Roundtable panelists are (l to r): William A. Stark, FDIC; Hal S. Johnson, BB&T; and Tanya S. Azarchs, Standard & Poor's.

Changes in the level and shape of the interest rate yield curve pose challenges for interest rate risk management in the U.S. banking sector. Based on a number of recent studies and analysis, there is general agreement that interest rate risk at institutions insured by the Federal Deposit Insurance Corporation (FDIC) is, for the most part, well managed. However, in spite of this general conclusion, there are a variety of approaches that individual financial institutions can take to manage this risk. Ultimately, the appropriateness of the techniques used by each institution tomanage its interest rate exposures can only be evaluated on a case-by-case basis. But given that all institutions are subject to the same market forces and have to make similar types of choices among analytical techniques and mitigation strategies, it is also useful to discuss interest rate risk from a macro perspective.

It was with this goal in mind that the FDIC convened a January 13, 2005, interest rate risk roundtable to discuss macro trends and techniques with leading industry experts. The panel of discussants included Tanya S. Azarchs, Managing Director, Financial Services Ratings, Standard & Poor's; Hal S. Johnson, Executive Vice President, Funds Management Department, BB&T; and William A. Stark, Associate Director of Capital Markets, Division of Supervision and Consumer Protection, FDIC. The roundtable was moderated by Richard A. Brown, the FDIC's Chief Economist.

The following is a summary of the roundtable discussion.

MR. BROWN: Good afternoon, and welcome to today's interest rate risk roundtable. Interest rates are a topic that economists know a lot about, but interest rate risk is another matter. This is a highly institutional area, involving a complex interaction between interest rate and yield curve trends and the unique financial positions held by each depository institution. A great deal of institutional and technical information is required to understand this issue, and so we are very fortunate to have with us today a panel of experts who are accustomed to assembling this information into useful analyses.

I would like to start out by asking each panelist to give a little of his or her own institutional perspective on interest rate risk, coming from a rating agency, a bank, and a regulatory perspective.

MS. AZARCHS: As a rating agency, I think our interests are rather well aligned with the interests of regulators in the sense that we care first and foremost about what regulators call safety and soundness, or what we call relative imperviousness to default. But the difference that we have as a rating agency is that we do not spend as long examining or analyzing each institution. We leverage off the examination work the regulators do.

That being said, we look into the issue of interest rate risk in the context of a broader initiative, which is to look at enterprise-wide risk management. We have a fairly intensive effort under way to do that. We look at market risk as it is expressed in trading risk. We think that is a larger risk than structural interest rate risk-taking for the institutions that are major trading houses, and even not-so-major trading houses. Credit risk continues to be of paramount importance to us. We still think that when banks fail, they typically fail the good, old-fashioned way, which is by making a mistake on credit risk. If I were to rank-order interest rate risk, I would put it third on our list, coming after market risk and credit risk.

When we look at interest rate risk, we are looking at very much the same things that the regulators look at. We look at corporate governance—that is, the policies, the procedures, and the communication that goes on amongst the various constituents at a bank. We also look at the methodology and the measurement of the risks. It is very difficult for us to communicate back to the outside world how we get comfortable with the interest rate risk-taking, because interest rate risk management is an art and not a science. We are fully cognizant of the assumption-driven, methodology-driven conclusions that one reaches from it.

Broadly speaking, we think that banks have done a fairly good job of managing interest rate risk over the years, during which the net interest margin, for example, has been fairly solid and uncorrelated to rate movements in the industry. But, we look into interest rate risk just to make sure the regulators do not miss anything.

MR. JOHNSON: I would concur with Tanya that interest rate risk probably would not be "Job One" at our institution. Credit risk holds that top spot. We look at a broader definition than interest rate risk; we call it market risk, and we do pay very close attention to it. We do not have a tremendous amount of interest rate exposure on our balance sheet. However, from a market perspective, we have a significant exposure to those things that we cannot necessarily control, and those are the things that we spend a lot of time with our team trying to figure out. We try to gauge what the next thing coming down the track is and what we can do to help either mitigate the effect or somehow offset it.

