In order to examine decentralization-relevant topics, it is necessary to examine most of the variables that are affected when decentralization occurs, and what kind of relationship exists between them. In other words, this is a discussion of the variables that influence decentralization. This may be called an influence model of decentralization, and it is the formation of such models that makes it possible to identify whether specific variables are influenced by each other and the degree of that influence, and also to identify how the influenced variables will change according to regular changes in value.

In this connection, Schneider (2003) states that decentralization will have positive or negative effects on economic growth, political stability and income inequality. In particular, many scholars argue that decentralization will cause competition between local authorities which will lead to improvements in the management of local resources, thus bringing growth to the local economy (Davoodi & Zou, 1998; Zhang & Zou, 1998). Meanwhile, others argue that decentralization is more than just strengthening resident participation and improving residential responsibility, and will lead also to strengthening national competitiveness. Discussion of the duty to bring about decentralization must incorporate the fact that when it is not limited to a dependent variable but also performs the role of an independent variable, its positive functions as a dependent variable can be improved.

There are not many studies that have discussed clearly the relationship between national competitiveness and decentralization. However, there are related studies, which can be divided into two categories (Choi, 2008). The first holds that local decentralization enhances government transparency, strengthens social capital, reduces regulation, and promotes economic development and growth (Xie et al., 1999; Oates, 1993). These arguments do not mention the term ‘national competitiveness’, and relate economic growth to governmental transparency and regulation, but given that these are included in the lower section of national competitiveness it may be said that there is a relation. These arguments maintain that through decentralization, local authorities will utilize their increased authority to alleviate regulation and in this way devise various methods for the revitalization of the economy, such as enterprise attraction, and consequently, strengthened local competitiveness will lead to strengthened national competitiveness.

In comparison, the second category of studies state that local decentralization will in fact weaken national competitiveness. According to this argument, decentralization will hamper economies of scale during the resource allocation process, and cause external effects and ‘nimby’ syndromes, and thus may diminish overall efficiency. It can be seen as being in line with the opinion that decentralization will have a negative effect on national competitiveness by triggering local imbalance and income imbalance. This argument states that decentralization will, especially in fiscal areas, cause unclear distributions in expenditure and tax revenue functions between central and local government, and may lead to an inefficient distribution of resources (Fukasaku and de Mello, 1998). Shleife (2000) also points out that fiscal decentralization in developing or underdeveloped countries can mislead to a political monopoly by groups of vested rights in the region. Thus, there are conflicting opinions on decentralization concerning whether it impedes or promotes economic growth and efficiency.

Nevertheless, major advanced countries have recently been moving towards decentralization and are utilizing it as a means of improve national efficiency and competitiveness. In addition, decentralization is also held to have a positive effect on ultimate life-goals such as increased happiness. Bjornskov et al. (2008) argue that many studies of the effects of decentralization implicitly assume that it increases welfare or happiness levels, only it is not yet known to what extent it may influence happiness. They assume that when decentralization occurs, politicians compete, various regulations and bureaucratic waste are reduced, reductions in regulation are promoted, and policy decisions become clearer as a result of much clearer information. Overall, local authorities respond more to the demands of residents and provide public services more efficiently than central government. Given these premises, Bjornskov et al. (2008) present a number of hypotheses, the most representative being that decentralization leads to greater happiness (Biornskov et al., 2008: 147). One notable point here is the premise that during the process of decentralization improving happiness, individuals perform ethical actions relating to the consequences that they all the time face, the degree of which can change depending on the amount of community spirit they show. This can be understood in terms of ‘social capital’, and it may be assumed that decentralization causes an increase in social capital, which then leads to an increase in happiness. Although institutional authorities are important as regards local autonomy, social capital, which may be seen as the communal competency of local residents, is also of importance. No matter how much authority is transferred institutionally, if local residents have low levels of social capital, that is, if they have no will to resolve problems and act ethically by themselves, this cannot lead to improvements in national competitiveness or happiness. In addition, social responsibility on the part of entrepreneurs is also important. The national competitiveness report by the IMD implies that corruption levels within a society are influenced by this, and that levels of entrepreneurial corruption also influence national competitiveness and happiness.

One study that has great significance regarding the relationship between variables relating to decentralization is that conducted by Fan et al. (2009), which describes the relation between decentralization and corruption. The authors state that if decentralization is brought about, many sectors of government expenditure will be managed by local government, and this is related to a decline in levels of corruption.

In summary, then, out of the variables relating to local decentralization, some are influenced by it while others influence it. In addition, some are discovered to have positive relationships while others have negative relationships. While national competitiveness and happiness may be assumed to be the ultimate dependent variables and decentralization is an independent variable, many other variables can influence these variables, and opinions about this are divided. In this study a hypothetical study model will be set which aims to reflect important policy variables such as decentralization, national competitiveness and national happiness as much as possible. This analysis using the hypothetical model is expected to provide results significant for policy.

