A better way to fix the banks

By Lowell Bryan and Toos Daruvala

Here’s a plan that could solve the toxic-asset pricing problem voluntarily—without requiring Uncle Sam to nationalize the whole industry—and make (pretty much) everyone a winner.

This short essay is a Conversation Starter, one in a series of opinions on topical issues. Read what the authors have to say, then let us know what you think.

In recent speeches, President Obama and Fed Chairman Ben Bernanke have said clearly that the United States will provide whatever capital is needed to keep US banks solvent and that they do not plan to nationalize even deeply troubled banks. This is good news, because economic recovery requires a healthy, profit-motivated US banking system.

However, there has largely been silence from the administration on taking any action to remove bad loans from bank balance sheets. Until the hundreds of billions of dollars of impaired assets that currently weigh down bank balance sheets are removed, credit flows will be restricted. If we wait for the banks to absorb the losses from these loans (net of recoveries), we will wait a long time, and the economic turnaround will be very slow in coming.

To date, plans to remove these bad assets have foundered on precisely how to do so—specifically, on how best to value the assets. Details matter. The Treasury has launched a program to use “stress tests” to identify, proactively, the bad assets on bank balance sheets, using financial models to project future loan values and loss rates under different economic scenarios. The intent is to get better answers about the extent of future losses and to better understand which banks are healthy. But what happens next? Which valuation method will be used to buy the bad assets? If prices are set too low, banks won’t sell the assets unless forced to do so. If prices are too high, the existing shareholders would be “unjustly enriched.” This dilemma is what stopped the first Paulson plan to remove toxic assets, and no amount of stress testing gets you around it.

Our aim is not to plunge into those political debates but rather to propose a broad answer to the valuation challenge under two basic assumptions: (1) forced asset sales are appropriate only for deeply troubled financial institutions, and (2) what’s essential now is a powerful and compelling voluntary program that motivates the rest of the banks to clean up their balance sheets before the deteriorating economy and forced write-downs in the future also cause them to become deeply troubled.

This risk cannot be overstated. McKinsey—as well as others, such as Goldman Sachs—estimates that US banks may currently hold as much as $2 trillion of impaired assets. Given the likely depth and duration of the recession, the losses on them could eventually exceed $1 trillion—on top of the $500 billion in losses already realized. Whatever the precise tally, the final reckoning is certain to be larger than many US banks can absorb out of common equity and from their earnings.

So what needs to be done? The answer first requires a brief detour into Accounting 101, which will also explain why the market in bad loans has so far been moribund. At present, assets on bank balance sheets are valued in either of two ways: fair value or hold-to-maturity value. Where possible, fair value uses mark-to-market accounting; however, absent the ability to determine a real market price, a mark-to-model approach must be used. In hold-to-maturity accounting, so long as principal and interest get paid under the terms of the original loan agreement, assets remain on the balance sheet at their original value. Most securities are valued using fair-value accounting (unless they are treated as long-term investments). Most loans use hold-to-maturity valuations. To date, the lion’s share of the mark-downs absorbed by banks have been on securities and loans subject to fair-value accounting. However, more than 60 percent of the credit on the balance sheets of US banks uses hold-to-maturity accounting, and it is within these assets that the bulk of future losses will occur.

Now assume you want to create a market for such impaired assets. The problem with fair-value accounting is that in the absence of a real, active market to set prices, the only alternative is to mark to someone’s model. But whose? Since private investors are motivated to make money, they want to use conservative assumptions to value securities. That, in turn, gives banks little incentive to sell—and so most have not. Absent government coercion, such a standoff will probably continue. As for the even bigger amount of potential bad loans that banks now value under hold-to-maturity accounting, the problem is that while you can actuarially foresee, under various assumptions, that some percentage of a portfolio will go bad, you can’t know which specific loans will default or how much of the original loan value you will recover. That too is a recipe for inaction.

To break the logjam, we propose that the government step in and establish a voluntary program to create a real market price and terms for the sale of bad assets. Rather than use modeling for valuation, the program would set discounts from either of the two basic approaches to accounting value, based on some recent past date (for instance, December 31, 2008). A reasonable level might be 10 percent off for securities already marked to fair value and 20 percent off for loans being held to maturity. Upon their sale to the government, existing shareholders would absorb the loss taken on the discount, and that loss of common stock value would be replaced by converting TARP1 preferred stock to nonvoting common (which would be vested with voting rights if sold to private parties).

Under this approach, the banks themselves could determine which assets to sell to the government, based upon its posted terms and conditions. If, over time, the government wanted to encourage sales, it could reduce the discount. If it wanted to discourage them, it could raise the discount. As time passed, the accounting valuation date would also be updated (for instance, from year-end 2008 to June 30, 2009).

The government would be partially protected from overpaying through this approach by its increased ownership of common stock in the bank, which means it would recover, as a shareholder, much of whatever it overpaid. If it underpaid, it would keep the gain. In addition, the government could provide an incentive for the banks, which should know these credit instruments best, to maximize the value of the assets they offload to it—say, by allowing them to earn a percentage of the subsequent asset recovery price as a servicing fee.

By our rough figures, if the government purchased $1.5 trillion in assets with an average 20 percent discount from accounting value ($300 billion), it would end up acquiring an ownership stake of some 36 percent in the industry as a result of the conversion of preferred stock to common or the injection of new common stock to make up for the equity lost through its discounted purchases. While that is a significant stake in the banking industry, it remains considerably less than what would occur under full-blown nationalization. And if restoring the ongoing-concern value of banks helped the industry’s valuation to regain its January 2007 level (about 20 percent less than its high), we calculate that the government’s shareholdings would be worth about $560 billion. Upon sale to private hands, this kind of gain would go a long way to recouping the costs of TARP.

There can, of course, be many variations on this proposal. Our point is that the government should be offering carrots, not just sticks, if it wants the banking industry to move quickly to get rid of bad assets—and to avoid the likelihood of even greater pain down the road.

About the author(s)

Lowell Bryan and Toos Daruvala are directors in McKinsey’s New York office.

More on Financial Services

The next-generation operating model for the digital world

Article - McKinsey Quarterly

What makes a CEO ‘exceptional’?

Article - McKinsey Quarterly

Three game changers for energy

Article - McKinsey Quarterly

How functional leaders become CEOs