“Assessment of the social expenditure impact on the economic growth in OECD countries”

Economic growth is exposed to many socio-economic factors that impact both the formation and allocation of resources. The theoretical part of this article discusses studies by various authors on the social expenditure impact on economic growth, the dependence of this influence on selected funding principles and social policy models. In the empirical part, using the Pooled Mean Group (PMG) procedure and the Fixed Effect Model, the impact of social expenditure on the economic growth in OECD countries is determined. An increased focus is put on assessing the long-term impact of the main types of social expenditures (public and private), based on different financing principles (distribution and accumulation), on the economic growth rates both in OECD in general and in the context of countries (based on the Esping-Andersen’s typology) grouped according to social policy models. The following conclusions are drawn: 1) an increase in the share of total social expenditures in the country’s GDP negatively affects economic growth; 2) an increase in the share of private social expenditures in the country’s GDP contributes to economic growth; 3) the obtained indicators of impact assessment are different depending on a social policy model chosen. The analysis is based on OECD panel data for the period 1980–2013. Mykhaylo Malyovanyi (Ukraine), Nataliia Ivanova (Ukraine), Kateryna Melnyk (Ukraine), Oleksandr Nepochatenko (Ukraine), Oleksandr Rolinskyi (Ukraine) BUSINESS PERSPECTIVES LLC “СPС “Business Perspectives” Hryhorii Skovoroda lane, 10, Sumy, 40022, Ukraine www.businessperspectives.org Assessment of the social


INTRODUCTION
Social protection of the population is one of the most important areas of any state activity, aimed first of all at stable functioning of society by protecting certain segments of the population from various socio-economic risks (unemployment, illness, poverty, old age, etc.). For that end, the state develops a highly-structured integral system of social protection, the elements of which differ between themselves in terms of operating environment and purpose. This enables to embrace population, which needs support from the state, as much as possible.
The well-developed social protection system makes it possible to effectively deal with poverty, which is one of the main problems of the world community. Therefore, in a postindustrial society based on a developed market economy, social protection plays an important role in ensuring a stable and steady development of the economy. The economic and social components of the state are closely interconnected and affect one another. At the same time, science and practice prove that social expenditures (the total cost of financing all social benefits and programs in the country within the system of social protection of the population financed by public and private funds) should be harmoniously related to the economic indicators of the country, which