The following three research questions are selected in this study:

The contextual range of this research embraces, first, a measurement of the decentralization levels of 34 OECD member nations; second, an investigation of the relation between decentralization levels and major policy variables; and third, an investigation of the relation between variables related to decentralization. Via these processes, the research aims to set out suitable directions for future decentralization policies in Korea. In the case of some OECD indicator data, data on other countries was not included, so 26 countries for which data were included were set as analysis targets. The time range was set at six years, from 2008 to 2013. Data on some individual indicators were omitted for certain periods, and in this case adjacent data were used.

Because the main targets of this research were decentralization, national competitiveness and national happiness, major national policy variables that were assumed to influence decentralization were used as parameters. GDP per person, restriction levels, social capital, social responsibility, policy transparency, corruption levels, income inequality, national competitiveness and happiness were selected as major variables that influence decentralization. Many sources of data were used to measure these variables, and any missing data were replaced with the most recent versions. Degree of happiness was measured by utilizing the World Happiness Report 2013 published by the UN, and national competitiveness was measured using the annual national competitiveness score published by the World Economic Forum.

Although decentralization in theory usually includes administrative, political and fiscal decentralization, in this research it was limited to fiscal decentralization. First, it is difficult to obtain data on administrative and political decentralization for different countries, and second, since fiscal decentralization has a high correlation with administrative and political decentralization, it was thought that it would be sufficient to utilize only fiscal decentralization. The indicators used to measure fiscal decentralization levels were of the proportion of total revenue of local taxes, and the proportion of total expenditure of local expenditure. However, in order to complement these quantitative indicators, the taxation autonomy indicators from the OECD were used, and by utilizing the data of each country presented in the OECD database the taxation autonomy index was calculated. However, while the OECD taxation autonomy indicators are applied to the taxation section, in this research it will be assumed that these indicators function in a similar way towards expenditure as well, and so they will be applied in that section equally. That is, since there can be restrictions on the expenditure of local authorities even if local expenditure level is high, restriction levels on expenditure were assumed to be similar to that on taxation.

Instead of the generally used fiscal decentralization indicators, the OECD has recently presented the level of discretionary power that local authorities can practically exercise in fiscal areas. In this connection, it classifies the discretion levels of local authorities regarding taxation according to nine grades, as follows:

The discretion of local authorities is strongest in the first category and increasingly less strong in the following categories. Therefore, in this research, by using values relating to local government authority in each individual country, counterweighted values will be set. Counterweighted values here are obtained by first assuming the nine categories as an interval scale, multiplying ① by 9 (discretionary power on tax rates and reduction), multiplying 8, 7, 6 by the lower categories in serial order, and finally multiplying ⑨ by 1. The sum of all individual countries will be considered as representing taxation autonomy. That is, the larger this value, the more discretion local authorities have on taxation. The value of the taxation autonomy indicators calculated here will be multiplied by the proportion of local taxes and will be used as the taxation decentralization level, and in the same context it will be applied to local expenditure costs as well in order to derive the expenditure decentralization level. The contents of the major variables to be used in this research are shown in Table 3.

1) Measurement of decentralization levels

As mentioned earlier, to measure decentralization levels a method of calculating the two indicators taxation decentralization and expenditure decentralization is selected. First, the two indicator values were obtained from the OECD database, then each indicator value was obtained using the weighted value of the taxation autonomy indicators. Then, the average of these two values was set as the value for fiscal decentralization level.

2) Measurement of the degree of influence between policy variables related to decentralization

A hypothetical research model for measuring this is presented below.

The steps for analyzing a hypothetical model are as follows. First, technical statistics regarding GDP per person, decentralization (taxation decentralization, expenditure decentralization), social responsibility, social capital, regulation levels, corruption levels, Gini coefficient, policy transparency, happiness and national competitiveness, and Pearson coefficients were obtained using SPSS 18.0. Second, a Structure Equation Modeling Analysis was conducted using AMOS 20.0. As a coefficient presumption method in the structure equation model, the Maximum Likelihood Method was used. The model’s evaluation methods include one using X²verification and another using the goodness-of-fit index. However, X²verification is sensitive to sample size and the null hypothesis is very strict, so the model was not evaluated solely on the X²verification. In this research, the simplicity of the model was considered and the goodness-of-fit of the model was evaluated through GFI, NFI, IFI, CFI, which had the standards of goodness-of-fit evaluation indexes established, and the appropriate standard is seen as .9 (Moon, 2009). Third, in case a hypothetical model should not be appropriate to the data, modifications to the model were made by confirming the modification index and adding routes with a correction index higher than 4. In this way, the best-fit model could be found while preserving the basic simplicity of the model. Also, an examination was carried out of the statistical significance of the routes between variables and its explanatory power (direct effect, indirect effect, total effect). Fourth, by using the bootstrap method, the kind of indirect effects that social responsibility, social capital, regulation levels, corruption levels, Gini coefficient, and policy transparency can have on the routes that start from GDP per person and decentralization level and lead to happiness and national competitiveness can be measured by focusing on the significance of indirect effects. The bootstrap method creates an experiential distribution of parameters when the parameter distribution is unknown, and is a useful means of obtaining values such as the approximate standard error, confidence interval, and significance probability, by estimating the sample distribution of all parameter estimation values included in the structure model (Kim, 2007).

