44th Congress of the European Regional Science Association: "Regions and Fiscal Federalism", 25th - 29th August 2004, Porto, Portugal
Many advocates of an active regional economic policy tend to consider regional sectoral concentration as something which can be created, as a recipe rather than just a development which sometimes occurs and sometimes not. Whether such a policy can be successful implicitly depends on a number of assumptions which are empirically tested in this paper: The first assumption is that regional concentration and regional specialisation foster growth. We test this hypothesis, using data for nine ICT industries in 97 German regions. According to preliminary results, the effect of concentration is positive as expected, but small and its significance depends on specification and industry. Second, the common perception is that East Germany is still a special case. This is clearly supported for the ICT industry, which lacks behind not only with respect to level, but also with respect to growth rates for 1998 to 2002. Third, if regional sectoral agglomeration is an advantage for regions due to technological externalities, for example, then the same reasons might lead to spillover effects between neighbouring regions. With Myrdal we call this a "spread effect". However, the opposite, called "backwash effect" by Myrdal, is also well possible: If a certain sector is growing in region A, that might be due to firms moving in from the neighbouring regions. While the spread effect is the more popular hypothesis, we find evidence for both effects, using standard spatial econometrics techniques. We also discuss the possible reasons why neighbourhood effects are positive in some ICT industries and negative in others. Forth, we investigate whether regional economic policy has a positive impact. Again using spatial econometrics techniques, not so standard in this case, our preliminary results show that the 16 German Laender seem not able to do change the path which is determined for the ICT industries by spatial and other variables.