Social costs and economic development relationship
The financial support for social protection is closely linked to a number of socio-economic factors, the change of which may directly or indirectly affect the sources of its financing. Therefore, as rightly pointed out by Cichon et al. (2004), the development of financial provision must necessarily take place taking into account the potential links between social protection and the economy. This allows not only to build a system of social protection balanced in terms of income and expenditures, but also to ensure the gradual socio-economic development of the entire country.
Social protection of the population is always closely related to the economy and its main determinants (labor market, wages (income), labor pro-ductivity, investment level, economic growth, etc.). At the same time, more and more scientists consider the social protection system not only from a social point of view, but also as a mechanism to stimulate the country's economic development. This requires a deeper analysis of the interconnection between socio-economic factors and social protection of the population, especially in the context of the development and use of financial resources. That said, this connection, as evidenced by the study, is of a mutual nature. That is, not only socio-economic factors affect social protection, but also its tools (various social services and benefits) exert pressure on the dynamics of certain factors. At the same time, this influence can be both positive and negative.
Researching this problem by various scientists ( Tridico & Meloni, 2018) made it possible to outline the principal tools, by using which the social protection system can potentially influence factors determining economic growth in the country (Table 1). Table 1 outlines only main tools, through which social protection directly or indirectly influences economic growth. The use of particular instruments has a positive or negative impact on the economic factors that shape the total economic impact. Thus, in particular, increasing the size of taxes or social contributions that are the source of financial resources for social protection will negatively affect the labor market (for example, due to the growth of its shadow component) and the amount of business profits, which ultimately reduces the size of investment resources for the economy and reduces revenues both to the budget system and to various state and non-state social insurance funds. Quite often, private investors are not encouraged to invest in regions where high social security contributions are applicable (Cichon et al., 2004).
At the same time, various instruments of social protection have a positive effect on the economy development: 1) the growth of volume and size of insurance premiums in the accumulation system of social insurance reduces the financial burden on public finances and stimulates the growth of investment in the economy; 2) the growth of social transfers contributes to rising incomes and poverty reduction, which ultimately, through the stimulation of aggregate demand, has a positive impact on the country's economic development; 3) reducing inequalities in income, increasing the efficiency of health services, increasing fertility contribute to improving human capital, which is the basis for the steady development of any country's economy; 4) active labor market policies contribute to reducing the number of unemployed persons, creating new jobs, increasing labor mobility and increasing labor productivity, etc.
It should be noted that some socio-economic factors that are catalysts of a positive impact on the country's economic development may often have a negative impact on other socio-economic determinants, which, in some cases, offset their positive effect. Therefore, in many scientific studies, the impact of social protection on economic growth is rather ambiguous.  Kostadinova, 2014), which involve the systematization of social policy models or typologies of countries based on social policy models, are largely based on the study by Esping- Andersen (1990).
Summarizing the existing scientific approaches to systematization of social policy models, the  It should be noted that the countries included in this group are at the stage of shaping their social protection systems, and therefore can change the type of social model in the near future. For example, the social system of Estonia, by its characteristics, already tends towards the Anglo-Saxon (liberal) model of social policy. 2 The Netherlands has previously been referred to the continental European social model, but now the country has some features (e.g., labor market characteristics) that resemble more closely the Scandinavian countries.
authors of the current article have distributed 33 OECD countries by six types of social models ( Table 2).
Despite the variety of social policy models common in OECD countries, the financial support for their social security systems can always be defined as a system of monetary relations of a distributive nature, in which, through taxes and special contributions, centralized and decentralized funds of future financial resources are formed for social benefits financing. The relations arising in the system of social protection with regard to the formation and use of financial resources are undoubtedly of a distributive nature. Therefore, under expanded reproduction, support for unemployable members of society, as well as those who need material or other assistance, is possible only at the expense of part of the added value or gross domestic product, which is distributed in their favor. Moreover, despite various conditions for the social protection systems functioning in the OECD countries, all of them are based on two basic principles of the financial resources development and use -distributive and accumulative ones.
In the current context of the world economy development and socio-demographic changes, the question arises: How to combine these principles and schemes of financing to preserve the existing level of social protection of the population on the one hand, and not to harm the country's economic development, on the other?
In order to describe the overall financial equilibrium of the social protection system, Cichon   The observance of this equilibrium in the medium and long term is violated by a number of factors that can negatively affect both via the reduction of money formation sources and through the costs increase: economic (economic growth, employment rate, wages, inflation rate, interest rate); demographic (population structure, active population, mortality and fertility rate, life expectancy); administrative or political (the structure of the social protection system, the size of the basic social benefits, the number and structure of the social assistance recipients), etc.
It should be also noted that the influence of any factor depends on the existing social model. Therefore, in order to minimize the impact of various risks (demographic, economic, social, and political) most countries now reform their social protection systems and seek a balance between the two principles of financial provision for social protection. This is especially true due to the demographic problems existing in most OECD countries. Complex demographic processes in the OECD countries are accompanied by a steady increase in the share of social spending in GDP (for example, in Italy, this share increased from 10.3% in 1980 to 30.0% in 2013, and in the UK, from 19.0% to 27.8%) (OECD Stat, 2018), which requires more and more additional financial resources each year to meet social needs, subject to a reduction in the main sources of funding.
Influenced by economic and demographic changes, countries gradually reformed their social protection systems and chose two approaches: 1) increasing the efficient functioning of the existing social protection system without more extensive use of the accumulative financing (private sector) (Austria, Belgium, France, Finland, Greece, Italy). In these countries, the share of public expenditures in total social costs has not only decreased, but has also increased; 2) expansion of the (private) accumulative financing principle, accompanied by a reduction in the share of public expenditure in total social expenditures (Canada, Chile, Iceland, Israel, the Netherlands, Sweden, Switzerland, United Kingdom, USA).
Under current conditions of economic, social and demographic development of the OECD countries, the increase in the share of private social expenditures financed by financial sources, which is formed based on the accumulative principle of financing, has a positive effect on the country's economic development. This is evidenced by the results of the analytical grouping, which made it possible to identify the relationship between the share of private social expenditures in total costs and the GDP per capita in OECD countries for the period from 1980 to 2013 (Table 3).
The analytical grouping results indicate that there is a correlation between the share of private social expenditures in the total expenditures and GDP per capita. The obtained empirical correlation between these factors was 0.413. η = This indicates that the connection between Y and the factor of X is moderate. To confirm the connection received, the statistical significance of the linkage strength indicator was checked. In order to check the null hypothesis that general correlation coefficient (empirical correlation relation) of a normal twodimensional random variable equals 0, with the competing hypothesis That is, the 17.06% variation in the effective indicator (regression coefficient) is due to the differences between the characteristics, and 82.94% is due to other factors.
However, this analytical grouping does not provide an opportunity to substantiate the positive effect of the increase in private social expenditures (based on the accumulative principle of financing) on economic growth, taking into account other factors (as evidenced by the resulting coefficient of determination) and the peculiarities of social models used in OECD countries. Therefore, in order to objectively and reasonably assess the impact of social expenditures (both public and private) on the economic growth and to define promising directions for the social protection systems development in the OECD countries, there is a need for a deeper and formalized analysis of this scientific challenge.

