-of-the-Doubt model in the first stage of analysis to evaluate countries’ pandemic management.
We also found several studies that extend the analysis of contextual variables under the DEA framework. Ramalho et al. (2010) suggested using fractional regression models as a second-stage measure. By combining fractional regression models with the dynamic generalised method of moments, e Souza and Gomes (2015) proposed a cross-sectional approach to measure the dynamic effects caused by contextual variables. Karagiannis (2015) extended the work of Banker and Natarajan (2008) to decompose the efficiency effect into simultaneous effects on technical efficiency and capacity utilisation. Johnson and Kuosmanen (2012) stated that DEA can be formulated as a constrained special case of the CNLS model and suggested a semi-nonparametric approach to estimate the coefficients of contextual variables through one-stage computation. Yu et al. (2024) proposed a centralised resource allocation model to assess the effect of contextual variables through one-stage computation. Shi et al. (2025) applied Shapley Additive Explanations to scrutinise the impacts of contextual variables. To the best of our knowledge, only Johnson and Kuosmanen (2012) built a simulation framework to compare the performance of their proposed model with the DEA+OLS approach. We suggest developing a more comprehensive simulation framework that incorporates varying levels of sample size and noise to better compare the performance of the aforementioned approaches.

5 Authorship network and organised conferences

To provide a comprehensive overview of Banker’s scholarly contributions, a bibliometric analysis was conducted based on 203 of his publications. The Scopus database served as the primary source for extracting relevant data collected in December 2024. Table 1 presents a chronological history of Banker’s affiliations with various universities and business schools across different states in the United States from 1980 to 2023. His academic journey began in 1980 at Harvard Business School in Massachusetts. Over the years, he made significant contributions to various universities, while spending much of his later academic career from 2005 to 2023 at Temple University’s Fox School of Business in Philadelphia. This extensive history showcases Banker’s dedication to academia and his contributions across multiple educational institutions in the United States.
Table 2 outlines Banker’s publications by type, totaling 203. It highlights that articles represent the majority with 133 ( ), followed by conference papers at 36 ( ). Reviews,
Table 1 Banker’s university affiliations (1980-2023)
Year Affiliation State
2005-2023 Temple University/Fox School of Business Philadelphia
2006 Arizona State University/W. P. Carey School of Business Arizona
2003-2005 University of California, Riverside California
2005 University of California/Anderson School of Management California
1992-2004 University of Texas at Dallas Texas
1989-1997 University of Minnesota/Carlson School of Management Minneapolis
1984-1996 Carnegie Mellon University Pittsburgh
1994 University of Minnesota/Carlson School of Management Minnesota
1980 Harvard University/ Harvard Business School Massachusetts
Table 2 Banker’s publications by type
Type Number of publications Percentage
Article 133 65.5
Conference Paper 36 17.7
Review 11 5.4
Edited Book of Conference Proceedings 7 3.4
Book Chapter 8 3.9
Editorial 7 3.4
Note 1 0.5
Total 203 100
conference proceedings, book chapters, editorials, and notes make up smaller portions of his works, demonstrating his diverse scholarly contributions across various platforms.
Table 3 presents Banker’s distribution of authorship positions in his publications. It shows that he was the first author in 159 papers ( ), indicating his primary contribution as the lead researcher. Additionally, he was listed as the second author in 28 papers ( ) and the third author in 14 papers ( ). For publications where he was listed as the fourth author or beyond, there were only two instances, representing 1% of his total publications. This breakdown illustrates the prominence of his role as the primary contributor or lead author in the majority of his scholarly works.
Table 4 outlines Banker’s top 10 co-authors based on the number of co-authorships they shared in scholarly works. It presents the respective co-authors’ names, the count of collaborative papers with Banker, their titles, affiliated universities or institutions, specific schools
Table 3 Banker’s authorship order in publications
First author Second author Third author Fourth author Total
Number of publications 159 28 14 2 203
Percentage 78.3 13.8 6.9 1.0 100
Table 4 Top 10 scholars who collaborated with Banker
Co-author No. of co-authorships Title University/Affiliation School/ Faculty Country
Chang, Hsihui 24 Professor of Accounting Drexel University Drexel University United States
Natarajan, Ramachandran Nat 13 Professor of Management University of Texas at Dallas The Naveen Jindal School of Management United States
Emrouznejad, A 11 Professor of Business Analytics University of Surrey Surrey Business School UK
Kauffman, Robert J 11 Professor of Information Systems Singapore Management University Singapore Management University Singapore
Cooper, William W 11 Professor of Operations Research University of Texas at Austin McCombs School of Business United States
Datar, Srikant M 10 Professor of Administration Harvard University Harvard Business School United States
Anderson, Mark C 10 Associate Professor of Accounting University of Calgary Haskayne School of Business Canada
Riedl, René 9 Professor of Digital Business & Innovation University of Applied Sciences Upper School of Management Austria
Pavlou, Paul A 9 Professor of Information Sciences University of Houston C. T. Bauer College of Business United States
Davis, Fred D.D 9 Professor of Information Technology Texas Tech University Rawls College of Business United States
Fig. 8 Banker’s co-authors network
or faculties within those institutions, and the countries where they are located. This table demonstrates the collaborative relationships that Banker established with these individuals across various universities, showcasing his partnerships within the academic community, particularly in the fields of accounting, management, business analytics, information systems, and operations research.
Figure 8 shows a network of co-authorships with Banker, where nodes represent coauthors, and distinct colours correspond to different clusters of authors. We utilised VOSviewer, a software tool developed by Van Eck and Waltman (2010), to create and visualise maps based on network data. The size of the nodes represents the number of co-authored publications. Banker is the central node in the network, and he has co-authored papers with many other researchers in several clusters. The clusters of authors, shown in different colours, represent researchers with common interests, locations, and/or expertise. For example, the navy-blue cluster contains scholars such as Cooper and Charnes, who made significant contributions to the introduction and development of DEA, while the phosphorus blue cluster includes researchers mainly in the accounting field. The fact that Banker is connected to researchers in multiple clusters suggests that he had a broad range of global collaborations and research interests. This is not surprising, given that he is a highly respected scholar who made significant contributions to multiple fields, including applied mathematics, accounting, and economics.
Table 5 Journal editorials co-authored by Banker
Title Journal Reference
Advances in Data Envelopment Analysis: Celebrating the 40th anniversary of DEA and the 100th anniversary of Professor Abraham Charnes’ birthday European Journal of Operational Research Emrouznejad et al. (2019)
Recent developments on the use of DEA in the Public Sector Socio-Economic Planning Sciences Ahn et al. (2018)
Business performance management under uncertain environments- II Journal of Centrum Cathedra: The Business and Economics Research Journal Charles and Banker (2017)
Business performance management under uncertain environments – I Journal of Centrum Cathedra: The Business and Economics Research Journal Charles and Banker (2016)
Data Envelopment Analysis in the public sector Socio-Economic Planning Sciences Emrouznejad et al., (2014a, b)
Efficiency and productivity: Theory and applications Annals of Operations Research Emrouznejad and Banker (2010)
Table 5 presents a compilation of journal editorials co-authored by Banker, including their titles, respective journals, and their references. Banker has made significant contributions to editorial pieces in esteemed journals, focusing primarily on DEA and its applications across various sectors. In collaboration with co-authors like Emrouznejad, Charles, Ahn, and others, he authored editorials celebrating milestones in DEA history, exploring recent developments, particularly in the public sector, and examining the theoretical aspects of efficiency and productivity.
Table 6 provides a list of conference proceedings co-edited by Banker across various international conferences on DEA, including the titles of the proceedings, the respective conference titles and references. These proceedings, spanning from 2012 to 2017, highlight the diverse applications and advancements in DEA methodologies discussed during significant academic gatherings worldwide. The titles reflect a wide range of topics, including the recent applications of DEA, its role in sustainable development, performance measurement, and theoretical explorations.
Table 7 represents Banker’s publication record across academic journals where he contributed three or more publications. This table shows his substantial contributions to various fields of study. It begins with the European Journal of Operational Research, which has the highest number of publications at 17, followed closely by Management Science with 15 publications. Table 7 also highlights the journals where Banker published more than half of his scholarly works , demonstrating his significant impact through these distinguished journals.
This data illustrates Banker’s diverse and extensive scholarly output, highlighting his influence across multiple academic fields and his active participation in advancing the knowledge and application of DEA and related methodologies.
Table 6 Conference proceedings co-edited by Banker*
Title Conference Authors/Editors ISBN
Recent applications of Data Envelopment Analysis 16th International Conference of DEA, June 2017, University of Economics, Prague, Czech Republic (DEA2017) Emrouznejad, Banker et al. (2017) 978185449 4337
Recent applications of Data Envelopment Analysis 14th International Conference of DEA, May 2016, Jianghan University, Wuhan, China (DEA2016) Emrouznejad, Banker et al. (2016) 978185449 4139
Data Envelopment Analysis and its applications 13th International Conference of DEA, August 2015, Braunschweig, Germany (DEA2015) Emrouznejad, Banker et al. (2015) 978185449 4979
Sustainable development and performance measurement International DEA Workshop, September 17-19, 2014, Hermosillo, Sonora, Mexico Banker et al. (2014) 978185449 4825
Theory and applications of Data Envelopment Analysis 12th International Conference of DEA, April 2014, University of Malaya, Kuala Lumpur, Malaysia (DEA2014) Emrouznejad et al. (2014a, b) 978185449 4870
Data Envelopment Analysis and performance measurement 11th International Conference of DEA, June 2013, Samsun, Turkey (DEA2013) Banker, Emrouznejad et al. (2013) 978185449 4771
Data Envelopment Analysis: theory and applications 10th International Conference on DEA, Natal, Brazil (DEA2012) Banker, Emrouznejad et al. (2012) 978185449 4375
Table 7 Journals with three or more publications authored by Banker
Journal/Source No. of publications % Publications % Cumulative
European Journal of Operational Research 17 8.7 8.7
Management Science 15 7.7 16.3
Annals of Operations Research 9 4.6 20.9
Accounting Review 7 3.6 24.5
Contemporary Accounting Research 7 3.6 28.1
Journal of Accounting and Economics 7 3.6 31.6
Information Systems Research 6 3.1 34.7
MIS Quarterly: Management Information Systems 6 3.1 37.8
Lecture Notes in Information Systems and Organisation 5 2.6 40.3
Journal of Accounting Research 4 2.0 42.3
Journal of Accounting, Auditing & Finance 4 2.0 44.4
Journal of Management Information Systems 4 2.0 46.4
Journal of Productivity Analysis 4 2.0 48.5
Proceedings of the Annual Hawaii International Conference on System Sciences 4 2.0 50.5
Communications of the ACM 3 1.5 52.0
IEEE Transactions on Software Engineering 3 1.5 53.6
Journal of Management Accounting Research 3 1.5 55.1
Table 8 Top ten most cited papers authored by Banker
Publication Journal No. of Citations
Banker et al. (1984) Management Science 11,547
Banker and Morey (1986) Operations Research 863
Banker (1984) European Journal of Operational Research 696
Anderson et al. (2003) Journal of Accounting Research 639
Banker (1993) Management Science 616
Banker and Natarajan (2008) Operations Research 559
Banker and Thrall (1992) European Journal of Operational Research 517
Banker et al. (2000) Accounting Review 490
Banker et al. (1998) Management Science 451
Banker and Morey (1986) Management Science 418

