3. Literature Review
3.2. Presidential Elections and Market Cycles
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.2. Presidential Elections and Market Cycles
Presidential elections and market cycles have been a topic of study for a long time, especially in the United States where different analysis have been conducted to identify cycles in the US markets (mostly indices), understand the reason of this occurrence and see whether or not the party in power has an impact on these cycles. Throughout times, different cycles have been demonstrated: the 28-day trading cycle that many markets appear to follow, the 10.5-month Futures cycle which is a good reflection of the “overall behavior” of commodity prices, the January effect stating that the market will follow the trend (rising or declining) seen in January for the rest of the year (with an accuracy rate of 86% between 1950 and 1993), the Juglar 9.2-year cycle existing in many areas of economic activity and finally the Kondratieff “wave”
which is a long-term cycle that only had the time to repeat itself three times in the stock market so far. Finally, the last one, that some authors like Wing-Keug Wong and Michael McAleer liked to call “the most popular cycle” is the four-year cycle also called presidential election or Kitchin wave. Different studies have been conducted to understand and interpret these 4-year cycles. Many analysts found significant the difference of the returns between the first and second half of the presidential term. In their study for example, Foerster and Schmitz found results that are consistent with most of the analyses that have been done on this topic: The stocks returns are lower the second year of the presidential term. Different explanations have been submitted for the cause of this presidential cycle, especially the assumptions made by Nordhaus (1975), Sachs (1988) and Willet (1989) stating that political parties manipulate business cycles to win elections as they have the incentive, after they have been elected, to pursue deflationary policies and on the other way around, pursue policies that would stimulate the economy to be elected again. Hibbs (1977) and Rogoff (1990) have also come out with
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
alternative explanatory theories that we will develop later in this part.
The first article I would like to introduce among all the studies that have been conducted to establish a relationship between presidential elections and the cycles occurring on the market was written by Marshall Nickles in 2004 under the name of Presidential elections and market cycles and is, I believe, very relevant for this literature review. The writer aims to analyze in this paper the relationship between the stock market cycles of the S&P 500 index in the United States and politics, more specifically the presidential elections. He used a “Peak-Trough”
method to identify the cycles of the S&P 500 which regroups the 500 top U.S companies and generally used as a benchmark for tracking the overall stock market. His analysis has been conducted between 1942 and 2002. It is relevant to say that the price data for the index (S&P 500) was compiled and averaged on a weekly but not on a daily basis. His first findings were that full cycles happen every 4.02 years in average for the index divided in “Bull” periods of 3.08 years, which is, in average, the period the market will rise and “Bear” periods of 0.94 year, which is, in average, the period the market will decline.
In his analysis, the author then tried to find a link, a relationship between the cycles of the index that approximates four years and the U.S presidential elections. He did his analysis focusing on the market troughs, identifying the year during the presidential term when the market bottomed.
The results of his study are very clear: Troughs occur at an average of 1.87 years into the presidential term (around 2 years). The writer does not tend to explain the reason why the troughs repeatedly occur at that specific time of the presidential term but wants to come out with a precise recurring pattern on financial market. The final purpose of the article is to give potential investment strategies that are based on the political cycles he identified in his study. In fact, he proposed two alternatives investment strategies based on the time the trough occurs in
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the presidential term. To better explain, he took two different periods of the cycle:
1) After the trough (27 months) - From October the 1st of second year of presidential term through December 31 of election year. This is the investment strategy number 1.
2) Before the trough (21 months) – From January the 1st of inaugural year through September 30 of second year of Presidential term. This is the investment strategy number 2.
The author realized a simulation for both strategies with an initial capital of 1,000$ with a recurring investment in the S&P 500 during the specific time period he allocated to the different strategies, and this, for each presidential election. His final results are eloquent: The first investment strategy brought 7,170% of return on the initial investment while the second one incurred a loss of 36 % over the 5 decades of study. It seems pretty clear that the link between market cycles and presidential terms can bring opportunities for investment strategies.
Nevertheless, the writer warns us from the possibility of other patterns and cycles, and explicitly tells that this paper does not intent to forecast the stock market.
This first paper has given us the main insights of what we should know about the presidential cycle. In this paragraph, we will focus on the revised version of the research paper Mapping the Presidential Election Cycle in US Stock markets, written by Wing-Keug Wong & Michael McAleer. This one will be relevant in our literature review as the main purpose of the writers was to show the close relationship between presidential election and the US stock prices, trying to identify cycles and point out the existence of a peak at a certain time of the presidential term period. Their study is also going further than the previous article analyzed in this literature review. In fact, they aim to explain the existence of those cycles and find an answer in what they
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
called “policy manipulation” by the politics. They also try to suggest which party, between the republican and democrats, may have greater cause to engage in this kind of “manipulation”. We will come back on these points later in our analysis.
