Economy and voting behavior

On Thursday, just ahead of local elections which have transformed into a referendum on the incumbent party, I participated in a very interesting workshop on the various determinants of voting behavior.
Four papers were presented during this workshop, organized by the Turkey Economy Foundation (TEK), at Boğaziçi University: “Elections and Public Sector Pricing” by Ercan Uygur; "Political Knowledge and Voting Behavior in Turkey" by Deniz Salman; "Economic Voting in Turkey: Some Recent Findings" by Cem Başlevent; and "What Outcome Should be Expected on March 30 and How the Outcome Should be Interpreted?" by Ali Akarca. Professor Uygur, the president of TEK, questioned if from 1980 to 2014 the prices of public goods like coal, petroleum, natural gas, electricity and water controlled by the government are raised before elections and released soon after. His answer is affirmative and this is a rather expected answer. However, what is less expected is the fact that the Justice and Development Party (AKP) government followed exactly the same pricing policy as the various governments in power from 1980 to 2003. That said, as Turkey became an open market economy from the 1980s, governments avoided creating a long-lasting deviation trend between prices of public goods and average inflation. In other words, they respected market rules except for short periods preceding elections. Salman, a political scientist from Boğaziçi University, tries to understand to what extent Turkish voters vote “correctly” using data from a survey conducted just after the June 2011 general elections. In political science jargon, “correct voting” means a vote that matches the political position of the voter defined by his/her thoughts on various issues -- captured by 19 questions in the survey -- with the party line expressed in the party program. One of the interesting findings in the paper is that only 53 percent of Turkish voters vote “correctly.” This average figure is 87 percent for the AKP and 41 percent for the Republican People's Party (CHP). Here we have, for sure, a political culture problem that deserves further research. Başlevent, an economist from Bilgi University, questioned to what extent voters hold the government responsible for changes in their economic conditions and to what extent this affects their votes. The paper uses very fresh data from a Konda survey on voting intentions done in February. The paper's econometric regressions show that worsening economic conditions negatively and sizably affect votes for the incumbent party. According to the survey, 31 percent of voters said their economic conditions had worsened (22 percent said their conditions had improved) and 80 percent of this 31 percent hold the government responsible. Nevertheless, the share of voters who voted for the AKP in the last elections who said their conditions had worsened was limited to 18 percent. The bottom line, in my opinion, is that we should not expect a sizable fall in the share of AKP votes (50 percent in June 2011) because of economic factors, though we do not know what the effect of the corruption scandal will be. The paper by Akarca from the University of Illinois, Chicago, is more ambitious since it predicts the share of votes for the incumbent party on March 30. Years ago, Akarca developed quite a sophisticated econometric model based on the data of 27 elections held in Turkey since 1950. I have no room to go into details but recommend readers the articles published by the author on this topic. Let me just say I am impressed by the prediction powers of the model. The model considers the credit or blame the government receives due to economic conditions, advantages and disadvantages of incumbency, political inertia and realignments. It also takes into account strategic voting by the electorate caused by election thresholds and their desire to balance the power of the government. According to Akarca, “the AKP should receive about 45 percent of the vote in the upcoming elections.” This limited loss of votes is particularly due to the fact that economic growth slowed but there is no recession. Taking into consideration the error margins for each determinant, Akarca adds that “a 95 percent confidence interval constructed puts the vote share somewhere between 42 and 48 percent.” I must warn that this prediction does not include the impact of the corruption scandal as pointed out by the author. It is up to your personal opinion to decide how much you have to reduce the share of votes as predicted by Akarca.