Financial resource of public social security expenditure, the rule of law, and economic inequality: international comparison of legal origins

Abstract This study indicates that the strength of the rule of law affects social security expenditure and its financial resources. However, legal origins affect the strength of the rule of law. We use data from 36 member countries of the Organisation for Economic Co-operation and Development. The legal origins affect the rule of law and have different marginal effects on public financial resources. Further, social security expenditures have a correction effect on economic inequality. However, the marginal effect of the social security expenditure on economic inequality differs based on its legal origins. Increasing social security expenditure affects correcting economic inequality to elaborate concerning all legal origins. Particularly in countries that adopted French legal origins, high social security expenditure further improves economic inequality. In contrast, increases in social security expenditures of countries with English legal origins do not improve economic inequality to a large extent. The results of this study suggest that differences in legal origins result in various types of capitalism.


Introduction
Developed countries are characterised as welfare states to varying degrees, with challenges in increasing social security expenditure. According to Wilensky (1975), developed countries exhibit a converging trend of developing as welfare states and simultaneously expand their social security systems and coverage 1 . Therefore, similarities can be found in the financial resources of these countries.
However, by observing the Organisation for Economic Co-operation and Development (OECD) member countries, we identified considerable public spending differences on social welfare policies and financial resources across countries. The Scandinavian countries (specifically Sweden) have significant tax and social insurance contribution expenses. However, the government provides substantial social security and is positioned as a 'big government.' Meanwhile, Anglo-Saxon countries, such as the United States and the United Kingdom, have formed market-based security systems that emphasise self-responsibility and are positioned as 'small government. ' Considering the welfare state's establishment and development process, based on Esping-Andersen's (1990,1999) theory of welfare regimes, we believe that the differences in each country's historical and cultural backgrounds led to the establishment of the various systems of the welfare states. He criticises the method of measuring the welfare state's degree of development by using public social expenditure; the theory of welfare regimes also created a de-commoditisation index using benefit levels and receipt requirements. The welfare regimes theory classified the welfare state into three types in decreasing order of de-commodification score: 'social democratic regime,' 'conservative regime,' and 'liberal regime.' 2 The three types of welfare states were, in our perspective, pioneering impact studies on the welfare state. According to Fazeli and Fazeli (2012) and Isakjee (2017), Esping-Andersen's insights provide implications for the arguments for the variety of capitalism made by Hall and Soskice (2001) and Amable (2003) 3 . These studies show that there are different types of capitalism in different countries. The idea of variety of capitalism differs from the considerations of Wilensky (1975). He argued that the factor of economic growth influences social security expenditures and that this expenditure converges by each country's stage of development.
Apart from Esping-Andersen (1990, 1999, another study that led to the diversity of capitalism is the legal origin theory proposed by La Porta et al. (1998Porta et al. ( , 2000Porta et al. ( , 2002 and La Porta et al. ( , 2013. The legal origins are categorised into English, French, German, and Scandinavian 4 . According to La Porta et al. (1998Porta et al. ( , 2000Porta et al. ( , 2002, the legal origins influence the institutions of creditor protection, minority investor protection, degree of law enforcement, and the structure of stock ownership in each country. The legal origin theory suggests that each country's civil, commercial, and securities laws differ based on legal origins. Ergungor (2004) characterised each country's financial markets as bank-based or market-based based on legal origin. Table 1 compares the classification of welfare regime theory with legal origins. Except for Switzerland, classified as a liberal regime, the classification of the welfare regime theory and legal origin theory is the same. Therefore, legal origins can be considered to influence the formation of the welfare state.
We examine the rule of law concerning legal origins 5 . The rule of law is a fundamental principle of the Anglo-American legal system that eliminates rule by tyranny and restrains power by law. According to the World Development Report (World Bank, 2017), the development of the rule of law requires the following three points: First, the shift from an informal and pluralistic system of law to a unified modern one. Second, acceptance of legal constraints on their power by powerful elites, and third, a country's successful adaptation of foreign legal systems for their purposes. Therefore, the rule of law is a principle based on constitutionalism, which aims to guarantee the rights and freedom of the governed by eliminating the rule of people and binding all governing powers by law. It is intricately linked to liberalism and democracy. Therefore, the degree of the rule of law may vary depending on the legal origin. Moreover, we assume that the degree of the rule of law is linked to the degree of quality of social security in each country.
