Investigating Influential Factors on Improving Poverty Conditions in Latin America

The objective of this study is to determine the spillover effects of various international supports for poverty in Paraguay. A vector autoregressive model is used to investigate a dynamic linkage among five components: prevalence of undernourishment, food intake, gross domestic products (GDP) per capita, primary school completion rate, and unemployment rate. We found that the primary school completion rate has the largest spillover effects for reducing poverty except GDP per capita. Supporting international agencies such as the Inter-American Development Bank can keep or invest more money in early education sectors, and Paraguay can obtain not only direct supports but also larger indirect effects.

still lacked accesses to adequate food, clean drinking water, and sanitation, even though global extreme poverty rates have decreased more than half since 1990.
For instance, Paraguay decreased its poverty rate from twelve percent in 1995 to less than three percent in 2015 ( Figure 1). In addition, gender is key to poverty rates; women in these under-developed countries are more likely to live in poverty conditions than men because of their uneven access to work, education, and property [1].
People living in poverty not only lack money (i.e., due to lower incomes) but also face other challenges. The United Nations Development Programme (UNDP) stated that poverty issues including women's illiterate and higher child mortality rate should be solved simultaneously. Additionally, poverty can involve not only a lack of necessities required for material well-being but also the denial of opportunities to live a tolerable life [2]. Based on the poverty dimensionality [2], a multidimensional social phenomenon, whose causes vary by gender, age, and other socio-economic contexts, was defined by [3]. Poverty has various causes, such as shortage of food or farmland, lack of access to clean water, unemployment, and/or lower education levels [1].
[4] conducted a study on relationships between income distribution and socioeconomic variables. He explored Malaysian poverty and income distribution using an enumeration survey focused on socioeconomic variables such as race, health, and standard of living for accessing poverty, and each dimension has multi-indicators. For instance, a year of schooling and child school attendance are in the educational dimension; mortality and nutrition are in the health dimension; and electricity, sanitation, water, floor (i.e., whether the household has a dirt, sand, or dung floor), cooking fuel, and assets are indicators of the standard of living dimension [5].
For estimating the poverty issues, [6] used a vector autoregressive (VAR) approach for examining the implication of unemployment and inflation on the poverty in Nigeria. Their VAR analysis includes poverty level, unemployment rate, and inflation rate. They found that an unemployment caused poverty in Nigeria, and recommended the government would expand educational programs to acquire the practical skills [6]. Another study on the poverty is [7].
They examined the nexus between the financial sector development and poverty reductions in Nigeria, and found that indirect economic effects of economic growth had the strongest influence on the poverty reduction in the short term, while it would widen the income inequality in the long term [7].

Poverty in Latin America
In this century, many under-developed countries in Southern Asia, North and The higher income inequality and the unemployment rate have led to the higher level of crime [10] [11]. As a result, Brazil has largely suffered high crime rates.
Argentina is another high-income country in the Latin America, which annual per capita income is over US $8000 [8]. Despite its wealth, Argentina has also been suffered from high level of the poverty though poverty rates fell forty percent in 1990 to twenty-two percent in 1994. However, income distribution has Similar to the other South American countries, Paraguay has historically suffered from poverty and plagued by corruption and political instability. According to the [12], approximately forty percent of Paraguayan population is poor, and twenty percent are extremely poor. The incidence of poverty illustrates variations across regions: Asuncion, the capital of Paraguay, is lowest and 21 percent, urban region outside central is 27%, urban central is 35.3% and rural household is 48.8% [12]. Paraguayan poverty is intense that over half of the poor and more than two-thirds of the extremely poor are located in the rural area. This occurred under the dictatorship of General Alfredo Stroessner from 1954 to 1969 [12]. During this term, agricultural growth was achieved through foreigner's agricultural migration policies by 1969. The Paraguayan government introduced a foreign currency and market-open policies, establishing industrial growth in the 1970s. These policies also led to the uneven distribution of agricultural land that 1.1% of the population owned approximately eighty percent of agricultural land (Table 2). In other words, over eighty percent of small farmers who owned land up to twenty hectors are only 6.2 percent of agricultural land in Paraguay, indicating that the smaller agricultural land led to the lower income for the farmer. Due to this historical background, the poverty has been widespread with increasing poverty gap in Paraguay [12] [13]. Journal of Human Resource and Sustainability Studies Organizations such as the International Bank for Reconstruction and Development (IBRD), the International Development Association, the IDB, the JICA and the UN have supported Paraguay to reduce the poverty. The largest supporter, the IDB, has given priorities to interventions in following six sectors: transportation and connectivity, water and sanitation, energy, productive development, finance, and public management. Total loans from the IDB were 1.57 billion US dollars in 2013, and total amounts of the IBRD loans and the IDA credits were 430 million US dollars in 2015. Due to these supports, the poverty rate has rapidly decreased in Paraguay after 2006 ( Figure 1). However, there are still many Paraguayan under the poverty, and it is important to improve the poverty conditions [12]. For reducing poverty, there have been various supports by international agencies. If there are spillover effects from one support in a certain field, it will lead to effective poverty reduction. Clarifying spillover effects are important both for Paraguay and supporting agencies. When we know indirect spillover effects of direct supports, international agencies such as the IDB or the UN can invest money more effectively. However, there is no study of clarifying indirect effects in Paraguay. Therefore, the objective of this study is to determine spillover effects of various international agencies' supports for poverty in Paraguay.