Right now, for instance, competitive loan pricing is a big factor in our marketplace as the rates on loans relative to the traditional indexes have fallen. We have seen that particularly in the mortgage market. Additionally, the supply of loans is not very robust right now, so that obviously is reflected in the pricing.

It is a real challenge trying to model interest rate risk.You model what you believe your balance sheet growth is going to look like, and as that growth changes or becomes different from what your expectations are, that also has an impact on your interest rate risk profile.

We are still a heavily margin-reliant institution, and so the interest rate risk management process is very important to us. Maintaining that margin and making sure that we use appropriate hedges to try to minimize our exposure to extreme interest rate events is important to us.

MR. STARK: It is always enjoyable to talk about interest rate risk, because it is a challenging area and hard to describe. When you make a loan and somebody does not pay you back, you know what your loss is. However, if interest rates move against your balance sheet position, it is not as straightforward a process to measure your loss.

Looking at interest rate risk from a supervisory viewpoint, I try to keep track of two things. First, I keep track of the type of assets banks are buying. Second, I watch management's behavior in periods of volatile interest rates in order to assess how they are managing their risks during these periods, to see if they have adequate controls, hedges, or mitigants in place and how those controls perform.

Regarding the type of assets that can present interest rate risk, exposure to mortgage-related assets is an obvious starting point. The mortgage loan is not a perfect instrument. From an investor's perspective, there is a "flaw" in the basic mortgage that is used here in the United States: it has an option in it.

When you invest in a mortgage, there are three things that can happen—and two of them are bad. One event is that rates go up and the value of the loan goes down. A second event is that rates go down and the customer pays off the loan. A third event is that rates stay exactly the same and you earn exactly what you thought you were going to earn.

As a result, this is an interesting category of assets to watch, because we know it creates unique challenges for the bank managers who are trying to manage the interest rate risk associated with these option instruments.

Finally, in an overall sense, I reach the same general conclusions as my fellow panel members that interest rate risk ranks secondary in a ranking of risks; it does not share the same level of concern as credit risk. Historically, we have found that credit risk is the primary cause of bank insolvencies, with rate volatility and related impacts to earnings and capital being an additive, but not primary, cause of failure.

MR. BROWN: I wanted to start getting into some detail by talking about modeling interest rate risk. Treatment of interest rate risk has made a lot of advances since the interest rate squeeze of the late 1970s and early 1980s. Mortgage lenders and other institutions have had to prepare for—and model for—volatile interest rates. Reliance on models is reassuring to many, especially to economists, but clearly models have their shortcomings. They have their challenges in terms of the accuracy of the inputs, the validity and testing of assumptions, and policies involving independent review.

I would like to hear your perspective on the art of modeling, and how reassured we should be with the results.

Tanya Azarchs makes a point, with Richard Brown at right.

MS. AZARCHS: Perhaps I am speaking more for myself than for S&P at large, but it raises all of my antennae when I hear that something is all model-driven, because models are always very sensitive to the assumptions that you put into them. We are very leery of models because of these critical assumptions that have to be made, and we see different banks make very different assumptions that drive the outcome of the sensitivity models they run.

We rarely see a bank, at least among the top 100 that we rate, that does not indicate in its disclosure a fairly modest amount of interest rate risk—2 percent, 3 percent impact or something like that. Yet banks make critical assumptions about the duration of indeterminate maturity liabilities, like demand deposits and savings accounts. The assumptions vary all over the board; for example, we see assumed durations of seven months or seven years. You can make all the arguments that you want that maybe different deposit bases have different behaviors, but they are really not that different, not different enough to warrant an assumption of seven months versus seven years.

