A New Look at the Equity Premium. A Survey of Collegiate Financial Faculty

Abstract

This paper deals with the difference of potential returns of diverse assets within a long-term period. Recent articles (e. g. Siegel/ Thaler 1997) investigated an unexplainable abnormality of long term returns of stocks in comparison to risk-free securities even if stocks were risk- adjusted. This phenomenon is called the Equity Premium Puzzle, which refers to the empirical fact that stocks have outperformed bonds over the last century by a surprisingly high large margin. For instance, if one had invested $1000 in Treasury Bills in 1925, where the money remained until 1995, one would have earned $12,720. On the other hand, if one had decided to invest the same amount of money in a portfolio of stocks within the same time period, one would have received $842,000, or 66 times as much money. Further articles of Siegel (1991, 1992) argue that the equity premium does not appear to be a recent phenomenon by analyzing three different time periods (1802-1870; 1871-1925; 1926-1990). Also MaCurdy/ Shoven (1992) document similar results.

With this paper we want to investigate how current investors make use of the disclosures of these papers. Since one assumption of capital markets is that any information is available to any investor in the world, any investor who invests on a long term basis would act irrational if she invests within the bond market, while she could earn a lot more money in the stock market. Furthermore this fact of the existence of not rational acting investors might show that capital markets are not that efficient as expected or information among investors are not shared sufficiently through markets as an earlier study of Noeth/Camerer/Plott/Weber (2001) showed. On the other hand if investors are aware of the Equity Premium Puzzle demand of stocks relatively to other assets will be disproportionately high. Again this might cause the abnormality of long term returns of stocks, since also demand affects stock prices and therefore returns.

In our study we created a questionnaire of about 22 questions. All questions which were applied were held superficially, so that the respondent had no clue which goal we were looking for with that survey. We distributed our questionnaire to all available investors of all different ages. However we tried to focus on those investors with a long time investment horizon, such as students, pupils or early workers in the end 20’s or less. Since the Equity Premium Puzzle is not only a national issue, we did our survey not only in the United States of America but also in Germany. An interesting result was that not many of the respondents were aware of the Equity Premium but want to apply this in future when considering another investment. This opens another opportunity for further research, since this behaviour will also affect the stock prices and returns. This can help to figure out the reason why stocks earns that abnormal higher return in comparison to bonds.

The Paper

This article presents the results of a survey of 116 finance professors regarding their assumptions and opinions of the equity premium puzzle. The survey was activated online on March, 1st and was available until March 15th 2007. During that period a slight crash occurred in the U.S. stock markets. However, all indices were able to recoup at least partially, as can be seen today.

Additionally, some earlier research by Fama and French in 2001, Shiller in 2000 and Welch in 2001 may have affected investor’s opinions about the equity premium. Welch investigated that the consensus of “finance professional’s” outlook concerning the equity premium was higher than the premium actually was. Thus, with a market downfall that may have resulted in finance professors being more pessimistic, we investigated further the comments made by Welch. In addition, we posed questions that may have further implications regarding other theories.

The true value of the equity risk premium is often at the center of constant discussion among academics and researchers. The value of the equity risk premium is the degree to which risky assets, or stocks, are expected to outperform relatively risk-free assets, such as bonds. Generally, there has been very little consensus concerning the value of the equity risk premium because of the role that varying, unobservable agent expectations play in shaping market portfolio and risk-free returns. The true value of the equity risk premium contends as “one of the most important but elusive quantities in finance” (Pastor & Stambaugh 2001). This can be duly noted as several plausible theories that attempt to explain the quantity are explored.

Research has been conducted to find historical, yet varying, views on the true value of the equity risk premium. These studies have evaluated the phenomenon mostly in the United States but have also touched on other countries as well. The value of the equity premium in the U.S. has ranged from 4.3% on average from 1870 to 1998 (Shiller 1989) and 5.8% during the entirety of the 20th century (Dimson 2000) to 6% during the 1889 to 1978 period (Mehra & Prescott 1985) to even between 8.2% and 8.6% in 1995 (Brealey & Myers 1996). Alternatively, in Germany, Canada, France, Japan, and the United Kingdom, the equity premium has been measured as low as 3% (Claus & Thomas 2001) and as moderately-high as 4.6% in Canada, 7.7% in France, and 4.9% in the United Kingdom for the entirety of the 20th century (Dimson 2000).

The first method used to describe the source of the equity premium involves the use of utility in consumer consumption patterns. Research has been conducted that depicts utility received by a consumer today is an elastic function of current consumption as well as future utility (Epstein & Zin 1989, 1991). The authors determined that covariance with market return and consumer consumption growth help to resolve systematic risk, or the varying returns on any two assets. Epstein and Zin conclude that investors achieve the higher rate of return on risky assets because of their effect on consumption growth as well as the covariance described above.

