By American University.
1. Introduction
The phrase “Wall Street versus Main Street” became commonplace during the recent financial crisis as a shorthand means of describing opposing sides in a variety of contexts, from blame attribution to beneficiaries of government intervention to investor protection. In the aftermath of the crisis, this characterization has become synonymous with the divide between the “haves” and “have-nots”, manifested for example in the sentiment apparent in the Occupy Wall Street movement.
It regularly appears in speeches of policymakers and politicians eager to address perceptions related to the faltering economy. In short, “Wall Street versus Main Street” distinguishes the views of financial insiders from those of the general population. Yet beyond the conventional wisdom and rhetoric, is there really a Wall Street/Main Street divide? On one hand, skeptics might appeal to theories of information flow to argue that financial market transactions are the result of individual investor decisions or standard finance theories that management decisions of publicly-traded firms are a direct result of the desire to maximize shareholder value. On the other hand, proponents would point to the low proportion of active investors in the Main Street population, citing evidence of low financial literacy rates (Lusardi and Mitchell 2011), few Americans holding stocks outside of a retirement portfolio (Poterba and Samwick 1995), and growing income inequality (Heathcote, Perri, and Violante 2009).
This paper examines the relationship between popular and market beliefs in an attempt to measure the magnitude of the Wall Street/Main Street divide. We find surprising similarities between beliefs about future stock market returns elicited from surveys of a representative sample of the U.S. population and market beliefs as calculated through option prices using the Black-Scholes model.
The main questions of interest are:
(1) Do Wall Street expectations influence the views of the general US population (Main Street),
(2) Does Main Street uncertainty mirror Wall Street uncertainty?
The approach taken in this paper draws on literature from finance, behavioral economics, econometrics, and survey methodology, specifically the following areas:
(1) the formation of equity return expectations,
(2) the information content of option-implied probabilities,
(3) the information content of subjective expectations, and
(4) the tendency of survey respondents to report focal points (clustering around rounded numbers) when asked probabilistic questions.
Researchers using survey data find substantial heterogeneity across individuals with regards to their expectations about future equity returns, although for any given individual, reported expectations appear to be relatively stable but by no means constant over time (Brennan, Cao, Strong, and Xu 2005; Dominitz and Manski 2011; Hudomiet, Kézdi and Willis 2011; Hurd and Rohwedder 2012).
Some of the interpersonal heterogeneity is attributable to differences in optimism among population subgroups: for example, women, African Americans and those in the lower education categories have less optimistic expectations relative to the overall population.
This expectations heterogeneity has in turn been used to explain heterogeneous investment in equities (Kézdi and Willis 2003, 2011). A separate literature gleans expectations of market participants from option price information, following the results from an early paper by Breeden and Litzenberger (1978) deriving state-contingent securities from combinations of options, thereby demonstrating that corresponding prices of these securities are implied by option prices: specifically, the price density of the state-contingent security evaluated at a strike price is equal to the second derivative of the price of a European call option with respect to the strike price.
A variety of functions for the price of an option are admissible in this context; the most often used one is the option pricing model described by Black and Scholes (1973) and Merton (1973) and we follow that convention in this paper. A number of researchers consider the inclusion of survey expectations in models of economic behavior (see Manski (2004) for a survey of this literature) and demonstrate that including probabilistic expectations can improve inference about economic behavior relative to models using only data on economic choices (revealed preference models).
Yet others show that survey responses do not exactly align with true expectations – for example, due to large clusters of responses occurring at “focal points” of the response distribution (e.g., Dominitz and Manski 1997; Hurd, McFadden, and Gan 1998; Kleinjans and Van Soest 2010) – and argue that adjustments to survey data to account for such aspects are necessary to improve inference (Bassett and Lumsdaine 2000; Lillard and Willis 2001).
While much of the literature comparing expectations to economic behavior is in the context of analyzing consumer intentions (e.g.,buying a new car, see Juster 1966), a similar approach considers expectations about equity markets and investors’ behavior (Hurd and Rohwedder 2012). The paper proceeds as follows: Section 2 describes the data construction and descriptive statistics of the main variables.
Section 3 describes the model.Section 4 examines the results of the regressions performed. Section 5 compares the predictive accuracy of Wall Street and Main Street beliefs. The final section concludes.
2. Data
The American Life Panel (ALP) provides a novel dataset with which to investigate the research questions. An internet panel with about 3000 active panel members, the ALP contains more than 200 survey modules (and associated survey weights designed to ensure a nationally representative population) administered by the RAND Corporation. A more detailed description of the ALP sampling frame, survey population, interview length, response rate and participation incentives is in Appendix A.
While some of the survey modules are stand-alone, others belong to periodically-repeated series (waves) on the same topic. This paper uses responses obtained from modules designed by Michael Hurd and Susann Rohwedder to investigate the effects of the financial crisis on American households, gathered from November 5, 2008 until March 10, 2011, corresponding to 25 waves of information.
Hurd and Rohwedder (2010) provide a detailed description of this series of modules; they are briefly summarized here. The first wave asks respondents about a wide range of topics such as labor force status, stock ownership, mortgage payments and expectations about the future. The second wave was conducted in February 2009 and subsequent waves have been conducted monthly from May 2009 onwards. Each module also contains demographic control variables such as age, race, gender, marital status, and education.
The final sample (after adjustments for, e.g., missing observations) consists of 47,488 surveys from 2,652 respondents (94.9% of the total number of surveys and 98.3% of the total number of 6 respondents) gathered over 364 survey days. The sample construction is also detailed in Appendix A1.
2.1 What Main Street thinks: survey expectations about future stock market performance As a proxy for the views of “Main Street”, the ALP elicits expectations about the stock market from survey participants via a series of questions, the first of which is the following (hereafter referred to as the “PositiveReturn” question):
“We are interested in how well you think the economy will do in the future. On a scale from 0 percent to 100 percent where “0” means that you think there is absolutely no chance, and “100” means that you think the event is absolutely sure to happen, what are the chances that by next year at this time mutual fund shares invested in blue chip stocks like those in the Dow Jones Industrial Average will be worth more than they are today?”
Respondents can give an answer ranging from zero to one hundred (the answer need not be an integer) to indicate the percentage chance of the event happening, as well as leaving the response blank. If a numerical response is not recorded to this question, the question is asked a second time, preceded by the additional instruction “You did not answer. Your answers are important to us. Please give us your best guess.”
To this second question, either (a) a response ranging from zero to one hundred, or (b) “don’t know” is recorded. The same structure of questions is repeated for a greater than 20% return and a greater than -20% return. For expositional ease, the questions referring to the probability of a positive return, a more than 20% return, and a more than -20% return will be referred to as PositiveReturn, >Plus20, and >Minus20, respectively. Using all three questions (when available) from above 47,488 surveys yields a total sample size for studying Main Street probabilities of 139,327 observations.
The phrasing of these questions may lead to differences in respondents’ interpretation (and hence the answers they give, see, e.g., Tversky and Kahnemann (1981)), since there is an implicit subjectivity associated with respondents’ understanding of “mutual funds shares” or “blue chip stocks like those in the Dow Jones Industrial Average (DJIA)”. For the purposes of this paper, however, it is necessary to assume that the responses given represent respondents’ subjective probability that the nominal (not inflation-adjusted) level of the DJIA in one year will be [greater / more than 20% greater / more than -20% greater] than the current level of the DJIA.
For each respondent, the current level of the index is assumed to be the closing level on the most recent business date prior to the date of interview, so that the response is assumed to be based on information available at the time of interview
For more about this report: go to: http://www.american.edu/cas/economics/news/upload/Lumsdaine-paper.pdf
