Designing a Dynamic Network Data Envelopment Analysis (DEA) Model for the Performance Evaluation of Bank Branch Management

Objective
Performance evaluation in the banking sector, and particularly within development banks, necessitates analytical frameworks that are both comprehensive and dynamic. This requirement stems from several distinctive features of development banking, including the multi-stage structure of banking operations, the existence of objectives that go beyond short-term profitability, and the intertemporal dependence of performance outcomes across successive periods. A review of prior empirical studies shows that a large share of the existing literature, especially in domestic research, has tended to focus on a single dimension of performance, most commonly either resource management efficiency or financial profitability efficiency, examined in isolation. Such approaches provide only a partial understanding of bank performance and fail to adequately capture the internal linkages between resource mobilization, cost structures, financial returns, and the realization of development-oriented goals. Moreover, limited attention has been devoted to analyzing these dimensions simultaneously within a dynamic framework that reflects the specific operational characteristics of development banks. In response to this gap, the objective of the present study is to conduct a comprehensive and intertemporal evaluation of managerial performance in the Cooperative Development Bank, focusing on its provincial management units. Specifically, the study aims to analyze the dynamic relationship between resource management efficiency, financial profitability, and the fulfillment of development objectives through the application of a dynamic network framework.
Methods
This study employs a Dynamic Network Slack-Based Measure (DN-SBM) model within the broader data envelopment analysis (DEA) framework. The proposed model conceptualizes banking operations as a two-stage sequential network consisting of a resource management stage followed by a financial profitability stage. The internal structure of the model explicitly incorporates intermediate products and carry-over variables to reflect the operational linkages between stages and the intertemporal dependence of performance. The empirical analysis is based on panel data from 31 provincial management units of the Cooperative Development Bank over the period 1401–1403 (2022–2024). Several methodological features distinguish this study. Deposits are disaggregated into low-cost and high-cost deposits and are treated as intermediate outputs of the resource management stage. Interest expenses associated with high-cost deposits are incorporated as an exogenous input in the profitability stage, reflecting their role in shaping financial outcomes. In addition, carry-over variables such as non-performing loans and branch location are included to capture dynamic effects that transmit performance across periods. The model also applies dynamic weighting to both stages and time periods based on their relative importance. To enhance the practical relevance of the analysis, slack analysis is conducted as a complementary tool to identify specific sources and magnitudes of inefficiency at the provincial level.
Results
The results indicate considerable heterogeneity in dynamic network efficiency among provincial management units of the Cooperative Development Bank. Only a limited number of provinces consistently remain on the efficiency frontier throughout the entire study period, while many others exhibit fluctuating performance over time. The slack analysis reveals that the sources of inefficiency vary substantially across provinces and across stages of operation. In some provincial units, inefficiency is mainly driven by excess operational and personnel costs, pointing to weaknesses in cost control and resource utilization within the resource management stage. In other cases, inefficiency is primarily associated with the profitability stage and is linked to elevated credit risk and higher levels of non-performing loans. The dynamic results further demonstrate that achieving efficiency in one period does not necessarily ensure efficiency in subsequent periods. This finding underscores the importance of carry-over variables in shaping performance trajectories and highlights the intertemporal nature of managerial performance in development banking.
 
 
Conclusion
The conclusions of this study suggest that the DN-SBM approach offers an effective and robust framework for evaluating the performance of development banks operating in multi-stage and intertemporal environments. By explicitly modeling internal processes, intermediate outputs, and carry-over effects, the proposed framework enables a more nuanced assessment of managerial performance than static or single-stage approaches. The ability to precisely identify inefficiency bottlenecks through slack analysis supports the formulation of targeted improvement programs and strengthens data-driven managerial decision-making. Overall, the findings indicate that dynamic network performance evaluation can assist managers and policymakers in better aligning development objectives with economic efficiency within development banking institutions.