Expenditure Analysis of Local Government and Regional Economic Development District/City of Ten Province in Sumatra Island Indonesia

The research problem is how to influence the local government spending, the number of local civil servants (PNSD), and the number of regional infrastructure to the local economic development districts/ cities in Sumatra island of Indonesia. While the purpose of the study is to estimate the model and analyze the influence of local government spending, the number of the civil servants, and the amount of infrastructure to the local economic development districts/cities in Sumatra island of Indonesia.Regional economic theory used is Keynesian model: Y = C + I + G + (X – M); A concept of Local Government Expenditure (local government spending) according to Wagner’s Law, Theory Peacock and Wiseman, Wagner, and Solow. While previous empirical research, among others is: (a) Susetyo (2003) and (b) Yanizar (2012). Research methods include: (a) the scope of the studies in this research is local government spending, the number of local officials (PNSD), local infrastructure (road length), and the regional gross domestic product (local economic). The location of observation as the unit of analysis is the district/city as much as 155 (comprising 121 districts and 34 cities) on the island of Sumatra, Indonesia; (b) The data type and data source, that the type of data used are secondary data and are equipped with primary data analysis uni; (c) The method of analysis in this research is quantitative descriptive and equipped with qualitative descriptive. The results of the research will be grouped into two categories: the results of descriptive analysis and quantitative analysis in the form of model parameter estimation data panel. The resulting model can be tested to get a good model, precise and consistent is by Chow and Hausman Test. Theoretically could have that amount of time series data is smaller than the district/city as a cross section preferably using random effect models is: Y = –2493,4550 + 5,7371X1 + 1,9784X2 – 1,2106X3 + e 1 ; where, Y = Gross Regional Domestic Product; X1 = Regional expenditure (local spending); X2 = Number of local civil servants; X3 = Length of road; e 1 = error term. Simultaneously (F-test) showed that the results obtained are the three variables observed were local spending, the number of civil servants districts/cities, length of roads in the district/city catastrophically affect positively and significantly. This value amounted to 317.2543 F-statistic is greater than F-table so the independent variable positive and significant effect to dependent variable, with the result of random effect model estimation. Partially (t-test), indicating that the effect of local spending to the regional gross domestic product is positive and significant. The value t-statistic of the local spending, the number of civil servants, the length of the local road each 17.85835; 12.79292; –2.482922 are greater than t-table so that a positive and significant influence.Coefficient of determination (R2 ) is 0.506860, or 50.68 percent and Adjusted R2 = 50.52 percent, means that the results can be explain by the model estimates a third variable and significant amounted percent, while the rest, or by 49, 32 percent is explained by factors outside the model. The best model after the Chow and Hausman test as well as the number of time series is less than the number of districts/cities, the best estimation models are models of random effect