The Impact of Individual Investment Behavior for Retirement Welfare: Evidence from the United States and Germany

Figure 3: Welfare Losses, ∆W0 / W0, for U.S. (SCF) Data;
Gender = 0 (male) or 1 (female), γ = 2, δ = 0.97, Age = 50, Education = Middle

--- put Figure 3 here ---

Figure 3 shows that the loss in welfare is usually larger for higher levels of labor
income. In general, the welfare loss due to variations in income is influenced by
three factors. First, higher income relative to net worth is associated with lower
savings (income crowds out savings). At higher levels of income, savings are often at
or near zero. Thus, for those with high income, asset allocation sometimes does not
matter at all (in case of zero savings) or does not matter much. This effect explains
why the 75% income quantile curve is lower at low levels of net worth. Second, for
those with higher income, the benchmark model predicts increasing shares of risky
asset holdings. But the coefficient for labor income/net worth from the regression
works in the opposite direction. Thus, the higher the income, the larger is the gap
between optimal and empirical risky asset shares, and the larger is the loss in
welfare. This effect explains why the 75% income quantile curve mostly lies above
the 50%, and the 50% above the 25% curve. Third, the loss in welfare is measured in
percent of current net worth. For low levels of net worth, most of the individual’s
overall expected lifetime wealth derives from expected future labor income (future
net worth). Thus, losses in welfare due to suboptimal asset allocation mainly stem
from investing future labor income payments suboptimally (as opposed to current
wealth for wealthy individuals). Relative to the considerably smaller level of current
net worth, the losses thus are proportionally larger. The last effect also explains the
horizontal order of the curves and, for the 25% and 50% line, the generally
decreasing tendency in net worth.

Varying net worth (and always implicitly the ratio of labor income to net worth)
produces a fourth effect that results in a decreasing shape of the curves. As stated
above, for increasing net worth (i.e., for a decreasing labor income-to-net worth
ratio), the benchmark model predicts lower risky asset shares (compare Figure 1).
The regression coefficient for the log of net worth predicts exactly the opposite.

24

<<   22   23   24   25   26   27   28   >>

More intriguing information