used in the literature.2
One difference between the two schemes is that capacity utilization and prices
are restricted not to respond contemporaneously to a monetary shock in the CEE
estimation. However, we provide an alternative to the estimation by CEE since, as
Canova and Pina [6] show, it is very hard to rationalize theoretically the zero re-
strictions imposed by triangular identification schemes. Any theoretical model, in-
cluding the one described below, predicts all endogenous variables to have nonzero
contemporaneous correlations through the impact of the fundamental shocks. As
the sign restriction method shows, once the zero restrictions are relaxed a mone-
tary shock can have important contemporaneous effects on real variables such as
the capacity utilization rate.
Nevertheless, we observe the two estimations to be fairly similar both in size
and persistence of the responses. In particular, the sign restriction estimation
implies that an expansionary monetary shock is associated with a reduction of
interest rates of about 50 basis points on impact and this variable stays below
average for about 9 quarters. This movement in interest rates is accompanied by
a sizeable and persistent effect on capacity utilization with a peak at around 7
quarters where the use of installed capacity rises by 0.7 percentage points. Finally,
prices react slowly.
Table 1 provides evidence on a possible channel through which the monetary
shock operates. It presents the results from regressing two policy variables, namely
the growth rate of the Fed holdings of government securities (FEDSEC) and the
growth rate of nonborrowed reserves (NBR), on a constant, four lags and the
monetary shock estimated through the sign restriction scheme described in the
previous paragraphs. A monetary shock that unexpectedly reduces interest rates
generates a statistically significant increase in both, the Federal Reserve holdings of
government securities as well as nonborrowed reserves.3 This finding supports the
idea that a monetary shock is associated with a change in the assets and liabilities
composition of the Fed’s balance.
In the next section we build a model to capture these empirical facts. In our
economy real assets may be used as productive capital. Agents with low levels
of wealth borrow assets using units of accounts created by commercial banks.
2For the CEE identification scheme a six variable VAR(4) was estimated which
included, in the following order, capacity utlization, the growth rate of the GDP
deflator, the growth rate of a commodity price index, the federal funds rate, the
growth rate of nonborrowed reserves and the growth rate of total reserves. The
monetary shock estimated with this scheme has a correlation of 71 percent with
the shock estimated with the sign restriction approach.
3 These results are robust to a wide variety of specifications that include the
variables in levels as well as other sources of shocks.