Some other assumptions that go into modeling are the prepayment of mortgage-related assets and the duration attributed to accrued receivables. There is also the issue with the large, complex banks where they sometimes carve out the trading book from the whole interest rate risk model. Trading books typically are short-funded because they are supposed to be short-lived assets, at least on the books, but whether you put in a large trading book or you take it out makes a big difference in terms of the results that you get from the model.

So, what are the right assumptions to make when modeling interest rate risk? The real question might be, what are the conservative assumptions? What will give you an investment portfolio that will not leave you high and dry if interest rates really take a big move?

MR. JOHNSON: I would certainly echo that the whole process of asset-liability management is an art, not a science. I think you have to look at your models as a series of tools that help you to build a circle around what your real exposures are and what your opportunities are.

We run a number of different models in our institution, and we look at all of the model outputs in the aggregate to determine the common themes that are impacting our institution and the unique things that some of the models show.

Modeling itself is extremely complex. We have ten people in our organization dedicated to running risk management models, and that is only one set of many models that we run in our bank. We run monthly processes and quarterly processes as we try to determine our market risk. We try to define what is going onwith the interest rate environment and with our competitive environment as it impacts items on our balance sheet.

Given that we are coming off the lowest long-term interest rates in a generation, I think you have to question whether historical prepayment models will be accurate in the current environment. You have to put some parameters around that and shock the models and determine some reasonable levels of variation around the expectation that the model has given you.

That is really where the art comes into the process. You have the science there in the form of a model. To turn that science into the art, you have to use common sense. You have to use your sense of what has happened in markets historically and what you think is different about the markets today. And you have to try not to get wedded to one particular model outcome.

MR. STARK: As regulators, we have asked the question, what type of model does a bank need to have? We arrived at the conclusion that the bank should have a model that accurately measures its interest rate risk in light of the risk profile of the particular institution. There is, accordingly, no one-size-fits-all interest rate risk model.

And I agree with Hal that the risk models are just tools, and that is all they will ever be. We sometimes get a little concerned with institutions that get so involved in the quantification process and the results that they begin to not question the outcomes. From a supervisory perspective, we focus more on the validity of the overall risk measurement system and validation process, including, specifically, the nature of the bank's stress testing and that it is looking at a range of scenarios.

One area that we are fairly sensitive to is mortgage prepayments. If 25 percent of an institution's balance sheet is mortgage-dependent, that means 25 percent of the balance sheet has to have prepayment assumptions assigned to it. These can be a range of assumptions, depending on the coupons and other factors. One of the things we have observed in the past is that institutions may not always keep their prepayment assumptions updated. Mortgage prepayments are directly dependent upon the current interest rate environment, and that interest rate environment changes over time.

MR. BROWN: Are there any good, off-site quantitative measures that you can use to readily assess, at least at a broad level, the interest rate exposure of an institution?

MR. STARK: There are, and I can tell you what we do at the FDIC. Nine years ago, our Capital Markets Group developed an off-site monitoring tool called Interest Rate Risk Standard Analysis (IRRSA). IRRSA uses data from bank and thrift Call Reports and targets seven different factors that our examiners look at as part of the examination preplanning process. If the IRRSA model sends out red flags, the examiners know they need to consider spending more time during the examination looking at interest rate risk in the institution. So we also have the ability to monitor banks off site back in my department using these tools, and we talk to the regions if we see things that are out of line.

MS. AZARCHS: One of the things that we look at is the mortgage asset, which we think is really difficult, if not impossible, to hedge 100 percent. We look at the concentration in mortgage-backed securities and mortgage servicing rights (MSR), as well as the amount of other comprehensive income that more sophisticated banks may have.

MR. JOHNSON: From an external perspective, I would look at the financial instrument disclosures about derivative use and the rate of change in that derivative portfolio from quarter to quarter. I think that tells you a lot about whether the bank is really hedging out longer-term risk or whether it is trying to create market opportunities.

