Ethnic diversity deflates price bubbles

Abstract

Markets are central to modern society, so their failures can be devastating. Here, we examine a prominent failure: price bubbles. Bubbles emerge when traders err collectively in pricing, causing misfit between market prices and the true values of assets. The causes of such collective errors remain elusive. We propose that bubbles are affected by ethnic homogeneity in the market and can be thwarted by diversity. In homogenous markets, traders place undue confidence in the decisions of others. Less likely to scrutinize others' decisions, traders are more likely to accept prices that deviate from true values. To test this, we constructed experimental markets in Southeast Asia and North America, where participants traded stocks to earn money. We randomly assigned participants to ethnically homogeneous or diverse markets. We find a marked difference: Across markets and locations, market prices fit true values 58% better in diverse markets. The effect is similar across sites, despite sizeable differences in culture and ethnic composition. Specifically, in homogenous markets, overpricing is higher as traders are more likely to accept speculative prices. Their pricing errors are more correlated than in diverse markets. In addition, when bubbles burst, homogenous markets crash more severely. The findings suggest that price bubbles arise not only from individual errors or financial conditions, but also from the social context of decision making. The evidence may inform public discussion on ethnic diversity: it may be beneficial not only for providing variety in perspectives and skills, but also because diversity facilitates friction that enhances deliberation and upends conformity.

Keywords: decision making; diversity; economics; markets; psychology; sociology.

Fig. 1.

The experiment. Participants were randomly assigned to markets that were ethnically homogeneous or diverse (Left). After they received the information needed to price stocks accurately, we assessed each participant’s financial skills individually, using 10 hypothetical market scenarios to establish a baseline of pricing accuracy (Center). Trading in a computerized stock market, each participant was free to buy and sell stocks and/or to make requests to buy (“bid”) or offers to sell (“ask”). All trading information was true, public, and anonymous: All participants could see all completed transactions and bid and ask offers (Right; see example in SI Appendix, Fig. S8). The data reflect actual prices in the sixth period of trading in two of the markets of study 1. The experiment did not involve deception.

Fig. 2.

Pricing accuracy in diverse and homogeneous markets across studies: (A) Southeast Asia and (B) North America. Pricing accuracy in trading (ex-post fit between market prices and true values) across diversity conditions and sites, measured by Haessel’s R². Higher score signifies higher pricing accuracy; the lower the score, the worse the accuracy, the greater the bubble. Error bars represent SEMs. Difference (across diversity conditions) in ex-post pricing accuracy in Southeast Asia = 0.302, t₍₂₁₎ = 3.059, two-tailed P < 0.01; in North America = 0.284, t₍₉₎ = 3.593, two-tailed P < 0.05. The results are robust whether using parametric or nonparametric statistical tests (SI Appendix). They are based on 2,022 market transactions by 180 individual traders in 30 markets, of which 16 were homogeneous and 14 diverse. Details are in SI Appendix, Table S1.

Fig. 3.

Average change in pricing accuracy in diverse and homogeneous markets. Average change is from ex-ante (pretrading baseline) to ex-post pricing accuracy (in actual trading). When negative, pricing accuracy deteriorates during trading; when positive, it improves. Error bars represent SEMs. Change in diverse markets: t₍₁₄₎ = 2.211, two-tailed P < 0.05. Change in homogeneous markets: t₍₁₆₎ = −2.944, two-tailed P < 0.05. The results are robust whether using parametric or nonparametric statistical tests (SI Appendix). They are based on 2,022 market transactions by 180 individuals in 30 markets, of which 16 were homogeneous and 14 diverse (SI Appendix, Table S1).