Methodology
Following [2] and [3], this study uses multi-dimensional aspects of Paraguayan poverty analysis for clarifying spillover effects of international supports for reducing poverty in Paraguay. We select four multi-dimensional aspects: food, economy, education, and work. First, this study divides a food dimension into two: food quality and food quantity. The prevalence of undernourishment is chosen as the food quality dimensional aspect, while the food intake is used as the food quality aspect. The prevalence of undernourishment measures the share of the population who has an insufficient caloric intake to meet the minimum energy requirements defined as a given population 3 . The food intake that indicates a national average food intake. Next, we choose dimensional aspects of economy, education, and work are the gross domestic product (GDP) per capita, the primary school completion rate, and the unemployment rate, respectively. The GDP per capita captures Paraguayan average income status, and the unemployment rate illustrates labor and economy. The primary school completion rate is the percentage of students completing the last year of a primary school. This captures the population of early education in Paraguay. Unlike the Malaysian case [4], this study does not include race because of the lack of Paraguayan data. Therefore, there are five indicators: prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate for revealing spillover effects of international agencies' supports in Paraguay. This study uses the VAR model to disclose the relationship among five indicators for reducing poverty in Paraguay since the VAR model can detect relationships among variables, can forecast future influences among each other, and can be relatively easy to estimate with small dataset [6] Table 3. The average prevalence of undernourishment during eighteen years sample period was 12.73%. This is slightly higher than the 2012 global total, 11% 6 . The mean of primary completion rate was 91.33% which was close to the 2012 global total, 91.4% 7 . The average Paraguayan unemployment between 1995 and 2012 is similar to that of global total, 6.19% 8 . The R program with the vars package was used for this estimation.

Results
We use the VAR model for the estimation of the spillover effects in the Paraguayan development supports. To interpret the economic implications, first, we estimate the VAR model with causality tests (Table 4). Then, we calculate impulse responses (Figures 2-6) and forecast error decomposition (Table 5) for analyzing the spillover effects of Paraguayan supports.
First, Granger causality tests were conducted to identify if any variables could be used in predicting other variables [15]. Table 4 shows the estimated VAR model results including Granger causality tests. The causality results indicate that the variables do not cause any of the other variables besides themselves. Therefore, all variables have a granger cause, indicating that prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate can be explained by the other variables.
Second, the impulse response function identifies the responses over time in all variables to a one-standard-deviation increase in one of the variables [16]. The system is triangularized with the variables ordered as prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate. The ordering for variables is based on prior knowledge and of the variable exogeneity and comparisons of alternative ordering [16]. Figures   2-6 illustrate the impact of a shock in one variable on the other variables. Figure   2 indicates that a positive shock in prevalence of undernourishment results in     Figure 2. Response of prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate to a shock in prevalence of undernourishment. Figure 3. Response of prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate to a shock in food intake. Figure 4. Response of prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate to a shock in GDP per capita.
K. Mitsumoto, K. Yamaura Figure 5. Response of prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate to a shock in primary school completion rate. Figure 6. Response of prevalence of undernourishment, food intake, GDP per capita, primary school completion rate, and unemployment rate to a shock in unemployment rate.
responses of food intake and primary school completion rate are small. Figure 3 shows a positive shock in food intake leads to the positive response by unemployment rate and small negative response by GDP per capita. Figure 4 depicts that a positive response in a GDP per capita results in the negative response by unemployment rate. Figure 5 indicates that a positive shock in primary school completion rate leads to a positive response by unemployment rate and a negative response by GDP per capita. Figure 6 shows that a positive shock in unemployment rate results in a negative shock in GDP per capita.
Finally, forecast error variance decompositions are used to interpret the VAR model analysis by many econometric studies [6] [15]. The variance decomposi-