MR. BROWN: In the last interest rate cycle, during 1994 and 1995 when the Federal Reserve raised short-term interest rates by 300 basis points in just over 12 months, we heard a lot about problems with structured notes, or notes with embedded options. Some of the problems appeared to arise from the use of structured notes by relatively unsophisticated investors, but there may have been broader issues with the ability to manage the risks in general. Are structured notes of any concern in the present environment?

MS. AZARCHS: There are a lot more structured notes now. During the mid-1990s, many of the structured notes were, in fact, mortgage-backed securities of various kinds, either cash instruments or synthetic versions of the same. Currently, there are other types of structured notes.

Collateralized debt obligations (CDOs) are extremely popular, but it is difficult to get the disclosure on how many of them there are. CDOs may essentially represent an investment—and think of it maybe as a mutual fund—which is a fixed-income, interest rate risk-taking mutual fund that tries to better the returns in the bond market in general. All the bank has on its balance sheet is the equity portion, so there is no earnings impact until the day of reckoning comes. A lot of the interest rate risk in the off-balance-sheet fund is not recognized, so there might be some of those kinds of concerns out there. [See Chart 1.]

MR. JOHNSON: I think there are a couple of issues that are tied to this. Many instruments on a bank's balance sheet have imbedded optionality, and that makes the modeling and the assessment of those instruments more difficult.

I think the mark-to-market issue is very interesting because you often have situations where one side of the transaction gets marked to market and the other side does not. If you properly structure a transaction, you may not have any real interest rate risk, but you might have an enormous amount of risk in that transaction from an accounting interest-rate-reporting standpoint. So I think you have to separate the economics from the accounting.

We do not currently use CDO products in our bank, but we do use other structured products like collateralized mortgage obligation products and mortgage-backed products, and the models for these products are fairly robust. Coming out of this low interest rate environment, is there some incremental risk there? Possibly so, but the state of the art in modeling has made these structured notes more manageable today than ten years ago, and I think the incremental risk is fairly low.

MR. STARK: One problem we have always had here in the regulatory community, going back to 1993 or 1994 when we wrote the first guidance of structured notes, was the definition of a structured note. What is a "structured note" is not transparent, and that can create a problem.

About a year ago I was speaking to a bankers' group, and I was engaged by the group of bankers about how tough our examiners were being on their structured note holdings. What we discovered had been taking place was that the structured note item in the Call Report was rising very rapidly in a lot of banks, and that this increase had been causing some examiner inquiries, but the type of structured notes banks were buying were relatively short-term, single step-up notes, which are fairly harmless. Because there is a great build-up of liquidity all across the banking sector, but particularly in many community banks, bankers were placing their money in these types of short-term investments, which one could argue are just as safe as a short-term callable bond. These are agency-issued, so there is no significant credit risk associated with them.

So, in September of last year, the FDIC issued some guidance to remind our examiners about the characteristics of structured notes and some of the advantages they have and advising them of the lack of complete transparency in the Call Report filings.

MR. BROWN: We hear an awful lot about the so-called carry trade—the leverage programs that institutions and various types of investors have with playing the yield curve. Currently, short-term interest rates are rising and the yield curve is getting shallower. Are there still concerns about the carry trade and investors or institutions having a problem unwinding those positions?

MS. AZARCHS: My concerns are not so much about unwinding those positions as just the dearth of further opportunities to use those kinds of positions to mismatch just a little bit and get some yield. We have seen the net interest margin in kind of a secular decline. I think actually the worst interest rate environment for banks is a sustained, very low interest rate environment, a bullish flattener kind of environment, where banks just cannot make any money on the free funds or the interest rate spread.

It is almost a relief to see rates start to go up a little bit. A steeper yield curve will probably be better for that profitability opportunity, as long as the mismatch is not outsized.

MR. JOHNSON: I agree with Tanya. I think the carry trade right now is a difficult thing to execute with the yield curve so flat. One of the outcomes that is least favorable to our bank would be either a continued flattening from where rates are or for rates to remain at these current levels. I think the general expectation of the banking industry was to have a little bit stronger increase in rates as we entered 2004, but the yield curve remained fairly flat last year.

