Money Illusion and Nominal Inertia in Experimental Asset Markets

Abstract

We test whether large but purely nominal shocks affect real asset market prices. We subject a laboratory asset market to an exogenous shock, which either inflates or deflates the nominal fundamental value of the asset while holding the real fundamental value constant. After an inflationary shock, nominal prices adjust upward rapidly, and we observe no real effects. However, after a deflationary shock, nominal prices display considerable inertia and real prices adjust only slowly and incompletely toward the levels that would prevail in the absence of a shock. Thus, an asymmetry is observed in the price response to inflationary and deflationary nominal shocks.

Keywords:

• Money illusion
• Nominal inertia
• Asset market bubble
• Nominal loss aversion
• Laboratory experiment

ACKNOWLEDGMENTS

We thank John Duffy and seminar participants at the Universities of Innsbruck, Melbourne, Sydney, New South Wales, Pittsburgh, the 2008 Economic Science Association European Meetings, and the 2008 Australian Experimental Economics Workshop for comments.

Notes

1. A particularly provocative suggestion is that the numerosity (or equivalently the value of each unit) of the medium of exchange can affect real decisions, and there is some evidence of such an effect (see, e.g., Gamble et al. [2002], Amado et al. [2007], Raghubir and Srivastava [2002], Svedsäter, Gamble and Gärling [2007]). The introduction of the Euro in 2002 represented a purely nominal shock to all prices in the economy. Nonetheless, Kooreman, Faber and Hofman [2004] and Cannon and Cipriani [2006] find that donations to charities experienced a real upward spike in the year of the introduction of the Euro. Weber et al. [2009] find that the medial prefrontal cortex (mPFC), which contains the terminus of much of the mesocortical dopamine pathway and which activates with the experience of reward, exhibits money illusion. That is, it exhibits a stronger BOLD response in an individual, the higher the income he receives in nominal terms, even when the real income he receives is held constant.

2. They find that when subject to nominal losses in a declining market, individuals set prices higher relative to the expected final price the more they paid for the house. These sellers also wait longer for their house to sell, and receive higher selling prices than other sellers.

3. Indeed, such downward wage and price rigidity is thought to influence the effect of macroeconomic policy and contribute to business cycles (see, e.g., Yellen and Akerlof [2006]).

4. The nominal and real components of a financial shock may not be distinguishable to an individual who does not know the underlying rate of inflation. Asymmetries in the functioning of the trading institution exist for example, in real estate markets, where typically the seller initiates the process of exchange by placing his house on the market and posting a price. In a real estate market, constraints bind asymmetrically, as the buyer is more likely to be cash constrained, while the seller always has an item to sell. Often a nominal shock, such as sudden inflation, may be an indication of unfavorable macroeconomic conditions or the result of poor government policy.

5. There is also experimental evidence, from markets for single-period goods rather than assets, that market interaction can fail to reduce and indeed even magnify the real effects of money illusion. Lian and Plott [1998] study an experimental macro economy with simultaneously operating input and output markets and a fiat money for trade. They subject the money supply to a nominal exogenous inflationary shock and observe a temporary increase in real output. Fehr and Tyran [2001] induce a nominal shock after a few periods of play of a pricing game. They find that nominal inertia, a tendency to price at preshock nominal levels, exists for individuals facing human but not computerized opponents. This indicates that there are market environments, in which money illusion on the part of a small number of individuals can affect responses of other individuals in such a way as to magnify the effect on real outcomes. In addition, the authors find that less nominal inertia, and thus smaller real distortion, is caused by a positive, or inflationary, nominal shock, than a negative, or deflationary, nominal shock. In a follow up paper, Fehr and Tyran [2007] find that money illusion can also have permanent effects through its influence on equilibrium selection when there are multiple equilibria.

6. Amado et al. [2007] note that “when one or more zeros are dropped from a currency, consumers rescale all prices relatively quickly rather than relearn them selectively through gradual exposure.”

7. The markets were thick and active. In the Deflation treatment, an average of 2.69 times the total stock of units was exchanged over the course of the 21 periods, and at least 1.60 times the total stock was exchanged in each session. In the Inflation treatment, the average quantity exchanged was 2.93 times the total stock, and the minimum was 1.61 times. For the control treatment the corresponding quantities were 4.35 and 0.96.

8. Indeed, after the deflationary shock, there is significantly more mispricing in real terms relative to fundamentals than in the other two treatments. Consider two measures of mispricing similar to those used in Haruvy and Noussair [2006] and Haruvy et al. [2007]. One measure is Total Bias, given by ∑_t (p_t – f_t ), where t is a market period, f_t is the real fundamental value in period t and p_t is the real median transaction price in period t. The second is Total Dispersion, given by ∑_t |p_t – f_t |. Prices and fundamentals are given in real terms. The Total Bias averages 605.7, 108.3, and 207.7 ECU in the Deflation, Inflation, and Control treatments, respectively, over periods 8–21. The Total Dispersion averages 614.7, 138.7, and 240.3 in the Deflation, Inflation, and Control treatments over periods 8–21. A Wilcoxon rank sum test, using each session as the unit of observation, rejects the hypothesis at the p < .05 level that Total Bias is equally distributed between the Deflation and the Inflation treatments, as well as between the Deflation and the Control treatments. The hypotheses that Total Dispersion is equally distributed between the Deflation and the Inflation treatments, as well as between the Deflation and the Control treatments, can also be rejected at p < .05. For both Total Bias and Total Dispersion, there is no significant difference between the Inflation and the Control treatments.

9. The ratios of average nominal prices of the pre-versus post-shock phase in individual sessions were 4.81, 8.11, 11.74, 11.81, 13.61, and 19.74.

10. One might conjecture that there could be a nominal wealth effect operating in the experiment. In the Deflation treatment, individuals may be trading at higher real prices after a shock because they have experienced a decrease in their nominal wealth. However, if such an effect were present, it would have to operate asymmetrically because otherwise it would suggest that trading would occur at lower real prices after an inflationary shock, which we do not observe.

11. Nominal loss aversion is also consistent with some other phenomena observed in the experimental asset market literature. For example, Haruvy and Schoenberg [2009] observe flat price trajectories in markets in which fundamental values decline over time. They attribute it to the accounting method they asked subjects to use and to imitation strategies. However, nominal loss aversion is also consistent with the price pattern that they observe.