6 Bibliometric analysis

6.1 Macro-level

Figure 9 illustrates the network of keywords used in Banker’s publications, each appearing at least three times, showcasing the interconnected themes within his research. The larger nodes within the network signify keywords used more frequently, indicating their prominence in his research. The clustering of keywords represents cohesive themes and subfields within his work, such as DEA, linear programming, and productivity in the earlier years, and innovation, finance, and commerce in more recent times. This shift in keywords over time reflects the
Fig. 9 Network of keywords used by Banker
evolving focus of his research, from traditional topics to more contemporary subjects within the field.
Table 8 presents the ten most cited papers authored or co-authored by Banker. It includes the titles of the publications, the respective journals where they were published, and the number of citations each paper has received. These papers, published across various prestigious academic journals such as Management Science, Operations Research, European Journal of Operational Research, Journal of Accounting Research, and Accounting Review, highlight the significant impact of Banker in advancing DEA models and applications.

6.2 Micro-level

In this section, we provide a detailed bibliometric analysis outcome of the three main research clusters identified in Banker’s work.

6.2.1 Analysis of the returns-to-scale topic

Figure 10 illustrates the citation trends of Banker’s key papers on RTS and MPSS. Banker et al. (1984) stands out as the most cited paper, demonstrating a continuous upward trend over nearly three decades, underscoring its foundational role in DEA literature. Banker (1984) is the second most cited paper, showing an upward trend, although with significantly fewer citations per year compared to Banker et al. (1984). While other papers in Fig. 10 (apart from those published in 1984) have fewer citations, the subjects of RTS and MPSS still represent a substantial segment of the DEA literature.
Fig. 10 Citation trends for RTS and MPSS related papers
Fig. 11 Citation trends for statistical analysis related papers

6.2.2 Analysis of the statistical analysis in DEA topic

Figure 11 depicts the citation trends of Banker’s key papers on statistical analysis in DEA. The most cited paper in this cluster is Banker (1993), which has shown a relatively stable citation trend over the last decade. Banker and Chang (2006) follow closely, with an upward trend in its initial years but stagnating in recent years. Overall, it appears that Banker’s contributions to statistical analysis in DEA have not achieved the same level of prominence as his papers on RTS.
These analyses highlight the impact and ongoing relevance of Banker’s research within the academic community, particularly in the areas of RTS, MPSS, and statistical methodologies in DEA. His work continues to influence and shape the direction of research in these areas, as evidenced by the sustained citation trends and evolving research focus over time.

