2. Literature review

Corak (Reference Corak and Rycroft2013a) looks at the intergenerational earnings elasticity between the United States and 24 other countries, and he uses the most common method of analyzing the intergenerational earnings elasticity between fathers and sons. For a recent review of the literature on intergenerational income mobility see Corak (Reference Corak2013b). Sometimes data is drawn from mothers and sons, fathers and daughters, mothers and daughters, and some combination of parents and children (Schnetzer & Altzinger, Reference Schnetzer and Altzinger2013). Schnetzer and Altzinger (Reference Schnetzer and Altzinger2013) only refer to their estimate of the intergenerational earnings elasticity for Austria as parents to children. Furthermore, concerns might arise due to different studies using data from different years of cohorts. Data for intergenerational income mobility are derived from two surveys for the United States: the Panel Study of Income Dynamics (PSID) and the National Longitudinal Surveys (NLS) of labor market experience (Solon, Reference Solon2002).

Andrews and Leigh (Reference Andrews and Leigh2009) examine the relationship between income inequality and mobility. They find evidence to suggest that there is indeed a negative association between inequality and mobility, that is, more income inequality is correlated with less income mobility; sons who grew up in countries that were more unequal in the 1970s had a worse chance for increased social mobility in the late 1990s.

Corak (Reference Corak and Rycroft2013a) analyzes the relationship between income inequality and mobility. Alan Krueger calls this relationship the ‘Great Gatsby curve’. Income inequality is captured by the Gini coefficient and is measured on the horizontal axis. Income mobility is captured by the intergenerational income elasticity measures and is displayed on the vertical axis. Gini coefficients are measured between 0 and 100 where zero means complete income equality and 100 means complete income inequality. The intergenerational income elasticity, β, measures the persistence of income from one generation to the next. That is, it measures how likely one individual is to follow in a parent's footsteps in relation to the income distribution. Thus, a lower β means an individual has lower earnings persistence and therefore higher income mobility. The positive trend line on the Great Gatsby curve illustrates an inverse relationship between income inequality and income mobility; countries with more income inequality have greater earnings persistence between generations and less income mobility.

The Great Gatsby curve has certain policy recommendations. If the relationship is such that the higher inequality in a country, the lower the mobility, then policies for higher social mobility are recommended to be accompanied by policies for more equal societies. In a policy recommendation, OECD (Organization for Economic Co-operation and Development; 2010) suggests that progressive tax systems and social transfer programs should not only make a society more equal but also strengthen the chances for individual social and economic advancement. Alternatively, economic regulations and ineffective government may explain declines in income mobility.

Policies consistent with improvements in economic institutions are hypothesized to allow for greater income mobility. If citizens are free to engage in commerce with others and if they do not fear their property will be stolen from them, then they should face better opportunities to experience the American Dream through income mobility. Specifically, regulatory constraints should have a large impact on income mobility. Indeed, this has been offered as an explanation of why income mobility (inequality) has decreased (increased) in the United States in the past few decades (Clark & Lawson, Reference Clark and Lawson2008).

In his inquiry on the mystery of capital, De Soto (Reference De Soto2000) finds that capitalism succeeds in the west and fails everywhere else. One contributing factor is that the barriers to entrepreneurship and acquiring private property are so tedious that citizens cannot improve their economic well-being. It took De Soto's team of researchers 289 days to register a legal business enterprise in Peru while working at it for six hours a day, filling out forms, standing in lines, and acquiring certifications necessary to operate. In total, the cost of legal registration was $1,231, which was 31 times the monthly minimum wage. Similarly, in the Philippines, the process to registering private property required 168 steps, involving 53 private and public agencies, and taking anywhere from 13 to 25 years. They also found it would take someone 5 to 14 years in Egypt and two years in Haiti (De Soto, Reference De Soto2000: 20–21). Therefore, it follows that sound economic institutions for commerce and entrepreneurship matter; sound rules allow individuals to prosper, and this is hypothesized to be reflected in greater income mobility.

