Equilibrium Asset Prices And Investor Behavior In The .

6m ago
248.51 KB
32 Pages

Equilibrium Asset Prices and InvestorBehavior in the Presence of Money Illusion Suleyman BasakLondon Business School and CEPRInstitute of Finance and AccountingRegents ParkLondon NW1 4SAUnited KingdomTel: 44 (0)20 7000 8256Fax: 44 (0)20 7000 8201Email: sbas[email protected] YanYale School of Management135 Prospect StreetP.O. Box 208200New Haven, CT 06520-8200United StatesTel: (203) 432 6277Fax: (203) 436 0630Email: [email protected] revision: July 2009 We are grateful to Bruno Biais (the editor) and three anonymous referees for valuable suggestions. Specialthanks to one referee for the detailed, constructive suggestions that benefited this work greatly. We also thankNick Barberis, James Choi, Ian Cooper, Wouter Denhaan, Stephan Dieckmann, Francisco Gomes, OwenLamont, Andreas Loffler, Anna Pavlova, Stephen Schaefer, Astrid Schornick, Lucie Tepla, Raman Uppal,Jessica Wachter, and seminar participants at London Business School, Yale, Western Finance Associationand European Finance Association meetings, where an earlier version of this paper was presented, for helpfulcomments. All errors are solely our responsibility.Electronic copy available at: http://ssrn.com/abstract 1280267

Equilibrium Asset Prices and Investor Behaviorin the Presence of Money IllusionAbstractThis article analyzes the implications of money illusion for investor behavior and asset pricesin a securities market economy with inflationary fluctuations. We provide a belief-basedformulation of money illusion which accounts for the systematic mistakes in evaluating realand nominal quantities. The impact of money illusion on security prices and their dynamicsis demonstrated to be considerable even though its welfare cost on investors is small in typicalenvironments. A money-illusioned investor’s real consumption is shown to generally depend onthe price level, and specifically to decrease in the price level. A general-equilibrium analysis inthe presence of money illusion generates implications that are consistent with several empiricalregularities. In particular, the real bond yields and dividend price ratios are positively relatedto expected inflation, the real short rate is negatively correlated with realized inflation, andmoney illusion may induce predictability and excess volatility in stock returns. The basicanalysis is generalized to incorporate heterogeneous investors with differing degrees of illusion.Journal of Economic Literature Classification Numbers: C60, D50, D90, G12.Keywords: Money Illusion, Asset Pricing, New Keynesian, Bounded Rationality, Equilibrium,Expected Inflation.Electronic copy available at: http://ssrn.com/abstract 1280267

1.IntroductionMoney illusion refers to “a tendency to think in terms of nominal rather than real monetaryvalues” (Shafir, Diamond and Tversky (1997)). Recognition of this seemingly simple confusiondates back to Fisher (1928) who argued that “Almost every one is subject to the ‘MoneyIllusion’ in respect to his own country’s currency.” Though the difference between nominaland real quantities is obvious, money illusion remains pervasive. As Akerlof (2002) notes, sevendecades after Fisher, “. rules of thumb involving ‘money illusion’ are not only commonplacebut also sensible—neither foolhardy nor implausible.” Despite its prevalence, money illusionhas largely been ignored by economists after the rational expectations revolution in 1970’s.1Yet, there has recently been growing, convincing support for money illusion. Shafir, Diamondand Tversky (1997) provide a psychological foundation for money illusion based on theiranalysis of survey data. These authors argue that although people are generally aware of thedifference between real and nominal values, they still often think of transactions predominantlyin nominal terms since, relative to real quantities, the nominal quantities are much moresalient. Fehr and Tyran (2001) demonstrate the role of money illusion in nominal pricerigidity by conducting a series of rigorous experiments. Interestingly, they also show thatmoney illusion arises only when payoffs are presented in nominal terms to the participants,that is, when nominal terms are more salient. The bias towards more accessible informationarises naturally, as argued in cognitive psychology (Kahneman (2003)). Moreover, as reasonedby Akerlof (2002), this bias can also be justified in many situations when losses from the biasare small.In this paper, we analyze money illusion in financial markets and do indeed demonstratethat the welfare loss of money illusion to an investor is small for typical environments, whileits impact on equilibrium can be considerable. Money illusion can then be interpreted asa form of bounded rationality (Simon (1968)) in financial markets: investors may partiallyoverlook inflation in their decision-making since the cost of this negligence is small. Thisresult parallels the main idea in the New Keynesian macro economics literature, which arguesthat small frictions and departures from full rationality may generate large fluctuations inthe real economy (Akerlof and Yellen (1985), Mankiw (1985)). In addition, our analysis alsoshows that some of the consumption and asset prices behavior can be better understood inthe context of money illusion.The presence of money illusion in financial markets has long been noted. Modiglianiand Cohn (1979) hypothesize that money-illusioned investors essentially discount future real1Early literature on money illusion emphasized its impact on labor supply and unemployment (Keynes(1936)).1Electronic copy available at: http://ssrn.com/abstract 1280267