6.2.3 Analysis of the contextual analysis topic

Figure 12 illustrates the contrasting trajectories in citations between two of Banker’s papers on second-stage analysis in DEA. Banker and Natarajan’s paper, published in 2008, initially experienced a surge in citations during its early years but has shown a decline in momentum in recent years. Conversely, Banker’s 2019 paper has enjoyed a consistent climb in citations, indicating sustained interest and relevance in the field. However, when comparing the number of citations overall, Banker’s paper on the second-stage analysis shows relatively fewer citations than the alternative approach proposed by Simar and Wilson (2007). For a comprehensive discussion on the alternative method and its bibliometric analysis, refer to Moradi-Motlagh and Emrouznejad (2022).
Fig. 12 Citation trends for second-stage efficiency analysis related papers

7 Conclusion

This paper explored the significant and enduring impact of Professor Rajiv Banker on the field of Data Envelopment Analysis (DEA). Throughout his esteemed career, Banker has been instrumental in shaping the trajectory of DEA, leaving a legacy that continues to inspire scholars, researchers, and practitioners alike.
Our comprehensive tribute to Banker’s lifetime contributions revealed a distinct and pioneering approach to DEA, focusing on three key clusters: Returns-to-Scale (RTS) and Most Productive Scale Size (MPSS) in DEA, Statistical Inference of DEA, and Contextual Analysis. Each of these clusters represents a paradigm shift within the DEA landscape, catalysed by Banker’s groundbreaking research and unwavering commitment to advancing the field.
In honouring Banker’s lifelong dedication to DEA, we recognise not only the intellectual depth of his contributions but also the inspiration he has provided to countless scholars. As the DEA community progresses, it does so on the solid foundation laid by Banker. Through this paper, we celebrate his enduring legacy and express gratitude for his transformative influence on Data Envelopment Analysis.
Acknowledgements The authors would like to thank the Editor and the anonymous reviewers for their valuable feedback on the previous version of this manuscript.

Declarations

Conflict of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is
not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

