Principles of Corporate Finance_ 12th Edition

(lu) #1

610 Part Seven Debt Financing


bre44380_ch23_597-617.indd 610 09/30/15 12:08 PM


mechanical credit scoring systems are used by banks to assess the risk of their corporate loans
and by firms when they extend credit to customers.
Suppose that you are given the task of developing a credit scoring system that will help
to decide whether to extend credit to businesses. You start by comparing the financial state-
ments of companies that went bankrupt over a 40-year period with those of surviving firms.
Figure 23.8 shows what you find. Panel (a) illustrates that, as early as four years before they
went bankrupt, failing firms were earning a much lower return on assets (ROA) than firms
that survived. Panel (b) shows that on average they also had a high ratio of liabilities to assets,
and Panel (c) shows that EBITDA (earnings before interest, taxes, and depreciation) was low
relative to the firms’ total liabilities. Thus bankrupt firms were less profitable (low ROA),
were more highly leveraged (high ratio of liabilities to assets), and generated relatively little
cash (low ratio of EBITDA to liabilities). In each case these indicators of the firms’ financial
health steadily deteriorated as bankruptcy approached.
Rather than focusing on individual ratios, it makes more sense to combine the ratios into
a single score that can separate the creditworthy sheep from the impecunious goats. For
example, William Beaver, Maureen McNichols, and Jung-Wu Rhie, who studied these firms,
concluded that the chance of failing during the next year relative to the chance of not failing
was best estimated by the following equation:^23

Log(relative chance of failure)

= −6.445 − 1.192(ROA) + 2.307 (^) ( liabilities____assets (^) ) − .346 (^) ( ____EBI T DA
liabilities
(^) )
Credit scoring systems should carry a health warning. When you construct a risk index,
it is tempting to experiment with many different combinations of variables until you find the
equation that would have worked best in the past. Unfortunately, if you “mine” the data in this
way, you are likely to find that the system works less well in the future than it did previously.
If you are misled by the past successes into placing too much faith in your model, you may
refuse credit to a number of potentially good customers. The profits that you lose by turning
away these customers could more than offset the gains that you make by avoiding a few bad
eggs. As a result, you could be worse off than if you had pretended that you could not tell one
would-be borrower from another and extended credit to all of them.
Does this mean that firms should not use credit scoring systems? Not a bit. It merely
implies that it is not sufficient to have a good system; you also need to know how much to
rely on it.
Market-Based Risk Models
Credit scoring systems rely primarily on the companies’ financial statements to estimate
which firms are most likely to become bankrupt and default on their debts. For small busi-
nesses there may be little alternative to the use of accounting data, but for large, publicly
traded firms it is also possible to take advantage of the information in security prices. These
techniques build on the idea that stockholders will exercise their option to default if the mar-
ket value of the assets falls below the payments that must be made on the debt.
Suppose that the assets of Phlogiston Chemical have a current market value of $100 and
its debt has a face value of $60 (i.e., 60% leverage), all of which is due to be repaid at the end
(^23) See W. H. Beaver, M. F. McNichols, and J.-W. Rhie, “Have Income Statements Become Less Informative? Evidence from the Abil-
ity of Financial Ratios to Predict Bankruptcy,” Review of Accounting Studies 10 (2005), pp. 93–122. Their model uses the technique
of hazard analysis. Another popular model, the Z-score model, uses multiple discriminant analysis. This was originally suggested by
Edward Altman and is described in E. I. Altman and E. Hotchkiss, Corporate Financial Distress and Bankruptcy, 3rd ed. (New York:
John Wiley, 2006).

Free download pdf