Additionally, research describing habit formation as a possible explanation for the equity premium puzzle has been proposed (Constantinides 1990). Similar to the work of Epstein and Zin, Constantinides contends that utility, in monetary terms, is actually a decreasing function of consumption in previous periods. This means that it takes more for a consumer to be happy today if he were to consumer more yesterday. This research concluded with the thought that should one’s demand for savings increase, the immediate effect would be a lowering of relatively risk-free investment yields, such as government bonds, notes, and bills.

One of the more prominent arguments for this type of explanation involved referencing psychological preferences in financial decision-making. The research depicted agents as having displayed loss aversion in their financial decision-making and consequent results (Benartzi & Thaler 1995). This simply implies that investors show significantly more hurt when faced with losses than more pleasure when faced with gains. Loss-aversion, then, can be used to explain a switch to relatively risk-free investments during periods of potentially weak risky investment performance. Consequently, the research can be followed to describe a now-required rise in risky investment returns in order to attract agents.

The most common explanation put forth by researchers involves the basic risk reward tradeoff. A study performed suggests, as is commonly taught, that the equity premium exists in conjunction with risk aversion on the part of investors (Mehra & Prescott 1985). Simply stated, this suggests that an investor’s avoidance of risk is positively correlated with his or her desire to invest in relatively risk-free assets. In a similar manner, research has also been conducted that concludes that stocks deserve higher returns because one will not invest without appropriate compensation for assets that are “far more risky” than government issues (Abel 1991).

Stock returns have also been modeled as leveraged claims against the firms invested in (Benninga & Protopapadakis 1990). These authors also suggest that, in order to adequately measure the equity premium, investors must possess a high amount of risk aversion. Further research follows these ideas and states that average returns of stocks co-vary more with consumption growth than do relatively risk-free assets (Kocherlakota 1996). This risk-aversion argument supports risk as the primary factor needing consideration when determining the equity risk premium.

Thus far, possible explanations haven’t focused much on the lower end of the premium spread, or the risk-free rate. Prominent research was conducted that suggested the equity risk premium has much to do with changing real returns of bonds during times of inflation (Siegel 1998). Siegel also concludes that stock returns will move with prices that also rise and will act as an effective hedge to inflation. In support of this, additional research described rising inflation as a determinant in increasing long-term interest rates and lower bond prices (Buckley 1999). Thus, relatively risk-free assets will return much less during inflationary times when compared to risky investments.

Several additional arguments exist for use in explaining the equity premium. Research has been done concerning limited market participation and its effect on the equity premium puzzle (Polkovnichenko 2002). It is derived from this research that those individuals who have a low level of income choose to not participate in the equity (more risky when compared to bonds) market. This lacking participation, it is concluded, results in limited sharing of risk over investment opportunities but fails to explain the equity premium puzzle. In further research, three-quarters of United States households were classified as non-stockholders, attributing themselves to the idea of limited market participation (Mankiw & Zeldes 1991).

Finally, the trading cost model has been presented as another solution to the derivation of the equity risk premium. Authors suggest that the higher costs associated with trading risky assets versus relatively risk-free assets exist due to the existence of transaction costs (Heaton & Lucas 1996). It is believed that these costs are one of the major causes of the relatively high historical equity premium over the risk-free rate (Fisher 1994). To compliment this, it has also been suggested that illiquidity, coupled with transaction costs, cause relatively risk-free assets to be overly invested in when compared to other, more risky, assets (Swan 2001).

This paper revisits earlier results by also considering expected performance of diverse stock indices coupled with different investment strategies. Our survey received 116 responses from finance professors in North America. The survey includes 20 questions which were divided into three sections. The first section dealt with investment strategies and accounted for eight questions. Section two discussed the equity premium puzzle, consisting of seven questions. And in the last section the respondent was asked to answer five questions concerning demographics. The survey is shown in Figure I. We used www.surveymonkey.com to host the survey online. The link to the survey was http://www.surveymonkey.com/s.asp?u=127943376178 and was only accessible to the finance professors that we contacted via e-mail during the timeframe described earlier.

We sent our email, which can be seen as Figure II, including the link required to access the survey to roughly 2,000 professors from universities in North America (U.S. and Canada exclusively). Thus, the response rate accounts for 5.8% percent. All of the respondents answered the questions appropriately so that we need not exclude any of the respondents. On March, 15th the survey was closed and the results were downloaded to a Microsoft Excel spreadsheet available directly from Survey Monkey’s user interface.

