Table 5: Partial R2s (in percent)
Variable |
Model I: 1992-2005 |
Model II: 1999-2005 | ||
df |
Partial R2 |
df |
Partial R2 | |
Environmental factors | ||||
Industry affiliation |
256 |
9.34* |
256 |
10.29* |
Location (district) |
439 |
3.12* |
443 |
2.77* |
Year-effects |
14 |
0.72 |
7 |
0.41 |
Firm-specific factors | ||||
a) Firm characteristics | ||||
Size category |
5 |
4.51* |
5 |
3.38* |
Production share in industry |
1 |
0.01* |
1 |
0.04* |
Number of owners working in the firm |
1 |
0.43* |
1 |
0.44* |
R&D intensity |
1 |
0.20* | ||
b) Outsourcing activities | ||||
Quota of material inputs |
1 |
1.27* |
1 |
1.41* |
Quota of external contract work |
1 |
0.74* |
1 |
0.77* |
Quota of external services |
1 |
0.03* |
1 |
0.17* |
Quota of temporarily employed labor |
1 |
0.01 | ||
Quota rents and leases |
1 |
0.00005 | ||
Sample selection control | ||||
Number of years observed |
1 |
0.02* |
1 |
0.10* |
Overall R2 |
21.78 |
21.11 | ||
Sum of all partial R2s |
20.19 |
20.00 | ||
Number of observations |
38,641 |
24,339 |
Notes: Dependent variable: productive efficiency; df is degrees of freedom;
statistical significance at the 1 percent level is indicated (*).
chemical industry. Industrial differentiation might also stem from differences in average quality
of inputs, the degree of implied product differentiation, or be due to characteristics of production
technology (e.g., Carlsson, 1972).
Third, firm size contributes about 20 percent to the model’s explanatory power. This finding
supports Hypothesis 3, and also confirms the results of other studies finding different efficiency
performance among different firm size classes (e.g., Alvarez and Crespi, 2003; Caves, 1992;
Torii, 1992). However, our results are in the opposite direction of the effects found in these
other studies: according to our analysis, firms become less efficient as size increases. Thus,
smaller firms are, on average, significantly more efficient than larger ones (Table 6). For ex-
ample, the group of firms with less than 49 employees is on average 15 percent more efficient
than the group of firms with more than 1,000 employees. Similarly, we find that relative size,
measured in terms of production share in total industry production, is negatively related to ef-
ficiency. Therefore, Hypothesis 3 holds with respect to the significance, but not with regard to
the direction, of the size effect.
14