The house money effect on investment risk taking: Evidence from Taiwan

Abstract

This paper investigates the effect of house money on the risk taking behavior of individual investors. When gains are more substantial, individuals tend to take greater risk. The house money effect seems to decline over time because the propensity for risk taking following gains is diminished with time. This study shows that when evaluating investment gains, the reference points for investors are adapted over time, with the current salient reference point being the highest stock price attained at a given time in the past. The empirical evidence suggests that the house money effect is actually discernible in the real world financial markets and not just in artificial laboratory experiments.

Introduction

Psychological studies have revealed that prior outcome affects feeling. However, does it affect decision-making strategy and risk taking in financial markets? The modern rationality-based economic theory suggests that prior outcomes do not affect investors' decisions. Investors should decide to invest only when future benefits are greater than future costs. Losses or costs that one may have experienced in the past but which are not expected to recur in the future are sunk costs and should not be part of the decision calculations. Therefore, the only question at issue is whether the anticipated return is in accord with the risk one takes. Prior outcomes of financial investments are sunk and irrelevant and should not influence one's risk taking behavior.

A competing view is that prior outcomes do affect decision-making. According to the psychological literature on the sunk costs effect, prior outcomes directly affect an individual when making real decisions. Indeed, numerous studies have established that sunk costs often influence decision makers.¹ Based on the findings on sunk costs, the argument is that people will always escalate their commitment to their chosen course of action, as opposed to cutting their losses and ending their questionable line of behavior. It therefore appears that individuals will tend to adhere to their losing ways and throw away good money in their attempts to rescue themselves.

The house money effect, which Thaler and Johnson (1990) first propose and document based on experimental evidence, refers to a pattern whereby people tend to take on increased risk subsequent to a successful investment experience. That is, prior gains lead to greater risk taking in subsequent periods. Gamblers refer to this feeling as “playing with the house's money.” After a large win, gamblers do not yet regard the new money as their own, and because they do not fully integrate their winnings with their own money, they act as if they are betting with the casino's money. People are more willing to take risks after a windfall, even when they do not conventionally lean toward risk taking. This phenomenon is observed essentially because after a gain, the feeling of satisfaction can cancel out the pain of a subsequent loss. It therefore appears that people become less risk-averse and have a greater propensity for risk taking behavior.

The experimental evidence on how prior gains and losses influence risk taking behavior is rather inconclusive, with a portion of the evidence supporting the house money effect² and several studies demonstrating otherwise (Clark, 2002). Barberis et al. (2001) develop a theoretical asset-pricing model that considers the house money effect. The settings on their model focus on the experimental evidence relating to the dynamic features of risk aversion in their attempt to exhibit and explain both the equity-premium and volatility puzzles.

Despite strong psychological foundations and extensive experimental evidence, little empirical evidence is currently available on the house money effect. Several recent empirical studies focus on intraday trading in the futures market. Frino et al. (2008) find evidence of the house money effect in the behavior of professional futures traders. Furthermore, Liu et al. (2010) examine the behavior of market-makers in the futures market and discover that morning gains lead to above average risk taking in afternoon trading. However, Locke and Mann (2003) find little change in risk taking after abnormal morning gains. Moreover, Coval, and Shumway (2005) argue that morning gains actually lead to decreased risk taking in the afternoon, while traders with morning losses are more likely to take above average afternoon risk.

We add to previous studies by analyzing the trading behavior of individual investors in the stock market and examining the variations in the house money effect over time and across gain sizes, as well as the impact on the house money effect from countervailing factors, such as the setting of reference points. The results reveal compelling evidence of the house money effect. Individual investors tend to increase their risk taking based on prior investment gains. Specifically, individual investors tend to buy stocks with higher volatility once they have experienced a prior gain.

To examine whether such behavior holds when investors experience relatively small prior gains or whether the gain is sufficiently large to be perceived as playing with the house's money, we analyze the house money effect based on gain size. Consistent with the results of Arkes and Blumer (1985) and Arkes et al. (1994), we find that the house money effect is associated with the size of the prior gain and that this gain must be substantial enough to be perceived as playing with the house's money.

Moreover, Arkes and Blumer (1985) and Gourville and Soman (1998) argue that the sunk costs effect may weaken over time. Conversely, Thaler (1980) and Kahneman et al. (1990) state that although a newly obtained asset may initially be viewed as a gain, it will gradually be incorporated into a person's wealth and became a part of the status quo. This paper argues that the same type of adjustment occurs in the case of investors selling stocks with gains.

We hypothesize that investors obtain considerable pleasure from substantial investment gains in the stock market and that they gradually adapt to such gains over time, which ultimately weakens the house money effect. Consistent with this hypothesis, we find that the house money effect attenuates with time, resulting in a greater willingness of investors to ignore a prior gain and far less willingness to increase risk taking in a later investment decision-making process.

Thaler and Johnson (1990) demonstrate that individuals have increasing risk tolerance when their wealth exceeds a specific reference point. This study examines the house money effect in light of this important issue. Kahnemn and Tversky's (1979) prospect theory depicts the shape of the value function around a certain reference point, but they do not specify its location.

Recent studies suggest that people tend to consider the average purchase price and the prior maximum price as their reference points in financial investments.³ This paper evaluates the role of different reference points associated with the house money effect. Our results reveal strong evidence of the house money effect irrespective of the reference point measures used. However, the house money effect associated with maximum stock price is more significant than that associated with other reference points. It could be that the feeling of happiness arising from any profit achieved over the prior high price is strong, and the effect of such feelings on risk taking behavior is also strong.

The organization of this paper is as follows: Section 2 discusses the data and methodology. Section 3 presents our empirical results. Section 4 summarizes the conclusions drawn from our study.