payoffs at nominal rather than real rates, causing the undervaluation of stocks in 1970’swhen inflation was excessively high. In the psychology literature, Svedsater, Gamble, andGarling (2007) highlight the role of money illusion in financial decision-making via a numberof experiments, and show that investors may be influenced by the nominal representationof stock prices when evaluating financial information. The existing analysis in the financeliterature is primarily limited to empirically understanding dividend-price ratios based onthis partial-equilibrium stock valuation framework (Modigliani and Cohn (1979), Ritter andWarr (2002), Campbell and Vuolteenaho (2004)). At present, we lack rigorous understandingof the financial economic implications of money illusion.Our goal is fill this gap and undertake an analysis of money illusion within a familiar assetpricing framework. Our strategy is to deviate only in one dimension, the incorporation ofmoney illusion, while retaining all other standard paradigms in finance (no-arbitrage, marketclearing). In particular, we study the implications of money illusion for investor behaviorand asset prices in a pure exchange, security market economy that is exposed to inflationaryfluctuations. To demonstrate the role of money illusion as clearly as possible, we employ thecommon constant-relative-risk-aversion (CRRA) preferences as the benchmark case with nomoney illusion. To the best of our knowledge, ours is the first attempt to directly embedmoney illusion into an investor’s decision-making and valuation framework.Towards that, we develop a belief-based formulation of money illusion which accounts forthe systematic mistakes in evaluating real and nominal payoffs. In our analysis, we considerthe general case of partial illusion. In a mildly inflationary environment, an investor maypartially overlook and may not bother to fully take into account of the impact of inflation onthe purchasing power of currency in his intertemporal choice. That is, he partially overlooksinflation in his decision-making, since ignoring inflation simplifies his decision-making processand only incurs a small welfare loss. To capture this idea, we employ a generalized version ofthe Modigliani-Cohn hypothesis, in which the illusioned investor discounts future real payoffsby a combination of nominal and real rates, with a degree of money illusion parameter controlling the bias towards nominal rates. We decompose this perception error in discounting into achange of measure, representing the money-illusioned investor’s beliefs, and an ensuing averagediscounting error, capturing the aspect that a money-illusioned investor tends to over-discountpayoffs in an inflationary environment. This tractable formulation nests the benchmark caseof a normal investor with no illusion, where the investor discounts real payoffs by real rates,as well as the Modigliani-Cohn case of complete illusion, where the investor discounts realpayoffs by nominal rates. The money-illusioned investor’s dynamic consumption-investmentproblem then takes place under his money-illusioned probability measure and the ensuingaverage discounting error, together which fully represent his perception bias.2