To conduct their research, they will use a spectral analysis to estimate the period of the cycles and adopt the EGARCH model to examine them, as they want to take into account the time dependence and the conditional heteroskedasticity in stock returns, what t-test and simple regression analysis ,which are normally used for this type of analysis, don’t.
The data used for their analysis is the S&P index, on a weekly basis from 1 January 1965 to 31 December 2003. The Data has been extracted from DataStream. Before realizing a “spectral analysis”, authors want to avoid bias on their results by checking the stationary property of the S&P index, its returns but also its logarithm. It is proved as only the stock returns are stationary.
Hence, the cycles will be analyzed with the returns series. The existence presidential election cycles is confirmed with the spectral analysis. It is shown that the stock index trough will occur during the second year (It concords with the first analysis of Nickles who found out a through was happening at an average of 1.87 year). Nevertheless, we also learn something new in this study. The peak is reached during the third of fourth year of the presidential term. It would suggest that the peak is reached right before the election event.
Those latest analyses come to confirm the one that have previously been done on this topic. In fact, in an early study of the relationship between stock market returns and the presidential election cycle, Allvine and O’neill built their analysis extracting data from the S&P 400 between 1948 and 1978 with the main purpose to analyze the returns of the index:
According to their findings, the first two years of the presidential term, the returns of the index
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
averaged respectively 0.6% and 0.7%, which is much lower than the returns observed in average for the second half of the presidential term (year 3 and 4). In fact, the results show average returns of 22,1% for the third year of the presidential term and 9.2% for the fourth year.
This discovery suggest the pattern of a presidential election cycle divided in two parts with an increase of returns (stock index price) during the second half of the term and a decrease of returns (stock index price) during the first half of the term. The average increase is the highest during the third year, according to their study. Five years after, the data used by Mr. Roger Huang showed that stock returns were systematically higher during the second half of a political term and this, from 1832 to 1979. In 1995, Gartner and Wellershoff identified this four-year cycle for the US stock price with an increase noted during the second half of the presidential term and a decreased noted during the first half. Their findings also show that this rule is applicable for different indices but also under both the rule of Republicans and Democrats. The presence of this presidential cycle is then confirmed, according to their study, no matter the political party ruling the country (applicable for the United States).
Democrats and Republicans are the two main parties in the United States and different comparisons have been done between them to see if the party in power had an impact on this four-year cycle around presidential elections. In the paper Presidential Politics, Stocks, Bonds, Bills, and Inflation Johnson, Chittenden and Jensen examine the returns of the S&P 500 index, from January 1929 to December 1996 and include the Democrats/Republicans factor in their analysis. The particularity of this study is that they use a wider range of data in order to compare them. They use both a small and large-stock index in order to incorporate the size effect in their analysis. Plus, they include bonds (intermediate and long term ones) to their study (former ones were basically using stocks): from corporations and the US government (T-bills).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The findings of this study show that the S&P 500 return is higher during Democrats’ terms than Republicans’ ones. Nevertheless, t-test and Wilcoxon test (parametric and non-parametric tests), used to analyze the relevance for the differences in returns, are found not statistically significant.
It seems then for the writers that there is no clear relationship between stock returns and the political party the president belongs to. Nevertheless, this statement, accurate for the S&P 500, is not true for the small-stock index. In fact, in their study, the analysts found that returns for small-cap stocks are significantly substantially higher during Democratic administration (around four times higher than during Republican administrations).
It seems, according to their results, that the equity and debt indices react differently to the political party in power. Indeed, the returns for debt index are over twice as high during Republicans than Democrats administration. The writers don’t give any approach to explain those findings but conclude their analysis comforting the previous analysis made regarding the four-year cycle: the returns are significantly higher during the second half of the presidential term. They add that the returns are higher, no matter the party elected but “are higher under Democrats elections”. According to them, this difference of returns, dividing the presidential term in two cycles of 2 years each, apply to the equity indices but is not significant when we consider the debt indices (bond returns).
In their study, Wing-Keug Wong & Michael McAleer also suggest that the second and third year intervention (from the government to stimulate the market) is significant only when the Republicans lead the country as it does not seem to be the case for the Democrats, according to their study. On the other way, the writers suggest in their study that the stock market seem to perform better under Democrats administrations.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
We have now seen that a four-year cycle has been identified and Mr. Hirsch focuses, in a research paper published in 2013 on the AAII Journal, on three main seasonal and cyclical patterns he uses to manage his portfolios. He reminds at the beginning that he is a “strong proponent of historical and seasonal market patterns” but he also warns us also saying that the history is not the same from year to year and these methods cannot be used to forecast the market with a complete accuracy. The main reason of the importance of this article in our literature review is the dedication of a part to the four-year presidential election/stock market cycle.
