Essays on discounting behavior and gambling behavior


This thesis consists of three independent chapters on the elicitation of individual discount rates and on the estimation of gambling prevalence in Denmark.

The first chapter, “Discount Rate Sensitivity to Background Consumption and Consumption Smoothing,” studies the sensitivity of individual discount rates with respect to background consumption and consumption smoothing. I use simulated choice data from standard decision tasks in time preference experiments and show that individual discount rates are sensitive to assumptions with respect to background consumption and consumption smoothing if the utility function is non-linear and exogenous. However, if discount rates and the utility function are elicited jointly, discount rates are robust to assumptions on time-invariant background consumption and consumption smoothing. The analysis clarifies mixed conclusions from previous studies and indicates which elicitation methods provide robust estimates of individual discount rates.

The second chapter, “Financial Wealth, Liquidity Constraints, and Discounting Behavior,” analyzes to what extent discounting behavior in experiments is correlated with financial wealth. Life cycle models of consumption and saving behavior associate higher discount rates with lower financial wealth and present bias with lower financial liquidity. We combine data from a field experiment in Denmark with administrative data on financial wealth and estimate exponential and quasi-hyperbolic discounting models. We find significantly lower discount rates for wealthy subjects than for poor subjects, but do not find any evidence of present bias and no significant association between present bias and liquidity constraints.

The third chapter, “Gambling Problems in the General Danish Population: Survey Evidence,” compares several survey instruments of gambling behavior and estimates the prevalence of problem gambling in Denmark. We administer surveys on gambling behavior together with standard survey instruments for alcohol use, anxiety, depression and impulsivity to 8,405 subjects. We estimate that 95% of the population has no detectable gambling risk, 2.9% has an early risk, 0.8% has an intermediate risk, 0.7% has an advanced risk, and only 0.2% can be classified as problem gamblers. Moreover, we find that gambling risk is positively correlated with alcohol use, anxiety, depression, and impulsivity.

Problem with this document? Please report it to us.