This study shows that the degree of social security expenditure depends on the rule of law, which depends on legal origins. We also investigate the dependence of social security expenditure on the legal origins concerning the rule of law and its impact on economic inequality. This study shows that differences in legal origins lead to different welfare states, leading to a diversity of capitalism. We recognise the importance of a social environment, adaptable with legal origins, and criticise to introduce policies of other countries without considering their legal origins.

Composition of Social Security Expenditures
This subsection analyses the component factor of social security expenditures to explain the dependent variables used by the regression analysis. The decomposing ratio of social security expenditures to the gross domestic product (GDP) into 'Ability of tax revenue,' 'Fiscal soundness,' and 'Fiscal sustainability' results in Equation (1).
where i is the country, and t is the year. Equation (1) shows the elements of the financial resources included in the ratio of Social Security expenditures to GDP. Social/GDP, the ratio of social security expenditure to GDP, is an indicator of the size of social security expenditure to the scale of the economy. Social/Tax, the ratio of social security expenditure to tax revenue, is an index showing the ability of tax revenue to disburse social security expenditure. Tax/debt measures the amount of public spending left over from past generations or the fiscal soundness in its current stage. The index measures the extent of government spending exceeding tax revenue over the years until the present. Debt/GDP, the ratio of debt to GDP, indicates the sustainability level of national finance.
In this study, we use the four types of variables included in Equation (1): the size of social security expenditure, the ability of tax revenue, fiscal soundness, and fiscal sustainability, as the dependent variables in the regression analysis. We estimate the relationship between these four types of variables and the rule of law considering each legal origin.

Economic Inequality
This subsection reviews the relationship between social security expenditure and economic inequality 6 . In a welfare state, income redistribution is implemented through social security expenditures, taxes, and social insurance premiums to correct economic inequality. According to Palda (1997), there is higher social security expenditure and higher tax revenues in a big government, however, most of the income is redistributed to the middle class. The redistribution of income to low-income earners is no more effective than in the case of a small government. Tanzi and Schuknecht (2000) investigate the income distribution by income class in 13 developed countries. They compare the redistributive function of taxes and social security with big government and small government. Consequently, they question the redistributive function of the big government. Although the welfare state corrects economic inequality, they report that the effects of income redistribution are not always equal. Korpi and Palme (1998) analysed 18 industrialised countries to investigate their social security and redistributive functions. In Sweden, Norway, and Finland, the scale of social security expenditures and benefits is large, and the effect of income redistribution is high. Contrarily, liberal regimes, including the United States and the United Kingdom, report a small scale of social security expenditure and no apparent effect on income redistribution. Tridico (2017) examines inequality theoretically and finds that while financialisation promotes inequality, public expenditure not only contributes to the reduction of inequality but also promotes economic growth. Source: Data for legal origin are referenced from La . Social/GDP is the ratio of public social security expenditure to GDP. The data source of Social/GDP is published by OECD Social expenditure statistics. The Rule of Law is published by the Worldwide Governance Indicators of The World Bank Group (2020). Note: Target countries are 36 OECD countries. English, French, German, Scandinavian denote type of legal origin. The data for each country is the average of the data from the years 2004 to 2018. The missing values used the data of the previous year. If the previous year's data also did not exist, it was treated as a missing value. The data used is unbalanced panel data.
Prior research suggests that income redistribution by the welfare state has different effects on big government and small government. Therefore, this study investigates the link between the size of social security expenditures and economic inequality. In the process, the analysis considers the impact of legal origins.

Data and models
This section explains the dataset and descriptive statistics used in this study. The data was collected from 36 OECD member countries, and Figure 1 summarises the country names, legal origins, Social/GDP ratio, and the Rule of Law. Social/GDP and Rule of Law for each country are based on the arithmetic mean of relevant data from 2004 to 2018. The classification of legal origins is per La   7 . Figure 1. indicates that the average of the rules of law is relatively high in the case of Scandinavian and English origins. However, it is relatively low in the case of French and German legal origins. Table 2 lists the variable definitions and data sources. We collected data from the OECD's (2020b) Social Expenditure Statistics, World Bank's World Development Indicators (Kaufmann & Kraay, 2020), World Bank Group's (2020) Worldwide Governance Indicators, and IMF's (2020) World Economic Outlook Database October 2019 to develop the variables required for analysis. Table 3 shows the descriptive statistics of the data used. In cases where data were missing, we used the value of the previous year. If there were no data for the previous year, it was considered a missing value.
