This paper proposes a dynamic slacks-based measure (DSBM) model for data envelopment analysis (DEA) to measure efficiency changes over time, considering carry-over activities between consecutive periods. The DSBM model extends the traditional DEA framework by incorporating carry-over activities into the efficiency evaluation, allowing for a more accurate assessment of decision-making units (DMUs) over time. The model distinguishes carry-over activities into four categories: desirable (good), undesirable (bad), free, and fixed. Desirable carry-overs are treated as outputs, while undesirable ones are treated as inputs. Free and fixed carry-overs represent discretionary and non-discretionary activities, respectively. The DSBM model is non-radial, enabling individual evaluation of inputs and outputs, unlike radial models that assume proportional changes. The model also considers three orientations: input-oriented, output-oriented, and non-oriented. The input-oriented model focuses on minimizing input excesses and undesirable carry-overs, while the output-oriented model aims to maximize output shortfalls and desirable carry-overs. The non-oriented model combines both aspects. The paper applies the DSBM model to evaluate efficiency changes in the power generation division of 50 electric utilities (41 U.S. and 9 Japanese) over seven years (1990-1996). The results show that the classification of carry-over activities significantly affects efficiency measurements. The dynamic model accounts for long-term investment trends, unlike traditional separate models that evaluate each year independently. The dynamic model provides more accurate and practical efficiency assessments by considering carry-over activities and long-term investment strategies. The study compares the results of the dynamic model with those of the traditional separate model, demonstrating that the dynamic model better captures long-term efficiency changes and avoids the limitations of single-year evaluations. The results indicate that the dynamic model is more suitable for evaluating long-term investment decisions and capital expansion policies. Future research directions include decomposing inefficiency into input, output, and link components, as well as extending the model to dynamic and network DEA frameworks.This paper proposes a dynamic slacks-based measure (DSBM) model for data envelopment analysis (DEA) to measure efficiency changes over time, considering carry-over activities between consecutive periods. The DSBM model extends the traditional DEA framework by incorporating carry-over activities into the efficiency evaluation, allowing for a more accurate assessment of decision-making units (DMUs) over time. The model distinguishes carry-over activities into four categories: desirable (good), undesirable (bad), free, and fixed. Desirable carry-overs are treated as outputs, while undesirable ones are treated as inputs. Free and fixed carry-overs represent discretionary and non-discretionary activities, respectively. The DSBM model is non-radial, enabling individual evaluation of inputs and outputs, unlike radial models that assume proportional changes. The model also considers three orientations: input-oriented, output-oriented, and non-oriented. The input-oriented model focuses on minimizing input excesses and undesirable carry-overs, while the output-oriented model aims to maximize output shortfalls and desirable carry-overs. The non-oriented model combines both aspects. The paper applies the DSBM model to evaluate efficiency changes in the power generation division of 50 electric utilities (41 U.S. and 9 Japanese) over seven years (1990-1996). The results show that the classification of carry-over activities significantly affects efficiency measurements. The dynamic model accounts for long-term investment trends, unlike traditional separate models that evaluate each year independently. The dynamic model provides more accurate and practical efficiency assessments by considering carry-over activities and long-term investment strategies. The study compares the results of the dynamic model with those of the traditional separate model, demonstrating that the dynamic model better captures long-term efficiency changes and avoids the limitations of single-year evaluations. The results indicate that the dynamic model is more suitable for evaluating long-term investment decisions and capital expansion policies. Future research directions include decomposing inefficiency into input, output, and link components, as well as extending the model to dynamic and network DEA frameworks.