3. Data and methodology

Data on political and economic institutions are available to test the hypothesis that fewer burdens and regulations allow for more income mobility. There are six World Wide Governance Indicators (WGI) described by Kaufmann et al. (Reference Kaufmann, Kraay and Mastruzzi2011): government effectiveness, control of corruption, voice and accountability, regulatory environment, rule of law, political stability, and absence of violence and terrorism. Since entrepreneurship is hypothesized to influence income mobility, only three of these variables are used to ascertain whether burdens and regulations stymie income mobility: (i) government effectiveness, (ii) regulatory quality, and (iii) the rule of law. According to the WGI, government effectiveness ‘reflects perceptions of the quality of public services, the quality of the civil service, and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies’. Regulatory quality ‘reflects perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development’. Rule of law ‘reflects perceptions of the extent to which agents have confidence in and abide by the rules of society, and in particular the quality of contract enforcement, property rights, the police, and the courts, as well as the likelihood of crime and violence’ (Kaufmann et al., Reference Kaufmann, Kraay and Mastruzzi2011). All three of these explanatory variables range from –2.5 to 2.5 where 2.5 indicates a country has a more productive regulatory environment, an effective quality of government, and a judicial and legal system that protects property rights. These three measures are then interpreted as follows: higher political, economic, and judicial quality of institutions as measured by the WGI is associated with higher numbers on the scale of –2.5 to 2.5.

Transparency International collects annual data on corruption known as the corruption perceptions index (CPI). It measures the perceived levels of public sector corruption in 176 countries and territories around the world by utilizing surveys from multiple, comparable sources, and it aggregates the results to construct the CPI index. The data are measured on a scale of 1 to 10 where 1 means very little corruption and 10 is the most corrupt. The CPI index by Transparency International was chosen for the analysis rather than the World Bank's control of corruption measure because the latter is another measure under the World Governance Indicator similar to government effectiveness and regulatory quality. Multiple sources and alternative institutional measures will do a better job of ascertaining the effects of institutional quality on income mobility.

To the best of my knowledge, there are currently 32 countries with an available β coefficient, the measure of intergenerational earnings elasticity derived from equation (1). Twenty-five of these estimates are available from Corak (Reference Corak and Rycroft2013a). The other seven come from other studies. In total there are 15 studies with at least one estimate of the intergenerational earnings elasticity for a country. These studies and their β coefficients are listed in Table 1. While there are some countries with only one β coefficient from one study, there are also quite a few countries with multiple β coefficients from multiple studies. For the empirical estimates derived in this study, intergenerational earnings elasticity estimates are taken from Corak (Reference Corak and Rycroft2013a) which provides β coefficients for 25 countries. The other studies have been listed in Table 1 in hopes that the data from these studies are useful to other researchers, but their estimates are not included in the scatter plots, correlation coefficients, and OLS regressions. The other intergenerational earnings elasticity estimates are not included in an attempt to minimize measurement error because they may use other combinations of cohorts, such as fathers and daughters, mothers and sons, and different time spans of study. Thus, empirical estimates derived from sources other than Corak (Reference Corak and Rycroft2013a) may actually capture differences in data collection rather than cross-country differences in income mobility. A more concerning issue is with the accuracy of the parental income. Zimmerman (Reference Zimmerman1992) has critiqued the increase in income mobility as a result of measurement error arising from attenuation bias. The results here are reported using income data from Corak (Reference Corak and Rycroft2013a) and it cannot be ruled out that attenuation bias is not present. However, Blanden et al. (Reference Blanden, Gregg and Macmillan2013), using National Child Development Survey (NCDS) income data, argue that their findings run counter to these claims because their data stem from observable characteristics, which are free from the influence of attenuation bias. While it should be noted that the data in this paper are gathered from Corak (Reference Corak and Rycroft2013a), this provides some evidence that increases in income mobility may arise for reasons other than attenuation bias.

Table 1.