First, we reviewed the relationship between Social/GDP and the Rule of Law in using a simple regression model, as shown in Figure 2. The value by country is the average of values from 2004 to 2018. The regression model was estimated using Ordinary Least Squares. The number of observations is 36. According to the estimation results in Figure 2, the relationship between Social/GDP and Rule of Law is positive and is significant at the 1% level. Figure 3 represents the variables in Equation (1). According to Figure 3, the social security expenditure among English origin countries is relatively low, and tax revenue is also low. The debt of countries with English legal origins is the same as that with German legal origins, but the Gini index tends to be high. As countries with French and German legal origins have high social security expenditure against tax revenues, they tend to have higher debt. However, countries with French legal origins exhibit more significant economic inequality than those with German legal origin. The social security expenditure and tax revenue levels are also high for countries with the Scandinavian legal origin; therefore, debt and economic inequality is low in this case.  (2020), OECD (2020a, 2020b), The World Bank (2020) and The World Bank Group (2020) in Table 2.
Based on the above, does the effect of the rule of law depend on legal origins? The following regression model was developed to confirm whether the rule of law differs depending on legal origins.
where i is the country, t is the year, e is the error term, and Control Variables1 includes the GDP per Capita and Population 8 . Additionally, this model includes a dummy variable for the year. As the legal origins dummy variable does not change during the target period, it cannot be measured by the fixed-effect model performed  after completely removing the effects of the time-invariant covariates. Therefore, the random effects model is used to estimate the model represented in Equation (2). The following estimations are made to analyse the relationship between the components of social security expenditure and the rule of law derived from the legal origins.
where, Social/GDP, Social/Tax, Tax/Debt, and Debt/GDP are the dependent variables, and u is the error term. Control Variables2 contains the variables Pension, Over65, GDP, Unemployment, and Inflation. We investigate the relationship between dependent variables and the interaction term. This model estimates whether the country's legal origin affects the rule of law, affecting public social security expenditure. We estimate this regression model with a random effects model. We must consider the ratio of the elderly population and pension funds as factors that affect social security expenditure. Considering the legal institution, differences in pension systems and tax systems affect the financial resources of social security expenditure. However, the primary sources of social security expenditure are tax revenue and pension insurance premiums. Moreover, the economic scale, unemployment rate, and inflation rate affect tax revenues. This model also includes year dummy variables.
Finally, we investigate the relationship between economic inequality and public social security expenditure by legal origins.
where i is the country, t is the year, and e is the error term. Here, the GINI index is the dependent variable, and Control Variables3 is the unemployment rate. We also insert the interaction term in Equation (4). This is because it is possible to estimate whether the difference in the legal origins that countries have adopted in the past will affect the level of social security expenditure, which will affect economic inequality. Here, as economic inequality is based on income inequality, the unemployment rate directly affects economic inequality. We also estimate this regression model with a random effects model 9 .

Estimation results and discussion
In this section, we explain the estimation results from Equations (2) to (4). Table 4 summarises the estimation results of Equation (2). The model excludes the legal origin, which is the base category of the dummy variable. The model uses balanced panel data without any missing variables. Accordingly, the number of observations is 540.
When compared with countries with English legal origin, the coefficient of French is negative and significant at the 1% level, and that of German is negative and significant at the 5% level. Neither the English and Scandinavian origin nor the French and German origin comparisons are significant. Compared with French, and German origins, the coefficient of Scandinavian is positive and significant at the 1% level in both cases. Table 4 indicates that the English origin has a better institution of the rule of law than the French and German origins. The English origin is statistically no different in the level of the rule of law from the Scandinavian origin. French and German origin have almost the same level of the rule of law. Scandinavian origin has a better institution of the rule of law than French and German origins. The rule of law is not developed in legal origins, which has a civil law background. Tables 5 and 6 show the estimation results of Equation (3). The dependent variables considered in these tables are Social/GDP and Social/Tax. The description of dummy variables is the same as in Table 4. The model uses unbalanced panel data with 530 observations 10 . In the model with Social/GDP as the dependent variable, all coefficients of the Rule of Law are positive and significant at the 1% level. In the English origin-based model, the coefficient of the interaction term on the Rule of Law Â French is negative and significant at the 1% level. Similarly, the coefficient of the interaction term on the Rule of Law Â German is negative and significant at the 1% level. However, the coefficient of the interaction term on the Rule of Law Â Scandinavian was not significant.
