Employment Location in German Manufacturing

Source: BfLR (1992). Because of data problems west Berlin is excluded from the dataset.Google Scholar

Cf. Pöschl (1992).Google Scholar

Cf. BfLR (1984).Google Scholar

See the appendix for the matching of different classifications.Google Scholar

Due to data protection rules the exact cut off point for no reporting is not zero but between zero and five. Therefore, these cases were treated as an employment of three.Google Scholar

Cf. Cowell (1995).Google Scholar

This applies to sectors 56 (public services) and 57 (social security services), according to the classification of the input-output table.Google Scholar

Provided the variables are bivariate normal, if the coefficient of correlation is larger than 0.193, it is significantly larger than zero at 5 % level of significance.Google Scholar

The official statistic distinguishes between firms as organizational units and establishments as localized units of production.Google Scholar

In difference to equation (2.56), the variables are not entered in deviations from the national mean. As a single industry is considered, its characteristics are simply picked up by the constant.Google Scholar

See footnote 44 in section 2.5.Google Scholar

In a study of regional firm formation Harhoff(1995a, 1995b) uses a quadratic specification of relative own employment and finds a decreasing positive effect which becomes negative at a higher level of own relative employment.Google Scholar

Today there are 327 districts. Five districts are excluded due to reforms of district territories (see appendix).Google Scholar

The reason here is that sums of employment are given for the districts one-digit industries at district level and for each two-digit industry for various sets of districts (see appendix).Google Scholar

See appendix for a list.Google Scholar

See Bernard / Durlauf (1996).Google Scholar

For the concept of leverage see, for instance, Davidson / McKinnon (1993).Google Scholar

See appendix for description of data and sources.Google Scholar

See Seitz (1996) for a recent study of suburbanization in German core-cities.Google Scholar

Henderson et al. (1995) report coefficients for a regression of current on lagged (log) employment between 0.365 and 0.647. cf. ibid. (1994), p. 1073.Google Scholar

Henderson et al. (1995) use the log of all other manufacturing employment and report significant coefficients between 0.223 and 0.986, cf. ibid., tables B3 and B4.Google Scholar

See Reynolds / Storey / Westhead (1994).Google Scholar

See Armstrong / Taylor (1993).Google Scholar

This result is quite different to Peschel / Bröcker (1988) and Bröcker (1089). They find no significant correlation between regional employment growth and industry-mix effects for the planning regions in West Germany for an even smaller time period. For comparison, the basic correlation between the actual and the predicted rate of employment growth in the present study is 0.27. When using the actual employment growth in percentage rather than in log differences the correlation reduces to 0.26. Provided the variables are bivariate normal, in the given case the critical value at 5 % level of significance is 0.094. Besides differences in the industry classification used the difference to the present study might arise from the focus on firms with at least 20 employees.Google Scholar

Cf. Bröcker (1989).Google Scholar

For a discussion of regional investment-promotion policy in Germany, see Franz / Schalk (1995) and the references listed there.Google Scholar

The coefficient of variation of the user cost of capital increases from 5.8 % in 1977, and 7.1 % in 1978 to a value of 15.8 % in 1989. (Own computations from the user cost of capital supplied by Franz / Schalk.)Google Scholar