References

Aggelopoulos, E., & Georgopoulos, A. (2017). Bank branch efficiency under environmental change: A bootstrap DEA on monthly profit and loss accounting statements of Greek retail branches. European Journal of Operational Research, 261(3), 1170-1188.
Ahn, H., Afsharian, M., Emrouznejad, A., & Banker, R. (2018). Recent developments on the use of DEA in the public sector. Socio-Economic Planning Science, 61, 1-3.
Alirezaee, M., Hajinezhad, E., & Paradi, J. C. (2018). Objective identification of technological returns to scale for data envelopment analysis models. European Journal of Operational Research, 266(2), 678-688.
Alperovych, Y., Hübner, G., & Lobet, F. (2015). How does governmental versus private venture capital backing affect a firm’s efficiency? Evidence from Belgium. Journal of Business Venturing, 30(4), 508-525.
Amirteimoori, A., Charles, V., & Mehdizadeh, S. (2023). Stochastic data envelopment analysis in the presence of undesirable outputs. Journal of the Operational Research Society, 74(12), 2619-2632.
Anderson, M. C., Banker, R. D. & Janakiraman, S. N. (2003). Are selling, general, and administrative costs “sticky”?. Journal of Accounting Research, 41(1), 47-63.
Aragon, Y., Daouia, A., & Thomas-Agnan, C. (2005). Nonparametric frontier estimation: A conditional quantile-based approach. Econometric Theory, 21(2), 358-389.
Asmild, M., Paradi, J. C., & Reese, D. N. (2006). Theoretical perspectives of trade-off analysis using DEA. Omega, 34(4), 337-343.
Assaf, A. G., & Gillen, D. (2012). Measuring the joint impact of governance form and economic regulation on airport efficiency. European Journal of Operational Research, 220(1), 187-198.
Assani, S., Jiang, J., Cao, R., & Yang, F. (2018). Most productive scale size decomposition for multi-stage systems in data envelopment analysis. Computers & Industrial Engineering, 120, 279-287.
Atwood, J., & Shaik, S. (2020). Theory and statistical properties of quantile data envelopment analysis. European Journal of Operational Research, 286(2), 649-661.
Badunenko, O., & Mozharovskyi, P. (2020). Statistical inference for the Russell measure of technical efficiency. Journal of the Operational Research Society, 71(3), 517-527.
Bandyopadhyay, S. (2011). In search of contextual variables in a stochastic DEA framework: Effect of regulation on efficiency of Indian cement industry. Journal of the Operational Research Society, 62(9), 1621-1637.
Banker, R. D. (1984). Estimating most productive scale size using data envelopment analysis. European Journal of Operational Research, 17(1), 35-44.
Banker, R. D. (1993). Maximum likelihood, consistency and data envelopment analysis: A statistical foundation. Management Science, 39(10), 1265-1273.
Banker, R. D. (1996). Hypothesis tests using data envelopment analysis. Journal of Productivity Analysis, 7, 139-159.
Banker, R. D., & Chang, H. (1995). A simulation study of hypothesis tests for differences in efficiencies. International Journal of Production Economics, 39(1-2), 37-54.
Banker, R.D ., & Chang, H. (2006). The super-efficiency procedure for outlier identification, not for ranking efficient units. European Journal of Operational Research, 175(2), 1311-1320.
Banker, R. D., & Maindiratta, A. (1986). Piecewise loglinear estimation of efficient production surfaces. Management Science, 32(1), 126-135.
Banker, R. D., & Maindiratta, A. (1988). Nonparametric analysis of technical and allocative efficiencies in production. Econometrica: Journal of the Econometric Society, 56, 1315-1332.
Banker, R. D., & Natarajan, R. (2008). Evaluating contextual variables affecting productivity using data envelopment analysis. Operations Research, 56(1), 48-58.
Banker, R. D., & Thrall, R. M. (1992). Estimation of returns to scale using data envelopment analysis. European Journal of Operational Research, 62(1), 74-84.
Banker, R. D., Potter, G., & Srinivasan, D. (2000). An empirical investigation of an incentive plan that includes nonfinancial performance measures. The Accounting Review, 75(1), 65-92.
Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models for estimating technical and scale inefficiencies in data envelopment analysis. Management Science, 30(9), 1078-1092.
Banker, R. D., Datar, S. M., & Kemerer, C. F. (1991). A model to evaluate variables impacting the productivity of software maintenance projects. Management Science, 37, 1-18.
Banker, R. D., Chang, H., & Sinha, K. K. (1994). Tests to evaluate the separability or substitutability of inputs to a production system.
Banker, R. D., Bardhan, I., & Cooper, W. W. (1996a). A note on returns to scale in DEA. European Journal of Operational Research, 88(3), 583-585.
Banker, R. D., Chang, H., & Cooper, W. W. (1996b). Equivalence and implementation of alternative methods for determining returns to scale in data envelopment analysis. European Journal of Operational Research, 89(3), 473-481.
Banker, R. D., Janakiraman, S., & Natarajan, R. (2002). Evaluating the adequacy of parametric functional forms in estimating monotone and concave production functions. Journal of Productivity Analysis, 17, 111-132.
Banker, R. D., Khosla, I., & Sinha, K. K. (1998). Quality and competition. Management science, 44(9), 1179-1192.
Banker, R. D., Cooper, W. W., Seiford, L. M., Thrall, R. M., & Zhu, J. (2004a). Returns to scale in different DEA models. European Journal of Operational Research, 154(2), 345-362.
Banker, R. D., Janakiraman, S., & Natarajan, R. (2004b). Analysis of trends in technical and allocative efficiency: An application to Texas public school districts. European Journal of Operational Research, 154(2), 477-491.
Banker, R. D., Chang, H., & Natarajan, R. (2005). Productivity change, technical progress, and relative efficiency change in the public accounting industry. Management Science, 51(2), 291-304.
Banker, R. D., Chang, H., & Natarajan, R. (2007). Estimating DEA technical and allocative inefficiency using aggregate cost or revenue data. Journal of Productivity Analysis, 27, 115-121.
Banker, R. D., Chang, H., & Lee, S. Y. (2010a). Differential impact of Korean banking system reforms on bank productivity. Journal of Banking & Finance, 34(7), 1450-1460.
Banker, R. D., Zheng, Z. E., & Natarajan, R. (2010b). DEA-based hypothesis tests for comparing two groups of decision making units. European Journal of Operational Research, 206(1), 231-238.
Banker, R. D., Cooper, W. W., Seiford, L. M. & Zhu, J. (2011). Returns to scale in DEA. Handbook on data envelopment analysis, pp. 41-70.
Banker, R.D., Emrouznejad, A., Lopes, A. L. M. & de Almeida, M. R. (2012). Data envelopment analysis: Theory and applications. In Proceedings of the 10th International Conference on DEA, Natal, Brazil.
Banker, R. D., Emrouznejad, A., Bal, H., Alp, I. & Cengiz, M. A. (2013). Data envelopment analysis and performance measurement. In Proceedings of the 11th International Conference of DEA, June 2013, Samsun, Turkey.
Banker, R.D., Emrouznejad, A., Vargas, F. & Flores, P. (2014). Sustainable development and performance measurement. In Proceedings of the International DEA Workshop, 2014, Hermosillo, Sonora, Mexico.
Banker, R., Natarajan, R., & Zhang, D. (2019). Two-stage estimation of the impact of contextual variables in stochastic frontier production function models using data envelopment analysis: Second stage OLS versus bootstrap approaches. European Journal of Operational Research, 278(2), 368-384.
Bardhan, I. R., Cooper, W. W., & Kumbhakar, S. C. (1998). A simulation study of joint uses of data envelopment analysis and statistical regressions for production function estimation and efficiency evaluation. Journal of Productivity Analysis, 9, 249-278.
Bessent, A., Bessent, W., Elam, J., & Clark, T. (1988). Efficiency frontier determination by constrained facet analysis. Operations Research, 36(5), 785-796.
Bobde, S. M., & Tanaka, M. (2018). Efficiency evaluation of electricity distribution utilities in India: A twostage DEA with bootstrap estimation. Journal of the Operational Research Society, 69(9), 1423-1434.
Boubaker, S., Le, T. D., Manita, R., & Ngo, T. (2023). The trade-off frontier for ESG and Sharpe ratio: a bootstrapped double-frontier data envelopment analysis. Annals of Operations Research. https://doi.org/ 10.1007/s10479-023-05506-z
Carvalho, P., & Marques, R. C. (2014). Computing economies of vertical integration, economies of scope and economies of scale using partial frontier nonparametric methods. European Journal of Operational Research, 234(1), 292-307.
Cazals, C., Florens, J. P., & Simar, L. (2002). Nonparametric frontier estimation: A robust approach. Journal of Econometrics, 106(1), 1-25.
Chang, K. P., & Guh, Y. Y. (1991). Linear production functions and the data envelopment analysis. European Journal of Operational Research, 52(2), 215-223.
Charles, V., & Banker, R. D. (2016). Business performance management under uncertain environments-I. Editorial. Journal of Centrum Cathedra: The Business and Economics Research Journal, 9(2), 90-91. https://doi.org/10.1108/JCC-01-2017-0003
Charles, V., & Banker, R. D. (2017). Business performance management under uncertain environments-II Editorial. Journal of Centrum Cathedra: The Business and Economics Research Journal, 10(1), 2-3. https://doi.org/10.1108/JCC-07-2017-018
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research, 2(6), 429-444.
Charnes, A., Cooper, W. W., & Thrall, R. M. (1986). Classifying and characterizing efficiencies and inefficiencies in data development analysis. Operations Research Letters, 5(3), 105-110.
