Taxes, corruption, and entry

Many studies, mostly pre-mid-2000s, proxied entrepreneurship as self-employment (see Ács et al. 2014; Bruce 2000, 2002; Gentry and Hubbard 2000; Parker 1996). Some found a positive relationship between tax rate and self-employment in developed countries. Evans and Leighton (1989a, b) and Blau (1987) studied the US context, and Bacher and Brülhart (2010) studied the Swiss context, both finding a positive relationship of tax progressivity and self-employment. However, some found negative, mixed, or insignificant effects. A negative influence of marginal tax rate on self-employment was found for Canadian microdata (Stabile 2004); similar findings emerged for the US and Canada (Schuetze 2000) and the UK (Parker 1996; Blanchflower and Oswald 1990). For Sweden, Fölster (2002) and Davis and Henrekson (1999) found a negative relationship between tax rates and self-employment, and Hansson (2012) found a negative relationship for both marginal and average tax rates. Another set of studies found no significance for tax rate and self-employment (Baliamoune-Lutz and Garello 2014; Bruce and Mohsin 2006; Parker 2003; OECD 2000). Mixed findings of different measures are not unusual: Robson and Wren (1999) studied OECD countries and found a positive relationship of self-employment with average tax rate, but a negative relationship with marginal tax rate. The distinction between self-employment (labor market trend) versus entry (new firm formation) has emerged, e.g related to educational level (Blanchflower 2004; Parker 2009) or conditions of unemployment and necessity (Arum and Müller 2004a, b). We do not examine self-employment in this study as we concerned with businesses most likely influenced by tax rates: new formal firms. For more on self-employment versus entrepreneurship, see Blanchflower (2004), Parker (2009) and Arum and Müller (2004a, b).

Value-added tax (VAT) is another tax policy tool which could enhance the desirability of corruption. Although many countries have adopted VAT, individuals might not register their business to lower tax liability, or register but submit personal expense invoices as business expense (Alm 2012; Gordon and Li 2009). Another strategy could be to buy or sell from firms owned by the same individual, within a country or overseas.

Two things are noteworthy about our time period, 2005–2011, which was restricted by data availability. First, results could be affected by the global recession, which occurred during this period. We accept this as a limitation which can be illuminated in future research as data availability improves. Second, as we detail in our discussion of the dependent variable, we use formal entry density to measure entrepreneurship. This standardized measure counts limited liability companies in a country using a well-defined process, so our measure captures registered firms which we infer to have been able to pay registration costs (and therefore would comply with tax and other regulations in the future), and are likely to have more financial resources than the self-employed or other entrepreneurs who act because they lack other opportunities (necessity entrepreneurship). The use of this specific measure for entrepreneurship, to some degree, protects our results from being affected drastically by the recession.

We are grateful to one of the reviewers of this paper for this suggestion.

We also conducted a set of checks using a third measure of corruption from the Heritage Foundation’s Index of Economic Freedom. This measure is derived in part with Transparency International (TI) and is correlated with the CPI measure (+92), so we do not report it here and consider the use of CPI adequate for our purposes.