3.1 Benchmark Regression
3.1.1 Benchmark Regression and Analysis of Results
Based on panel data of 28 countries which are varying in history, culture and economic development level from 1989 to 2019, an empirical test will be done to analyse the relationship between financial development level and income inequality. Regression equation used here is
Ginii,t = θ1 + η1 FDi,t−1 + γ1Xi,t−1 + µ1i,t−1 (1)
Results are shown in Table I and have passed the test of heteroscedasticity. According to results in Table I, financial development has a significantly negative effect on Gini Index. So countries with more developed financial markets are fairer than those less developed countries. For control variables, GDP per capita is negatively related to Gini Index while political rights are positively related. Thus, if a country wants to narrow domestic income gap, it’s helpful to improve economy and pay more attention to the extent of free and fair elections, political pluralism and participation, as well as a functioning government.
3.1.2 Discussion on problems that may exist so far
Even though simultaneity and heteroscedasticity have been avoided, there are still other problems that may cause endogeneity in the model and affect accuracy of results.
(1) Autocorrelation
Autocorrelation means that there are correlation relationships between different expected values of error terms, then estimated values of the model will be biased. Reasons for autocorrelation may include data of financial development and inequality level might fluctuate with business circle, the hysteresis effect of economy and setting errors of the model. For the sake of avoiding this problem, error terms in this model are assumed to be error terms with white noises. This means each element of error terms has an identical, independent and mean-zero distribution.
(2) Measurement error
Measurement error is the difference between measured value of a variable and its true value. It can be divided into random error and systematic error. Random errors are resulted from those unstable factors when gathering data and can be reduced by repeating the experiment. Yet systematic errors are not introduced by chance, rather affected by inaccuracies oriented in the system. No matter financial development or inequality level is easily influenced by changes in outside environment, thus both type of measurement error might affect the accuracy of results. Data used here are taken from official reports and worldwide famous research papers, which is helpful to limit negative effects of measurement error in a certain level.
(3) Omitted variable
The relationship needed to be studied is how financial development affects inequality, yet there may be other variables that can affect financial development and inequality level at the same time. To reduce errors caused by omitted variables, four control variables are used here. As discussed before in 2.3, these control variables include: growth rate of GDP per capita, natural logarithm of real GDP per capita, the level of democracy as well as political rights in a certain country.
In addition to actions that have been taken so far, the method of instrumental variable will also be used to increase the accuracy of empirical research and it will be analysed in the following sections.
3.2 Suitability of the Instrumental Variable
3.2.1 Relevance of Instrumental Variable
As discussed before in 2.4, legal origin of a certain country is used as the instrumental variable here. Legal rules that regulate financial industry adjust the relationship between financial transactions and supervision. Perfect financial legal system is supposed to include rules on financial market entry, operation scope, self-regulations and protections on depositors as well as investors.
As shown in Figure II, La Porta et al. (1997, 1998) concluded that legal origins, which can be resulted from historical and colonial factors, affect protection on investors. Protection on depositors and investors is an important part of financial supervision and helps to improve the efficiency of financial system, while poor investor protections will lead to smaller and less active financial market. Thus, legal origin of a given country has an influence on its financial development level.
3.2.2 Exogenous restriction of instrumental variable
In addition to relevance, another requirement that a suitable instrumental variable needed to be met is exogeneity. In model of this paper, exogenous restriction means that legal origin does not affect inequality directly. It only has an influence on inequality through affecting development level of financial markets. In commom sense, commom laws put more importance on published judical opinions. While codified statutes predominate in civil law systems. Disparities in these two legal systems may influence economic performance differently. For most Latin American countries, they didn’t choose which type of law system to adopt by themselvels. Their legal origin is usually affected by colonial history and oriented from European countries who used to colonize them. Thus the acceptance of legal origin can be considered to be objective.
What’s more, legal origin cannot affect inequality directly, rather through laws and regulations which are determined by the certain type of legal origin. For example, countries with French law origin like Brizial and Chile tend to have smaller and less active financial markets than America and England, whose legal origin is English. As explained before, more developed financial market is beneficial for individulas as well as enterpreneus to get external finance more easily and it is also beneficial to reduce inequality.
3.3 Two-Stage Least Squares Regression
Taking legal origin as the instrumental variable, 28 countries are divided into four groups, which are countries with English legal origin, French legal origin, German legal origin and Scandinavian legal origin. Regression equations used here are
\(\widehat{\text{F}\text{D}}\) i,t−1 = β0 + β1Zi + γXi,t−1 + εi,t−1 (2)
Ginii,t = θ2 + η2\(\widehat{\text{F}\text{D}}\)i,t−1 + γ2Xi,t−1 + µ2i,t−1 (3)
Results are shown in Table II and Table III.
Based on results of DWH-test, the hypothesis that financial development level is an exogenous variable is rejected. Thus, it’s necessary to accept the method of instrumental variable. According to the results of first stage, F-value is 79.0484, which is greater than 10. What’s more, Stock/Yogo-test has also passed. So the instrumental variable used here is not a weak instrument, it’s suitable and does help to increase the accuracy of regression model.
As shown in table III, financial development has significantly negative effect on Gini Index. So higher financial development level in the past would lead to lower inequality nowadays and accords with argument of this paper. With the development of financial system, more individuals and firms are able to get external credits through financial intermediaries. For individuals who couldn’t get access to financial markets before, they are capable of getting information about investment projects now and share fair opportunities with those rich people. Advantages of financial innovations are open to each person as well. What’s more, the improvement and completeness of financial contracts are helpful to protect rights and interests of all participants in financial markets. As a result, poor people can also enjoy the benefits brought by investments that both have low risks and high returns, which is helpful for capital accumulation and inequality reduction. Similarly, with easier access to credit, people who have talented ideas but limited finance before are able to not only invest in financial products, but also set up the entrepreneurship now. The establishment of more and more firms creates job opportunities and increases demand for labour. At the same time, poor people can use that newly earned external finance to educate themselves or their children. This can contribute to the accumulation of human capital and increase numbers of skilled labour force in the society. The development of financial markets benefits both the demand and supply aspect of labour markets and is helpful for the sustainable development of society.
When referring to the influences of legal origin, according to the results of Table II, the coefficient of French legal origin is only 2.0762, which is quite lower than other types of legal origin. As discussed before in 2.4, most Latin American countries are deeply affected by French civil law because of historical and colonial factors. Also, they are suffering from severe inequality problems nowadays. According to database of this paper, nearly all the Latin American countries have a Gini index higher than 0.5, some are even over 0.6. For the income share aspect, Felix Rioja (2008) found that income of the richest people is almost 20 times of that for the bottom part of the society. The negatively related relationship between financial development level and income inequality can help to partly explain the relatively high level in Latin America.