RESEARCH AIM
The purpose of this study is to assess the impact of social expenditure on the economic growth both in general and in terms of its types (public and private) in OECD countries, taking into account the social policy model features.

Description of the economic growth model
The calculations made in this research are based on the neoclassical model of economic growth, developed by Robert Solow (1956), an American economist, and independently, by the Australian economist Trevor Swan (1956). It was built on the basis of an aggregative two-factor Cobb-Douglas production function. The model uses indicators that change significantly only in the long run as the labor productivity (RL) dependence on the capital employment ratio (E) (Grodsky, 2015).
In the context of scientific and technological progress in the economy, the Solow model could not simply be based on new data on the elasticity indicator in the production function and, therefore, since the 1980's, attempts have been made to develop a theory of growth towards creating endogenous models that contain internal impulses of increased returns of production factors. Paul M. Romer (1986), an American economist, began to consider three-factor production function, instead of two-factor one, with labor power, capital goods and "knowledge". Another American, Robert E. Lucas (1988), proposed to consider the "human capital" factor as the result of learning in the production process. Mankiw, Romer, and Weil (1992) have developed a growth model (MRW model), containing the "human capital" factor, which made it possible to formally consider it as an economic development model.   Additionally, a basic and expanded model of economic growth for groups of countries is calculated based on their social model (OECD countries are grouped according to social models in Table 2).

Data
Economic and statistical surveys were conducted in 33 OECD countries (except for Chile and Turkey) for the period 1980-2013 (902 observations) 3 . This period was chosen taking into account two factors: 1) the availability and completeness of all data necessary for econometric calculations (the main factors were chosen based on the economic growth model proposed by Bassanini et al. (2001); 2) this period is characterized by a gradual shift away from the concept of the "welfare state" and the gradual growth of the accumulative financing in the social protection systems of the OECD countries.

Variables used in empirical studies
Basic variables used in the empirical analysis: Additional variables included in the empirical analysis to assess the social expenditure impact on the economic growth:

