What Drives the Productive Efficiency of a Firm?: The Importance of Industry, Location, R&D, and Size

empirical findings of Albach (1980), Caves and Barton (1990) and Hoskisson et al. (1994), but
is counterintuitive since it seems as though R&D should lead to improved products or cost re-
duction (Aghion and Howitt, 1992; Grossman and Helpman, 1991). One explanation for this
odd finding may be that there can be a considerable time lag between R&D spending and R&D
results (Helpman, 1992). If this is the case, R&D expenditure is simply an additional cost at the
time it is incurred, thereby reducing productive efficiency at that time, whereas the benefits can
be appropriated only later. Unfortunately, we cannot test for longer time lags since information
on R&D activity is available in our data for only the last six years. In addition, R&D is risky
and a considerable share of projects are likely to fail, thus possibly making it an inefficient use
of resources, no matter what time period is examined. We also find that most outsourcing activ-
ities enhance efficiency, which goes toward proving 4, however, the effect of R&D is negative,
which contradicts this hypothesis. Moreover, the partial R²s for both variables are of fairly
small magnitude. In sum, then, Hypothesis 4 must be rejected.

Finally, the year dummy variables are not significant.¹⁸ Since we are looking at the average
efficiency of firms, this is not surprising: some firms improve their efficiency, others become
less efficient. The resulting net effect is zero. This explains why we do not find an improvement
of average efficiency over time, a finding in support of Hypothesis 5.

4.3 Subgroups of different efficiency performance

To obtain a more detailed understanding of the factors that contribute to the observed efficiency
differences between firms, we conduct the analyses for three subgroups: (i) the 10 percent least
efficient firms (“worst performers”), (ii) the 10 percent most efficient firms (“best performers”),
and (iii) firms with an efficiency level between these groups (“medium performers”). The partial
R²s and parameter estimates appear in Tables 7 and 8, respectively. Each of these tables contains
six models. We first present the analyses of three subgroups for the period 1992-2005; the
remaining results refer to the same subgroups for the later period, 1999-2005.

The results for the subgroups show that the significance as well as the relative importance of
certain influences differ tremendously across the three different groups of firms. In particular,
many of the previously statistically significant effects are no longer import. Several of these
findings deserve special mention.

¹⁸Parameters are not reported here to conserve space, but they are available upon request from the authors.

16

<<   16   17   18   19   20   21   22   >>

More intriguing information