3) Policy simulation

A policy simulation based on the system dynamics method was conducted on the results produced from the hypothetical model. This policy simulation is based on the system dynamics methodology, and estimates how dependent variables change by altering decentralization levels. The simulation method is aimed at understanding and explaining complicated situations from a dynamic feedback perspective. In addition, it is the methodology and framework used for establishing a computer model that is based on understanding and testing how situations formed of complicated causal relationships can dynamically change (Jung et al., 2005: 220). The parameter of individual variables is estimated through the optimization method, and to achieve this the POWERSIM simulation program is utilized. Via this policy simulation method, changes to national competitiveness and national happiness following changed values in decentralization can be estimated.

The values for taxation decentralization and expenditure decentralization levels were used to measure decentralization levels, and, as mentioned above, the value of taxation autonomy index was applied to calculate such values. First, the weighted values of discretion on taxation by central and local government as calculated by the OECD was used. For the ten categories, counterweighted values per categorical percentages are given, then the values are added. Assuming that these ten categories are interval scales, the category that shows the highest local government discretion is multiplied by 10, the second-highest category is multiplied by 9, and so on. The value calculated by this process will be named the ‘weighted value on taxation autonomy’. The results of recalculating distribution levels by applying them to the ratio of local taxation and expenditure levels of OECD nations in 2012 are seen below.

Table 4 shows that the taxation autonomy weighted values for 26 OECD nations are on average 0.76, while Korea’s is slightly lower than that, at 0.67. Meanwhile the table shows the ratio in 2012 of local taxes to total taxation. As Table 4 shows, the average for 26 nations in 2012 was 17.40, while that for Korea was 15.81, slightly below the average value.

1) Descriptive statistics

Table 6 shows the basic statistical value of the variables used in the research. The annual values from 2008 to 2013 are processed in a pooling method. However, relative decentralization values were not used for decentralization indicators, local tax and local expenditure ratios being used instead.

Verification results of the model on in terms of how GDP per person and decentralization produce happiness and national competitiveness via social responsibility, social capital, regulation levels, corruption levels, Gini coefficient and policy transparency are shown in Table 9.

Local decentralization in general includes administrative, political, and fiscal decentralization. However, it is very difficult to compare administrative or political decentralization levels across countries. Though it is also difficult in the case of fiscal decentralization as well, comparisons here are relatively easier to make. In this study, the main focus was on fiscal decentralization, and comparisons were made by deriving decentralization scores of different countries by applying weighted values on indicators such as the discretion of local authorities on taxation, as well as utilizing traditional methods which consider the ratio of local taxes to total tax income or the ratio of local expenditures to total government expenditure. The results show that the fiscal decentralization level of Korea is estimated to be about 90 percent of the average OECD level. When one uses the weighted values of taxation autonomy related to the discretion of taxation as standards, Korea’s score becomes 88 percent compared to the OECD average, similar to the previously mentioned score. Also, since taxation decentralization levels are lower than the OECD average relative to expenditure decentralization, there is a need, especially, to strengthen decentralization in this area.

In addition, the results of an analysis of the influences decentralization has on dependent variables such as national competitiveness or national happiness show that although decentralization does not directly influence national happiness, not only does it have direct and indirect effects on national competitiveness but it actually affects this the most out of all policy variables.

From this analysis a number of policy implications can be drawn. First, there is a need to properly recognize local decentralization levels. There is a need for objective problem recognition in relation to the current level of Korea compared with that of other countries, instead of only emphasizing local decentralization. When one focuses on fiscal decentralization, the taxation and expenditure of local authorities in Korea fail to reach the average for the 26 OECD countries, being only 90 percent of this.

Second, there is a need to strongly promote decentralization, as it positively improves national policy variables. National competitiveness is directly influenced by decentralization: the effect on national competitiveness is stronger than that of any other variable. Not only that, but decentralization also directly increases social responsibility, fosters social capital, and brings positive effects to deregulation. Furthermore, it is estimated to have a direct effect on the improvement of policy transparency.

Even when one considers indirect effects, decentralization increases social capital, and this increase leads to an increase in national happiness. In addition, it is estimated that decentralization increases social capital, and this leads to an indirect increase in national competitiveness. As can be seen in these processes, decentralization itself has direct or indirect positive effects on national competitiveness and happiness, and therefore, strongly promoting decentralization in governmental policies is required. It is also found that deregulation, which has recently been emphasized by the Korean Government, can be improved through the promotion of decentralization.

Third, even if decentralization is promoted, it is necessary to put policy effort into increasing variables, such as social capital, which perform the role of intermediate parameters. Even if improvement here cannot be brought about instantly, efforts to improve social capital, such as ethical actions and mutual trust, are needed to improve the level of national happiness.