Our main findings are as follows. First, at a partial equilibrium level, a money-illusionedinvestor’s real consumption is decreasing in the price level (holding all else fixed). To see theintuition, recall that a normal investor’s optimal consumption plan is such that his marginalutility is proportional to the cost of consumption in a given state. Consequently, the investorconsumes less in states for which consumption is relatively expensive, and more otherwise.When the price level is increased, a normal investor understands that there is no change inreal prices, and so does not alter his consumption. A money-illusioned investor, however,is confused between nominal and real quantities, and so perceives the consumption to bemore expensive, thereby consuming less. This prediction is consistent with the experimental evidence in Raghubir and Srivastava (2002), who show that an individual decreases hisexpenditure when faced with higher nominal price, and vice versa for lower nominal price.Second, since the effects of money illusion at an individual investor level are similar acrossinvestors, it is natural to expect those effects to show up at the aggregate level. To investigatethe impact of possibly widespread behavior of money-illusioned investors in financial markets,we move to a general equilibrium setting. Our focus is on the implications for asset prices andtheir dynamics. Equilibrium in the presence of money illusion emerges rich in implicationsand is able to generate various predictions, pertaining to inflation and asset price behavior. Inparticular, the real long term bond yields and the dividend price ratio are positively relatedto expected inflation under money illusion. The reason is that, as discussed above, moneyillusion induces an investor to consume less when the price level increases. Thus, higherexpected inflation tends to lower future demand for consumption, and hence makes futureconsumption“less valuable” in equilibrium. Since the stock and long term bonds are claimsto future consumption, higher expected inflation leads to lower stock and bond prices, orequivalently to higher dividend price ratio and long bond yields.The implication of a positive relation between expected inflation and dividend-price ratiois consistent with the Modigliani and Cohn (1979) hypothesis. Further supportive empirical evidence for this implication is recently documented by Ritter and Warr (2002), Sharpe(2002), Campbell and Vuolteenaho (2004), Cohen, Polk and Vuolteenaho (2005), Chordiaand Shivakumar (2005), Basu, Markov and Shivakumar (2005). More recently, in the context of housing prices, Brunnermeier and Julliard (2008) also provide evidence supportingthis argument. Piazzesi and Schneider (2008) combine the Modigliani-Cohn argument withdisagreement about inflation, and argue that both high and low inflation can lead to houseprice frenzies. Also in the context of housing markets, Genesove and Meyer (2001) documentthat home sellers appear to be averse to realizing nominal losses.While the Modigliani and Cohn hypothesis is intuitive, by assumption it is silent on the3

impact of money illusion on bond yields. This issue is addressed in our equilibrium modelin which interest rates are endogenously determined, and long bond yields are positivelyrelated to long term expected inflation, via a mechanism similar to the one for dividend-priceratio. Our bond yield implication is also supported by the empirical findings in both U.K.data (Barr and Campbell (1997)) and U.S. data (Sharpe (2002)). More importantly, ouranalysis provides a micro foundation for money illusion by showing that partially overlookinginflation has a large impact on security prices but only a small utility cost. This resultparallels that in Fehr and Tyran (2001), who show that a small amount of individual-levelmoney illusion can cause considerable price rigidity at the aggregate level, with the impactof money-illusioned agents being amplified by rational agents’ strategic responses. Both theiranalysis and ours underscore the importance of money illusion by demonstrating that a slightdegree of illusion can induce a large impact on equilibrium. We also demonstrate how moneyillusion may induce stock return predictability and excess volatility. The dividend-price ratiodepends on expected inflation under money illusion, and hence time variation in expectedinflation naturally generates time variation in dividend-price ratio, and so induces stock returnpredictability and higher volatility.Finally, the baseline economic setting is extended to feature multiple investors with different degrees of money illusion, in order to obtain further insights as to the effects of moneyillusion, as well as to demonstrate the robustness of our main results. Now, the equilibriumis additionally driven by the stochastic distribution of consumption across investors, leadingto richer dynamics. Here, the degree of money illusion in the economy is endogenously determined by the price level and the initial wealth distribution across investors. An importantimplication is that the real short interest rate and realized in