Excerpt
Table of Contents
1 Introduction
2 What Theory predicts about Households Stockholding Decision
3 Contributions to Explain Limited Stock Market Participation
3.1 Deviation from Standard Expected Utility
3.2 More Restricted Borrowing Constraints
3.3 Participation and Entry Costs
3.4 Lack of Financial Awareness
3.5 Financial Illiteracy
4 Achieve More Stock Market Participation
4.1 Advertisement
4.2 Targeted Financial Education
4.3 Cost Reduction
5 Conclusion
Bibliography
1 Introduction
The field of household finance is confronted with the phenomena called “stock market participation puzzle” (Haliassos, 2003, p. 32) which describes that most households do not hold stocks, despite that there are higher expected returns on stocks than on risk free assets (Haliassos, 2003, p. 31-32). According to Haliassos and Bertaut (1995), 75% of households in the United States do not own stocks directly although returns on equity exceed the returns on risk free assets. This issue does not only concern households themselves, financial intermediaries, and stock issuers. It is also relevant for questions concerning privatisation, asset pricing, and tax rates on capital gains (Haliassos and Bertaut, 1995, p. 1110). The participation rate can directly affect the equity premium as argued by Mankiw and Zeldes (1991), p. 99, p. 109-110, and Heaton and Lucas (1999), p. 227-232.
Stock market participation in this seminar thesis refers to households participating in the market by holding stocks, excluding retirement accounts. The purpose is to demonstrate five factors discussed in financial literature that contribute to solve the stock market participation puzzle or reasons why limited stock market participation can be observed1: Deviation from standard expected utility (Haliassos, 2003; Epstein and Zin, 1990; Haliassos and Hassapis, 1999; Diecidue and Wakker, 2001; Haliassos and Bertaut, 1995), more restricted borrowing constraints (Constantinides, Donaldson, and Mehra, 2002; Cocco, Gomes, and Maenhout, 2005; Haliassos, 2003), participation and entry costs (Guiso and Japelli, 2004; Gollier, 2002; Haliassos and Michaelides, 2003; Paiel- la, 2001; Vissing-Jorgensen, 2002; Haliassos, 2003), lack of financial awareness (Guiso and Jappelli, 2004), and financial illiteracy (van Rooij, Lusardi, and Alessie, 2007; Arrondel, Debbich, and Savignac, 2012; King and Leape, 1987).
There are more factors affecting the participation decision that are discussed in financial literature, for example trust in the stock market (Georgarakos and Pasini, 2011, p. 722; Guiso, Sapienza, and Zingales, 2008, p. 2592), intelligence (Grinblatt, Keloharju, and L innainmaa, 2011, p. 2161-2162), gender effects (Barber and Odean, 2001, p. 289), religion (Renneboog and Spaenjers, 2011, p. 124), and the internet (Bogan, 2008, p. 208). However, this seminar thesis focuses on the five factors mentioned above.
After showing how and why these factors keep households away from the stock market, three possible solutions in order to achieve more stock market participation will be of- fered: Advertisement (Grullon, Kanatas, and Weston, 2004; Banks and Tanner, 2002; Merton, 1987; Guiso and Jappelli, 2004; Haliassos, 2003; Haliassos and Bertaut, 1995), targeted financial education (van Rooij, Lusardi, and Alessie, 2007; Bayer, Bernheim, and Scholz, 2009; Haliassos, 2003), and cost reduction (Alan, 2006; Guiso and Jappelli, 2004; Haliassos and Michaelides, 2003; Haliassos, 2003).
The final section summarizes the main results and broaches some additional issues that still require theoretical and empirical research in order to get a better understanding of the stock market participation decision of households and to find effective ways to achieve more participation.2
2 What Theory predicts about Households Stockholding Decision
According to Haliassos (2003), given that a household maximizes its expected utility and is risk-averse, financial theory under the standard assumptions3 predicts that this household is always willing to invest a marginal amount in stocks if returns on stocks are higher than on the risk-free asset (Haliassos, 2003, p. 32). While this can be proven technically as in Haliassos and Bertaut (1995), p. 1113-1114, Haliassos (2003) pro- vides an intuitive explanation why investing a marginal amount in stocks is optimal for an expected utility maximizing household. In addition to assessing assets according to their expected returns, a risk-averse household also assesses all assets according to their effect on the “consumption stream” (Haliassos, 2003, p. 32). Given two conditions the household will prefer the risky asset over a riskless asset in order to maximize its ex- pected utility. The first condition is that the expected return on the risky asset exceeds that of the riskless asset. The second condition is that the household values the risky assets “contribution to the expected return on the portfolio” (Haliassos, 2003, p. 32) more than its risk contribution to the consumption stream. For a household that holds no stocks, initially, consumption is not correlated with stock returns (there is no risk con- tribution of stocks to consumption) and adding a marginal amount of stocks does not add consumption risk. Therefore a non-stockholding household would prefer adding a marginal amount of stocks to adding a marginal amount of the riskless asset. Participat- ing in the stock market is preferred to not participating at all if stocks offer a premium over the risk free asset, regardless of the level of risk aversion.4 This holds even when the household bears any kind of “background risk”5 (Haliassos, 2003, p. 33), as long as the risk is not correlated with stock returns, which is always the case for a non- stockholding household. Even when a household faces borrowing constraints6 it will not end up having zero stockholding. This is because positive stockholding can be fully financed through borrowing as long as the size of total assets matches the size of out- standing debt so that the non-negative net wealth restriction is met (Haliassos, 2003, p. 32-33, p. 37). As Campbell (2006) states, according to financial theory all households, independent of their level of risk aversion should hold stocks when the premium on the risky asset is positive. Observing little stock market participation can be attributed to the absence of some standard assumptions (Campbell, 2006, p. 1568), and five of them are going to be discussed in the following section three.
