**Conflict of interest**

The author declares no conflict of interest. The funders had no role in the design of the study; the collection, analyses, or interpretation of data; the writing of the manuscript, or the decision to publish the results.

*Energy Policy*

*p-values.*

**Table 8.**

at the 1% significance level. Therefore, the fixed effect model is appropriate for the panel regression analysis. Further, to test the validity of the panel GMM estimation, this study performs a Sargan-Hansen test for the exogeneity of the instrumental variables. From Hansen J's statistical results, the number of instrumental variables

*Note: 1. \*\* and \* indicate significance at the 1 and 5% levels, respectively. 2. The values between parentheses are* 

**Method EF Panel GMM**

Number of observations 987 940 Adjusted R-squared 0.9850 0.9840

J-statistic 0.0122 Prob(J-statistic) [0.9119] Instrument TFP(−1)

Hausman test 17.086\*\* Prob (Hausman) [0.0043]

**Coefficient Standard** 

*Constant* (*δ*) −0.434\*\* (0.196) 0.240 (0.264) *δLN* 0.215\*\* (0.080) 0.331\*\* (0.084) *δOR* 3.751\*\* (0.328) 4.781\*\* (0.368) *δTFP* 0.707\*\* (0.008) 0.550\*\* (0.037) *δCDD* 0.001 (0.003) 0.000 (0.003) *δHDD* 0.000 (0.008) 0.001 (0.008)

**error**

**Coefficient Standard error**

The signs for all the variables are consistent under both models. The sign for the establishment size is positive, meaning establishments with a larger number of employees have a higher electrification rate. Further, the higher the proportion of offices, the greater the electrification rate. It is also noteworthy that the sign of an establishment's productivity is positive. This indicates that an increase in the establishment's productivity is proportional to that in the electrification rate. The magnitude of the coefficient on productivity is between 0.475 and 0.676, and it significantly influences the electrification rate. Neither cooling nor heating degree

In sum, the establishment scale and productivity are closely related to the electrification rate, which may influence energy efficiency. That is, productivity improves energy efficiency through an increase in electrification at factories and business establishments. Therefore, the efforts to increase the office productivity

This study analyzed the energy efficiency levels and their determinants in Japan's industrial sector using an energy demand frontier function. To the best of the author's knowledge, this is the first attempt to do so. Energy intensity has been traditionally used as a proxy for energy efficiency and depends on economic

is appropriate and satisfies the condition of heteroskedasticity.

*Panel estimation results on determinants of the electrification rate.*

days are statistically significant.

could improve energy efficiency.

**4. Conclusions and policy implications**

**98**

*Energy Policy*
