Essays on behavioural economics and household finance
This thesis studies the impact of behavioural biases and limited cognition in the domain of household finance, in particular financial preparation for retirement. A special focus is placed on the resulting implications for policy-making. These questions are of particular importance in the current policy environment, which endows individuals with much responsibility for their economic decisions. Appropriate regulation and consumer protection appear to be an essential objective, and for two important reasons. First, the available financial products have become very complex and difficult to compare. Second, there is strong evidence for low levels of financial sophistication as well as for the impact of behavioural biases on financial decisions. Thus a helpful design of choice environments and regulatory policies requires a thorough understanding of the mechanisms underlying consumers' decision-making in a marketplace. In Chapter 1, I modify the two-system model of self-control by introducing cognitive costs of decision-making in order to account for the difficulty of planning for retirement. The resulting possibility of (rational) inaction allows to generate a range of predictions supported by empirical evidence. Non-savers are characterised by poor financial self-control, high cognitive costs, and/or low incomes. Automatic enrolment into pension schemes has a substantial effect on participation (the 'default option effect'), but its impact on aggregate saving remains ambiguous as some individuals are 'forced' to save while others become `discouraged' and adhere to potentially low default contribution rates. As a result, automatic enrolment reduces cross-agent variation in wealth accumulation. A stylised numerical application of the model produces substantial default effects and generates a U-shaped relationship between the default contribution rate and aggregate saving. Savings behaviour may also be adversely affected by the presence of behavioural biases. In Chapter 2, I model a market for pension products in which a pension provider interacts with a present-biased individual. Agents who are aware of their future present bias ('sophisticates') are offered efficient savings contracts independent of the magnitude of the bias, while contracts offered to agents who are not fully aware of their future present bias ('naifs') are distorted in a way that exacerbates their forecasting errors. Importantly, such exploitative contracts can be either 'inefficiently cheap' (low-yield, low-fee) or 'inefficiently expensive' (high-yield, high-fee), depending on whether the income or the substitution effect of an interest rate change dominates in the agent's utility function. Chapter 3 constitutes a numerical extension of Chapter 2. To examine the quantitative importance of contractual design and choice, I introduce the interaction with a pension provider into a numerical life-cycle framework with hyperbolic discounting. A sizeable forecasting errors of a naïve agent affect the savings contract offered in market equilibrium in the predicted way. Under the benchmark calibration, the equilibrium contract is Pareto inefficient, lowers agent's wealth at retirement by 10%, and generates a consumption-equivalent welfare loss of 0.17%.