Charnes, A., Gallegos, A., & Li, H. (1996). Robustly efficient parametric frontiers via multiplicative DEA for domestic and international operations of the Latin American airline industry. European Journal of Operational Research, 88(3), 525-536.
Chen, C. M., & Delmas, M. A. (2012). Measuring eco-inefficiency: A new frontier approach. Operations Research, 60(5), 1064-1079.
Chen, K., & Zhu, J. (2019). Computational tractability of chance constrained data envelopment analysis. European Journal of Operational Research, 274(3), 1037-1046.
Cooper, W. W., & Tone, K. (1997). Measures of inefficiency in data envelopment analysis and stochastic frontier estimation. European Journal of Operational Research, 99(1), 72-88.
Cooper, W. W., Huang, Z., & Li, S. X. (1996). Chapter 13 satisficing DEA models under chance constraints. Annals of Operations Research, 66, 279-295.
Cooper, W. W., Huang, Z., Lelas, V., Li, S. X., & Olesen, O. B. (1998). Chance constrained programming formulations for stochastic characterizations of efficiency and dominance in DEA. Journal of Productivity Analysis, 9, 53-79.
Cooper, W. W., Park, D. K. S., & Ciurana, P. J. T. P. (2000). Marginal rates and elasticities of substitution with additive models in DEA. Journal of Productivity Analysis, 13, 105-123.
Cooper, W. W., Deng, H., Huang, Z., & Li, S. X. (2002). Chance constrained programming approaches to technical efficiencies and inefficiencies in stochastic data envelopment analysis. Journal of the Operational Research Society, 53, 1347-1356.
Dai, S., Kuosmanen, T. and Zhou, X. (2022) Partial frontiers are not quantiles. arXiv preprint arXiv:2205. 11885.
Dai, S., Kuosmanen, T., & Zhou, X. (2023). Generalized quantile and expectile properties for shape constrained nonparametric estimation. European Journal of Operational Research, 310(2), 914-927.
Daouia, A., & Gijbels, I. (2011). Robustness and inference in nonparametric partial frontier modeling. Journal of Econometrics, 161(2), 147-165.
Daouia, A., & Simar, L. (2005). Robust nonparametric estimators of monotone boundaries. Journal of Multivariate Analysis, 96(2), 311-331.
Daouia, A., & Simar, L. (2007). Nonparametric efficiency analysis: A multivariate conditional quantile approach. Journal of Econometrics, 140(2), 375-400.
Dia, M., Takouda, P.M. & Golmohammadi, A., (2022). Assessing the performance of Canadian credit unions using a three-stage network bootstrap DEA. Annals of Operations Research, 1-33.
Du, K., Worthington, A. C., & Zelenyuk, V. (2018). Data envelopment analysis, truncated regression and double-bootstrap for panel data with application to Chinese banking. European Journal of Operational Research, 265(2), 748-764.
e Souza, G. D. S., & Gomes, E. G. (2015). Management of agricultural research centers in Brazil: A DEA application using a dynamic GMM approach. European Journal of Operational Research, 240(3), 819-824.
Emrouznejad, A., & Amin, G. R. (2009). DEA models for ratio data: Convexity consideration. Applied Mathematical Modelling, 33(1), 486-498.
Emrouznejad, A., & Banker, R. D. (2010). Efficiency and productivity: theory and applications. Annals of Operations Research, 173(1), 1-10.
Emrouznejad, A., Anouze, A. L., & Thanassoulis, E. (2010). A semi-oriented radial measure for measuring the efficiency of decision making units with negative data, using DEA. European Journal of Operational Research, 200(1), 297-304.
Emrouznejad, A., Banker, R. D., Lopes, A. L. M., & de Almeida, M. R. (2014a). Data envelopment analysis in the public sector. Socio-Economic Planning Sciences, 48(1), 2-3.
Emrouznejad, A., Banker, R., Munisamy, S. & Arabi, B. (2014b). Theory and applications of data envelopment analysis. In Proceedings of the 12th International Conference of DEA, April 2014, University of Malaya, Kuala Lumpur, Malaysia.
Emrouznejad, A., Banker, R.D., Ahn, H. & Afsharian, M. (2015). Data envelopment analysis and its applications. In Proceedings of the 13th International Conference of DEA, August 2015, Braunschweig, Germany.
Emrouznejad, A., Banker, R., Ray, S. C. & Chen, L. (2016). Recent applications of data envelopment analysis. In Proceedings of the 14th International Conference of DEA, May 2016, Jianghan University, Wuhan, China.
Emrouznejad, A., Jablonský, J., Banker, R. & Toloo, M. (2017). Recent applications of data envelopment analysis. In Proceedings of the 15th International Conference of DEA, June 2017, University of Economics, Prague, Czech Republic.
Emrouznejad, A., Banker, R. D., & Neralic, L. (2019). Advances in data envelopment analysis: Celebrating the 40th anniversary of DEA and the 100th anniversary of Professor Abraham Charnes’ birthday. European Journal of Operational Research, 278(2), 365-367.
España, V. J., Aparicio, J., Barber, X., & Esteve, M. (2024). Estimating production functions through additive models based on regression splines. European Journal of Operational Research, 312(2), 684-699.
Esteve, M., Aparicio, J., Rabasa, A., & Rodriguez-Sala, J. J. (2020). Efficiency analysis trees: A new methodology for estimating production frontiers through decision trees. Expert Systems with Applications, 162, 113783.
Färe, R., Grosskopf, S., & Lovell, C. K. (1985). The measurement of efficiency of production (6th ed.). Springer Science & Business Media.
Färe, R., Grosskopf, S., Lovell, C. K., & Yaisawarng, S. (1993). Derivation of shadow prices for undesirable outputs: a distance function approach. The Review of Economics and Statistics, 75, 374-380.
Färe, R., Grosskopf, S., & Lovell, C. K. (1994). Production frontiers. Cambridge University Press.
Fernandez, C., Koop, G., & Steel, M. F. J. (2002). Multiple-output production with undesirable outputs: An application to nitrogen surplus in agriculture. Journal of the American Statistical Association, 97(458), 432-442.
Førsund, F. R. (1999). The evolution of DEA-The economics perspective. University of Oslo Working paper, Oslo, Norway.
Førsund, F. R. (2018). Economic interpretations of DEA. Socio-Economic Planning Sciences, 61, 9-15.
Førsund, F. R., & Hjalmarsson, L. (2004). Calculating scale elasticity in DEA models. Journal of the Operational Research Society, 55(10), 1023-1038.
Førsund, F. R., Hjalmarsson, L., Krivonozhko, V. E., & Utkin, O. B. (2007). Calculation of scale elasticities in DEA models: Direct and indirect approaches. Journal of Productivity Analysis, 28, 45-56.
Førsund, F. R., Kittelsen, S. A., & Krivonozhko, V. E. (2009). Farrell revisited-Visualizing properties of DEA production frontiers. Journal of the Operational Research Society, 60, 1535-1545.
Frisch, R. (1964). Theory of production. Springer Science & Business Media.
Fukuyama, H. (2000). Returns to scale and scale elasticity in data envelopment analysis. European Journal of Operational Research, 125(1), 93-112.
Fukuyama, H. (2003). Scale characterizations in a DEA directional technology distance function framework. European Journal of Operational Research, 144(1), 108-127.
Giménez, V., Prior, D., Thieme, C., & Tortosa-Ausina, E. (2024). International comparisons of COVID-19 pandemic management: What can be learned from activity analysis techniques? Omega, 122, 102966.
Giraleas, D., Emrouznejad, A., & Thanassoulis, E. (2012). Productivity change using growth accounting and frontier-based approaches-Evidence from a Monte Carlo analysis. European Journal of Operational Research, 222(3), 673-683.
Golany, B., & Yu, G. (1997). Estimating returns to scale in DEA. European Journal of Operational Research, 103(1), 28-37.
Gong, B. H., & Sickles, R. C. (1992). Finite sample evidence on the performance of stochastic frontiers and data envelopment analysis using panel data. Journal of Econometrics, 51(1-2), 259-284.
Grosskopf, S. (1996). Statistical inference and nonparametric efficiency: A selective survey. Journal of Productivity Analysis, 7, 161-176.
Halická, M., & Trnovská, M. (2021). A unified approach to non-radial graph models in data envelopment analysis: Common features, geometry, and duality. European Journal of Operational Research, 289(2), 611-627.
Hatami-Marbini, A., Emrouznejad, A., & Agrell, P. J. (2014). Interval data without sign restrictions in DEA. Applied Mathematical Modelling, 38, 2028-2036.
Johnson, A. L., & Kuosmanen, T. (2012). One-stage and two-stage DEA estimation of the effects of contextual variables. European Journal of Operational Research, 220(2), 559-570.
Johnson, A. L., & McGinnis, L. F. (2008). Outlier detection in two-stage semiparametric DEA models. European Journal of Operational Research, 187(2), 629-635.
Jradi, S., & Ruggiero, J. (2019). Stochastic data envelopment analysis: A quantile regression approach to estimate the production frontier. European Journal of Operational Research, 278(2), 385-393.
Kang, H. J., Kim, C., & Choi, K. (2024). Combining bootstrap data envelopment analysis with social networks for rank discrimination and suitable potential benchmarks. European Journal of Operational Research, 312(1), 283-297.
Kao, C., & Hwang, S. N. (2011). Decomposition of technical and scale efficiencies in two-stage production systems. European Journal of Operational Research, 211(3), 515-519.
Kao, C., & Hwang, S. N. (2019). Efficiency evaluation in the presence of undesirable outputs: The most favorable shadow price approach. Annals of Operations Research, 278, 5-16.
Karagiannis, R. (2015). A system-of-equations two-stage DEA approach for explaining capacity utilization and technical efficiency. Annals of Operations Research, 227, 25-43.
