Optimal DB-PAYGO pension management towards a habitual contribution rate (Q2212147): Difference between revisions

A defined benefit pension scheme is considered where the pension depends on the present salary level and the date of the entrance to the scheme. All participants enter at the same age \(a_0\) and get retired at the age \(\tau\). The density of the entrants at time \(t\) is modelled as \(n(t) = n_0 e^{\rho t}\) with some \(\rho < 0\). Let \(s(m)\) be the probability that a person being alive at age \(a_0\) still is alive at age \(m\). In the model, persons die in a deterministic way, such that the density of \(m\) year old in the scheme at time \(t\) is \(n(t - (m-a_0)) s(m)\). The salary level \(L(t)\) is modelled by a Brownian motion. Moreover, there is a deterministic growth of salaries, so that the salary of an \(m\) year old at time \(t\) is \(L(t) e^{w (m-a_0)}\). In addition, there is a financial market modelled by a Black-Scholes market driven by the same Brownian motion as the salary level. That is, financial market and salaries are perfectly correlated. The pension scheme can choose the amount \(\pi(t)\) invested in the financial market and the rate \(c(t)\) at which the members of age \(m \in [a_0,\tau]\) contribute. Thus the aggregate contribution to the scheme is \(c(t) L(t) I(t)\), where \[ I(t) = \int_{a_0}^\tau n(t-(m-a_0)) s(m) e^{w (m-a_0)} \;d m\;.\] The benefit for a retiree at time \(t\) is \(b_0 e^{\zeta t}\) for some \(\zeta >0\). The accumulated outflow rate is \(B(t)\), calculated similarly as the contributions. Thus the wealth of the pension scheme follows the SDE \[ d X(t) = (\pi(t)(\mu-r) + r X(t) + c(t) I(t) L(t) - B(t))\; d t +\sigma \pi(t) \;d W(t)\;,\] with \(X(0) = x\). A goal is to keep the contributions stable. One thus defines the target contribution rate as a weighted average of the past contribution rates \[ \psi(t) = \psi_0 e^{-\gamma t} + \nu \int_0^t e^{-\gamma(t-s)} c(s)\;d s\] for some \(\gamma > 0\). Thus, the target is a weighted average of the past contributions. The value of strategy is defined as \[ V(x) = \min_{c,\pi} E\Bigl[ \int_0^T \lambda e^{-r s}(c(s) - \psi(s))^2\;ds + (1-\lambda)e^{-r T} (X(T) e^{-r T} - x)^2\Bigr]\;.\] That is, the final wealth target is the initial wealth invested in the riskless asset. The problem is solved by a martingale method and illustrated by some examples. The work may be seen as a first approach to the problem. To make it more realistic, the salary level and the financial market should be driven by a two dimensional Brownian motion, the benefits should be linked to the present salary level and the salary level at retirement, the entrance age to the scheme should be different for different members, and also the demography should be stochastic. Moreover, it would make more sense to link the target level to the salary level than to a deterministic goal. Also, the quadratic utility for the final wealth may lead to the strange situation that money is thrown away in order not the get a too large wealth. It would be interesting to use different utility functions.