Does natural disaster influence people׳s risk preference and trust? An experiment from cyclone prone coast of Bangladesh

Abstract

Natural catastrophic events may have enormous negative effects on economic growth. People affected by the disaster might be risk averse because of anxiety about the future uncertainty of economic returns. The purpose of this empirical study is to highlight the effect of natural disasters (specifically coastal cyclonic storm surges) on individuals׳ risk preference and level of trust. This study also aims to disentangle risk propensity from trust. It reveals that natural disasters can significantly reduce people׳s risk-taking attitudes, whereas the catastrophic events have no influence on trusting behavior. The study suggests that risk attitudes are significantly negatively correlated with trust.

Introduction

Global climate change, sea level rise and the increased intensity of natural catastrophic events (like tropical cyclone, drought, and flood) have tremendous negative impact on human society and have attracted the attention of the global community [1]. Natural disasters often cause massive physical and economic damage. Consequently, natural disasters can have a negative impact on GDP [2], [3]. Developing countries are more vulnerable than other countries to the increasing likelihood of natural catastrophic events because of their often high population size compared to their underlying natural endowments, economic constraints and limited adaptive capacity. For instance, Bangladesh is considered one of the most vulnerable countries in the world to climate change and sea level rise [4]. Low-lying Bangladesh is now under tremendous pressure from devastating coastal cyclonic storm surges resulting from global climate change. Recently, cyclone Sidr, which hit the Bangladeshi coast in 2007, and Aila, which struck in 2009, caused huge physical damage and cost the lives of several thousands of coastal citizens of Bangladesh. Devastating events also occur in the developed world; Hurricane Katrina took a heavy toll on the citizens of New Orleans, USA, in 2005. In fact, about 2.7 billion of world׳s coastal inhabitants will be severely affected by natural disasters caused by global climate change [5]. Consequently, about 40% of the world׳s populations live in locations where the risk of economic and physical damage as a result of climate change is quite high. As already mentioned, millions of people have been affected by sudden natural disasters like cyclonic storm surges. This raises the question of how natural catastrophic events affect individual risk-taking behavior. However, there is a dearth of empirical studies to answer this query. Some economists (see e.g., [6], [7], [8]) assume that an individual׳s risk preference is not only influenced by a single risky situation but also individual exogenous or background risks. For example, the economic analyses commonly focus on individuals׳ behavior when faced with single risky decisions such as whether to buy or sell an asset with an uncertain return [9]. In real life, however, people always face multiple risks, and such background risks may influence an individual׳s decision under risk. Therefore, it is very important to understand whether and how individuals׳ risk preferences are actually affected by background risk parameters.

The term “risk” has been heterogeneously defined by scholars of multiple fields like economics, psychology, sociology, management, marketing and finance. Risk is typically defined as the extent to which the outcome of a decision is uncertain [10]. The outcomes of a risky event could be positive or negative. However, scholars have mixed opinions on the inclusion of negative and positive outcomes in the concept of risk. For instance, according to Sjoberg [11], risk is mainly related to the probability of negative events. Rousseau [12] defined risk as the perceived probability of loss. MacCrimmon and Wehrung [13] also focus on the negative aspects of risk. Conversely, several scholars consider both positive and negative domains in their definitions of risk. According to March and Shapira [14], risk is a situation in which one should expect the possibility of both negative and positive outcomes. Douglas [15] provides a similar definition of risk. For the purpose of this study, we use March and Shapira׳s definition of risk.

According to Butler and Cantrell [16] and McKnight et al. [17], trust is the context of personal characteristics that inspire positive expectations in a counterpart. James [18] characterizes trust as a risky action taken by an agent in an economic transaction under uncertainty or informational incompleteness with the anticipation that the other agent of the transaction will not behave opportunistically. Trust is particularly important because most transactions involve an element of trust and because trust influences economic success and growth [19]. There is evidence that countries with higher levels of societal trust tend to exhibit higher economic growth and investment relative to GDP [20]. Some scholars argue that the act of trusting is equivalent to the act of taking a risk (see e.g., [21]), so it is interesting to understand how risk attitudes influence people׳s levels of trust. There is no doubt that risk preference and trust behavior are important deciding factors in most economic decisions, as any economic transaction is dependent on risk and trust. We use James׳s [18] definition of trust for our present research.

