The effect of political elections and referendums on stock returns: The case of the ISE

In an efficient market, a stock?s price should reflect quickly all publicly-disclosed information about the underlying firm. There may be many different information types about the firms such as dividend and earnings announcements, mergers and acquisitions, etc. The impact of these types of information on common stock returns has been the subject of many studies in finance literature. On the other hand, the number of studies examining the impact of political events on stock returns is relatively limited.The relationship between financial markets and the political elections has been attracted the attention of the scholars studying in the area of finance and numerous studies have been examined this relationship. The reason for this interest is based on the importance of choosing the incumbent/ prime minister who will affect the country's foreign relations, legislative - executive organs and the economy. Changes in the outcome of the presidential elections and in the composition of the government affect the stock markets at the same time. The stock markets can be considered as the barometer of a country?s economy (Wong & McAller, 2008). Therefore, the first part of the study examines the literature review which are done on the relationship between political elections and stock prices:Niederhoffer and all. (1970) studied the market behavior around the United States presidential elections. They analyzed whether the market prefer Republicans to Democrats due to a belief in the Wall Street. They found out that the average changes in the market was %1.08 to - %0.23 when a Democrat party won and %1.30 to -%1.13 when a Republican party won. According to Furio and Pardo (2010), with the study of Niederhoffer and all. The number of the studies examining the relationship between the presidential elections and the stock market behavior has increased.Rilley and Luksetich (1980) studied the 1900-1976 period in order to analyze the stock market behavior around the presidential elections. In the introduction part of their study, a number of existing beliefs in the Wall Street are discussed. These beliefs were related with the market does not like uncertainty, the market will go a downward trend with the change of the incumbent. The result of the study showed that with the during the victory of the Republicans, the stock prices are higher compared to Democrats. However, although the election results may influence stock prices, there is an uncertainty about the economy's performance is affected by the ruling party.Herbst and Slinkman (1984) analyzed the stock market behavior in the Unites states stock market for the period starting from January 1924 to December 1977. They used the expression ?mirror? and ?molder? for the stock markets. They found the existence of 48?month election period in addition to 24-month election period. But the peak point of the 24 month election period is not the same with the elections. Further evidences from U.S. stock market are provided by Huang (1985) .Foerster (1994) analyzed the effect of American election on the Canadian stock market behavior. In the result of the study, it is found that the response of the Canadian stock to American elections is harder that the response to Canadian elections.Pantzalis et.al. (2000) investigated the behavior of the stock market indices for 33 countries during the political elections from 1974 to 1995. They found a positive abnormal return during two weeks before the elections. Santa-Clara and Valkanov (2003) studied the presidential election cycles and the stock market behaviors in U.S.The most important study examining the relationship between the stock market behavior and the elections in Turkey has been done by Mandacı (2003). The aim of that study was to evaluate the impact of general elections in Turkey on ISE-100. In the study 20 November 1991, 24 December 1995, 18 April 1999 and 3 Novermber 2002 dated general elections have been analyzed. Abnormal returns were observed three days before and two days after the general elections on 18 April 1999. in the first, second and fourth day following the general elections on 3 November 2002 a similar trend has been observed. However, for the majority of the days before or after the elections, a statistically meaningful abnormal return was not observed.The second part of the study consists of the political history of Turkey and the political elections of Turkey. Modern Turkey has a democratic tradition marred by several periods of instability and authoritarian rule (Coleman, 2010). These periods have been examined period by period until the last general election. In summary, after a long instability period, the government in Turkey has been formed by a conservative party called as Justice and Development Party (AKP) in the last three general election. The most important opposition party is Republican Society Party (CHP). By the examination of political history, according to Country Watch editors, the political risk index of Turkey has been determined as 7. This is important in the scope of political risk.In the third part, within the context of this study, political risk refers to the uncertainties resulting from the events having strong political implications such as local and general elections and referendums. Specifically, the aim of this study is to investigate the impact of local and general elections and referendums on the stock returns of the firms listed in Istanbul Stock Exchange (ISE) 100 and also to find out whether any abnormal return exists around the election dates. The sample period starts in January 1989, the date when ISE started to trading, and ends in September 2011 including the general elections 20 October 1991, 24 December 1995, 18 April 1999, 3 November 2002, 22 July 2007 and 12 June 2011; local elections 27 March 1994, 18 April 1999, 28 March 2004 and 29 March 2009; referendums 21 October 2007 and 12 September 2010. The data files provided by ISE have been used for obtaining the daily returns of ISE 100 index. The econometric methodology of event studies suggested by MacKinlay (1997) has been employed.According to z- test result of the abnormal returns around the election dates, in the 95% confidence interval, before and after the general election 1991, there is not any significant difference in the stock returns. 10th day before the 1995 general election and the 1st, 2nd and 12nd day after the election have significant difference in the stock returns. But the 10th day before the election and the 1st day after the election have negative abnormal returns and the rest are positive.The 4th and the 3rd day before the 1999 general election and after the 2nd day, there are significant difference in the stock prices. The 4th day before the election has negative abnormal return. The 11th day before and the 1st , 2nd, 4th day after the 2002 general election, there are abnormal returns. All of those abnormal returns are in the positive direction. In the 2007 general election, the abnormal returns 7th and 2nd day before and 1st and 4th day after the election are significant. Only the 4th day after the general election is negative abnormal return. In addition, there is no significant abnormal return before and after the 2011 general election.Coming to local elections, the 3rd day before the 1994 local election, and also the 5th, 6th, 7th and 8th day after the election, in the 95% confidence level, there are abnormal returns. Only the abnormal returns before the election has negative direction. The 1999 local election has been realized in the same day of 1999 general election and has the same results of the general election. There is not any abnormal return before and after the 2004 local election. Only the 8th day after the 2009 local election has positive abnormal return. Finally, before abd after the 2007 and 2011 referandums, there is not any abnormal returns.When cumulative average abnormal returns (CAAR) are calculated by the help of the t-test, CAAR are especially higher after the elections dates in the other words after the event dates. In the 90 % confidence interval, for the (0,+5) period after the election has significant abnormal returns. For the local elections, (0,+10) and (0, +15) periods are significant for both 90 % and 95 % confidence level.As a result, abnormal returns are observed in the ISE-100 returns around the election dates for the event windows including 15 days before and after the event dates. Abnormal returns are observed especially around the political elections excluding 1991 and 2011 general elections. There are also abnormal returns for 1994 and 1999 local elections in the post and pre-event periods. In addition, there is no abnormal return around the referendum dates. The cumulative average abnormal returns are statistically significant especially for the post-event periods ((0,+5) period for general elections and (0,+10) period for local elections). Considering the abnormal returns that are analyzed before and after the elections, rational invertors can take position. In addition, the rational investor may build a position according to the type of the government. He may have a different position if the government is formed by a coalition or by one party.