Kerstens, K., & Eeckaut, P. V. (1999). Estimating returns to scale using non-parametric deterministic technologies: A new method based on goodness-of-fit. European Journal of Operational Research, 113(1), 206-214.
Khezrimotlagh, D. (2022). Simulation designs for production frontiers. European Journal of Operational Research, 303(3), 1321-1334.
Khezrimotlagh, D., & Zhu, J. (2023). The role of unobserved units in two-stage network data envelopment analysis. Journal of the Operational Research Society, 74(5), 1275-1285.
Kneip, A., Park, B. U., & Simar, L. (1998). A note on the convergence of nonparametric DEA estimators for production efficiency scores. Econometric Theory, 14(6), 783-793.
Kneip, A., Simar, L., & Wilson, P. W. (2015). When bias kills the variance: Central limit theorems for DEA and FDH efficiency scores. Econometric Theory, 31(2), 394-422.
Kneip, A., Simar, L., & Wilson, P. W. (2016). Testing hypotheses in nonparametric models of production. Journal of Business & Economic Statistics, 34(3), 435-456.
Kneip, A., Simar, L., & Wilson, P. W. (2021). Inference in dynamic, nonparametric models of production: Central limit theorems for Malmquist indices. Econometric Theory, 37(3), 537-572.
Koenker, R., & Bassett, G., Jr. (1978). Regression quantiles. Journal of the Econometric Society., 23, 33-50.
Kontolaimou, A., & Tsekouras, K. (2010). Are cooperatives the weakest link in European banking? A nonparametric metafrontier approach. Journal of Banking & Finance, 34(8), 1946-1957.
Korhonen, P. J., Soleimani-Damaneh, M., & Wallenius, J. (2011). Ratio-based RTS determination in weightrestricted DEA models. European Journal of Operational Research, 215(2), 431-438.
Korostelëv, A. P., Simar, L., & Tsybakov, A. B. (1995a). Efficient estimation of monotone boundaries. The Annals of Statistics, 23, 476-489.
Korostelëv, A. P., Simar, L., & Tsybakov, A. (1995b). On estimation of monotone and convex boundaries. In Annales de l’ISUP, 39(1), 3-18.
Krivonozhko, V. E., Førsund, F. R., & Lychev, A. V. (2014). Measurement of returns to scale using non-radial DEA models. European Journal of Operational Research, 232(3), 664-670.
Krivonozhko, V. E., Utkin, O. B., Volodin, A. V., Sablin, I. A., & Patrin, M. (2004). Constructions of economic functions and calculations of marginal rates in DEA using parametric optimization methods. Journal of the Operational Research Society, 55, 1049-1058.
Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.). The University of Chicago Press.
Kuosmanen, T. (2006). Stochastic nonparametric envelopment of data: Combining virtues of SFA and DEA in a unified framework. MTT Discussion paper no. 3/2006, MT
Kuosmanen, T. (2008). Representation theorem for convex nonparametric least squares. The Econometrics Journal, 11(2), 308-325.
Kuosmanen, T., & Johnson, A. L. (2010). Data envelopment analysis as nonparametric least-squares regression. Operations Research, 58(1), 149-160.
Kuosmanen, T., & Johnson, A. (2017). Modeling joint production of multiple outputs in StoNED: Directional distance function approach. European Journal of Operational Research, 262(2), 792-801.
Kuosmanen, T., & Kortelainen, M. (2012). Stochastic non-smooth envelopment of data: Semi-parametric frontier estimation subject to shape constraints. Journal of Productivity Analysis, 38, 11-28.
Kuosmanen, T., Post, T., & Scholtes, S. (2007). Non-parametric tests of productive efficiency with errors-invariables. Journal of Econometrics, 136(1), 131-162.
Kuosmanen, T., Johnson, A. & Saastamoinen, A. (2015). Stochastic nonparametric approach to efficiency analysis: A unified framework. InData Envelopment Analysis: A Handbook of Models and Methods, pp. 191-244
Kuosmanen, T., & Zhou, X. (2021). Shadow prices and marginal abatement costs: Convex quantile regression approach. European Journal of Operational Research, 289(2), 666-675.
Lee, C. Y., & Wang, K. (2019). Nash marginal abatement cost estimation of air pollutant emissions using the stochastic semi-nonparametric frontier. European Journal of Operational Research, 273(1), 390-400.
Leleu, H. (2013). Shadow pricing of undesirable outputs in nonparametric analysis. European Journal of Operational Research, 231(2), 474-480.
Liao, Z., Dai, S., & Kuosmanen, T. (2024). Convex support vector regression. European Journal of Operational Research, 313(3), 858-870.
Lin, S. W., & Lu, W. M. (2023). Efficiency assessment of public sector management and culture-led urban regeneration using the enhanced Russell-based directional distance function with stochastic data. Journal of the Operational Research Society, 75, 1-19.
Lotfi, F. H., Jahanshahloo, G. R., & Esmaeili, M. (2007). An alternative approach in the estimation of returns to scale under weight restrictions. Applied Mathematics and Computation, 189(1), 719-724.
Lozano, S. (2011). Scale and cost efficiency analysis of networks of processes. Expert Systems with Applications, 38(6), 6612-6617.
Lozano, S., & Villa, G. (2010). Gradual technical and scale efficiency improvement in DEA. Annals of Operations Research, 173, 123-136.
Mergoni, A., Emrouznejad, A., & De Witte, K. (2025). Fifty years of data envelopment analysis. European Journal of Operational Research. https://doi.org/10.1016/j.ejor.2024.12.049
Michali, M., Emrouznejad, A., Dehnokhalaji, A., & Clegg, B. (2023). Subsampling bootstrap in network DEA. European Journal of Operational Research, 305(2), 766-780.
Mitropoulos, P., Talias, MA., & Mitropoulos, I. (2015). Combining stochastic DEA with Bayesian analysis to obtain statistical properties of the efficiency scores: An application to Greek public hospitals. European Journal of Operational Research, 243(1), 302-311.
Moradi-Motlagh, A., & Emrouznejad, A. (2022). The origins and development of statistical approaches in non-parametric frontier models: A survey of the first two decades of scholarly literature (1998-2020). Annals of Operations Research, 318(1), 713-741.
Nguyen, B. H., Simar, L., & Zelenyuk, V. (2022). Data sharpening for improving central limit theorem approximations for data envelopment analysis-type efficiency estimators. European Journal of Operational Research, 303(3), 1469-1480.
Odeck, J. (2009). Statistical precision of DEA and Malmquist indices: A bootstrap application to Norwegian grain producers. Omega, 37(5), 1007-1017.
Olesen, O. B., & Petersen, N. C. (2016). Stochastic data envelopment analysis-A review. European Journal of Operational Research, 251(1), 2-21.
Olesen, O. B., Petersen, N. C., & Podinovski, V. V. (2022). Scale characteristics of variable returns-to-scale production technologies with ratio inputs and outputs. Annals of Operations Research, 318(1), 383-423.
Papaioannou, G., & Podinovski, V. V. (2023). Multicomponent production technologies with restricted allocations of shared inputs and outputs. European Journal of Operational Research, 308(1), 274-289.
Pastor, J. T., Ruiz, J. L., & Sirvent, I. (1999). A statistical test for detecting influential observations in DEA. European Journal of Operational Research, 115(3), 542-554.
Pastor, J. T., Ruiz, J. L., & Sirvent, I. (2002). A statistical test for nested radial DEA models. Operations Research, 50(4), 728-735.
Perelman, S., & Santín, D. (2009). How to generate regularly behaved production data? A Monte Carlo experimentation on DEA scale efficiency measurement. European Journal of Operational Research, 199(1), 303-310.
Pham, M., Simar, L., & Zelenyuk, V. (2024). Statistical inference for aggregation of Malmquist productivity indices. Operations Research, 72(4), 1615-1629.
Podinovski, V. V. (2004). On the linearisation of reference technologies for testing returns to scale in FDH models. European Journal of Operational Research, 152(3), 800-802.
Podinovski, V. V. (2017). Returns to scale in convex production technologies. European Journal of Operational Research, 258(3), 970-982.
Podinovski, V. V. (2019). Direct estimation of marginal characteristics of nonparametric production frontiers in the presence of undesirable outputs. European Journal of Operational Research, 279(1), 258-276.
Podinovski, V. V. (2022). Variable and constant returns-to-scale production technologies with component processes. Operations Research, 70(2), 1238-1258.
Podinovski, V. V., Førsund, F. R., & Krivonozhko, V. E. (2009). A simple derivation of scale elasticity in data envelopment analysis. European Journal of Operational Research, 197(1), 149-153.
Podinovski, V. V., & Førsund, F. R. (2010). Differential characteristics of efficient frontiers in data envelopment analysis. Operations Research, 58(6), 1743-1754.
Podinovski, V. V., Ismail, I., Bouzdine-Chameeva, T., & Zhang, W. (2014). Combining the assumptions of variable and constant returns to scale in the efficiency evaluation of secondary schools. European Journal of Operational Research, 239(2), 504-513.
Podinovski, V. V., Chambers, R. G., Atici, K. B., & Deineko, I. D. (2016). Marginal values and returns to scale for nonparametric production frontiers. Operations Research, 64(1), 236-250.
Ramalho, E. A., Ramalho, J. J., & Henriques, P. D. (2010). Fractional regression models for second stage DEA efficiency analyses. Journal of Productivity Analysis, 34, 239-255.
Resti, A. (2000). Efficiency measurement for multi-product industries: A comparison of classic and recent techniques based on simulated data. European Journal of Operational Research, 121(3), 559-578.
Rosen, D., Schaffnit, C., & Paradi, J. C. (1998). Marginal rates and two-dimensional level curves in DEA. Journal of Productivity Analysis, 9, 205-232.