Substantiating the econometric research technique
In this study, panel data are used for econometric calculations. Thanks to the special structure, the panel data allow to build more flexible and meaningful models and getting answers to questions that are not only accessible within, for example, spatial data-based models. There is an opportunity to take into account and analyze the individual differences between economic units, which cannot be made as part of standard regression models. Often, unobservable factors are correlated with other variables. As part of regression models, this means that the unobservable factor is a significant variable in the model and its exclusion leads to shifting estimates of other parameters. In other words, panel data models allow for more accurate parameter estimation (Grodsky, 2015).
To calculate panel data, the following economic and statistical methods are usually used: • Pooled OLS Regression; • Fixed Effect or LSDV Model; and • Random Effect Model (Random Effect Regression).
When working with panel data, there is always a problem which model (pooled regression, fixed or random effect model) should be selected. At the content level, the difference between models can be interpreted as follows. The pooled model assumes that economic units do not have individual differences, and in some simple situations such an assumption is justified. In a fixed effect model, it is assumed that each economic unit is unique and cannot be considered as the result of random selection from a certain general population. This approach is fair when it comes to countries, large regions, industries, and large enterprises. If the objects fell into the panel as a result of a sample from a large population, then a model with a random effect is acceptable. Small firms, households, and individuals may provide an example. It should, however, be emphasized that in similar situations (especially for a small number of economic units), there may be a question of the existence of individual differences (Baltagi, 2005).
There are statistical criteria that can partially solve the problem of choosing a model using standard hypothesis testing techniques. To select the model that is most adequate for the analyzed data, a pair comparison of the estimated models has been carried out. The following results were obtained using the EViews 9 program (Tables 4-6).
The determination coefficient in the pooled regression model is negligible (R-squared -0.018676) to assume that the chosen model properly describes the relationship between economic growth and selected factors. In addition, the pooled regression model received negative coefficients at the human capital logarithm, which does not correspond to the socio-economic nature of the indicator.
The Hausman and Wald tests yielded following results:   Bassanini et al. (2001), this method allows to minimize the disadvantages of the above-mentioned methods and taking into account the impact of socio-economic factors in the long run, considering the particularities of the countries included in the analysis.
It should be noted that none of the economic-statistical methods makes it possible to eliminate all problems arising in the panel data analysis. Therefore, in order to obtain more reliable results, two methods of panel data analysis are used in this research: The Fixed Effect Model and the Pooled Mean Group (PMG) procedure. Table 7 presents the calculation results of the of social expenditure impact on the economic growth using the Pooled Mean Group (PMG) procedure. Impact assessment is carried out by gradual expansion of the model through the inclusion of new factors (net total social expenditure, public social expenditure, and private social expenditure). At the same time, calculations were carried out both for the entire population of the country and only for the able-bodied population (aged 15 to 64).

