Abstract
We implement an endogeneous switching‐regression model for labour productivity and firms’ decisions to use business‐to‐business (B2B) e‐commerce. Our approach allows B2B usage to affect any parameter of the labour productivity equation and to properly take account of strategic complementarities between the input factors and B2B usage. Empirical evidence from 1,460 German firms shows that there is a simultaneous relationship between labour productivity and the adoption of B2B. Firms deciding to use B2B e‐commerce employ their input factors more efficiently than non‐B2B users. Conversely, firms refrain from engaging in B2B probably because they expect the cost of B2B adoption will not be sufficiently compensated by productivity gains.
Notes
1. For further details see Bertschek and Kaiser (Citation2004). The GAUSS code for the Maximum‐likelihood function can be downloaded at http://www.ulrichkaiser.com/papers/orga.html.
2. The sectors that were included in the study are listed in detail in the Appendix.
3. As Germany’s largest credit rating agency, Creditreform has the most comprehensive database of German firms at its disposal. Creditreform provides data on German firms to the Centre for European Economic Research (ZEW) for research purposes.
4. The relation between the introduction of new ICTs and the need for organisational changes in the firm in order to achieve positive productivity effects is examined for instance by Bresnahan et al., (Citation2002) and is also discussed by Brynjolfsson and Hitt (2000).
5. A detailed description of the ICT‐sector is given in the Appendix.
6. Note that for labour input, the estimated coefficients displayed in Table are negative since they correspond to γ − 1. Thus, adding 1 to the estimated coefficients yields the partial output elasticity of labour.
7. Calculating the productivity differences, it is important to note that the results should only be interpreted qualitatively rather than quantitatively, because we had to approximate ICT capital and non‐ICT capital by the value of the respective investments. This shortcoming especially affects the estimated difference between the two regimes with respect to the constant terms, since this difference reflects inter alia the difference in capital stocks that is not covered by investments.
8. Recall further that the descriptive analysis in Table already revealed that the means of (relative) ICT investment are not significantly different in the two regimes with and without B2B e‐commerce. The estimation results of the two alternative specifications are available from the authors on request.
9. These figures are based on the firms’ own information. Using the definitions of the System of National Accounts, the reported export shares of many firms would be different. For example, every sale of a retailer to a foreigner, a tourist, for example, accounts for export of a retail trade service according to the official definitions. Transactions like this are, however, not very likely to be considered by firms when responding to the question whether or not they export.
10. Note that the selection equation is estimated in the reduced form so that the parameter a in equation (Equation8) is not estimated directly. However, since the variables of the productivity equations are jointly (and individually) significant, it can be concluded that the adoption of B2B is influenced by productivity differences.