Infrastructure and entrepreneurship

Abstract

This paper is one of the first studies to examine the link between infrastructure and entrepreneurship. Because infrastructure can enhance connectivity and linkages that facilitate the recognition of entrepreneurial opportunities and the ability of entrepreneurs to actualize those opportunities, a hypothesis is developed suggesting that startup activity is enhanced by infrastructure. However, not all types of infrastructure have a homogeneous impact on the entrepreneurial decision, so that a second hypothesis is developed suggesting that certain types of infrastructure which facilitate connectivity and linkages among people are more conducive to startup activity. The empirical results suggest that startup activity is positively linked to infrastructure in general, but that certain specific types of infrastructure, such as broadband are more conducive to infrastructure than are highways and railroads. Finally, we hypothesize that the types of infrastructure have varying influences in different sectors. Our empirical analyses support this view and we conclude that particular infrastructure policies can be used to facilitate regional startup activities and, furthermore, to foster startup activities in desired industries.

Appendix

See Tables 3, 4 and 5.

Table 3 Descriptive statistics

Table 4 Correlations

Table 5 Tobit regression for high technology manufacturing

1.1 Calculation of broadband penetration index

In contrast to publicly available information on the other types of infrastructure, details on broadband rollout and its availability are not publicly available in Europe, at least not on a county level. The impact of broadband on startups is twofold: First, broadband enables entrepreneurs to access knowledge without being close to a knowledge incubator and accessing a broad range of customers beyond regional proximity. Second, the transmission rate, i.e., the quality of broadband, is the key pre-condition to provide completely new types of services. In this second regard, broadband itself is a platform for completely new and innovative types of entrepreneurs whose business model completely depends on the high transmission rate of broadband. For our cross-infrastructure comparison, we focus on the impact of broadband availability on startups. Broadband was introduced in Germany mainly starting in 2000. Since then, no substitute infrastructure, such as cable or mobile infrastructure has been available which could have influenced the distribution of broadband in Germany.

We measure broadband availability using the date of the upgrade of MDFs. An MDF connects the local loop infrastructure and the backbone infrastructure. Thus, each MDF belongs to only one area with one common area code. As soon as an MDF is upgraded to provide access to higher-speed backbone infrastructure, this MDF can be used for local broadband access.¹¹ Households and companies with standard subscriber lines connected to an upgraded MDF by standard copper lines can directly switch to broadband without major physical arrangements.

Main distribution frames are not directly related to a particular county but serve households and companies also across county borders. Similarly, the capacity of MDFs is limited. Therefore, multiple MDFs are required to provide telecommunication services to a larger county. Taking into account these obstacles, we approximate broadband availability with the so-called broadband penetration index per county. Based on the area code, we know the subscriber lines per area and we also know the subscriber lines per MDF. Thus, we calculate the share of upgraded MDFs in an area as

$$shbb_{mt} = \frac{{\sum\nolimits_{a} {\# sl_{amt} } }}{{\# sl_{mt} }}$$

which is the share of subscriber lines at upgraded MDFs per total subscriber lines in the area m at time t.

We use households as the linking pin between area codes and counties as households is the majority of broadband users and as households usually own one subscriber line. In so doing, we calculate the average broadband penetration based on the areas covering a county as follows:

$$PI_{it} = \sum\nolimits_{s} {\frac{{\# hh_{ist} }}{{\sum\nolimits_{n} {\# hh_{\text{int}} } }}} shbb_{st}$$

with $\frac{{\# hh_{imt} }}{{\sum\nolimits_{n} {\# hh_{\text{int}} } }}$as the share of households in county i with subscriber lines to MDFs in area m.