earticle

영어

Risk aversion and time preference are fundamental variables when people make financial decisions, but there are numerous opinions for how to estimate them. Researchers frequently estimate the demand of financial products with an ordinary survey, where subjects provide self-perceptive information. Given that behavioral financial questions are included in the simple questionnaire, the subjects may misunderstand such questions, or may not be able to completely understand the importance of the concepts driving them. Therefore, subjective bias could be implicitly problematic in many surveys, rendering their results untrustworthy. Although researchers can draw their conclusions based on the data collected from the survey, the economic results may be distorted or prove insignificant. This study takes the behavioral and experimental economics explored in Kahneman and Tversky (1979) and applies those theories to a financial product, such as a single payment annuity. This is a relatively new product in Korea’s aging population, thus demand for it has not been analyzed in the Korean context, despite the popularity of such products in the U.S. (Mitchell and Utkus, 2003). Following the method of eliciting risk aversion proposed by Holt and Laury (2002), we develop a new measure of risk aversion that is modified using the quantile normalization (Bolstad et al., 2003), and use the measure to check for subjective bias when choosing an annuity product. Time preference is also very critical in financial decisions. We propose a time preference measure that follows Coller and Williams (1999) and use it as an objective measure of the subject’s time preference. We design an experiment in which the subjects are guided by the instruction and payoff schemes, and follow several sessions. The participants choose a lottery from a pair of lotteries whose payoffs and probabilities are different, revealing the participants’ risk aversion preferences. In the next experiment on time preference, the participants choose a lottery from a pool of several featuring different interest rate payoffs with higher and lower payment spans, depending on the interest rates. During the experiment, the participants’ choices reveal their time preferences for our analysis. In addition to the results of the aforementioned experiments, we collect the participants’ background information, such as financial status, financial literacy, and intelligence. Given the dataset, we can then find that the subjective answers from the survey are not significantly related to the financial decision on single payment annuity. This allows us to verify that the subjective bias exists, and confirm that self-perceptive surveys are not trustworthy in choosing financial products. We also compare the self-perceptive answers and the experimental measures to test which methods are more reliable and significant. The results show that neither approach significantly determines time preferences and risk aversion in some model specifications. We construct the bias variables of risk aversion, time preference, life expectancy, financial literacy, and intelligence using the differences between the subjective answers and experimental results. Thus, when we execute econometric models with the bias variables of interest, we discover that the biased self-perceptive and experimental measures of the time preference, risk aversion, life expectancy, financial literacy, and pension literacy are significant determinants of the demand for single payment annuity. Therefore, the preferential and behavioral variables are critical in choosing financial products such as annuities, rather than the self-perceptive answers in the survey. We also run a Heckman two-step regression of the premium of the single payment annuity and find that financial status and portfolio are the most crucial determinants, regarding the moneywise incentive. Finally, the time preference and risk aversion experiments we design reveal that behavioral and preferential biases are critical in choosing financial products. Therefore, we conclude that researchers should take bias variables seriously when they estimate the demand for financial products, even after controlling for other variables.

한국어

본 연구는 일시납 연금 가입수요의 결정요인을 확인하기 위해 실험연구를 진행하였다. 결정요인으로는 개인의 금융상품 선택에 있어서 가장 중요한 요인인 위험회피(risk aversion) 정도와 시간선호(time preference)에 초점을 두었다. 40~59세의 한국인 500명을 대상으로 하여 설문과 실험 두 가지 방법을 모두 실시하여 개인들의 위험회피와 시간선호 변수를 추정하였다. 이를 통해 피설문자의 답변이 가질 수 있는 편의들을 실험환경을 통하여 획득한 정보와 비교해 보고, 일시납 연금의 가입수요에 적합한 변수 선정과 추정방식이 무엇인지 검증하였다. 연구의 주요 결과는 다음과 같다. 첫째, 설문 답변을 통한 변수들은 실험을 통해 추정한 변수들과 상당한 차이를 보여 주관적 편의 (subjective bias)의 문제를 내포하고 있음을 확인하였다. 둘째, 실험을 통해 추정한 변수들 중 미래에 대한 시간선호는 일시납 연금 가입의 중요한 결정요인이지만 위험회피 정도는 유의하지 않았다. 설문을 통한 본인자각적 답변들은 일시납 연금 수요의 결정요인이 아니었다. 셋째, 본인자각적 답변과 실험결과의 차이로 정의된 미래선호, 위험회피, 기대여명, 금융지식, 연금지식의 편의 변수는 일시납 연금 가입의향과 유의한 관계를 보였다. 위의 결과들은 금융상품의 잠재적 수요에 있어서 행태적, 선호적 편의가 중요한 영향을 미칠 수 있음을 의미한다.