Essays On Risk and Time with applications in insurance and finance

The predictions of theoretical models of human decision-making should agree with empir- ical and experimental evidence. In that respect, some fields of economics have embraced more plausible representations of preferences than others. For example, while standard in finance, temporally risk-averse preferences are still uncommon in the life-cycle literature, although they greatly improve its predictions. Chapter I introduces the issues that this dissertation touches on regarding the modeling of intertemporal preferences under uncertainty Chapter II discusses the allocation of aggregate longevity risk within a pension system when insurance markets are perfect. For a given expected budget, temporally risk-averse individuals prefer contributions and benefits to depend on the evolution of aggregate mor- tality rates rather than being fixed. The optimal allocation therefore transfers some risk to individuals, even though pension providers have access to perfect insurance markets. Indeed, this offers a “hedging” strategy where the prospect of a shorter life (e.g. emerging diseases) entails higher consumption while the prospect of a longer life is combined with lower consumption levels. Chapter II illustrates the benefits of applying models borrowed from finance to the life- cycle literature. In addition, empirical and experimental research refines our understand- ing of human behavior, triggering further improvements in the modeling of decision- making processes. Recent experiments indicate that individuals are more averse to im- mediate risk than to delayed ones. Because it involves risk attitudes and temporal delays, this bias may affect the decision to purchase insurance for long-run risks such as long-term care insurance. Chapter III considers individuals whose propensity to take risks decreases when old. Unaware of their changing taste for risk, naives delay the purchase of long-term care insurance, although this is a dominated strategy. Sophisticates, on the other hand, do not delay their purchase and, provided the elasticity of intertemporal substitution is below unity, end up better insured than naives. Delay-dependent risk aversion may also affect the pricing of financial assets. In fact, recent empirical evidence suggests that risk premia are sometimes higher for short-term yields than for long-term ones. Existing asset pricing models cannot reproduce this feature. Chapter IV contains a consumption-based model with a dynamically inconsistent rep- resentative agent whose risk aversion decreases with the delay. When this agent is naive, this model’s achievements are similar to those of existing asset pricing models: Equity risk premia are high and interest rates are low. The innovation is that the term structure of equity premia is decreasing