Public cluster policy and neighboring regions: beggar-thy-neighbor?

ABSTRACT

Research on public cluster policy has largely taken a perspective evaluating firm performance or local cluster performance, almost neglecting spillover effects on neighboring regions. This study evaluates the effects and performance of public cluster policy in three ways: firstly, by evaluating public cluster policy per se; secondly, whether positive effects are shaped as a consequence of the ‘picking-the-winner’ competition or by the subsidizing effects afterwards; and finally, whether effects of public cluster policy spill over to neighboring regions or are mainly bounded locally. Based on a unique panel dataset encompassing all German labor market regions and covering a 15-year period, we apply difference-in-difference estimations and quantile regression techniques to identify and separate the different effects. Our results confirm positive cluster effects of the chosen industries, but also show that positive externalities are spatially limited. Policy-makers should be aware of the local boundedness of public cluster initiatives and possible adverse ‘beggar-thy-neighbor’ effects.

KEYWORDS:

• Public cluster policy
• beggar-thy-neighbor effects
• local spillovers
• policy evaluation

JEL Classification:

• R11
• R58

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. See Kuratko and Menter (2017) for a more in-depth description of recent public policies in Germany, especially the leading-edge cluster competition.

2. Notice that Berlin, the largest city and metropolitan area in Germany with over 3 millions of inhabitants and the highest in-migration rate of all cities in Germany, belongs to the ‘non-neighbouring regions’.

3. The central assumption of the DID estimation refers to unobserved differences between the treatment and control groups which should be the same over time in the absence of the treatment. An assumption that may hold for Germany as a rather balanced economy (see Martin 2012).

4. The 0.20 quantile divides the dataset into two parts, whereas 20% of the included firms have productivity rates less or equal the 0.20 quantile and 80% of the firm have higher productivity rates.

5. The variables UIC_wcwr as well as UIC_wcor have to be dropped in the second estimation approach (models VII–XII), since only the value zero can be assigned to all respective regions.

6. Within our time period from 1998 to 2012, the overall regional productivity in Germany has increased by more than 20%. The significant and negative signs of the 0.20 and 0.40 quantile regressions, i.e. low endowment regions, as well as the significant and positive signs of the 0.80 and 0.90 quantile regressions, i.e. high endowment regions, emphasize the outstanding economic role of high-performing cluster regions. These regions serve as the growth engines concerning economic growth and productivity.