تفاوت های بین کشوری در نرخ خود اشتغالی: نقش نهادها

This paper examines the role of institutional variables in determining the large disparities observed in self-employment rates across OECD countries. We develop a simple model analysing the role of taxation and tax evasion opportunities. This model predicts tax evasion opportunities to have an unambiguous positive impact on self-employment, while taxation can either spur or reduce the self-employment rate depending on the country attitude towards tax evasion. We find empirical support for the model predictions using a panel of OECD countries. We also show that the selfemployment rate depends negatively on the share of workers in the public sector, while we do not find any robust relationship with employment protection legislation. We find a positive correlation with product market regulation and a negative one with the unemployment benefit replacement rate, but their relevance is sensitive to model specification.

Until recently, most cross-country labour market analyses have focused on paidemployment, while research on self-employment has been mostly conducted at the micro-level.1 Yet, self-employment rates show great cross-country variation, which is likely to affect labour market performances and certainly deserves to be understood. Recently, also due to a resurgence of self-employment in countries that had previously experienced a steady decline in the self-employment rate (OECD, 2000), several studies have tried to fill this gap. Cross-country analyses include Acs et al. (1994), who focus on differences in factor endowments (capital per worker), industry composition and demographic characteristics and Blanchflower (2000) who studies the role of labour market conditions and worker characteristics. Other cross-country studies (Staber and Bogenhold, 1993, Robson and Wren, 1999, Parker and Robson, 2000 and OECD, 2000) and several country specific contributions have investigated the role of institutional variables, such as taxation and labour market regulation. The empirical literature has shown a rather robust, negative relationship between the per capita GDP level and self-employment rate, so that richest countries typically have a lower incidence of self-employment. This stylised fact, taking per capita GDP as a proxy for capital per worker, can be rationalised theoretically by the model of Lucas (1978), which predicts a decline in returns to entrepreneurship relative to wages, as economies become more capital-intensive. Moreover, certain industries, like private sector services, have been shown to display higher self-employment rates than others, so that changes in industry composition and cross-country differences in industry specialization could explain changes over time in self-employment spread and cross-country differences in self-employment rates (Acs et al., 1994). These factors, however, do not seem to account for the persistence of the observed large cross-country differences. Following a rather vast literature, we thus study the role of a selected number of institutional variables, with a particular focus on the size of the public sector and taxation. Previous cross-country empirical studies have almost ignored the potential role of the public sector in crowding-out self-employment opportunities. In fact, public sector expansion can both tilt distribution of workers between industries toward public services where self-employment has no role (e.g. general administration), and crowd out private business in sectors like education and health where the State typically gets involved. We provide evidence that these effects have a significant negative impact on self-employment in a panel of OECD countries. As to taxation, whose impact on self-employment has received most attention, we show with a simple model the theoretical relevance of considering jointly tax rates and tax law enforcement. Previous empirical literature has mostly assumed a positive relationship between the level of taxation and self-employment, on the ground that self-employed workers would have higher opportunities to hide their income from the tax authorities. A tax increase would thus increase the incentive to become self-employed in order to avoid taxation (Blau, 1987, Evans and Leighton, 1989a, Evans and Leighton, 1989b, Parker, 1996 and Shuestze, 2000). From a theoretical point of view, however, this negative relationship cannot be taken for granted. If the income of self-employed workers is more sensitive to individual effort than wage income, a tax increase will hurt self-employed workers more than employees (Robson and Wren, 1999).2 This effect could offset the benefit of greater opportunities of evading taxation, so that the impact of a tax increase on the incidence of self-employment is theoretically ambiguous. In fact, a few empirical studies have provided evidence of a negative relationship between tax rates and self-employment (Davis and Henrekson, 1999 and Fölster, 2002). Moreover, in the policy debate, it is often stated that higher tax rates are detrimental to entrepreneurial activities. We argue that to understand the relationship between taxation and self-employment we should distinguish between the effects of taxation per se and the effect of tax law enforcement. In the simple model we develop, assuming that tax evasion is easier for self-employed workers, an increase in the probability of being detected by the tax authorities unambiguously reduces the incentive to become self-employed, irrespective of the level of taxation; cross-country differences in the enforcement of the tax legislation could thus explain some of the differences in self-employment rates. Moreover, according to the model, tax rate changes can affect in a different (even opposite) way countries characterised by high or low enforcement of their tax legislation. Assuming that self-employed income is more sensitive to individual choices (say effort), if irrespective of employment status tax evasion were not allowed, an increase in the tax rate would unambiguously shift workers from self-employment to paid-employment (incentive effect). On the contrary, in countries where the detection probability is low and possibly lower for self-employed workers, a tax increase can be expected to spur self-employment, or at least we could expect the incentive effect to be offset by higher tax evasion opportunities. We test the predictions of our model on a panel of OECD countries finding empirical support for all of them. We also investigate the role of employment protection legislation, of product market regulation and of the unemployment benefit replacement ratio. Like Robson (2003), we find that the cross-section correlation between self-employment rates and EPL does not hold in a multivariate context. We do find evidence of a more robust positive correlation between product market regulation and self-employment and a negative one with the unemployment benefit replacement ratio. These results, however, are sensitive to the model specification. The paper proceeds as follow. The next section analyses trends over time in selfemployment and the role of industry composition in explaining cross-country differences in self-employment rates. In Section 3, we discuss the relationship between selfemployment and institutional variables and we develop a simple model of employment choice, investigating the role of taxation. In Section 4, we present our empirical results, which are summarised in the last section.

Self-employment rates differ in a substantial and persistent way across countries. We have shown that industry distribution of employment plays a minor role in explaining such large disparities and we have argued that institutional variables, together with still strong differences in capital endowment are the most likely factors behind differences in self-employment rates. We have analysed the empirical link between self-employment rates and a set of institutional variables that have been proposed as being related to the share of self-employed workers, among them the size of the public sector. The role of this variable had not been thoroughly analysed in a cross-country perspective before, and we have shown it to have a negative significant impact on self-employment rates. Moreover, we have developed a simple model that studies the potential role of tax evasion opportunities in shaping the incentives workers face when choosing their job. According to the model, provided that self-employed income is more sensitive to individual effort than paid-employment salary, a tax increase hurts self-employed workers more than employees; therefore, unless self-employment offers sufficiently higher tax evasion opportunities, a higher tax rate reduces the incentive to enter self-employment. On the contrary, if the tax evasion opportunities of self-employed workers are sufficiently high, a tax increase will encourage growth in self-employment. According to the model, higher tax evasion opportunities have a positive and unambiguous impact on self-employment irrespective of the tax level. Our empirical analysis seems to support these conclusions. Our proxy for the tax evasion opportunities, the Corruption perception index, is positively related to selfemployment and our pooled estimates show the correlation with this indicator to be quite robust to the inclusion of suitable controls. In addition, the tax and social contribution rate shows the expected asymmetric impact the model predicts. As to labour and product market regulation, we find that the relationship between employment protection legislation and self-employment does not hold in a multivariate context. Somewhat more robust seems to be the positive relationship with the product market regulation and the negative one with the unemployment benefit replacement rate. However, their impact is sensitive to the model specification and to the controls included in the regressions.