Especialización sectorial, capital humano e ingreso regional en Ecuador

Resumen:El desarrollo de Ecuador medido por el ingreso per cápita, presentó avances en los últimos años pero distribuido de manera desigual en el territorio nacional. El presente trabajo tiene por objetivo examinar como el nivel promedio de capital humano y la especialización sectorial, incrementan la diferencia regional del ingreso per cápita de los 221 cantones de Ecuador. Para corregir la endogeneidad entre el capital humano e ingresos, se estimaron regresiones con variables instrumentales. Mediante el uso de datos de corte transversal y técnicas de econometría espacial, se encontró que un mayor nivel promedio de capital humano y la mayor especialización en el sector terciario, incrementan el ingreso regional.Abstract:The regional income disparities are an issue of wide interest in the academic environment and in the design and implementation of economic policies. Real income per capita is an approximate measure of development. Countries where per capita income is not distributed equally over the national territory, present challenges associated with efficiency and lack of opportunities for the inhabitants of less developed regions. In Ecuador, there are significant spatial differences in the level of per capita income at the provincial level and it is most evident at the cantonal level. The data show that regions with high incomes are specialized in manufacturing and services, and its population has a high average level of schooling. This reality suggests that human capital and the sectorial specialization determine the level of income and the consequent regional disparity. In this context, the objective of this research is to examine the effect of the average level of human capital and sector specialization in regional income gap of the 221 cantons of Ecuador. We use cross-sectional data from 2010, obtained from the National Institute of Statistics and Census (INEC) and Central Bank of Ecuador (BCE). In addition, we analyze the existence of spatial dependence of income. This dependence arises due the interaction between regions, caused by the effects of spills of knowledge and the distribution of economic activities in the territory. The methodology is divided into three stages for two separate sets of regressions. In the first stage, we estimate several Ordinary Least Squares (OLS) regressions. In the first set of regressions, the dependent variable is the logarithm of Gross Value Added (GVA) as a measure of income, and the independent variable is the average schooling as a measure of human capital. In the second set of regressions, the dependent variable is also the VAB per capita, while the independent variable is the coefficient of sectorial specialization of the canton i, where i = 1,2 ..., I. In addition, we added instrumental variables to avoid the omission of other relevant variables in explaining income. In the second stage, we correct the bias of the estimates due to endogeneity between the cantonal VAB and average schooling by using instrumental variables. The instrument used for average schooling is the average schooling of parents in each canton. Recent advances in the field of regional studies and the development of spatial econometric techniques can capture the effect of the interplay between space units. Here, in the third stage we performed a robustness analysis of the results obtained in step 1 and 2 using spatial econometric techniques. We determine the relevance of the application of SAR (Spatial Autoregressive Model), SEM (Spatial Error Model) and the combination of both SARMA model in spatial models. SAR models consider how the income of a region is affected by entering neighboring regions and SEM can capture the effect of omitted variables with spatial dependence on revenue from neighboring cantons. The results are listed below. First, as it increases the coefficient of specialization in manufacturing and services, VAB per capita levels are higher. Second, the instrumental variable for human capital was appropriate for the model, we determined that the cantons with higher average years of schooling in the population have higher levels of GVA per capita. Third, using spatial techniques, we show the existence of spatial dependence on the used variables, so we estimate spatial econometric models SAR, SEM and SARMA. These models ensure consistency of the results obtained in the previous two stages. Therefore with the SEM model, we conclude that the income of a region also depends on variables omitted in neighboring regions with spatial dependence as transport, trade, flow of people and capital, etc. On the other hand, the SAR model was significant in the relationship between the secondary sector and income and the model relating the human capital and income. That is, if a county has a high industrial development, it can exchange knowledge through their workers with neighboring regions and increase the income of both. However, when we adding control variables SAR model is not significant in both relationships. This is because in Ecuador, the industrial sector is underdeveloped and is concentrated in a few regions. In this sense, this paper addresses two implicit biases in the OLS regressions for the context of the investigation. The bias arising from endogeneity and the omission of spatial interdependence between territorial units. Our results support two implications of economic policy. First, you cannot exploit the development potential of the cantons without increasing investment in human capital. Sectorial economic policies that encourage specialization in manufacturing and services can increase the production of the cantons of Ecuador. Second, ignore the role of spatial interaction between the territorial units in the design and implementation of public policies can lead to problems of economic inefficiency. The results show that the specialized cantons in the secondary and tertiary sectors have higher incomes, while specialization in the primary sector generates low income. The only natural resource revenue generator is oil. However most of these revenues are distributed throughout the country and are not invested in local development. In addition, the cantons with higher average level of human capital, they will have more income too. These results show low specialization in the cantons, is affected by a low-skilled work. Therefore the production structure generates differences in the level of income of the regions as well as the unequal distribution of human capital.