In the French origin-based model, the coefficients of the interaction terms Rule of Law Â German and Rule of Law Â Scandinavian are not significant. The coefficient of the interaction term Rule of Law Â Scandinavian is positive and significant at the 10% level. All coefficients of Rule of Law are positive and significant at the 1% level in the model with Social/Tax as the dependent variable. In the English origin-based model, the coefficient of the interaction term on the Rule of Law Â French is negative and significant at the 5% level. In the same model, the coefficients of the interaction term on the Rule of Law Â German and the Rule of Law Â Scandinavian are not significant.
In the French origin-based model, the coefficient of the interaction term, Rule of Law Â German, is not significant. However, the coefficient of the interaction term, Rule of Law Â Scandinavian, is positive and significant at the 1% level, while that of Rule of Law Â Scandinavian is positive and significant at the 5% level. Table 7 shows the marginal effect of the interaction term estimated in Table 5. In legal origins with higher marginal effects, the dependent variable tends to increase as  Table 6 shows the model's estimation results using Tax/Debt and Debt/GDP as the dependent variables. For the model using Tax/Debt as the dependent variable, the coefficient of the Rule of Law in the German origin-based model is negative and significant at the 1% level. In the English origin-based model, the coefficient of the interaction term on the Rule of Law Â French is not significant, while that of the Rule of Law Â German is negative and significant at the 1% level. The coefficient of the interaction term on the Rule of Law Â Scandinavian is not significant. Note: ÃÃÃ , ÃÃ , Ã , each significant at the 1%, 5%, 10% level. The value in parentheses is the t value.
Year is a year dummy if the dummy included in the model are described as Yes. Obs. means the number of observations. The random-effects model is used for this estimation. Since the dummy variable that indicates the legal origin does not change during the target period, it cannot be measured by the fixed-effect model that is performed after completely removing the effects of the time-invariant covariates. Source: Prepared by the author.
The coefficient of the interaction term on the Rule of Law Â German is negative and significant at the 10% level in the French origin-based model. However, the coefficient of the interaction term on the Rule of Law Â Scandinavian is positive and significant at the 10% level while that of Rule of Law Â Scandinavian is positive and significant at the 1% level. Table 8 shows the marginal effect of the interaction term estimated in Table 6. In legal origins with higher marginal effects, the dependent variable tends to increase as the Rule of Law increases. In the model where Tax/Debt is the dependent variable, the marginal effect of the interaction term is 0.585 for Scandinavian origin, 0.286 for English origin, À0.114 for French origin, and À0.490 for German origin. In the model where Debt/GDP is the dependent variable, the marginal effect of the interaction term is 0.478 for German origin, 0.179 for French origin, À0.182 for English origin, and À0.690 for Scandinavian origin.
Next, we discuss the estimation results from Tables 5 to 8. The scale of social security tends to be high, especially in countries where the rule of law is high and English and or Scandinavian legal origins are adopted. The ability to cover social security expenditure using tax revenue is lower in countries where the rule of law is Note: The marginal effects were calculated using the sum of the Rule of Law coefficient and the interaction term coefficient. If neither the coefficient of the interaction term nor of Rule of Law is significant, it is denoted as Not significant. In this table, the model selected has a highly significant level for the coefficient of the legal origin or the interaction term. The column of Country describes the legal origin, which has a significant level of higher marginal effect. If both the coefficient of interaction term and of Rule of Law are significant at the 5% level or below, it is denoted as a . Source: Prepared by the author. Note: The marginal effects were calculated using the sum of the Rule of Law coefficient and the interaction term coefficient. A column of Country describes the legal origin, which has a significant level of higher marginal effect. If neither the coefficient of the interaction term nor of Rule of Law is significant, it is denoted as Not significant. In this table, the model selected has a highly significant level for the coefficient of the legal origin or the interaction term. When the coefficient of interaction term and Rule of Law are significant at the 5% level or below, it is indicated as a .
