El voto económico de los Andaluces: Una aproximación dinámica por municipios

Resumen: Aunque está muy extendido que muchos ciudadanos votan o no al partido en el gobierno en función de criterios económicos, lo cierto es que, en muchas ocasiones, los investigadores no encuentran una medida clara y amplia del denominado voto económico. Parte del problema puede estar en la metodología utilizada. Por este motivo, proponemos la utilización de datos agregados (a un nivel municipal) y de estimadores econométricos dinámicos que puedan recoger la influencia de la tendencia temporal, y no sólo de los datos económicos coyunturales, en la obtención de resultados electorales. Nuestros resultados constatan la existencia de voto económico en Andalucía en sus dos versiones, retrospectivo y prospectivo. Abstract: The objective of this work is to study the socio-economic factors that have determined the economic vote in the 761 Andalusian municipalities during the six general elections (Spanish Congress of Deputies) held throughout the period 1989-2008. The term “economic voting” is used in the literature to refer to voters who condition their vote choice on personal or general economic conditions. Specifically, two areas can be distinguished in the theory of economic voting: the retrospective theory of voting and the prospective theory of voting. The retrospective theory of voting (Fiorina, 1981; Rogoff and Sibert, 1988; Rogoff, 1990; Alesina and Rosenthal, 1995) looks to the past and postulates that individuals evaluate the past performance of the incumbent party. That is, voters reward or punish the incumbent party depending on whether they believe that the policies implemented have served their own or the community’s interests. In contrast, the prospective theory of voting (Frejohn, 1986; MacKuen et al., 1992; Erikson et al., 2002) looks to the future and holds that voters have a personal economic interest arising from their economic conditions or social class, and will vote for parties and political leaders whose election promises are more similar to their own, that is, voters will elect representatives who defend policies that they believe will serve their own or the community’s interests.However, this distinction between past and future in practice is not so simple. First, retrospective voting, that is, the degree to which voters perceive that economic conditions have improved during the incumbent party’s term in office, may also be distorted by individuals’ own economic conditions and ideology (Campbell et al., 1960; Key 1966; Hibbs, 1977; Kramer, 1983; Popkin, 1991; Zaller, 1992; Mutz, 1993; Maravall and Przeworski, 2001; Evans and Andersen, 2006; Lewis-Beck, 2006; Ladner and Wlezien, 2007; Gerber, 2009). According to Kramer (1983), individual-level, survey-based studies can yield inconsistent data because it is not known how individuals acquire knowledge of the economic reality, and suggests that it is better to work with aggregate and observed data.Second, prospective voting is not based exclusively on voters’ current economic conditions, but also depends on their previous economic conditions and experiences such that the credibility assigned to candidates is weighted by factors which have been formed throughout voters’ lives (Campbell et al., 1960, 1966; Duch and Stevenson, 2006). This difficulty suggests that it is better to work with dynamic panels than with data from a specific electoral process.For these reasons, in this paper we study the economic voting in Andalusia with three methodological decisions that represent a clear differentiation from other works: a) a panel with data aggregated at the municipal level, b) is constructed dynamic-econometric estimators for panel, and c) the study focuses in a Spanish region, since the Spanish case is particularly interesting in that the weight of economic voting is abnormally low relative to other Western democracies (González, 2009; Mata López, 2009; Fraile and Lewis-Beck, 2010).To quantify the associations between vote choice and socio-economic variables, we use official statistical data drawn from the Multi-Territorial Information System of Andalusia (SIMA), which is managed by the Statistics and Cartography Institute of Andalusia. This freely accessible database includes time series of variables broken down at the municipal level for the years 1989, 1993, 1996, 2000, 2004, and 2008, all of which correspond to election years to the Spanish Congress of Deputies. The dependent variables are the Spanish Socialist Party (PSOE) vote share aggregated by municipalities, and, alternatively, the People’s Party (PP) vote share obtained in elections to the Spanish Congress of Deputies. The independent variables are demographic variables, income, wealth, voter turnout, mortality, labour market variables, public pensions and public schools, among others. We have also constructed the growth rate for all the variables mentioned and have been used in the econometric models.Using a fixed effects model for panel data, we have measured the correlation between the vote for the PSOE when it is the incumbent party and the independent variables; and between the vote for the PP when it is the opposition party and the independent variables. The results show that there is a certain complementarity between the municipalities that vote for the PSOE and those that vote for the PP. Specifically, the PSOE vote share tends to be higher in municipalities with higher unemployment rates, higher mortality rates, and a larger proportion of younger people. In contrast, the PSOE’s municipal vote share is inversely correlated with the following municipal-level variables: voter turnout, population, relative population, middle-aged population and older, average age of the population, population living outside the city or town, non-contributory pensions, number of vehicles per capita, number of taxpayers, and higher per capita income. Respect to the PP, his best electoral outcomes are associated with the highest values of the following variables: voter turnout, total and relative population, middle-aged population, population over 65, mean age of the population, population living outside the city or town, proportion of population with non-contributory pensions, and the three rates of income and wealth (vehicles per capita, per capita income and number of taxpayers). The worst outcomes for the PP are correlated with high values for the following variables: female population, unemployment rate and younger relative population.We have also estimated dynamic models of the vote for the PSOE and the PP with panel data. Next, we summarize the main findings.During the period analysed, the PSOE had a vote share of 47.5%. However, voting dynamics were negative in the limit of significance so that in each election the PSOE obtained 24% less votes than in the previous election. The PP had a vote share of 25% and its voting dynamics were positive at 14.7%In municipalities where a larger percentage of the resident population lives outside the city or town, the PSOE vote share is lower although at the limit of significance. In municipalities where per capita income is higher, the PSOE tends to obtain a lower vote share. This variable was not significant for the PP.In municipalities with a higher number of vehicles per capita -a proxy for the consumption level of municipalities-, both the PSOE and the PP usually obtain a higher vote share, but the PSOE benefits more from the vote share than the PP.An increase in the unemployment rate has a negative effect on the vote share of the PSOE when it is the incumbent party, and has a positive effect on the PP when it is the opposition party. However, the unemployment rate rewards the PP more than it punishes the PSOE. This type of voting behaviour provides evidence of retrospective economic voting.In short, the results allow us to deduce that in the municipalities of Andalusia, local economic conditions and improvements in such conditions are correlated with the vote share obtained by the incumbent party and the main opposition party. In other words, there is both retrospective and prospective economic voting.