Rosko, M. D. (2001). Impact of HMO penetration and other environmental factors on hospital X-inefficiency. Medical Care Research and Review, 58, 430-454.
Ruggiero, J. (1996). On the measurement of technical efficiency in the public sector. European Journal of Operational Research, 90(3), 553-565.
Ruggiero, J. (1998). Non-discretionary inputs in data envelopment analysis. European Journal of Operational Research, 111(3), 461-469.
Ruggiero, J., & Vitaliano, D. F. (1999). Assessing the efficiency of public schools using data envelopment analysis and frontier regression. Contemporary Economic Policy, 17(3), 321-331.
Sahoo, B. K., Khoveyni, M., Eslami, R., & Chaudhury, P. (2016). Returns to scale and most productive scale size in DEA with negative data. European Journal of Operational Research, 255(2), 545-558.
Sahoo, B. K., Zhu, J., Tone, K., & Klemen, B. M. (2014). Decomposing technical efficiency and scale elasticity in two-stage network DEA. European Journal of Operational Research, 233(3), 584-594.
Schmidt, P. (1985). Frontier production functions. Econometric Reviews, 4(2), 289-328.
Seiford, L. M., & Thrall, R. M. (1990). Recent developments in DEA: The mathematical programming approach to frontier analysis. Journal of Econometrics, 46(1-2), 7-38.
Seiford, L. M., & Zhu, J. (2003). Context-dependent data envelopment analysis-Measuring attractiveness and progress. Omega, 31(5), 397-408.
Shephard, R. W. (1970). Theory of cost and production fictions. Princeton University Press.
Shi, Y., Charles, V., & Zhu, J. (2025). Bank financial sustainability evaluation: Data envelopment analysis with random forest and Shapley additive explanations. European Journal of Operational Research, 321(2), 614-630.
Shiraz, R. K., Hatami-Marbini, A., & Emrouznejad, H. F. (2020). Chance-constrained cost efficiency in data envelopment analysis model with random inputs and outputs. Operational Research-An International Journal, 20, 1863-1898.
Simar, L. (2007). How to improve the performances of DEA/FDH estimators in the presence of noise. Journal of Productivity Analysis, 28, 183-201.
Simar, L., Vanhems, A., & Wilson, P. W. (2012). Statistical inference for DEA estimators of directional distances. European Journal of Operational Research, 220(3), 853-864.
Simar, L., & Wilson, P. W. (1998). Productivity growth in industrialized countries (No. UCL-Université Catholique de Louvain). Université Catholique de Louvain.
Simar, L., & Wilson, P. W. (1999). Estimating and bootstrapping Malmquist indices. European Journal of Operational Research, 115(3), 459-471.
Simar, L., & Wilson, P. W. (2000a). A general methodology for bootstrapping in non-parametric frontier models. Journal of Applied Statistics, 27(6), 779-802.
Simar, L., & Wilson, P. W. (2000b). Statistical inference in nonparametric frontier models: The state of the art. Journal of Productivity Analysis, 13, 49-78.
Simar, L., & Wilson, P. W. (2002). Non-parametric tests of returns to scale. European Journal of Operational Research, 139(1), 115-132.
Simar, L., & Wilson, P. W. (2007). Estimation and inference in two-stage, semi-parametric models of production processes. Journal of Econometrics, 136(1), 31-64.
Simar, L., & Wilson, P. W. (2020). Technical, allocative and overall efficiency: Estimation and inference. European Journal of Operational Research, 282(3), 1164-1176.
Simar, L., & Zelenyuk, V. (2020). Improving finite sample approximation by central limit theorems for estimates from data envelopment analysis. European Journal of Operational Research, 284(3), 1002-1015.
Simar, L., Zelenyuk, V., & Zhao, S. (2024a). Inference for aggregate efficiency: Theory and guidelines for practitioners. European Journal of Operational Research, 316(1), 240-254.
Simar, L., Zelenyuk, V. & Zhao, S., (2024b). Statistical inference for Hicks-Moorsteen productivity indices. Annals of Operations Research, 1-29
Singh, S., & Ranjan, P. (2018). Efficiency analysis of non-homogeneous parallel sub-unit systems for the performance measurement of higher education. Annals of Operations Research, 269, 641-666.
Soleimani-Damaneh, M. (2013). Another approach for estimating RTS in dynamic DEA. Journal of Productivity Analysis, 39(1), 75-81.
Soleimani-Damaneh, M., Jahanshahloo, G. R., & Reshadi, M. (2006). On the estimation of returns-to-scale in FDH models. European Journal of Operational Research, 174(2), 1055-1059.
Soleimani-Damaneh, M., & Reshadi, M. (2007). A polynomial-time algorithm to estimate returns to scale in FDH models. Computers & Operations Research, 34(7), 2168-2176.
Staat, M. (2002). Bootstrapped efficiency estimates for a model for groups and hierarchies in DEA. European Journal of Operational Research, 138(1), 1-8.
Sueyoshi, T. (1999). DEA duality on returns to scale (RTS) in production and cost analyses: An occurrence of multiple solutions and differences between production-based and cost-based RTS estimates. Management Science, 45(11), 1593-1608.
Sueyoshi, T., & Aoki, S. (2001). A use of a nonparametric statistic for DEA frontier shift: The Kruskal and Wallis rank test. Omega, 29(1), 1-18.
Sueyoshi, T., & Goto, M. (2012). Returns to scale and damages to scale on US fossil fuel power plants: Radial and non-radial approaches for DEA environmental assessment. Energy Economics, 34(6), 2240-2259.
Sueyoshi, T., & Goto, M. (2013). Returns to scale vs. damages to scale in data envelopment analysis: An impact of US clean air act on coal-fired power plants. Omega, 41(2), 164-175.
Sueyoshi, T., & Sekitani, K. (2005). Returns to scale in dynamic DEA. European Journal of Operational Research, 161(2), 536-544.
Taleb, M., Khalid, R., Ramli, R., Ghasemi, M. R., & Ignatius, J. (2022). An integrated bi-objective data envelopment analysis model for measuring returns to scale. European Journal of Operational Research, 296(3), 967-979.
Talluri, S., Narasimhan, R., & Nair, A. (2006). Vendor performance with supply risk: A chance-constrained DEA approach. International Journal of Production Economics, 100(2), 212-222.
Tan, Y., Wanke, P., Antunes, J., & Emrouznejad, A. (2021). Unveiling endogeneity between competition and efficiency in Chinese banks: A two-stage network DEA and regression analysis. Annals of Operations Research, 306(1), 131-171.
Tone, K. (2001). On returns to scale under weight restrictions in data envelopment analysis. Journal of Productivity Analysis, 16, 31-47.
Tortosa-Ausina, E., Grifell-Tatjé, E., Armero, C., & Conesa, D. (2008). Sensitivity analysis of efficiency and Malmquist productivity indices: An application to Spanish savings banks. European Journal of Operational Research, 184(3), 1062-1084.
Tsionas, M. G. (2021). Optimal combinations of stochastic frontier and data envelopment analysis models. European Journal of Operational Research, 294(2), 790-800.
Tsionas, M. G. (2022). Convex non-parametric least squares, causal structures and productivity. European Journal of Operational Research, 303(1), 370-438.
Tsionas, M. G., & Philippas, D. (2023). Measures of global sensitivity in linear programming: Applications in banking sector. Annals of Operations Research, 330(1), 585-607.
Van Eck, N., & Waltman, L. (2010). Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, 84(2), 523-538.
Varian, H. R. (1984). The nonparametric approach to production analysis. Econometrica: Journal of the Econometric Society, 579-597.
Varian, H. R. (1985). Non-parametric analysis of optimizing behavior with measurement error. Journal of Econometrics, 30(1-2), 445-458.
Wang, H., Pan, C., Wang, Q., & Zhou, P. (2020). Assessing sustainability performance of global supply chains: An input-output modeling approach. European Journal of Operational Research, 285(1), 393-404.
Wang, Y., Wang, S., Dang, C., & Ge, W. (2014). Nonparametric quantile frontier estimation under shape restriction. European Journal of Operational Research, 232(3), 671-678.
Wang, K., Wei, Y. M., & Huang, Z. (2016). Potential gains from carbon emissions trading in China: A DEA based estimation on abatement cost savings. Omega, 63, 48-59.
Wheelock, D. C., & Wilson, P. W. (2008). Non-parametric, unconditional quantile estimation for efficiency analysis with an application to Federal Reserve check processing operations. Journal of Econometrics, 145(1-2), 209-225.
Wu, J., & An, Q. (2013). Slacks-based measurement models for estimating returns to scale. International Journal of Information and Decision Sciences, 5(1), 25-35.
Wu, F., Wang, S. Y., & Zhou, P. (2023). Marginal abatement cost of carbon dioxide emissions: The role of abatement options. European Journal of Operational Research, 310(2), 891-901.
Yu, A., Zhang, H., Liu, H. C., Shi, Y., & Bi, W. (2024). Dynamic centralized resource allocation approach with contextual impacts: Analyzing Chinese carbon allocation plans. Annals of Operations Research, 341(1), 451-483.
Zelenyuk, V. (2013). A scale elasticity measure for directional distance function and its dual: Theory and DEA estimation. European Journal of Operational Research, 228(3), 592-600.
Zelenyuk, V. (2020). Aggregation of inputs and outputs prior to data envelopment analysis under big data. European Journal of Operational Research, 282(1), 172-187.
Zhang, Q., & Yang, Z. (2015). Returns to scale of two-stage production process. Computers & Industrial Engineering, 90, 259-268.
Zhou, P., Zhou, X., & Fan, L. W. (2014). On estimating shadow prices of undesirable outputs with efficiency models: A literature review. Applied Energy, 130, 799-806.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

