Essays in the economics of the retirement decision-making

Living longer and in a better health condition should be seen as a chance. However, the increase in the life expectancy is often presented as a source of concern. Indeed, combined with the decline in the fertility rate, the increase in the life expectancy gives rise to the aging of the population that represents one of the main challenges of the twenty-first century since it will put the financial health of the Welfare State under pressure in many countries. Moreover, living longer requires more income and particularly more retirement income. Consequently, citizens of those countries will have probably to rely more on themselves than in the past to get this income. Therefore, it is essential that the young generations are able to make plans for the future in order to have enough resources during their retirement. However, the elaboration of these plans poses two issues. First, the young generations should have all the necessary information to build these plans. Second, once these plans are built, the young generations have to follow them. In this regard, the economic literature provides some upsetting evidence. First, some empirical studies show that the discount rate is not constant over time but is higher for short horizons than for long horizons (e.g. Thaler (1982)). These studies suggest that the discount function would be (quasi-)hyperbolic and not exponential as it is generally assumed in the literature dealing with intertemporal choices. When a person displays a quasi-hyperbolic discount function, it means that this person is too optimistic concerning her future behaviours what can lead her to make time-inconsistent choices due to self-control problems. Second, several studies show a lack of knowledge about pension and Social Security benefits among the population of some countries (e.g. Gustman and Steinmeier (2001)). In conclusion, this evidence suggests that the required information to make plans for the future is not widespread among the population and that people do not follow these plans once they are build because of self-control problems. Therefore, the misperception of the retirement income and quasi-hyperbolic discounting are two factors that could affect the standard of living of the elderly. This thesis is a collection of five independent chapters addressing these issues. Chapter 1 is a short note describing hyperbolic and quasi-hyperbolic discounting. Chapter 2 examines how quasi-hyperbolic discounting influences the retirement decision with respect to exponential discounting. In a 3-period model in which a person has to decide to work or to retire in period 2, we investigate whether quasi-hyperbolic discounting leads to early or later retirement compared with exponential discounting. With respect to exponential discounting, quasi-hyperbolic discounting implies that the agent cares relatively more about the remote future than about the near future. For the young generation, retiring later implies to incur a higher effort cost that takes place in the near future to benefit from a higher income that partially takes place in the remote future. Thus, quasi-hyperbolic discounting induces the young generation to plan to retire later. On the other hand, for the middle-aged generation, retiring later implies to incur an effort cost now to benefit from a higher income in the near future. That is why quasi-hyperbolic discounting induces the middle-aged generation to retire earlier than planned. Therefore, a person who displays a quasi-hyperbolic discount function may retire later or earlier than a person who displays an exponential discount function. However, the impact of quasi-hyperbolic discounting on the retirement decision depends on the degree of awareness of the potential self-control problem and on the tools available to deal with it. In this regard, if the person perceives the potential self-control problem and if she is able to commit her future retirement choice, quasi-hyperbolic discounting leads to later retirement with respect to exponential discounting. On the contrary, if she does not perceive the potential self-control problem or if she is unable to commit her retirement choice, the impact of quasi-hyperbolic discounting on the retirement decision is indeterminate. In conclusion, the retirement decision depends effectively on the discount function we consider and quasi-hyperbolic discounting, even if it does not provide a general explanation for early retirement, may explain, in some cases, this phenomenon. Chapter 3 provides a deep analysis on how quasi-hyperbolic discounting affects the retirement decision. This chapter investigates how the bias for the present that is inherent to quasi-hyperbolic discounting influences the retirement choice. As in Salanié and Treich (2006), the two impacts of the present bias, i.e. the self-control effect and the discounting effect, are considered. Contrary to Chapter 2, we consider that the workers can retire at any age. We show that the self-control effect leads to early retirement while the discounting effect leads to later retirement. Indeed, if the person cares less about the future, she saves less what induces her to retire later. On the contrary, her lack of self-control induces her to retire earlier than expected. The global impact of the bias for the present on the retirement decision is therefore indeterminate. Nevertheless, if a person is able to commit her retirement choice, she avoids early retirement since the self-control effect does not play. However, we show that perceiving the potential self-control problem is not sufficient to attenuate early retirement. The literature on quasi-hyperbolic discounting generally considers that a person who perceives the potential self-control problem tries to commit her future choices or determines her current actions by taking into account her effective future reactions to these actions (Strotz (1956)). Cremer et al. (2007) and Bassi (2008) show that is also possible to design a public transfer scheme through majority voting in order to deal with self-control problems. Following this idea, Chapter 4 shows how a Pay-As-You-Go (PAYG) pension can emerge in order to solve the self-control problem regarding the retirement decision. If an agent perceives the potential self-control problem but if a commitment technology is not available, he can make a time-consistent retirement choice by contributing to a PAYG pension system in which the contributions made are lost in the case of early retirement. Therefore, a PAYG pension system can emerge even if it is a dominated saving device since contributing to such a pension system can be a way to constraint the future retirement choice by increasing the cost of early retirement. This chapter identifies the conditions under which such equilibrium arises and compares this voting equilibrium to the one emerging in an economy populated by agents who do not perceive the potential self-control problem. Finally, several studies show a lack of knowledge about pension and Social Security benefits among the population in many countries. However, other studies show that the knowledge of these benefits become more precise for people close to their retirement what suggests that a progressive learning process is at work (e.g. Rohwedder and Kleinjans (2006)). While this lack of knowledge should have sizeable effects on the consumption level during the retirement and on the retirement age, empirical studies find that this misperception has a low impact on the retirement decision and on the consumption level during retirement (e.g. Gustman and Steinmeier (2001), Rohwedder and van Soest (2006)). Chapter 5 investigates the theoretical implications of this misperception of the pension benefit and this progressive learning on the retirement and consumption decisions. The main result is that the impact of the misperception on these outcomes may be ambiguous if only two retirement ages are possible. This could rationalize why empirical studies only find a limited impact of the misperception of the retirement benefit on these outcomes. In addition, this misperception may explain simultaneously early retirement and the Retirement-Consumption Puzzle, i.e. the unexplained downward shift in the consumption path around the retirement age.