5 Empirical Results
5.1 Benchmark Estimates of Financial Market Frictions
Figure 7 shows the evolution of the estimated bankruptcy cost parameter μ over our
sample period. From early 1997 to the end of 1999, the point estimate of μ appears
to be quite stable, hovering in the range between 0.08 and 0.16. One clear exception
during this period is the substantially larger estimate for 1998Q4. This jump in
estimated bankruptcy costs reflects the turbulence in financial markets following the
Russian default and the LTCM crisis in the late summer of that year. Smoothing
through the 1998Q4 spike, the average estimate of μ during this period is remarkably
close to 0.12, the value chosen by BGG in the steady-state calibration of their model.
Our estimates are also within the range of bankruptcy costs estimated by Altman
(1984) for a sample of industrial firms that declared bankruptcy in the mid-1970s.23
The bursting of the stock market bubble in the spring of 2000 significantly de-
pressed equity valuations, causing an increase in corporate leverage. By the onset of
the last NBER-dated recession in March 2001, credit spreads also had widened sig-
nificantly. In the context of the BGG model, however, the rise in leverage apparently
was insufficient to account fully for the runup in credit spreads. A part of the increase
in credit spreads during this period reflected an increase in the external finance pre-
mium, as our estimates of ω*t only rose slightly (see equation 6).24 The increase in
the external finance premium also did not come about from higher σ'*t’s, the point to
which we return later. Instead, it stemmed from an increase in bankruptcy costs, as
evidenced by our estimates of μ, which more than doubled over this period.
After declining moderately over the course of the 2001 downturn and into early
2002, the estimated μ rose sharply in the latter half of the year. This increase likely
reflected a further plunge in share prices and the balooning of credit spreads, as
the post-Enron wave of corporate governance scandals and heightened geopolitical
tensions that preceeded the invasion of Iraq rattled investors’ confidence. For this
period as a whole, our estimates of bankruptcy costs are much closer to the average
23Altman’s (1984) estimates of bankruptcy costs include both the direct and indirect costs and
average between 11 percent and 17 percent of the value of the firm. Direct costs—explicit admin-
istrative costs paid by the debtor during the reorganization/liquidation process—were taken from
the bankruptcy records of individual firms. Measures of indirect costs, namely lost profits, were
estimated.
24For example, the sales-weighed median ω*t rose from 0.218 to 0.236 in 2000.
23