از دست دادن کوتاه بینی و تجربه بازار
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|16000||2014||13 صفحه PDF||سفارش دهید||8500 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Economic Behavior & Organization, Volume 97, January 2014, Pages 113–125
We probe the boundaries of myopic loss aversion (MLA) theory through market treatments designed to reduce the MLA effect. Our market settings separate investment commitment from information frequency, display a running average asset value and explore the influence of participant experience. The market-based results suggest MLA persists with inexperienced participants despite efforts to mitigate MLA. Prices in markets with returning participants do not display an MLA effect. However, the same experienced participants individually succumb to MLA in an allocation setting immediately following the market. Overall, our results suggest that, while market experience mitigates the MLA effect, participants do not transfer these results to other settings.
This paper examines the ability of markets to overcome myopic loss aversion (MLA). This ability of markets to overcome the MLA effect has received little attention in prior MLA research. The notable exception is Gneezy et al. (2003), which provides initial evidence that markets do not dispel individual MLA bias. The inability of Gneezy et al. (2003) markets to overcome MLA is curious given evidence that markets can overcome individual biases (Forsythe et al., 1992), and that market structure can drive biased traders to equilibrium (Jamal and Sunder, 1996). We employ a series of modifications to Gneezy et al. (2003) market design to further explore when market conditions influence the MLA effect. We make four key modifications to Gneezy et al. (2003). First, we hold trading commitment constant across information frequency treatments by allowing trading every period, while comparing trading results with information frequency every period verses every fourth period. This design better reflects a typical market setting and allows us to focus on whether information frequency alone can drive MLA in a market setting. Second, we expand the number of trading periods to offer the market mechanism more time to overcome the MLA bias and move toward equilibrium. Third, we incorporate a treatment prominently displaying the average asset value, along with the periodically reported asset value. The average value aggregates and summarizes previous information and frames the information in a manner that should reduce the participants’ heterogeneous beliefs about asset value and decrease the participants’ cognitive costs to estimate the assets expected value. Finally, we explore the role of experience on MLA's effect by inviting a set of participants to return for a second set of experiments. These experienced subjects enable us to test whether market experience reduces bias in a manner that parallels research showing that trading experience can reduce behavioral biases (List, 2003). We find inexperienced participants succumb to MLA in our markets. Average trade values are lower when we provide information every period rather than summed and provided every four periods. This basic finding parallels Gneezy et al. (2003). The fact that participants still succumb to MLA with trading commitment and information frequency separated suggests that information frequency is sufficient to create the MLA effect. Furthermore, the treatment displaying average asset value does not mitigate the MLA effect. In contrast, we find that markets with experienced participants who return for a second market session do not display an MLA effect – mean prices are not significantly different across information frequencies. However, these same experienced participants do not overcome MLA in an allocation session immediately following the market session. The convergence in prices appears to be due to the power of the double auction to converge prices to equilibrium and not necessarily due to participants learning to overcome MLA bias. Our results suggest market experience alone does not transfer to other settings. Further research is needed to explore the conditions under which market experience reduces the MLA's impact. This paper proceeds as follows. First, we provide background on MLA and theory to explore both the general and our specific market conditions that potentially mitigate individual MLA biases. Section 2 describes our market setting and the subsequent allocation treatments. Section 3 describes our results, and in Section 4 we conclude. Our conclusion includes potential limitations and opportunities for further research.
نتیجه گیری انگلیسی
This paper makes several contributions. First, holding investment commitment consistent by allowing participants to trade in each period across FULL and AGGR conditions does not mitigate MLA. This suggests that information frequency alone can produce MLA in a market, which is particularly concerning in corporate settings where investment commitment is often independent of information frequency. Second, we find no evidence that lowering the participants’ costs to calculate the expected value of the underlying asset, nor that prominently framing the expected value mitigates MLA. This finding suggests that even providing year-to-date information in more frequent financial reports is unlikely to mitigate the impact of MLA on market prices. Third, we find intriguing evidence that experienced participants overcome the effects of MLA in a market setting. Nonetheless, we are unable to fully assess whether the convergence in prices is due to the power of the double auction to move prices toward equilibrium or whether participants learn to overcome MLA on an individual basis. That said, there is some indication that the market does not teach individual participants to overcome MLA. In the allocation setting, experienced participants continue to allocate less to risky shares under the FULL treatment, consistent with MLA. Thus, experience seems to mitigate the pricing impact of MLA, but the inability of the same participants to overcome MLA in the allocation setting suggests that market experience alone does not teach individuals to undo MLA's effects. Our research comes with a few caveats. First, we use single-period assets and re-endow participants every period in the market and allocation designs. This approach is consistent with prior research. However, the ecology we are interested in informing with our research consists, in many cases, of multiple-period assets that are not re-endowed. We made this design choice to minimize the susceptibility found in prior experimental economics research to trading bubbles with multiple-period assets (Smith et al., 1988). We leave future research to explore the impact of MLA on more frequent reporting with multi-period assets. Second, we do not provide the underlying distributional detail to participants but instead have them infer them from a history of draws. This approach increases the challenge to subjects in that they must infer the underlying value of assets. Some have argued that without disclosing the distribution from which the earnings are drawn to participants the information in the FULL treatment is fundamentally different than the AGGR treatment. We use our approach to better capture natural trading ecologies where aggregated data would not include any information about the period-by-period results. We believe the results from this more realistic setting would not differ significantly if distributional data were provided. In fact, Gneezy et al. (2003) provide the underlying distribution to participants in their market study and find results similar to ours. Future research could explore this possibility further. Finally, we provide evidence that experience impacts the MLA effect in a market setting but are unable to fully explain its cause and effect in the market. We believe future research is needed to explore both the extent to which experienced traders can reduce MLA effects and the conditions under which such experience reduces the MLA bias in prices.