The writer starts to remind what has been found by most of the persons studying this cycle, as we explained in our literature review. In his own words, “presidential elections every four years have a profound impact on the economy and the stock market. Wars, recessions and bear markets tend to start or occur in the first half of the term, with prosperous times and bull markets in the latter half. This pattern is most compelling”. In his analysis, he uses this time a different index, the Dow Jones. Nevertheless the results are totally similar as the ones previously found by the authors we discussed: since 1939, the third year of the presidential term has the best performance. We can even say there have been no Dow Jones average losses in the year preceding the presidential election.
What makes this article more special is the second part of the article where the author wants to explain the reason of this four-year cycle. For him, the government “manipulates” the economy
“to stay in power” and it has, obviously, a direct impact on the stock market. It seems that governments tend to implement the difficult initiatives, laws and policies at the beginning of their term. The reason for this is that they will still have the other half of the term to “catch-up”
in the eyes of the electors. According to Mr. Hirsch, the technique consists in jiggling fiscal
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
policies to get more disposable income, social security benefits and federal spending while having lower interest rates and inflation. The main purpose is to make the economy looks good to people so that they can vote again for incumbent candidate. It seems then that the movements of the four-year cycle would be explained by the intervention of the US government after the midterm congressional election (especially on fiscal policies) in order to make the economy better, or I would say, looking better than it actually is, to get as many voters as possible.
On the same way, in an updated analysis published in 2012, Nickles (in collaboration with Granados) wants to explain the reason of this four-year cycle as he did not do it in his first paper of 2004 which was, as we have seen in this literature review, an analysis conducted to identify market cycles and recommend some investment strategies. In this new paper, the authors remind that the fiscal policies can be defined as “an increase or decrease of taxes or government spending”. And it seems again pretty clear for the writers that when the presidential election approaches (second half of the term), the government proposes tax reductions and an increase in spending for some specific government programs and this, in order to gather the maximum electors and votes before the new election. They also suggest that governments can also ask some help from the Fed, reducing interest rates and increasing money supplies.
As we can see in some of the research paper selected to develop our literature review, some authors tried to understand the reason of occurrence of this four-year cycle. It was then important to focus more precisely on the three main theories that have been proposed since the first proposal made by Nordhaus in 1975:
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
The opportunistic political business cycle model
This model aims to show that the government in power (also called incumbent) tries to manipulate the economy. The reason of this manipulation can be found in the quest of re-election for the current government. Expansionary monetary policy would be used before the presidential election in order to increase the economic activity and get people votes. This is first Nordhaus (1975) who developed this model, later supported by Sachs (1988) and Willet (1989).
According to them, the government stimulates the economy before the presidential election period and pursues deflationary policy after. A strong assumption to the theory proposed by Nordhaus is that people (voters) are “myopic” and their behavior is “backward-looking”. It means the voters’ behavior depends on the recent past economic performance so as their expectations regarding inflation. Considering those elements, the current incumbent could impact the inflation-unemployment cycle (cf. Phillips and the inverse relationship between employment and inflation) to make it reach a peak prior to the presidential election (lower unemployment and higher inflation) and as a result, get the votes from the majority.
Nevertheless, this model has received several critics. The first one is that the incumbent has the control of monetary policy. The principle of independence of the central bank is then not respected. The second critic relies on the fact that the model supports the irrationality of the voters. Finally, reducing unemployment with inflation surprises is also questioned.
The political budget cycle
This model, developed by Rogoff in 1990 differs from the previous one as it focuses on a fiscal side but not on a monetary one. Indeed, Rogoff makes a clear distinction between the role of the incumbent and the central bank regarding monetary policy: they are independent. The author
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
explains in his political budget cycle model that the incumbent use fiscal policy (not monetary at the difference of the previous one) to have a positive impact on the economy and at the same time, on voters’ choices for the upcoming election. There are different ways to impact economy through fiscal policies. The following examples are the most common ones: increasing government investment and reducing tax.
The partisan theory
The first approach of this theory has been realized by Hibbs in 1977. Alesina continued its development in the late 1980s. This model differs from the political business cycle in the way that it assumes a political system with two parties. Both of them have different characteristics and promote different policies as they have different positions on economic issues. The difference can be found between the “right-wing” and “left-wing” party in the way that the left-wing party is considered having greater target for inflation than the right-wing party. In
The first approach of this theory has been realized by Hibbs in 1977. Alesina continued its development in the late 1980s. This model differs from the political business cycle in the way that it assumes a political system with two parties. Both of them have different characteristics and promote different policies as they have different positions on economic issues. The difference can be found between the “right-wing” and “left-wing” party in the way that the left-wing party is considered having greater target for inflation than the right-wing party. In