β coefficients from Corak (Reference Corak and Rycroft2013a) and other studies on intergenerational mobility

The Gini coefficient is another explanatory variable used to capture the Great Gatsby effect, and it is taken from Deininger and Squire (Reference Deininger and Squire1996). This Gini is subject to a couple of criticisms (Atkinson & Brandolini, Reference Atkinson and Brandolini2001; Galbraith & Kum, Reference Galbraith and Kum2005). Most notably, the Gini for China is lower than in Europe, which is a striking result. However, the results presented here are robust to the substitution of the Gini measure used by the World Bank.Footnote ¹ The distance to the equator is also included as an additional measure of economic institutional quality (Hall & Jones, Reference Hall and Jones1999). It is hypothesized that countries located farther away from the equator set up better economic institutions.Footnote ²

The Great Gatsby curve and the relationship between institutional measures and income mobility are depicted in Figure 1 below. The correlation coefficient of income inequality and the intergenerational earnings elasticity is 0.47, which is the lowest correlation coefficient among all five measures. In addition, there are four institutional measures hypothesized to be associated with income mobility: (i) government effectiveness, (ii) regulatory quality, (iii) rule of law, and (iv) corruption.

Figure 1.

Great Gatsby curve.

Rule of law facilitates entrepreneurship through the channel of the protection of property rights and an equal enforcement of contracts. When property is better protected under the law and there is legal remediation for contract violation, entrepreneurs have a greater incentive to start their own business; Coase (Reference Coase1937) suggests that individuals join firms in order to reduce transactions costs. However, if transactions costs associated with contract enforcement are reduced and property is more secure, then individuals have a greater incentive to work for themselves. Similarly, the Kirznerian entrepreneur is more likely to possess a desire to discover profitable opportunities when he may reap the benefits from his alertness. Rule of law is located at the top of Figure 2, and it has a correlation coefficient of –0.66. This suggests that the security of property rights is positively correlated with income mobility as measured by intergenerational earnings elasticity. The relationship between corruption and income mobility is depicted at the bottom of Figure 2. Méon and Sekkat (Reference Méon and Sekkat2005) suggest that corruption can either sand or grease the wheels for economic development, depending on the institutional structure of the economy. In countries with strong economic institutions, corruption is expected to stymie entrepreneurship because bribes and additional bureaucratic payments increase the cost and time of entrepreneurship. This relationship boasts the highest correlation coefficient of 0.70; more corruption is associated with higher earnings persistence and lower income mobility.

Figure 2.

Rule of law, corruption, and intergenerational mobility.

Countries with a superior regulatory quality are expected to facilitate entrepreneurship through the promotion of private sector development. Regulatory quality is depicted at the top of Figure 3. The correlation coefficient is –0.52, which suggests a lower regulatory burden is associated with lower earnings persistence and higher income mobility. Finally, government effectiveness is located at the bottom of Figure 3. Similar to the regulatory quality, an effective government is expected to promote entrepreneurship by supporting good economic policies and independence from the political process. Its correlation coefficient with income mobility is –0.62; an effective government is associated with less income persistence and higher income mobility. All five of these measures and the correlation coefficients are illustrated in Figure 4 below.

Figure 3.

Regulatory quality, government efficacy, and intergenerational mobility.

Figure 4.

Correlation coefficients of income inequality, institutions, and income mobility. Note: The correlation coefficients are calculated as the correlation between intergenerational earnings elasticity and five measures: income inequality, regulatory quality, government effectiveness, rule of law, and corruption.

Table 2 illustrates the descriptive statistics for the variables used in this study. The dependent variable is a measure of income mobility using collected data on intergenerational income elasticity measures. The values for the intergenerational earnings elasticity range from –0.15 to –0.69 corresponding to Denmark and South Africa, respectively. Typically, these values are positive as shown in Table 1 and Figure 1, but in order to make the results easier to interpret, the signs have been flipped for the intergenerational earnings elasticity measure in the regression output. The way to interpret the income mobility measure is simple; it measures the degree of earnings persistence from one generation to the next. Higher values for the intergenerational income elasticity translate into more persistence in earnings and consequently less income mobility. Since smaller values mean more income mobility, it makes more sense to make them negative. This would allow larger numbers to express more income mobility as it is easier to interpret. Thus, income mobility measures are multiplied by negative unity.

Table 2.

Descriptive statistics

GDP = gross domestic product; PISA = Programme for International Student Assessment.