I do think there are things banks can do, and one of the things we are doing right now is looking at the scenario where interest rates remain at these levels and we do not get a strong increase in rates. We are looking at some of the things we can do to help to mitigate the potential impacts on our balance sheet.

One strategy, for instance, is to take a mortgage product, put it on your balance sheet, and not hedge it out for the first year, then use a forward-starting swap to hedge it beyond the first year. You get somewhat of a carry trade for the first 12 months, and you have protected yourself from a flat or down interest rate environment. If rates move up, the hedge protects you after the first year. And if rates rise and you lose some margin on the position, for an asset-sensitive institution there are a number of other things on the balance sheet that will significantly improve the performance of the bank as rates rise.

We are looking at a number of strategies like this as the kinds of opportunities available to us, depending on where we see rates going over the course of the next six months.

MR. STARK: The carry trade really is making money off interest rate risk. Generally, it is mismatching long assets and short liabilities with some kind of overnight repo or something with a longer-dated asset and playing the positive yield curve.

We understand that the business of banking is managing risk at a profit, and this is a form of risk being managed many times by institutions. Hal described a strategy where banks can manage the risk, building a wider spread in the first year and then backing it up after a year with some form of hedge. As regulators, we have no real problems with that. But we need to ensure that everybody in the bank's management and board of directors understands the risks that are being taken by the bank, that the bank has sufficient capital, and that it has sufficient management resources to keep track of the bank's risk positions over time. Interest rate risk is dynamic and can turn around on you.

MR. BROWN: Let's talk about mortgage lenders. With 80 percent of the mortgage loans having been written since 2002, and with a similar percentage of those loans being fixed-rate mortgages, we are clearly in uncharted waters. Are mortgage lenders significantly more at risk for rising rates now than they were five years ago?

MS. AZARCHS: We do not really think so, although there probably are more mortgage assets as a proportion of total assets on banks' books. But we are told the nature of that risk is not all that different than five years ago. We are told that banks have generally kept the adjustable-rate product on their balance sheet and securitized the fixed-rate product. So the 30-year fixed-rate product is, broadly speaking, not on the balance sheet. [See Chart 2.]

However, the adjustable-rate products are not quite as adjustable as they used to be, with some of the new hybrid products that are around—the three-year, five-year, and seven-year fixed-rate loans that then go to adjustable rates. These hybrids should be fairly easy to hedge, so the real issue goes back to how well they are hedged.

MR. JOHNSON: I do not think that this period is any more risky, and I think the interest rate risk models that are out there do a pretty good job. Certainly, from the ability to hedge there is a full complement of products in the marketplace to use.

I think the risk that exists today is in the potential for a significant disconnect in the prepayment expectations versus the prepayment realities over the next five to seven years. If the hedges and the underlying assets do not end up matching up over that period of time, banks could end up with a portion of their income with either a negative carry or generating a greater benefit. To me, that is the risk.

MR. STARK: I agree. I do not think there is any more risk. There are some micro issues like how to model the newer hybrid products. I imagine getting the models to fully capture the behavioral characteristics of novel, hybrid products to reflect accurately the way they truly behave is a challenge at times for bankers.

MR. BROWN: I would like to move on to deposit pricing, where in the third quarter of 2004 we saw the initial effects of the Federal Reserve raising short-term interest rates. For large banks—with total assets over $10 billion—we saw their cost of funds go up 15 basis points, but for the smallest groups of banks—with assets under $100 million—their cost of funds only went up 2 basis points.

What do you see going forward in terms of the stability of core deposits if the short-term rates continue to increase as anticipated?

MS. AZARCHS: We think that the volume of deposits will remain with the banks, because they have been relatively sticky and have not changed very much in a correlated way with interest rates. The real issue is the pricing on some of the discretionary deposits and deposits in general. I think what we are seeing now is a positive stickiness in pricing in a rising rate environment, where the deposit rates have been a little bit slow to move up.

