Is TARP II enough?

A commentary in the VoxEU Debate Financial rescue and regulation

Posted By Michael Pomerleano , The World Bank on 11 February 2009

Is TARP II enough?
By Michael Pomerleano, Harald Scheule, and Andrew Sheng
The US Treasury has announced a Financial Stability Plan (a successor to the original TARP program) to help purge banks of their toxic assets. The government would achieve this by creating financial incentives for other players in the private sector to buy up the troubled assets. The basic idea is to lend government money through the Fed (or guarantee borrowings via FDIC) at a suitable spread over Libor to anyone willing to buy toxic assets from the banks. The private partners might, for example, be hedge funds or pension funds. The program would require the buyers to put up anywhere from 20 to 50 percent of their own capital and would force the buyers to bear first losses out of this capital, but the government would have no recourse to the buyers beyond that initial capital.

The plan has virtue in two respects: First, it recognizes that only the private sector can fully unlock the value of toxic debt, and second, it would remove the toxic assets from the balance sheets of the banks. However, the plan’s ultimate success will depend on the incentives it creates and the detailed arrangements made for its implementation.

Is the plan enough?
On the basis of macro, top-down calculations, we estimate that the US economy suffered a wealth loss roughly equivalent to 100% of GDP in 2008. Fed Flow of Funds data indicate that, at the start of the year, US households and corporations sector held real estate assets valued at the equivalent of 225% of GDP, while stock market capitalization was another 145% of GDP. Real estate values subsequently declined by an average of about 20% in 2008 while the stock market fell by some 40%. US households had suffered a loss of net worth equivalent to US$7 trillion (or nearly 50% of GDP by the third quarter of 2008), so the balance of the total loss of wealth would be borne by the corporate and banking sector. Since FDIC data showed that bank capital was only $1.3 trillion or under 10% of GDP, and there is general expectation that the asset bubble may deflate still more, it is not surprising that more bank losses are on the cards.
We think that the losses to the banking system could well be in the region of $3-3.5 trillion, based on two considerations. First, The IMF’s Global Financial Stability Report, released on January 28, 2009, estimates the potential losses of U.S.-originated credit assets held by banks and others at $2.2 trillion (up from $1.4 trillion in October 2008 due to worsening credit conditions) . If we add to this the $752 billion of losses that have been recognized to date (according to Bloomberg), we would arrive at a total of $3 trillion.
Secondly, for comparative purposes, we looked at the Japanese experience during their financial crisis of the 1990s. The fiscal recapitalization of the Japanese banks cost the equivalent of 24% of GDP. An equivalent proportional bill for the US could be around $3.3 trillion. It is true that the Japanese real estate bubble was larger than that in the US, but on other hand the Japanese banks did not suffer losses from holding toxic derivatives at that time.
Why are the current losses likely to be under-estimated? The devil is in the details.
Our recent study suggests that, to date, most of the losses that have been recognized are in the trading books of banks, which are marked to market, and for which the losses are thus transparent . But the trading book is only a fraction of a bank’s assets. A larger portion-- over 80% of assets-- are booked as banking assets. This are where the banks’ valuation and provisioning methodologies, which rely on external ratings and internal models, may still be under-estimating the losses to a serious extent.
First, the allowance that banks make for loan losses for credit risk is mostly modeled through-the-cycle, and is measured as averages over past business cycles. The use of averages rarely provides adequate provisioning for credit risk at a risky point-in-time. Moreover, current risk models are calibrated based on data from 2000 onward, which may be inadequate to provision for credit losses in a possible prolonged downturn. There have been two major downturns over the past three decades: 1981-1982 and 1990-1993. However, the level of present nonperforming loans (“NPLs”) so far recognized by the banks is only half the level of NPLs experienced in 1982, whereas the present globalized contraction is far more severe than the 1982 recession.
Second, these through-the-cycle provision models are also used by the rating agencies. Since the rating agencies are likely to downgrade banks’ assets further as the cycle deteriorates, these models will end up with pro-cyclical provisioning over time.
Since Basel II allows banks to use internal risk models to calibrate their need for equity capital, the use of such models has created a self-selection problem: namely that the banks which had the greatest desire to expand also had the greatest incentive to incur the cost of building these internal risk models that would allow them to skimp on provisioning and capital. As we know, VaR models have gros