The moderating role of stock markets in the bank competition-entrepreneurship relationship

We review some of these studies in the next section

Freixas and Rochet (1997) provide a review of traditional models of Industrial Organization (IO) consistent with this view.

Many studies investigate the impact of bank competition on credit availability to firms in general (see Love and Martinez-Peria 2015 for the most recent paper on the subject). Our interest in this paper is on the implications on entrepreneurship. Thus, without claiming to be exhaustive, we only mention studies relevant to that.

Non-structural measures of bank competition have already been used to address other questions, including the impacts of competition on firms’ access to finance in general (see Love and Martinez-Peria (2015) for a recent contribution).

A concise description of the three measures, and the computation procedures are presented in

A thick market—as opposed to a thin market— is one with a high number of buyers and sellers, which increases the liquidity of stocks—that is the extent to which assets are bought and sold at stable prices on the market.

Tables recapitulating the list of variables, their definitions, sources, and key descriptive statistics are provided in Appendix B.

Most of institutional data (World Bank Doing Business data) are only available from 2004, while data constraints on bank competition measures limit our sample period to the year 2010.

This is adopted for the sake of comparison or consistency across countries of different legal and economic systems.

The GEM project is a popular study in entrepreneurship research, which provides survey-based data on entrepreneurial activity across countries all over the world. The key country indicator of this dataset is the Total Entrepreneurial Activity which is defined as the percentage of 18–64 population who are either nascent entrepreneur or owner-manger of a new business. This indicator clearly could not be used, as it does not capture actual business creation. The related new business ownership measure does not fit the purpose either as it captures businesses aged up to 42 months. Besides, countries which participate in the survey differ from year to year. Thus, from the perspective of our study which exploits variation across countries and within countries, this is not an optimal choice. The COMPENDIA database focuses on the Organization for Economic Co-operation and Development (OECD) countries (see Van Stel 2005 for documentation on this database). It provides annual data on self-employment which includes owner-managers of both unincorporated and incorporated businesses. The drawbacks of this dataset in the context of our study are the fact that the business ownership indicator includes unincorporated companies (not to mention the emphasis on the occupational status) and the measure is a “stock” variable.

We investigate the issue using the COMPENDIA dataset in the robustness section.

Note that as of January 2017, Bankscope is no longer available and has been replaced by Orbis Bank Focus.

Data on employment protection laws (EPL) are available for OECD countries. However, historical worldwide data on labor market rigidities are not available in the DBD nor elsewhere.

We also considered specifications where the impact of competition is allowed to be non-monotone. The non-monotonicity effects were insignificant.

A simultaneity bias could occur if newly registered businesses launch their IPO during the first year of operation, which is very unlikely.

Our key results are robust to experimenting with a number of other plausible control variables, including unemployment and lagged GDP growth.

Recall that the bank competition variable BCO (Boone indicator, Lerner index and adjusted Lerner index) is an inverse measure of competition.

For the sake of readability and compactness, only estimates of the key variables—including significant control variables —are reported. The full estimation tables are available upon request.

The estimations also confirm that bank competition indicators are strongly related to bank credit, as also shown in Fig. 2 and Table 7. In particular, increased bank competition leads to higher levels of bank credit.

Although the interest here is on testing heterogeneous effects of impacts of bank competition (including the moderating effect of stock markets), we also run a version of Eq. (4) in which the variable “stock market size (SMS)” is equally allowed to present differentiated impacts. The results are generally robust to this specification.

This is also true for the baseline models—where we do not control for bank concentration.

Full estimation results are available upon request.

While Lerner (1934) specifically addresses issues surrounding the measurement of monopoly power, one can easily extend it to any market situation where firms have some sort of market power.

According to the quiet life hypothesis (Hicks 1935), monopolistic firms may trade potential rents in exchange for profit and cost inefficiencies.