Un modelo no paramétrico de evaluación de la eficiencia en la gestión tributaria aplicado a las Delegaciones territoriales españolas

Resumen:En este trabajo se mide el desempeño de la gestión tributaria española, para 2004-2015, mediante un análisis envolvente de datos, de dos etapas. En la primera, se estiman los indicadores de eficiencia, de las Delegaciones territoriales, correspondientes a las CCAA de régimen común; que por término medio alcanza el 85 %. En la segunda etapa, mediante el modelo de regresión censurado, se identifica una relación significativa de la eficiencia, con el PIB y con el nivel de educación. Adicionalmente, mediante la regresión logística, se establece la probabilidad de ser eficiente, a partir de los valores futuros de las variables ambientales.Abstract:Introduction The function carried out by the tax public Administration, which consists in obtaining resources with which to finance the public Sector, must be carried out in an efficient manner to legitimize its actions over time and to encourage the fulfillment of obligations of the taxpayers. The application of the spanish tax regimen in the decentralized field, has gained an important role as a result of the strong process of delegation of the public actions, entrusted to the Autonomous Communities, initiated with the Constitution spanish of 1978. Therefore, in this paper we measure the performance of the tax management carried out by the territorial Delegations of the spanish autonomic tax administration, for the period 2004-2015, by means of a semi-parametric analysis of two stages. The aims of the research are to estimate the levels of relative technical efficiency, reached by the territorial delegations, in the management of the taxes with the most important incomes. These taxes are the property tax, the inheritance and donations tax, the tax on onerous patrimonial transmissions and documented legal acts; and the taxes on the game. The tax collection obtained by the previous taxes, represent in 2014, 77% of the collection obtained by the taxes with tax autonomy, wich managed by the autonomous regions. Also, the research aims to identify the factors that cause efficiency levels, by means of parametric regression techniques. Methodology This study uses a semi-parametric methodology of two stages, which consists in applying, in a first stage, the data enveloping analysis -DEA- of Charnes, Cooper and Rhodes (1978), to estimate the levels of technical efficiency relative, of the 46 territorial Delegations, corresponding to the 15 spanish autonomous regions, financed by the common regimen. The variables used in the research are obtained from the General Budgets of the State. An entry is used, which is the personnel that carries out the management of the transferred taxes. As well as 5 outputs that are, the number of records instructed, the number of providences issued, the number of complementary liquidations of the property tax, the number of autoliquidations sent of the tax on onerous patrimonial transmissions and documented legal acts, the number of files sent of the inheritance and donations tax; and the number of self-liquidations presented by automatic machines and devices. These variables are representative of the functions that compose the tax procedures carried out by the autonomous tax administrations in the management of the ceded taxes, which are the management, the fundraising, the inspection and the application of the sanctioning regime. In the second stage, through the Tobit regression model (Tobin, 1958), the environmental factors that affect the efficiency levels are identified. The model dependent variable is the level of efficiency obtained in the first stage of the investigation. The returns considered in the study are the environmental factors that represent the economic context, measured by GDP per capita and with the number of mortgages constituted; the situation of fiscal policy, measured with public debt as a percentage of GDP; the social context, determined by potential contributors and with the level of training achieved by society; the geographical environment, measured with the km2 of the regions; the legislation in force in the period of analysis, established by organic law 3/2009 of 18 December, on the financing of the autonomous Communities; and the form of organization of the tax management, carried out through a tax Administration or through an autonomous tax Agency. Additionally, through the binary logistic regression model, a predictive model is established that reports the probability that the autonomous public administration will be efficient, based on the future values of the environmental variables, wich are significant in the study. Main results The results of the DEA show an average technical efficiency in the tax management of 85%, for the 2004-2015. This means that for the autonomous tax administrations to achieve optimal behaviors with the amount of work factor used, they should expand their outputs radially by 17.64%. In addition, the minimum level of technical efficiency obtained is 33%, which is reached by the territorial delegation of Ourense. Also, the results show a tendency of the average technical efficiency decreasing, during the 12 years of the period analyzed. The Tobit model determines a significant direct relationship of efficiency with GDP and the level of higher education obtained by the population. It is to be hoped that greater tax capacity of the region will achieve a higher level of efficiency. Also, there is a inverse significant relationship between efficiency and the variables public debt and km2. This implies that a lower level of efficiency in management could lead to lower levels of tax revenue. This deficit could be financed with higher levels of public debt. For its part, a greater extension of the territory where the management is carried out, can produce more efforts by the tax managers, reducing their levels of efficiency. Conclusions In this paper we study the tax management of spanish regional territorial Delegations during the 2004-2015. The technical efficiency levels estimated with the DEA are 85%, on average. The delegations that are considered reference for the remainder in each of the 12 years of study are, Salamanca in 2004, Asturias and Ciudad Real in 2005, Asturias in 2006, Asturias in 2007, Madrid in 2008, Madrid in 2009, la Rioja in 2010, Tarragona in 2011, Barcelona in 2012, Segovia and Valencia in 2013, Barcelona and Málaga in 2014; and Malaga in 2015. According to Tobit's regression, the variables that denote direct significance with efficiency are the GDP and the higher education level reached by the population. The factors that have an inverse significance with efficiency are the public debt and the geographical extension, measured with the km2 of the regions.