  1. Victor Podinovski
    V.Podinovski@lboro.ac.uk
    Vincent Charles
    c.vincent@qub.ac.uk
    Amir Moradi-Motlagh
    amoradi@swin.edu.au
    1 Surrey Business School, University of Surrey, Guildford GU2 7XH, UK
    2 Loughborough Business School, Loughborough University, Leicestershire LE11 3TU, UK
    3 Queen’s Business School, Queen’s University Belfast, Belfast BT9 5EE, UK
    4 Swinburne University of Technology, Melbourne, Australia
  2. From a definitional perspective, the applied directional Benefit-of-the-Doubt model can be regarded as a DDF model without explicit inputs. However, from a computational perspective, it can be treated as a conventional DDF model due to the presence of undesirable outputs and the absence of inputs.
    The sample size for the simulation tests in Johnson and Kuosmanen’s (2012) study is set at 100 , which is categorised as a ‘medium-sized’ sample in Banker et al. (2019). Their proposed model demonstrated better performance than the DEA + OLS approach in simulation testing.
    In addition to Scopus, Google Scholar is used to extract seven conference proceedings co-edited by Banker.
  3. *Banker also co-organised or served as an invited speaker at several other DEA conferences, including: DEA2002 (invited speaker; Moscow, Russia), DEA2004 (invited speaker; Birmingham, UK), DEA2007 (coorganiser; Hyderabad, India), DEA2009 (organiser; Philadelphia, USA), DEA2010 (invited speaker; Beirut, Lebanon), DEA2011 (invited speaker; Thessaloniki, Greece), DEA workshop 2013 (co-organiser; Hebrew University, Israel), DEA2018 (co-organiser; Wuhan, China), DEA2019 (co-organiser; Calgary, Canada)