Table 11: Parameter estimates of selected variables: determinants of the dynamics of firm effi-
ciency
|
Model I: 1992-2005 |
Model II: 1999-2005 | |
|
Firm-specific factors | ||
|
a) Size category | ||
|
Less than 49 employees |
-0.002 (-1.34) |
-0.002 (-1.18) |
|
50-99 employees |
-0.000236 (-0.16) |
0.0001 (0.10) |
|
100-249 employees |
0.001 (0.67) |
0.001 (0.83) |
|
250-499 employees |
0.001 (0.89) |
0.001 (1.08) |
|
500-999 employees |
0.001 (1.03) |
0.001 (1.14) |
|
More than 1000 employees | ||
|
Production share in industry |
-0.001 (-0.14) |
-0.001 (-0.25) |
|
Number of owners working in the firm |
-0.0002 (-0.62) |
-0.0002 (-0.49) |
|
R&D intensity |
0.02** (2.02) | |
|
b) Outsourcing activities | ||
|
Quota of material inputs |
0.00006 (0.61) |
0.00005 (0.46) |
|
Quota of external contract work |
0.004* (3.8) |
0.004* (3.71) |
|
Quota of external services |
-0.003* (-2.5) |
-0.003* (-2.35) |
|
Quota of temporarily employed labor |
0.003 (0.34) | |
|
Quota rents and leases |
0.000009 (0.67) | |
|
Year Dummies | ||
|
D1992 |
0.001 (0.50) | |
|
D1993 |
-0.003 (-0.91) | |
|
D1994 |
0.001 (0.49) | |
|
D1995 |
0.00004 (0.01) | |
|
D1996 |
0.00009 (0.03) | |
|
D1997 |
-0.0002 (-0.05) | |
|
D1998 |
-0.0001 (-0.04) | |
|
D1999 |
-0.001 (-0.41) |
-0.0008 (-0.35) |
|
D2000 |
-0.002 (-0.59) |
-0.002 (-0.63) |
|
D2001 |
0.002 (0.58) |
0.002 (0.69) |
|
D2002 |
0.002 (0.80) |
0.002 (0.97) |
|
D2003 |
0.001 (0.48) |
0.001 (0.54) |
|
D2004 |
-0.003 (-1.23) |
-0.003 (-1.26) |
|
D2005 |
0.004* (2.29) |
0.004* (2.84) |
|
Number of observations |
3,147 |
3j16 |
Dependent variable: θi , notes of Table 6 apply.
Hjalmarsson, 1999). The parameter θi indicates whether a firm’s efficiency increases (θi > 0)
or decreases (θi < 0) with time t . Therefore, in this part we extended the translog production
function framework by including firm-specific time trends. We performed this analysis only
for firms with at least 10 observations in order to obtain more reliable estimates of θi . We also
refrained from including a quadratic time trend in the translog production function, as the high
collinearity between the linear and quadratic time trends leads to imprecise estimates of both
21