From the reports we are getting, I think the effect of the interest rate increase has already run its course. They may move up more in lock-step, but the value of the free funds is the thing we are looking to see turn positive as rates continue to move up.

William Stark (left) and Hal Johnson.

MR. JOHNSON: I think in general core deposits ought to actually be better off with the rates moving up. I think it has been challenging for people to find a home for their money that really earned them much.

We do think that demand deposit account balances will decline, but I am actually surprised that there has not been more of a runoff to this point. So far, our demand accounts have been pretty sticky, but our expectation would be that as rates begin to move up and alternative uses for that cash among our commercial clients improve, we will see some of that money leave the bank.

MR. STARK: I find it interesting that banks in the Southeast part of the country appear to be able to lag by maybe 25 percent the Federal Reserve's short-term rate increases. But in the Midwest there is a little more competition for the money, and banks are having to raise their rates. It is interesting to see how rate increases can affect the regions differently. Also, the big banks have a much larger wholesale funding component, so their cost of funding could be going up more dramatically than the community banks and those banks that rely on core deposits.

MR. BROWN: I would like to move on to hedge accounting. I know that there may be some issues out there in terms of the accounting standards applied to hedge positions and whether or not these standards help or hurt the ability to undertake a macro hedging strategy. Is there a sense in which the accounting rules make it difficult for institutions to optimally hedge their interest rate risks?

MS. AZARCHS: Well, I do think the accounting rules make hedging more difficult. There is a sort of a double responsibility that every bank treasurer has—to think about what the real economic hedge is and what the accounting impact of it could be. Some of the more efficient hedging strategies no longer make it in terms of qualifying for hedge effectiveness under FAS 133, as well as the issue of the mortgage servicing rights.

I do think it puts a spoke in the wheel in many cases, and maybe makes banks prefer the less efficient way of hedging, which would be the cash instrument rather than the synthetic derivative expression of the same. Maybe banks do less hedging because of the fear of the accounting. Right now I think the largest risk is the accounting risk, and the fear that it will be misperceived if there is a change in accounting method from hedge accounting to mark-to-market accounting and an attendant loss. That has a very negative effect on shareholders, on the stock market price, and things like that where people just see accounting risk and head for the exits.

MR. JOHNSON: Accounting rules clearly do make it difficult for institutions to optimally hedge their risks, and in a significant way. If you look at accounting for things like the MSR asset, it is totally devoid from the actual cash flow of that asset. Additionally, when you have assets like the MSR asset and commercial mortgage-backed security pipelines, where the hedge is marked to market but the asset remains as the lower of cost or market, it makes the hedging extraordinarily difficult, because you no longer have an economic offset at the end of your accounting period.

So I think the accounting rules do have an impact, and I think they are hurting our ability to effectively hedge and take out the kinds of risk we want to, because we have to be sensitive to the potential accounting outcomes.

MR. STARK: I am not the accountant in the crew here, but I agree with what I hear from Tanya and Hal. It is disturbing when accounting rules and economic reality differ, and it is generally going to lead to a bad result. From a safety and soundness viewpoint, we have an interest in avoiding a bad result, so these are areas that we monitor carefully.

MR. BROWN: Tanya, as an analyst, do these developments make it harder to evaluate the institution?

MS. AZARCHS: It does cause some loss of transparency. Hedge accounting makes it difficult to track what is happening from a cash flow sense, a liquidity sense, or an operational sense. Technically, when we get to full mark-to-market accounting, the income statement will only be the residual reconciliation between the balance sheet at Time A and the balance sheet at Time B. However, when evaluating the soundness of the bank and the health of its franchise, it is extremely important for us to see what is happening operationally, and that is getting more and more difficult to do.

Securitization rules are another area of distortion from what is really happening operationally. All we can see is the change in the mark-to-market of the residuals, but we do not know what is happening to the pool of assets that underlies that.

