«Der Open-Access-Publikationsserver der ZBW – Leibniz-Informationszentrum Wirtschaft The Open Access Publication Server of the ZBW – Leibniz ...»
This literature has been reviewed in Bertelsman and Doms (2000), and highlights the role of several supply-side production factors that determine productivity at the ﬁrm level. For instance, technology shocks, management skills, R&D investments, among others, are shown to aﬀect ﬁrm level productivity in traditional production function estimation approaches.
In addition, Syverson (2004) studies the impact of demand-side factors that determine diﬀerences in observed productivity levels. In particular, he ﬁnds evidence of the negative impact of product substitutability within an industry on the dispersion of its productivity distribution. The lower the product substitutability within an industry, the more low productivity ﬁrms are able to stay in business and the higher the corresponding dispersion of the industry’s productivity distribution.
This literature has also considered the role of ICT in determining ﬁrm level heterogeneity.8 In that literature, the estimated positive impact of ICT on productivity is recovered as the coeﬃcient on a proxy for ICT capital from a production function estimation. That is, ICT is shown to increase the central tendency of the conditional productivity distribution but is silent about the eﬀect of ICT on the dispersion of this distribution. Moreover, negative and positive impact of ICT are averaged out, making impossible to identify the potential negative impact of ICT and its consequences for ﬁrm strategies.
This paper presents a measure of ﬁrm heterogeneity that explicitly accounts for the dispersion of the productivity distribution, allowing the analysis of the role of ICT use on ﬁrm heterogeneity. In particular, ﬁrm heterogeneity is deﬁned as the deviation of a ﬁrm’s productivity level from a given industry benchmark. In consequence, the adopted measure exhibits two main advantages with respect to the existing literature. On the one hand, it permits to directly account for the role of ICT on ﬁrm heterogeneity in a way that cannot be inferred See Bresnahan et al. (2002), Brynjolfsson and Hitt (2003), and Bloom et al. (2008).
from existing production function estimation. On the other hand, the economic literature has shown how ﬁrm heterogeneity (as previously deﬁned) actually explains a great variety of strategic decisions at the ﬁrm level as the following literature review shows.
The second strand of literature related with the present paper analyzes the impact of ﬁrm heterogeneity on ﬁrm strategies. Although there is a voluminous literature in this area, the relevant work for the present paper corresponds to the analyses that considered similar measures of ﬁrm heterogeneity from an empirical perspective. For instance, ﬁrm decentralization decisions are theoretically and empirically explained by the role of ﬁrm heterogeneity (Acemoglu et al., 2007) and the nature of the relationship between innovation and competition depends on the level of ﬁrm heterogeneity (Aghion et al., 2005).
More speciﬁcally, deﬁning the distance to the technological frontier as the diﬀerence between a ﬁrm’s productivity level and the highest productivity level observed in the same industry (i.e. a measure of ﬁrm heterogeneity), Acemoglu et al. (2007) develop a model to analyze the relationship between the diﬀusion of new technologies and the decentralization of ﬁrms.
They show, theoretically and empirically, that ﬁrms closer to the technological frontier are more likely to choose decentralization. The intuition behind this result states that these ﬁrms deal with newer technologies about which there is less information available. In consequence, a decentralized structure allows them to beneﬁt from better informed managers (as opposed to principals).
In addition, and developing additional measures of ﬁrm heterogeneity, the paper shows how ﬁrms in more heterogeneous environments are also more likely to be decentralized because ﬁrm heterogeneity makes learning (i.e. how to exploit a new technology given the experience of others) more diﬃcult. In a sequence of papers, Acemoglu et al. (2003), Aghion et al. (2005), Acemoglu et al. (2006) and Aghion et al. (2006) use similar measures of heterogeneity in order to study, among other topics, the relationship between innovation, entry, credit constraints and competition. Essentially, these papers highlight the role of ﬁrm heterogeneity as a main driver of industry evolution.9 In a diﬀerent approach, Chun et al. (2008) directly estimate the impact of ICT use on ﬁrm heterogeneity for a panel of U.S. ﬁrms from 1971 to 2000. They ﬁnd that elevated heterogeneity in ﬁrm performance (i.e. variability in labor productivity) is positively and signiﬁcantly correlated with the use of ICT (i.e. ICT capital stock). The results also show that See Bartelsman et al. (2006) for an analysis of industry evolution using similar measures of ﬁrm heterogeneity.
ﬁrm heterogeneity is associated with faster productivity growth. They argue that the results provide evidence of creative destruction (i.e. increased competition) at the ﬁrm level. That is, through their use of ICT, more productive ﬁrms displace less productive ﬁrms. However, their results can only be recovered at the industry level and do not permit the analysis of their impact on ﬁrm strategies. This paper contributes to this literature by explicitly considering the impact of ICT on productivity heterogeneity at the ﬁrm level, analyzing the consequences for speciﬁc ﬁrm strategies relevant to the process of creative destruction (i.e.
3 Empirical Analysis
The analysis is based on two waves of a business survey carried out by the Centre for European Economic Research (ZEW) corresponding to the years 2003 and 2006 (ZEW ICT Survey).
The data set is a representative sample of the German manufacturing and service sector, and contains detailed information on the economic characteristics, performance and ICT use for 4,400 ﬁrms in each wave. Table 1 provides some descriptive statistics for the year 2003.
