Thank you for inviting me to offer a few observations on some design and implementation issues we all face with regard to macroprudential supervisory policies. [Note]

Perhaps a positive offshoot of the recent severe financial crisis is that it has galvanized policymakers to expedite work on macroprudential supervision and regulation. The sense of urgency has helped regulators make significant progress on increasing bank capital buffers and has sharpened the focus on liquidity risk. These efforts are important, and should improve the strength and resiliency of the financial system when fully implemented.

[Slide 2] Over the last year I have sometimes made a comparison between new and needed financial regulations and the regulations developed in the 1800s to prevent the "great fires" that had plagued many major cities. One thing that greatly reduced the risk of such fires was the development and enforcement of tough new building codes. This meant using building materials that were less flammable, and raising structures that made it far less likely that a building would catch fire - or that would suppress the spread of a fire, should it nonetheless occur. In many respects I see enhanced capital and liquidity rules as equivalent changes in our financial system, which should make financial institutions far less susceptible to "financial firestorms."

Another aspect of the solution to great fires was the move to hiring professional fire fighters with adequate equipment and sufficient access to water to quickly respond to a fire. This reduced the risk that if a fire occurred, it would spread in some catastrophic way.

The importance of response time is highlighted in the questions that get asked when you take out fire insurance on your house. They ask how close you are to the nearest fire station, and how close you are to a fire hydrant. The questions are important because the answers impact the extent of the damage a fire can cause. They also impact the likelihood that a fire at a neighbor's house could spread and impact your home.

Response time is also critically important in financial regulation and in limiting financial damage. Today I am going to briefly discuss one type of financial "first response" - the ability to retain capital in the banking system by cutting bank dividends when problems first emerge.

Given the evidence in the data, I will highlight the fact that in relation to dividends, bank management and supervisors responded quite slowly in the United States during the financial crisis. I will suggest that macroprudential supervision should include early detection of problems, and in a troubled banking system the reduction of dividends to retain capital.

The Timing of Dividend Reductions

Figure 1 shows that in August 2007, as the financial crisis emerged, banks became willing to lend to each other at one-month maturity only at rates dramatically different from the previous year. While the increase in August 2007 looks a bit less spectacular than the spike that followed the events of the fall of 2008, it was nonetheless a dramatic change. The economic implications of emerging financial problems contributed to the significant response in monetary policy.

But as Figure 2 highlights, the 2007 events did not lead to similarly significant changes in supervisory policy. The dividends on common stock declared by the largest banking organizations (the 19 SCAP [Footnote 1] participants and others) actually increased in the fourth quarter of 2007, and did not show dramatic reductions until after the financial crisis hit a crescendo in the fall of 2008.

Figure 3 puts in perspective the cumulative cash dividends of the largest banks by comparing them to the Capital Purchase Program (CPP) funds they received.

[Slide 6] This suggests that if dividends had been halted at the SCAP banks once the LIBOR rate rose, nearly $80 billion would have been retained as capital. This represents close to 50 percent of the CPP funds used to recapitalize these banks in the fall of 2008. Clearly a proactive approach to dividend retention could have substantially reduced the need for an emergency infusion of public funds.

Financial Indicators of Emerging Problems

Regulatory dividend policy was focused on the ability of banks to pay dividends out of earnings. However, earnings declines occurred later than the financial market responses to emerging problems. In addition, banks' management and boards of directors rarely are proactive in reducing dividends, because of concerns that it will signal that they are experiencing more severe financial problems than their peers. With earnings somewhat slow to react to emerging problems, and boards reluctant to signal problems, most of the large banks were very slow to react to the crisis with reduced dividends.

Similarly, banks' supervisory ratings were also slow to react, as shown in Figure 4. The measure of the share of assets in so-called "problem" banks [Footnote 2] - the blue line - responds only with a delay. A similar pattern emerges if one looks at the percentage of bank assets in the three lowest supervisory ratings (CAMELS ratings of 3, 4, or 5), although it captures a much higher percentage of total bank assets. In contrast, the credit default swap spread for the four largest U.S. banking organizations - the green line - began to climb in the fall of 2007 and continued to climb throughout 2008 and into 2009. I should note that we use the four largest banks simply because of the availability of historical credit default swap data, but our modeling suggests this is representative for the largest banks more generally.

Figure 5 shows that a similar pattern emerges if one uses the market capitalization relative to assets of the four largest banking organizations. [Footnote 3] Like the credit default swaps, the measure (the green line) began to decline significantly in the fall of 2007 and continued to decline through 2008 and into 2009. Again this financial indicator of bank health moved far more quickly than regulatory ratings (see the blue line) or dividend reductions.

Figure 6 highlights another issue - that as financial markets took a bleaker view of banks' financial prospects (in the top chart), banks had begun to tighten credit standards (in the bottom chart). By the end of 2007, credit standards had tightened substantially, relative to where they were in the beginning of that year. Banks seek to maintain reasonable capital-to-assets ratios and loans are, of course, assets for banks. If capital is eventually reduced due to loan write-downs, and banks reduce lending to maintain capital-to-asset ratios, the "collateral damage" to those who would like to borrow from "supply-constrained" banking institutions can impose a significant cost to the economy more broadly. So the impact of only slowly reducing dividends, rather than retaining capital in the banking system, extends well beyond losses to stockholders and potentially taxpayers.

Policy Implications

[Slide 10] Even from the perspective of microprudential supervision, more proactive dividend reduction would have ameliorated some of the costs of the financial crisis. Rather than basing regulatory dividend policy on accounting constructs that tend to be slow moving and backward looking, regulatory policy should prompt supervisors to look forward and thus more proactively seek reductions in dividends in appropriate circumstances. In particular, given the financial-market signals of impending problems that I have highlighted today, I believe that supervisors need in such circumstances to be more proactive in seeking capital retention through reduced dividends at banking organizations.

[Slide 11] Of course, signaling problems can make it difficult for supervisors to mandate reductions at individual institutions. When a wide variety of indicators suggest potential problems in the banking sector, signaling problems could be avoided if, hypothetically, all banks were required to reduce dividends - or all banks with outsized exposures. For example, banks with large real estate exposures in loans and securities could in such a situation be required to reduce dividends before accounting losses were realized. Such a policy might very well reduce the credit-availability problems that have unfortunately been a hallmark of this financial crisis.

In retrospect, empirical analysis suggests that bank management and supervisors had an opportunity to retain capital once problems became identifiable, but before the full extent of the crisis emerged in the fall of 2008. In the spirit of learning and applying lessons, one must consider whether this dynamic could repeat itself - or indeed is repeating itself. For example, in some parts of the world today, sovereign risk exposure seems problematic and elevated credit default swaps and low stock market valuations have not prevented some banks from continuing to pay dividends.

In sum, I suggest that more aggressive policy may not only reduce potential taxpayer exposure to the banking sector, but also insulate the broader economy from the sorts of loan-supply shocks that can provide significant headwinds to an economic recovery.

Thank you.