MR. BROWN: I want to move on to our bottom line question here, and, for the FDIC, many times the bottom line relates to scenario analysis. There are two interest rate scenarios I would like to consider.

First, let's consider a baseline scenario. Federal funds futures are implying that short-term interest rates will increase another 100 basis points by August and maybe another 50 basis points by the end of the year. So we are anticipating steady, moderate increases in short-term interest rates. What effect is that likely to have on the industry and individual institutions with regard to interest margins, securities gains, and credit quality? Are there any bad outcomes that we should be worried about?

MS. AZARCHS: For Standard & Poor's, one of the first things we think about is what economic scenario goes along with that. Assuming that we are getting a strong economy with good growth in discretionary income, the credit quality problems that could come from rising interest rates may be dampened.

The greatest area of concern is the amount of consumer debt in the high-growth areas of home equity lending and other forms of unsecured or even secured consumer debt. In a high-rate environment, what is going to happen to that from a credit quality point of view? You need to have a really strong economy to offset any potential credit fallout in these areas.

In terms of net interest margin, I think interest rate margins would be slow to recover to their historic norms under what we might call a bearish flattener. If the Federal Reserve raises rates but the long end stubbornly refuses to go up, there is going to be continued pressure on the margin. But I think there would be even more pressure if rates stayed very low and flat, a bullish flattener scenario.

MR. JOHNSON: If we see a continued flattening yield curve, I agree that you have to look at what is happening to the economic scenario. If we do not see more robust loan demand and some relief on asset pricing, that is somewhat of a bearish scenario for 2005 from a bank earnings perspective.

A good outcome would be a steepening yield curve, a little bit more robust economy, more loan demand, and a little less pressure on loan pricing. That is our preferred economic scenario.

MR. STARK: In the supervisory business, we focus on banks' risk management processes given that bank's risk profile, and we try to steer clear of second-guessing bank management's interest rate forecasts. We are concerned primarily about any extreme position in a bank where there is not sufficient capital. We do not differentiate whether it is an extreme long or an extreme short position. If the bank does not have sufficient capital and rates go the wrong way, the bank can get into trouble and may not be acting in a safe and sound manner. That is what we try to evaluate at all times.

MR. BROWN: Let's talk about an extreme case here—the worst case scenario. The United States is running a current account deficit in the neighborhood of $500 billion to $600 billion, and yet we have long-term yields around 4.15 percent today, which is pretty low.

A lot of the slack is being picked up by the purchases of U.S. Treasury and agency securities by foreign central banks, especially in Asia. That represents an official policy that might change at some point if those central banks decided not to accumulate more dollar assets. If that were the case, there is the potential that we could see the dollar decline, and we could see long-term interest rates shoot up pretty dramatically. If long-term interest rates shoot up to 6 or 7 percent, or even higher in some very adverse scenario, what effect would that have on the banking industry? Which institutions are going to be affected the worst, and how?

MS. AZARCHS: When we look at stress tests that banks do in their asset-liability management, we see those types of scenarios being tested to some extent. I think higher rates are survivable as long as the economy does not decline and the asset liability managers hedge appropriately.

What we are not necessarily seeing tested, and I wonder why not, is a stagflation scenario. In a very stagnant, high-unemployment kind of economy, the credit risks in the consumer sector would be extremely large. That is when I think we would see the consumer sector decline.

MR. JOHNSON: I agree. I think it becomes a potential credit issue at that point, and I think the issue of inflation and stagflation is one that we have to keep a very close eye on. If we see those two things beginning to show up on the horizon, we need to be able to react, because neither of them is particularly attractive for banks.

MR. BROWN: I want to thank you all for your willingness to talk about these issues and share your perspectives with us. It has been very interesting and helpful to us here at the FDIC.

Lynne Montgomery provided editorial assistance for this article.

Photographs are by Sally Kearney and Mary Ledwin Bean.