In general, and in line with similar data sets at the ﬁrm level, the surveyed ﬁrms evidence a great variability with respect to sales (in millions euros), number of employees and labor productivity calculated as the ratio between sales and number of employees. In addition, the empirical distribution of the reported sales, as well as the number of employees appear to be left skewed. The median size of the surveyed ﬁrm corresponds to 50 employees, whereas the average is about 337.7 employees. This indicates the presence of few very large companies in the data.
Analogously, the median value of sales is 7 millions with an average of 131.7 millions. The same also holds for the distribution of labor productivity where the median and the mean are
0.13 and 0.24 millions per employee, respectively. This skewness present in the distributions of the reported data is consistent with empirical analyses at the ﬁrm level. Table 1 also includes information about the use and intensity of ICT. The intensity in the use of ICT is measured by the percentage of employees working mainly with a PC (PCW) and is nearly equally distributed around the diﬀerent percentiles. In addition, the data also provides information about diﬀerent ICT software applications, namely enterprise resource planning (ERP), supply
chain management (SCM) and customer relationship management (CRM).
Table 2 shows how the use of these diﬀerent ICT applications are related with diﬀerent ﬁrm characteristics. These ﬁrm characteristics (e.g. sales and number of employees) vary if the ﬁrm uses ICT more intense than the median (i.e. PCW greater than 50%) or introduced ICT applications. Speciﬁcally, the table shows that the intense use of ICT, as well as the introduction of diﬀerent ICT applications, is correlated with higher ﬁrm performance in terms of sales, number of employees and labor productivity. However, it should be noted that standard deviations are high through all the data, which indicates a high level of heterogeneity.
For example, the surveyed ﬁrms evidence 338 employees on average, whereas companies using ERP, SCM or CRM tend to be larger with 518, 518 and 627 average number of employees, respectively. The same can be observed in terms of sales. Firms using ICT applications or using ICT more intensively were more likely to exhibit higher sales. The average level of sales over all ﬁrms is 131.7 millions euros, whereas companies using CRM, SCM or ERP evidence 205.9, 222.8 and 209.2 millions, respectively. In general, these statistics show that ﬁrms using ICT more intensively tend to be larger than their counterparts.
Moreover, ﬁrms that use ICT more intensively and that introduce ICT applications do not only tend to be larger, but more eﬃcient than other ﬁrms. In terms labor productivity (i.e.
ratio between sales and number of employees), it can be seen that the average is 0.24 million euros per employee. This is higher for ﬁrms using ICT more extensively. Speciﬁcally, companies using ICT more intensively (i.e. PCW greater than 50%) exhibit a labor productivity of 0.31 millions euro per employee. Moreover, companies using CRM exhibit on average 0.27 millions euro per employee, while those using SCM and ERP evidence 0.26 millions and 0.27 millions per employee, respectively.
In addition to these general characteristics, the data also contain information regarding the innovation activities of the sampled ﬁrms. For instance, as a proxy for innovation incentives, the data contain information on the fraction of employees working on R&D activities in 2006 (i.e. R&D intensity). The mean value of this variable is 0.17 with a median equal to 0.09, indicating a left-skewed distribution.
In addition, information on innovation outputs is also available. This includes dummy variables indicating whether the ﬁrm introduced product or process innovations during the periods 2001-2003 and 2004-2006, the number of such innovation introduced during 2004-2006, the percentage of sales reported in 2006 that are derived from the product innovations introduced during the period 2004-2006 (mean: 26.32, median: 20) and the percentage of cost reductions achieved in 2006 from the introduction of process innovations during the period 2004-2006 (mean: 9.49, median: 8).
The measure of ICT intensity, percentage of employees working with a PC (PCW), is positively correlated to all measures of innovation. This does not hold for the ICT applications considered. For instance, ERP use is related to a slightly higher amount of the value of process innovations, while the value of product innovations slightly decreases from 27.4% to 25.8%. The fraction of employees working on R&D activities (i.e. R&D intensity), however, is decreasing from 21.7% to 14.9% for ﬁrms that adopted ERP. This is diﬀerent for the case of CRM use. The value of product innovations, the value of process innovations and R&D intensity are positively related with CRM adoption.
From the data set, the measure of ﬁrm heterogeneity at the ﬁrm level is constructed as the absolute value of the deviation of a given ﬁrm labor productivity, with respect to the median labor productivity in its corresponding industry. Deviations are taken in absolute values to derive a measure of heterogeneity associated with the dispersion of the productivity distribution. Labor productivity is computed as the ratio between sales and number of employees reported by the ﬁrms for a given year.10 Note that by taking absolution values positive, as well as negative deviations are treated equally. Table 3 presents a quantile regression analysis of the ﬁrms’ speciﬁc deviations (i.e.
without taking absolute values) that further motivate the adopted measure of ﬁrm heterogeneity previously deﬁned. The table shows that when used intensively, ICT induce positive deviations for quantiles above the median. That is, the marginal eﬀect of ERP, conditional on high values of PCW, is positive for quantiles above the median (i.e. columns 4-5), increasing the ﬁrm speciﬁc’s deviation. In contrast, for quantiles below the median, ICT use reduces such deviation (i.e. column 2).