all of their short-term assets instead of rolling them over. Left to their own
devices, savers violate the standard intertemporal consumption condition,
U
0
(c
1
) = rU
0
(c
2
); by consuming too much, too soon when the interest rate r
turns out high. This commitment problem distorts savers’initial allocation
of wealth between the two assets and that misallocation lowers the return to
savings and savers’welfare.
2
Savers can overcome their commitment problem, or at least reduce it,
by locking their money in a bank before they are tempted to consume it,
i.e., before liquidity needs are realized. These "piggy bank" intermediaries
limit withdrawals in the high return state, thereby increasing the amount
reinvested in those states. Savers with early liquidity needs lose ex post, of
course, but they are better o¤ ex ante by delegating investment decisions to
an intermediary.
This commitment function is very di¤erent from the other roles attributed
to intermediaries. Our intermediaries are certainly not providing risk sharing
or insurance. In fact, consumption is more volatile with the "piggy bank"
so its commitment function is valued only by savers who are not too risk
averse. Nor are these intermediaries producing information; the assets here
are more like public securities that savers could hold directly and trade freely
as liquidity needs and interest rates change. Nor, …nally, are these banks pro-
viding liquidity; in some sense the problem here is really too much liquidity,
and the solution is too lock up funds in a intermediary that limits liquidity
in some states.
Many real world intermediaries serve the basic commitment function of
getting liquid assets out of peoples’hands before the more impulsive savers
can spend at regrettable times. Pension funds, insurance annuities, mutual
funds, and banks–the big four intermediaries–may all serve this commitment
role, quite apart or incidentally to their more familiar roles. Many of the
contracts these institutions sell to savers penalize the early withdrawal of
funds one way or another (or in certain events) as do the intermediaries here.
Bank CDs are the obvious example. Callable CDs, a recent bank innovation,
even charge penalties that are e¤ectively contingent on interest rates, very
much like the contracts intermediaries o¤er here.
3
2
To be clear, the commitment problem in our model does not stem from time inconsis-
tent preferences, as in Laibson (1997)
3
Issuers can call the CD over a speci…ed interval, either by paying o¤ the CD holder or
rolling it over at the new market rate. All else equal, banks will excercise the call when
rates have fallen and pass when rates have risen, so the reinvestment decisions by banks
2