3 Contributions to Explain Limited Stock Market Participation
3.1 Deviation from Standard Expected Utility
From section two, neither any kind background risk faced by the household, nor the risk attitude can be a reason for the decision not to participate in the stock market (Halias- sos, 2003, p. 33). So risk aversion itself, defined by the curvature of the utility function (Haliassos and Hassapis, 1999, p. 8), cannot explain zero stockholding (Haliassos, 2003, p. 33).
Haliassos and Hassapis (1999) present how departures from standard expected utility models, using non-expected utility that exhibits a different kind of risk-aversion, can serve to answer why stockholding is so limited. Their focus is on the “rank-dependent utility” (Haliassos and Hassapis, 1999, p. 6) which exhibits “first-order risk aversion” (Haliassos and Hassapis, 1999, p. 6). The basic idea is that inclusion of outcome rank- ing can lead to zero stockholding. Consider a household that ranks outcomes from worst outcome to best. The ranking of the outcomes again depends on the households’ actions. High returns are assessed as good when the household takes a long position on stocks. When the household can short stocks high returns are assessed as bad. Outcomes are equally ranked when stockholding is zero (neither long nor short). At this point the ob- jective function alters (because preferences change) resulting in “non-differentiability” (Haliassos and Hassapis, 1999, p. 7) of the objective function.7 This means that first- order conditions cannot be used as usual to derive the optimal solution. The respective indifference curves exhibit a “kink” (Haliassos and Hassapis, 1999, p. 7). For a house- hold with zero stockholding the kink can result in preferring zero stockholding to either a long or short position (Haliassos and Hassapis, 1999, p. 6-9). This is because the trade-off between risk and return at zero stockholding is not perceived as smooth (Hali- assos and Bertaut, 1995, p. 1120). In other words, the household sticks at the point where a reversal in outcome ranking takes place which means it sticks at zero stock- holding because this is the reversal point of outcome ranking (Haliassos, 2003, p. 34- 35). This behaviour is called first-order risk aversion8 and is able to justify zero stock- holding (Haliassos, 2003, p. 35; Haliassos and Hassapis, 1999, p. 7). Epstein and Zin (1990) show that first-order risk aversion is capable of explaining behaviour that is not in line with behaviour predicted by standard utility models. They state that there are two observations that support the validity of first-order risk aversion: Specialization in the risk free asset despite the return premium on the risky asset, and the demand for full insurance given small marginal risky asset holding (Epstein and Zin, 1990, p. 397).
However, rank-dependent utility and the resulting first-order risk aversion as presented here only imply preference of zero stockholding when there is no background risk like labour income risk (Haliassos and Hassapis, 1999, p. 7). When such risk is present the occurrence of a reversal in outcome ranking (change of the objective function) does not necessarily takes place at zero stockholding (H aliassos, 2003, p. 35). The household sticks at the point where a reversal in outcome ranking takes place and with labour in- come risk this point (the kink) is not generally where stockholding is zero (Haliassos, 2003, p. 35).9 Although rank-dependent utility cannot fully explain zero stockholding when background risk is present, it can still “lower predicted stockholding considera- bly” (Haliassos, 2003, p. 35). Diecidue and Wakker (2001) emphasize that rank- dependent utility is capable to capture real-world decision making because “rank- dependence is a psychologically realistic phenomenon” (Diecidue and Wakker, 2001, p. 284). However, the solution to the stock market participation puzzle does not solely lies in preferences (Haliassos, 2003, p. 35).
3.2 More Restricted Borrowing Constraints
As stated in section two, borrowing constraints in the theoretical framework cannot ex- plain zero stockholding (Haliassos, 2003, p. 37). In order to justify non-participation with borrowing constraints more restrictions are needed as for example short sale con- straints. A household that is denied for a credit (to buy stocks) will probably also not get a short sale permission at the stock exchange. Then this household is not able to buy stocks through the credit and is also not able to short sale stocks, so will end up having zero stockholding. The combination of borrowing plus short sale constraints can lead to zero stockholding (Haliassos, 2003, p. 37-38).
