Regardless of whether the U.S. banking system would not presently be able to meet its liabilities with its assets, there is another question: assuming that banks are allowed to extend and pretend for a long enough period of time, will they ultimately be able to accumulate enough retained earnings in the years ahead to cover eventual loan losses? In other words, is it possible that everything will be OK if we just look the other way long enough?
From my perspective, it depends on what "OK" means. Simply in terms of long-term solvency - assets being ultimately able to meet liabilities - my impression is that yes, given enough time, retained bank earnings should cover the losses on existing loans. Indeed, it's possible that banks might be able to report fairly healthy "operating earnings" to investors, and then somewhat more quietly write off losses as "extraordinary" charges over a period of years. This type of outcome is beginning to look possible, because investors evidently don't mind repeatedly having their pockets picked as long as "operating earnings" come in above analyst estimates.
Unfortunately, in that sort of world, the economy would likely be hobbled for a long period of time, as Japan has discovered over the past couple of decades. With banks focused primarily on survival and recapitalization, retained earnings would be directed to making the existing liabilities whole, rather than contributing to productive new investment.
In January, the London Economist reported a study by the McKinsey Global Institute, noting "Assigning the odds of further deleveraging is not the same as gauging its likely economic impact. To do that, the study looks to history. It finds 32 examples of sustained deleveraging (at least three consecutive years in which ratios of total debt to GDP fell by at least 10%) in the aftermath of a financial crisis. In some cases the debt burden was reduced by default. In others it was inflated away. But in about half the cases—which the report regards as the most appropriate points of comparison—the deleveraging came through a prolonged period of belt-tightening, where credit grew more slowly than output. The message from these episodes is sobering. Typically deleveraging began about two years after the beginning of the financial crisis and lasted for six to seven years. In almost every case output shrank for the first two or three years of the process. (Countries which defaulted or inflated their debt away saw bigger recessions at first, but had higher output growth than the belt-tighteners by the end.)"
So to the extent that "extend and pretend" is successful in averting insolvency concerns, it will also tend to weigh down lending activity, as resources are allocated toward servicing existing debt burdens on bad assets, rather than toward new lending for productive activity. The most efficient outcome is always for lenders who provide capital to take losses if the loans go bad. That sort of market discipline is the only way to ensure that capital gets allocated properly. This is not the world that we have lived in over the past year, as policy makers have pledged public money to make private bank bondholders whole, regardless of how irresponsibly the banks allocated the money. But it is important to recognize that this policy comes with longer term costs.