from age x to age 64 at the end of each year t. In later periods, from age 65 to T, Lt is
replaced by a deterministic (government) pension income that stays constant in real
terms. Thus, the retirement age is exogenously fixed at age 65.
The maximization problem is given by:
subject to consumption constraints:
max
αt,Ct
E0(U(C)),
(3)
C0=W0-S0
Ct=St-1(1-αt-1)Rf+St-1αt-1Rt-1+Lt-1-St∀t∈{1,2,...,T-x},
144444424444443
Wt
subject to borrowing constraints:
0 ≤ St ≤ Wt , (5)
and subject to no-short-sale constraints:
0 ≤ αt ≤ 1. (6)
3.2 Calibration
In this section we calibrate our model for U.S. and for German individuals. We
report the choice of our benchmark parameters, but also give alternative values that
will be used for sensitivity analyses later in the paper. Table 1 summarizes the
calibration.
The individual’s preferences are described by setting the constant of relative risk
aversion γ to 2 (alternatively to 1 or 3), the subjective discount factor δ to 0.97
(alternatively to 0.95 or 0.99), which are typical values found in intertemporal
optimization models (see, e.g., Laibson, Repetto, and Tobacman, 1998).
For the U.S. survival probabilities, we use the United States Life Tables 2003 (see
Arias, 2006); for German survival probabilities, we use the Life Table for Germany
2002/2004 from the German Federal Statistical Office (see Federal Statistical Office,
10