In Constantinides, Donaldson, and Mehra (2002), borrowing constraints are the reason for little stock market participation by the young. The young are not willing to reduce their current consumption in order to save in form of stocks. But they are willing to bor- row (in view of labour income in the future) in order to do both buy stocks and increase current consumption. However, this is not possible because borrowing constraints do not allow the size of outstanding debt being larger than the size of the asset holdings, so that the young rather end up with no stock holding. Not being able to borrow against income that is expected in the future is a realistic restriction (Constantinides, Don- aldson, and Mehra, 2002, p. 291). This is in line with Cocco, Gomes, and Maenhout (2005), who state that non-participation of young households is a solid empirical finding and the young households’ decision not to participate in the presence of borrowing con- straints is endogenously driven (Cocco, Gomes, and Maenhout, 2005, p. 520-521).
Haliassos (2003) stresses that zero stockholding can be explained by borrowing con- straints at best if a household initially holds no assets. But for a household having a lot of cash or savings in the bank, positive stockholding is predicted. So among the middle- income to high-income and wealthy households there should be positive stockholding, but it is observed that only a few participate in the stock market. Even though borrow- ing constraints can be an obstacle for non-asset holding, young and impatient house- holds that may also face short sale constraints, for households that do not have these characteristics there must be other factors affecting the participation decision negatively (Haliassos, 2003, p. 38).
3.3 Participation and Entry Costs
According to Guiso and Japelli (2004), entry costs do constitute a participation hurdle for not wealthy households. In the presence of entry costs, stock market participation is only beneficial when the excess returns on stocks overcompensate the entry costs. The relationship between wealth and the probability of stock-owning is predicted by entry costs: Those with wealth above a certain threshold do participate, while those beyond do not. The thesis that high entry costs are responsible for non-participation is supported by the evidence that participation is strongly positive correlated with households’ wealth (Guiso and Jappelli, 2004, p. 7). Haliassos and Michaelides (2003) show how entry costs can justify zero stock holding, although they are paid only once and allow access to the stock market forever. They derived an upper limit (threshold) of the entry costs that keeps households out of the stock market. The computed “threshold entry costs” (Haliassos and Michaelides, 2003, p. 162) tend to be small,10 which means even small entry costs are enough to discourage households from participating in the stock market. One crucial reason for this result is that “households are likely to spend a sub- stantial fraction of their time at levels of normalized cash on hand that do not justify any stockholding” (Haliassos and Michaelides, 2003, p. 164). Another reason is that house- holds tend to be impatient and therefore accumulate only a small amount of assets after entering the stock market.11 Households therefore rather choose a lower level of con- sumption smoothing and holding fewer assets than a higher level of consumption smoothing and holding more assets. This supports the evidence that even when having access to the stock market, households do not necessarily hold stocks in every period, which lowers the benefit from entering the stock market. In the case that the benefit from entering is low, small entry costs are sufficient to prevent households from partici- pating in the stock market (Haliassos and Michaelides, 2003, p. 162-166, p. 175). As Haliassos (2003) states, when planned investment is little or limited, the benefits from entering are small, so that low entry costs can discourage from participating. This sug- gests that the “entry cost explanation of the participation puzzle could apply to a wide range of households” (Haliassos, 2003, p. 40).
[...]
1 Stock market participation being limited is frequently stated in financial literature, for example in C a mpbell (2006), p. 1568.
2 This seminar thesis does not describe in detail all the empirical frameworks and methods which were applied by some of the sources to come to specific conclusions. There is also no purpose to demo n- strate the theoretical foundations behind some methods. It rather concentrates on the evidence about the stock market participation decision and the respective implications which result from some of the empirical studies. Details on empirical frameworks and theoretical models can be taken from the re- spective sources.
3 The assumptions include that there are no transaction costs, that investors are fully aware of available assets and their risks and returns, and that depending on the risk attitude the investor allocates its wealth between the risk free and the risky assets (Guiso and Jappelli, 2004, p. 7).
4 Risk aversion does not determine whether stocks should be included or not, but once stocks are included in the portfolio, risk aversion determines the share of included stocks (Haliassos, 2003, p. 33).
5 The household is not able to insure itself against background risk. One example for background risk is uncertain or risky labour income (Haliassos, 2003, p. 32).
6 When net wealth is restricted to be non-negative (Haliassos, 2003, p. 37).
7 For graphical and technical illustration, see Haliassos and Hassapis (1999).
8 Unwillingness to trade off risk for higher expected returns (Haliassos and Bertaut, 1995, p. 1120 ). Technically, first-order risk aversion occurs at the points where the utility function or the indifference curves are not differentiable (Segal and Spivak, 1990, p. 117, p. 119).
9 In an example with two states (good, bad) that lead to either high or low stock returns, and either high or low labour income, the reversal in outcome ranking takes place at the (positive) stockholding level where portfolio income in the good state minus portfolio income in the bad state exactly offsets labour income in the good state minus labour income in the bad state (Haliassos, 2003, p. 221). Further in- sights about the relationship between labor income risk and stock market participation can be gained by the life-cycle model presented in Gomes and Michaelides (2005) where risky labor income can lead to the decision not to participate (Gomes and Michaelides, 2005, p. 897).
10 Relative to the mean annual labour income (Haliassos and Michaelides, 2003, p. 162).
11 They build only a small buffer of assets to smooth fluctuations in consumption (Haliassos and Mich-aelides, 2003, p. 166).