Some of the results needed to be “cleaned” in such a way that various answers were formatted appropriately so that responses were uniform. Cleaned results were then converted into yet another Microsoft Excel spreadsheet. For example, a typical case which required “cleaning” was professor estimates of the three leading U.S. indices’ developments. Some respondents answered with a percentage increase of the indices while others answered with an index point increase. These heterogeneous responses were standardized into point-based expectations. Percentage increases were based on a DJIA level of 12,400, an S&P 500 level of 1,400, and a NASDAQ level of 2,500.

After the answers were converted into the new Microsoft Excel spreadsheet, we ran partial regressions among all variables using SPSS, where variables correspond with the questions posted on the survey. This was done to disclose both correlations among the variables and correlations between the variables, all the while seeking general information regarding the equity premium puzzle. Since some of the questions implicated similar meanings we reduced some of our variables to be inclusive. Cases of this included respondents answering Question 10 with either Daily updates or Daily updates coupled with desktop tracking. These two answers were considered similar.

The survey was anonymous. That is, solely the Microsoft Excel spreadsheet and its contents from www.surveymonkey.com were downloaded and analyzed. This spreadsheet included the summary of all answers posed in the questionnaire and nothing else. We had no access to any personal information of any type aside from what was provided us by participants in the original survey.

The answers we were seeking that revolved around the equity premium puzzle, were limited in scope. Obviously, when receiving 116 responses from a base of 2,000 possible response candidates, our research was not as conclusive as we would have liked. We sought to gather enough information to validate previous theories concerning a large equity premium occurring during times of bullish markets. The markets, however, hiccupped slightly during the time of this survey after experiencing relatively sizable gains. The DJIA had even recently topped 12,000 and had captured several record-breaking values.

By measuring investment strategies couple with assumptions concerning the equity premium, stock returns, and T-Bill returns we also hoped to garner more insight into the actual varied sizes of the equity premium during similar economic times. By measuring expectations and assumptions from the “top” down, that is from the more educated and familiar, we looked for ways to compliment previous theories that have attempted to explain fully the equity premium puzzle. Again, our response rate was a limitation. We did, however, gather interesting secondary-data from the respondents of the equity premium survey.

One measure that we have been able to derive from our survey is that of asset allocation in actively managed investment portfolios. Percentage allocation was measured in four categories: stocks, bonds, liquid assets (e.g. treasury bills), and other (e.g. real estate). On average, respondents reported allocations of stocks at 66.47%, bonds at 13.79%, liquid assets at 11.19%, and other investments at 8.60%. Asset allocation can be viewed below in Figure III.

Insert Figure III about here.

To compliment this, on average, respondents expected a 19.79% increase in the point value of the DJIA, a 21.68% increase in the point value of the S&P 500 and an 18.36% increase in the point value of the NASDAQ. Given recent sentiment and expectations that stock indices will continue to exhibit strong performance, it is likely that these investors are demonstrating low risk aversion during times of bull markets.

The survey also measured the frequency to which respondents balanced and rebalanced their investment portfolios to match their respective asset allocations. Frequencies were categorized into five groups. These included: weekly, monthly, quarterly, annually, and other (e.g. never, seldom). Of the 116 respondents, only 2 rebalanced weekly, 4 monthly, 16 quarterly, 49 annually, and 45 other. These results are summarized below in Figure IV.

Insert Figure IV about here.

These results imply long-term investment strategies amongst collegiate finance faculty. Additionally, long-term strategies provide little opportunity for rebalancing should the equity premium change rapidly; it would take at least an entire year for most of these respondents to react. To compliment this, the survey also measured the extent to which respondents were turning over their portfolios.

Portfolio turnover was measured from less than 10% to over 100% with 10% intervals (e.g. < 10%, 10%, 20%, etc.). Overwhelmingly, 64 respondents responded with less than 10% portfolio turnover. Also, 15 responded with 10% turnover, 13 with 20%, 9 with 30%, 4 with 50%, 2 with 70%, 1 with 80%, 2 with 100%, and 3 with greater than 100%. Again, this implies a long-term investment horizon. The results of this question are summarized below in Figure V

Insert Figure V about here.

The survey also measured investment horizon. And, as expected, the term for most respondents was long to very long. Horizon was categorized as either very short-term, somewhat short-term, intermediate-term, long-term, and very long-term. Long-term and very long-term comprised 93 of the responses (47 long-term and 46 very long-term), while 14 invested at the intermediate horizon, 6 at the somewhat short horizon, and 3 at the very short horizon. The results are summarized below in Figure VI.

Insert Figure VI about here.

So, who is it that is so long-term oriented in their investment allocation, turnover, and horizon? We asked the 116 respondents who was in charge of their portfolios. The survey included choices of full-service stock brokers; 17.24% of respondents, discount stock brokers; 56.03%, certified financial planners; 6.03%, other financial advisor (e.g. yourself); 19.83%, and none; 11.21%. Results are summarized below in Figure VII.

[...]