RESULTS
In general, the model effectively characterizes the socio-economic factors influence on the economic growth, the main factors are significant and the received characteristics correspond to their socio-economic content. It should be also noted that positive coefficients have been obtained in the model that characterize the effect of the growth rate of the able-bodied population on the pace of economic growth per capita, which is not expected. As Arjona et al. (2002) note, "higher population growth negatively affects the GDP per capita increase". However, these coefficients were statistically insignificant, and, therefore, the impact of this indicator on the economic growth is rather ambiguous. The coefficients of the general and public social expenditure growth are negative and significant, while those for the private expenditure growth are positive, which in general confirms the hypothesis on the positive impact of the increase in the share of social expenditures, financed from sources, which are formed according to the accumulative principle of financing.
Approximate coefficients for the selected indicators have been obtained using also another research method -Fixed Effect Model. It should be also noted that when using the Fixed Effect Model, all variables were obtained with expected signs, although in this method, the growth rate of the able-bodied population is a negligible factor ( Table 8).
The results obtained using two evaluation methods showed a negative impact of the growth rate of social expenditure on the economic growth rate. An increase in the social expenditure growth rate by 1% in the long-term perspective will reduce the economic growth rate by 0.05-0.08%. That is, if the share of total social expenditure in OECD countries' GDP grows from 21.1% (2013 figure) to 22.1%, GDP growth will decrease by 0.38%. At the same time, the increase in the share of investment in GDP by 1% leads to GDP growth by 0.2%, and the increase in the number of schooling years by 1 -by 0.69%.
With that, the calculation results showed a positive effect of the growth rates of private social expenditure on the economic growth compared to the state social expenditure. So, if the increase in the rate of state social expenditure by 1% leads to an economic growth reduction by 0.084-0.189% (the increase in the share of public social expenditure in the OECD's GDP from 21.1% in 2013 to 22.1% will reduce the economic growth by 0.89%), the growth of private spending by 1% leads to an acceleration of economic growth by 0.006-0.011% (if the share of public social expenditures in OECD GDP is increased from 2.6% in 2013 to 3.6%, then it will affect the GDP growth by 0.42%).
The results on the impact of public and private social expenditures can be interpreted through differences in approaches to both the formation and use of social protection funds. In the first case (public social expenditure), the formation  Note: Standard errors are in parentheses. * statistical significance level is * -10%; ** -5%; *** -1 %. The results of the Pooled Mean Group (PMG) procedure are calculated for both short and long-term dynamics, but the coefficients included in the table describe the long-term dynamics. and use of financial resources for social protection negatively affects economic development. Funds are formed mainly at the expense of either insurance premiums in compulsory state social insurance funds or at the expense of taxes and dues to the budget system. All this creates a tax burden on business, which manifests itself both through the reduction of financial resources for investment activities in private structures and through the informal (shadow) economy formation. At the same time, the current proceeds are predominantly intended to finance current social expenditures.
In the second case (private social expenditure), the formation of financial resources though carried out at the expense of contributions, which increases the cost of business structures, but the social expenditure financing is carried out at the expense of accumulated funds. Due to this principle of financing, considerable financial resources are being drawn up, which are aimed at investing. Thus, for example, the total assets of the OECD NPF increased from 104.1% of GDP in 2006 to 125.7% of GDP in 2016. In terms of value, the total assets of the OECD NPF increased from USD 25.34 trillion in 2006 to USD 38.14 trillion in 2016 (OECD, 2017).
The calculations performed using the Pooled Mean Group (PMG) procedure allowed to assess the impact of public and social expenditures on social policy models ( Table 9).
As a result of the calculations, the following findings were obtained: with an increase in the share

CONCLUSION
Modern social protection systems in OECD countries need to be reformed because of the social and economic problems facing the world community in recent decades (the economic growth slowdown, complex demographic situation accompanied by population aging, international challenges, in particular, mass migration, etc.   Note: Standard errors are in parentheses. * means statistical significance level at 10 ; ** -at 5%;*** -at 1%. The Pooled Mean Group (PMG) procedure results are calculated both for short-and long-term dynamics, but the coefficients included in the table describe long-term dynamics.
that, different approaches to the development and use of financial resources in the social protection system have been developed -in some countries the share of public expenditure prevails, while in others, it is private.
Using economic and mathematical methods it was revealed that an increase in the share of social expenditure in the country's GDP negatively affects the economic growth rates. At the same time, the increase in the share of private social expenditure (funds that are formed using the accumulation principle) contributes to the economic growth, while the increase in the share of public expenditure (the formation and use of funds is mainly based on the distribution principle) negatively affects economic growth in the long run. It has been found that the level of influence of social expenditure on economic growth varies depending on the chosen social policy model. With that, it was found that the greater the share of public expenditures in the total social costs, the higher the size of the estimated coefficients.
For study, the econometric methods of the Pooled Mean Group and the Fixed Effect Model have been used, which, according to the authors calculations and the opinion of other scientists (Bassanini et al., 2001;Arjona et al., 2002), are most suitable for the analyzed data set. The calculations were carried out based on the neoclassical model of economic growth, which included, in addition to social expenditure, investment, human capital, labor, and convergence coefficient. Using data from the OECD countries for the period from 1980 to 2013 makes it possible to assume that the results obtained are sufficiently significant and objectively and actually characterize the relationship between social expenditure and economic growth.
The results obtained suggest that in the long term, OECD countries need to stabilize the share of social expenditure in GDP (and reduce it if possible) and develop private social protection systems based on the accumulation principle of funding.
According to the authors, further studies on this issue should focus on assessing the impact of a particular social policy model on achieving the main objectives of the social protection system: reducing poverty, reducing income inequality, etc. Also, to justify the most optimal combination of public and private social expenditures, it is necessary to expand the range of factors included in the proposed model of economic growth.