Return distributions of equity-linked retirement plans under jump and interest rate risk (Q362051)

A retirement plan is studied, where payments are made at deterministic points of time. The contributions by the insured are guaranteed as benefit at the time of retirement. The riskless rate of interest is modelled by a Hull-White short rate process. The risky asset is modelled as a jump-diffusion process with double exponential jumps. Three investment schemes are studied. For the `actuarial method', the amount \(F_t\) needed to match the guarantee is invested in a riskless bond, the rest in the risky asset. For the `constant proportion portfolio insurance' (CPPI), the amount \(\min\{m(P_t - F_t), P_t\}\) is invested in the risky asset, where \(P_t\) is the value of the portfolio at time \(t\) and \(m > 1\) is a leverage factor. For the `stop-loss method', the capital is invested in the risky asset, until the lower target value is met. Then all the capital is invested in the riskless bond. In the case of contributions at different times considered in the paper, capital once invested in the riskless bond remains in the bond; so no capital is taken from the riskless bond. If at the time of a contribution the value of the portfolio exceeds the amount \(F_t\), the contribution is invested in the risky asset, until the lower target value \(F_s\) is met (again). A simulation study for three scenarios of the drift parameter determines the distribution of the value of the portfolio at the retirement date and the average equity exposure in the cases with and without managing costs. Moreover, it is investigated, how often the guarantee is not met and therefore there remains a financial gap by the different strategies. Of course, the actuarial method is secure, but has in the average the lowest performance. For the CPPI, there only rarely remains an investment gap. The most risky strategy is the stop-loss method, because the amount invested into the risky active is not reduced when the lower target value is approached. In an appendix, properties of the considered processes are obtained. Moreover, it is explained how the parameters for the simulation study were estimated from data.