The Gini coefficient is measured on a scale of 0 to 100 where 0 means completely equal wealth and 100 means one person owns all of the wealth. The minimum in the 25 country data is 17.8 and the maximum is 62.3 with a mean Gini coefficient of 35.428. The largest Gini coefficient corresponds to South Africa and the smallest Gini coefficient corresponds to China. Gini data is taken from Deininger and Squire (Reference Deininger and Squire1996) with data observed around 1990 and gathered as personal income. However, the results in all tables and specifications are robust to the World Bank's Gini measure in 2006. These results are reported in Appendix Table 2. Data on per capita income and the growth of per capita income are taken from the Penn World Tables (Heston et al., Reference Heston, Summers and Aten2012). Data on the growth of per capita income are for the 10 years from 1996 to 2006. Data on the level of per capita income are gathered in 2006. Growth rates and levels are included for the convergence of incomes to steady-states (Barro & Sala-i-Martin, Reference Barro and Sala-i-Martin1992; Barro et al., Reference Barro, Mankiw and Sala-i-Martin1995; Sokoloff & Engerman, Reference Sokoloff and Engerman2000).

Kirzner (Reference Kirzner1979) argues that private property rights, free enterprise, and sound money foster an environment where subconscious learning can occur and the alert entrepreneur discovers these profitable opportunities. Excessive regulations which curtail business activity and consequently make it harder for poorer people to become entrepreneurs are barriers to increasing income mobility. If this is true, removing some of these barriers, regulations, and red tape are posited to improve income mobility. This leads to the first hypothesis.

Proposition 1: Institutional quality that allows entrepreneurs the chance to flourish is associated with greater income mobility.

Another alternative has been suggested by OECD (2010). The idea captures the Great Gatsby relationship. Subsequently, they suggest a policy recommendation of increasing social transfers which should reduce income inequality and increase income mobility. These social transfers can be thought of as increased government spending on education and healthcare. It is hypothesized that education expenditures by the government allows economically disadvantaged groups the opportunity to increase their earnings capacity by receiving an education. In a study of US states, Mayer and Lopoo (Reference Mayer and Lopoo2008) find that high spending states possess greater income mobility than low spending states. Similarly, healthcare expenditures help to ensure that citizens in society are healthy, and facilitate movement between income brackets for economically disadvantaged citizens. Several studies support this claim. Case et al. (Reference Case, Angela and Paxson2005) suggest that a large share of the intergenerational transmission of socioeconomic status works through the channel of parents’ income to their children's health. Eriksson et al. (Reference Eriksson, Bratsberg and Raaum2005) report that the intergenerational earnings elasticity in Denmark falls by 25%–28% when controlling for their parental health status. Hertz (Reference Hertz2007) finds that the relationship between parental income and health status explains 8% of intergenerational correlation of income in the United States. Similarly, poor health explains 5%–10% of the variation in an offspring's income mobility (Esping-Anderson & Wagner, Reference Esping-Andersen and Wagner2010). For a larger review of the literature, see Nolan et al. (Reference Nolan, Esping-Andersen, Whelan, Maitre and Wagner2010).

Proposition 2: Countries with greater spending on education and healthcare should be associated with greater income mobility.

In order to test these two hypotheses, an empirical specification using OLS is employed, and the form of the equations estimated is specified as follows:

(2)\begin{equation} {\rm Income}\,{\rm Mobility}_i = \alpha + \beta _1 {\bf X}'_i + \beta _2 {\bm \delta }'_i + \varepsilon _i \quad {\rm for}\quad i = 1 \ldots n,\end{equation}

where the dependent variable, Income Mobility, refers to the intergenerational income elasticity; X′ is a vector of institutional variables that are of interest in explaining the variation in income mobility, and it includes rule of law, corruption, regulatory quality, and government effectiveness; δ′ is a vector of control variables that may be important in analyzing the relationship between institutional quality and income mobility. Included in this control vector are the growth of per capita income, the level of GDP per capita, the Gini coefficient, distance to the equator, government expenditures on education, and government expenditures on healthcare. Government expenditures on education and healthcare are taken from the World Bank in 2006 and are measured as a percentage of Gross Domestic Product (GDP). In addition, I have used the internationally comparable Programme for International Student Assessment (PISA) data as an alternative measure of education. This specification does not change any of the effects of institutions on income mobility and it is reported in Appendix Table 3.

Table 3.

Effect of institutions on income mobility

Note: |t| statistics in parentheses.

Degree of freedom corresponds to the critical value, α = 0.05 of 2.11.

† denotes thousands of miles.

*p < 0.10, **p < 0.05, ***p < 0.01.