Predicting Bankruptcy

Maureen F. McNichols of Stanford Business School and several co-authors have conducted a fascinating study regarding corporate bankruptcies.  They have examined the usefulness of financial statement analysis as a tool for predicting bankruptcy.  They analyzed data from 1962-2002 for thousands of publicly traded companies.  They found that, over time, financial statement analysis (traditional ratio analysis and the like) became less useful as a means of predicting corporate bankruptcies.   Note that the analysis was still quite useful, just not as effective at predicting bankruptcy as it was back in the early to mid-1960s.  Why might that be the case?  The scholars offer several suggestions.  First, companies restate earnings more frequently today than they did in the 1960s.  That would suggest a higher frequency of earnings manipulation of earnings today.  Second, many tech companies spend a significant portion of sales on research and development.  Those investments do not make it onto the balance sheet in the way that capital investments in property, plant, and equipment do.   As a result, ratios become less useful in predicting bankruptcy.  Finally, more firms have negative income today than in the early 1960s.  When firms lose money in a particular year, it becomes much harder to predict what will happen to them in the following years.   Yet, losses in a particular year don't necessarily mean a bankruptcy is in the future.