we consider a Counterfactual scenario with a time-invariant μ set equal 0.12, the
benchmark calibration of BGG. We then use model-implied σ↑t and ω*t computed at
the time-varying NLLS estimate of μ. As shown in Figure 9, time variations in μt are
necessary to generate economically significant swings in the model-implied external
finance premium. The counterfactual scenario, in fact, implies very small movements
in the external finance premium over the entire cross-section of firms. Such small
movements are difficult to reconcile with the sharp drop in capital spending during
the latest economic downturn, unless one is willing to assume a very large elasticity
of investment with respect to the external finance premium.
Nevertheless, one might argue that the sharp increase in the external finance
premium during the latest recession stemmed instead from an increase in the idiosyn-
cratic risk. In fact, according to Figure 2 and 3, a bigger external finance premium can
be associated with both a bigger bankruptcy cost parameter μ and higher volatility
of the idiosyncratic risk σ . As shown in Figure 10, however, the runup in the ex-
ternal finance premium cannot be attributed to a large increase in the idiosyncratic
risk. The model-implied volatility parameter σ for the sales-weighted median firm is
fairly constant over the business cycle, despite a noticeable increase in the expected
default probability.26 The relatively stable σ appers to confirm our account of the lat-
est economic downturn, according to which to explain the large movements in credit
spreads and the external finance premium, you do need substantial time variation in
the bankruptcy cost parameter μ.
26In terms of Figure 2 and 3, the intuition is as follows. Small changes in the volatility of idiosyn-
cratic risk cause large and unlikely variations in credit spread and external finance premium. On
the contrary, significant changes in the bankruptcy parameter imply relatively smaller variations in
credit spread and external finance premium.
29