Scholars of various fields like economics, sociology, and psychology have used two distinct methodologies to measure people׳s risk preference and their level of trust. One group of economists measured risk preference by questionnaire surveys including hypothetical lottery options. The individual׳s trust was measured by attitudinal survey questions from the General Social Survey (GSS). The survey-based approach has been criticized by a number of experimental economists who contend the questionnaire approach usually fails to create a real-life scenario. Therefore, the results may not elicit the individuals׳ actual risk preference and trust behavior. These economists (see e.g., [22], [23], [24], [25], [26]) considered lottery games and trust games (played with real money) to measure individuals׳ risk preference and trust behavior, as they believed that playing with real money favored the creation of real risk and trust situations for the individuals.

The trust game (introduced by Berg et al. [27]) is also known as the investment game. Several experimental economists used the trust game to measure trust behavior. In the traditional trust game, there are two groups: a trustor group (A) and a trustworthy group (B). To set up the game, each person of group A is given a certain amount of money. The players in group A individually decide how much money to send to their anonymous counterpart (a player in group B). Players in group A are informed that each monetary unit sent by them to their anonymous counterparts will be tripled by the time it reaches their partners. The players in group B are then asked to send back the tripled money to their respective counterparts. Players in group B are also informed that they are free to send any amount of the received money, or they can keep the whole amount. Similarly, in a one-shot investment game, there is no opportunity for investors to retaliate or for recipients to build reputational capital. According to game theory, it is predicted that the second mover (trustworthy) will not send any amount of money, as he/she has no economic incentive to do so. The first mover (trustor) realizes this; according to a Nash equilibrium prediction, the first mover will send no money to his anonymous counterpart. Therefore, sending money for the first mover is a risky decision. However, a Pareto improvement is possible if the first mover sends an amount of money to the receiver and the receiver returns at least one third of the tripled amount received. Thus, an individual׳s risk preference is expected to have an impact on the individual׳s expression of trust; risk-seeking individuals are likely to be more trusting and contribute more in the investment game. Contrary to the Nash equilibrium prediction, a number of studies confirm that in a one-shot trust game, a significant number of trustors choose to trust, and a large number of players also return a positive amount of money to the trustor [25], [26], [27].

To the best of our knowledge, Schechter [28] is the only person who designed an experiment to disentangle risk aversion from trust (with a risk game containing a similar payoff to the trust game), in this case with Peruvian villagers. She reported that risk attitudes are highly related to trust. The Peruvian participants played a risk game (lottery) first and then the trust game (with a similar payoff). This could influence the subjects to assume the trust game to be a gamble as well [28]. Thus, one might argue that the finding of the Peruvian experiment is not robust. Moreover, trust games played with non-student subjects at the field level have been criticized by Nannestad [29]. In his recent review article, Nannestad [29] argues, “When nonstudent subjects are studied, it is often difficult to determine what population they represent, and to what population one can hence generalize experimental results in a straightforward manner. The obvious solution to this problem is replication of the experiment with different subjects and an otherwise identical setup.” In that paper, Nannestad [29] also points out that “in general there are still too few examples of such replication studies, and hence the scope of validity of the results from quite a few experimental studies on trust remains an unsettled issue.” This inspired us to extend Schechter׳s work by conducting lottery and trust games with different subjects (with an alternation of game orders). In our study, the risk game (based on a lottery) and the traditional trust game proposed by Berg et al. [27] were played with coastal shrimp farmers of Bangladesh. We used a similar methodology to that of Schechter [28] in Peru. We think an important advantage of this game design is that it is easily explainable to the subjects outside the usual convenient sample of university students. Our sample played both risk and trust games in two different orders: for five villages, we ran the risk game first and then the trust game, and vice-versa in the other five villages. This allowed us to fill in the gap of the Peruvian study done by Schechter and to establish any effect of game order on the subjects.

The objectives of this study are as follows: (1) to disentangle risk preference from trust and (2) to assess how natural disasters (in our case tropical cyclonic storm surges) can influence individuals׳ risk preference and trust.