When only the coefficient of the interaction term is significant under the 5% level, it is indicated as b .
high, and Scandinavian or English legal origins are adopted. By comparing the tax revenue and debt, we identify countries that adopted English and Scandinavian legal origin, where the rule of law is high, and debt tends to be low. Moreover, from the perspective of fiscal sustainability, there is a tendency for debt to be higher than GDP in countries where the rule of law is higher with German and French origin. Table 9 shows the estimation results of Equation (3). The description of dummy variables is the same as in the previous model. The model uses unbalanced panel data with 527 observations and uses GINI as the dependent variable. The results show that all coefficients of Social/GDP are negative, and the English origin-based model is significant at 10%, the French origin-based model is significant at 1%, and the German origin-based model is significant at the 1% level. In the English origin-based model, the coefficient of the interaction term on Social/GDP Â French is negative and significant at the 1% level. In contrast, the coefficients of the interaction terms related to German and Scandinavian are not significant in the same model. In the French origin-based model, the coefficient of the interaction term on Social/GDP Â German is positive and significant at the 1% level while that of Social/GDP Â Scandinavian is positive and significant at the 5% level. Nonetheless, the coefficient of the interaction term, Social/GDP Â Scandinavian, is not significant in the German origin-based model. Table 10 shows the marginal effect of the interaction term estimated in Table 9. In legal origin with a higher marginal effect, the dependent variable tends to be higher as Social/GDP increases. The marginal effect of the interaction term is À0.025 for English origin, À0.026 for German origin, À0.043 for Scandinavian origin, and À0.091 for French origin. The estimated results of Tables 9 and 10 are interpreted here. Increasing social security expenditure corrects economic inequality across all legal origins. Especially in countries that adopted French legal origins, high social security expenditure ameliorates economic inequality. In contrast, an increase in social security expenditures of English legal origin countries contributes little to resolving economic inequality. Figure 4 presents a graph to clearly compare the marginal effects calculated by each regression model. As seen in Figure 4, the increase in social security expenditure could lead to higher taxes in the English legal origins and a fiscal deterioration in the French and German legal origins. Furthermore, while an increase in social security expenditure reduces economic inequality, their marginal effect varies with the legal origins. Note: The marginal effects have calculated the sum of the Social / GDP coefficient and the interaction term coefficient. The column of Country describes the legal origin, which has a significant level of higher marginal effect. If neither the coefficient of the interaction term nor of Social / GDP is significant, it is denoted as Not significant. In this table, the model selected has a highly significant level for the coefficient of the legal origin or the interaction term.
If both the coefficient of interaction term and Social / GDP is significant at the 5% level or below, it is indicated as a . Source: Prepared by the author.

Conclusion and implication
This study analyses data from 36 OECD countries and finds that social security expenditures and its components depend on the degree of the rule of law, which depends on legal origins. We find that the impact of social security expenditure on economic inequality depends on each country's legal origins. Especially in countries where the degree of the rule of law is high, the size of social security in countries that adopted English and Scandinavian legal origins tends to be higher. The ability of tax revenue is lower and fiscal soundness (ratio of tax revenue to debt) is higher in countries with a high degree of the rule of law and have adopted Scandinavian and English legal origins. This indicates more tax revenue and less debt in these countries.
In terms of fiscal sustainability, countries with a higher degree of the rule of law and German and or French legal origins are likely to have a higher debt to GDP ratio. Concerning all legal origins, increasing social security expenditure has the effect of correcting economic inequality. Particularly in countries that adopted French legal origins, high social security expenditure further ameliorates economic inequality. In contrast, increases in social security expenditures of countries with English legal origin do not improve economic inequality to a large extent. Therefore, a government official needs to perceive the factors of their legal origin and the extent of the rule of law. It is then necessary to estimate how much social security expenditure is needed to reduce economic inequality. If the legal origins and or degree of the rule of law are different, the required social security contributions would be different to correct economic inequality.
A limitation of this study is that the model is estimated based on the historical indicator of the legal origin. However, it is considered that the impact of path dependency affects modern public spending and the rule of law. Therefore, the rule of law is influenced by legal origins, influencing public spending, and economic inequality. The suggestions in this paper are as follows. Depending on the legal origins, an increase in social security expenditure has different effects on reducing economic inequality. This phenomenon occurs because different legal origins have different levels of the rule of law. This difference suggests that increasing social security expenditure without considering the legal origins and the level of the rule of law may have little effect. Conversely, the improvident policy may lead to a deterioration of public finances and an increase in taxes, which may not have the desired effect.
We recognise the importance of a social environment, adaptable with legal origins, and criticise to introduce policies of other countries without considering their legal origins. Notes 1. According to Wilensky (1975), a country is considered a welfare state if its social security expenditures exceed a certain percentage of its GDP. Additionally, the study examines the convergence theory of public spending on social welfare programs, which states that no matter what political and economic regime, cultural and historical background, or affluent society a country has, its social structure will become similar. Beblavy (2010) showed empirically that economic development is the main factor of increase in social expenditure.