Thank you, Mr. Chairman. I’ll be referring to the charts that were just circulated. Fixed-income markets in the intermeeting period were characterized by a high degree of volatility, as market participants readjusted their expectations for the future path of monetary policy in response to the Committee’s June 25 statement, comments from various Committee members emphasizing near-term growth prospects, the generally positive economic data, and a relatively strong corporate earnings season.
The top panel on the first page graphs the U.S. three-month deposit rate in black and the three-month deposit rates three, six, and nine months forward in the dashed red lines. In the weeks before the June meeting, rates were trending down, as further easing by the Committee was anticipated in light of the various comments by Fed officials about the risk of deflation and the need to take aggressive measures to counteract it. The 25 basis point easing move at the June meeting—less than the 50 basis point cut many had expected—combined with language that suggested a better outlook for the economy, led markets to re-price short-term interest products. Many concluded that the easing cycle was perhaps finished and that there was only a small probability that the Fed would take so-called unconventional monetary policy actions. That move toward higher expected interest rates accelerated on July 31 with the release of the second-quarter GDP data, which came in higher than expected, and the continuing strength in some of the manufacturing and service sector surveys.
Treasury yields had declined sharply in May and into early June, based on the same set of expectations. Market participants expressed their views by putting on carry trades along the Treasury yield curve as well as in other sectors such as swaps, agencies, and mortgages. As these expectations were revised, the liquidation of positions—aided by some very strong technical factors that I’ll get to in a second— accelerated, and yields rose along the curve. The two-year note traded to an intraday high of 1.93 percent on August 1, creating an unusually wide spread between its yield and the funds rate, especially taking into account the various statements by Committee members that the funds rate was likely to stay low for a long time. The two-year note yield subsequently eased back a little to 1¾ percent. Similarly, the yield on the ten-year note moved to an intraday high of more than 4½ percent from about 3.1 percent or so. The bottom chart graphs the “breakeven” rate on ten-year TIPS since January 2000, calculated by subtracting the TIPS yield from the nominal yield. The breakeven rate has traversed a well-worn range of between 125 and 225 basis points. I would make two points. First, despite the market’s apparent fixation on deflation during the last few weeks and months, the breakeven rate drifted down only slightly and did not get far below 175 basis points before reversing recently. Second, the recent rise in nominal rates has been mostly an increase in real rates, which could be seen as consistent with expectations for an improved macroeconomic environment.
Still, the question remains: What were the factors that led to the sharp selloff in fixed-income markets? While incorrect expectations for policy played a role, at least as large a role was probably played by the dynamics of the mortgage-backed securities market. Unlike earlier cycles, the mortgage market is now larger than the Treasury market. Anecdotal reports indicate that a large percentage of MBS holdings are actively hedged by fund managers, all of whom are using fairly similar models.
The top left panel on page 2 graphs the distribution of MBS coupons as of December 31, 2002. The largest concentration of coupons was at 6½ percent. The shaded area to the right represents a very rough estimate of the portion of the distribution of mortgages that were candidates for refinancing based on December’s average mortgage rate of 6.05 percent. In short, the shaded area represents the portion in which the homeowner’s option to prepay is in the money. These estimates are crude and require the usual health warnings, but the trends are interesting and revealing nonetheless. The top right panel depicts the comparable distribution as of May 31, 2003. By then the thirty-year fixed-rate mortgage had come down to about 5½ percent and was still falling as we were heading into June. Nearly the entire distribution was in the money. However, with the backup in yields, mortgage rates rose even faster than other rates. By late July, the thirty-year fixed rate was reported by Freddie Mac in its survey to be at 6.14 percent. The most recent distribution of the coupons, which is as of June 30, gives a picture of the degree to which mortgages went from being in the money to being out of the money. Adding up the three bars in the unshaded area of the middle left panel indicates that a total of about $1 trillion of mortgages was suddenly out of the money.
Now, as rates were falling, fund managers were buying Treasuries as fast as possible to hedge against a shortening of duration caused by prepayments. Once rates began to back up, the risk was an extension of the portfolio. At that point the model suggested selling Treasuries to keep duration from increasing. Since the structural position of the mortgage market is well known, some speculators were no doubt front-running those flows and adding to the liquidity demands. The middle right panel graphs an estimate of the duration of the universe of fixed-rate mortgage- backed securities. In a span of about five weeks, that duration increased from about one-half year to more than three years. The increase was the largest ever seen over such a short period of time. The bottom panel on page 2 attempts to quantify the hedging needs generated by the backup in yields. Each bar represents the estimated purchase or sale of ten-year Treasury equivalents for each half-month period since the beginning of the year. The red line is the ten-year Treasury yield. As with most estimates, these figures are to be treated gingerly. The calculations assume that the entire MBS universe is actively hedged. That is not the case. But even if only half is managed in this way, the numbers are still very large. On the other hand, this graph does not include whole loans held by banks, thrifts, and the GSEs, some of which we know are managed actively. Nor does it include the activities of the mortgage servicers.
As the hedging in the mortgage-backed market reached something of a climax in late July, the activities of MBS investors increasingly pressured the swap market, which is the other major hedging vehicle for mortgage investors. The top panel of page 3 graphs the ten-year swap spread, which jumped more than 30 basis points in late July. Scattered reports of spotty liquidity circulated during this period. But for dealers and some investors the important point is not necessarily the level of the swap spread itself but the rate of change, since risk-management models make assumptions about the speed with which positions can be hedged.
The middle panel shows the standard deviation of the twenty-day change of the swap spread going back to 1993. Typically this measure of volatility was very small, less than 1 basis point, until 1998 when it suddenly spiked to about 4 basis points in the aftermath of the Russian default. In recent years, swap spreads have become somewhat less volatile as the spread narrowed. In July the volatility shot up again to a level comparable to or higher than that seen in 1998 and mid-2000. Occasional rumors circulated again about some bad positions and some losses in the dealer community, and liquidity was reported to have decreased as demand for fixed payer swaps rose. Nevertheless, the markets generally worked well. And as shown in the top panel, the spread itself has dropped more than 20 basis points from its peak. Mortgage and agency spreads also widened suddenly in late June and July. That is not surprising given the shifts in the mortgage market and, in the case of GSEs, the entity-specific news that buffeted them. On the other hand, the corporate market and, for that matter, the equity markets were not noticeably affected. The bottom panel graphs the investment-grade corporate spread year-to-date. Although it has ticked up in recent days, it moved slightly lower over the intermeeting period and is substantially down from January 1. The high-yield and EMBI spreads in the bottom right panel show broadly similar—and benign—patterns. U.S. equity markets on balance are slightly higher and traded in a narrow range over the period.
Turning to page 4, the reversal in our bond markets is not isolated. This page graphs the ten-year yield since April 15 for the United States, Germany, Canada, Sweden, Australia, Japan, and Mexico. All of these sovereign yields reached their lows in mid-June and then rose together. While part of the explanation may be that the economies are more interrelated, the more convincing explanation may be that bond markets are more interrelated and are especially sensitive to U.S. yields. I have just two minor points to add. First, the ten-year yield is now higher than that of the German Bund for the first time since May 2002. Second, the more than doubling of the ten-year JGB yield—from about 43 basis points to about 110 basis points at the peak—created much discussion about the losses on the books of Japanese banks. But it was little more than discussion, in part because the rally in the NIKKEI, it was argued, offset some or all of those losses. More recently, ten-year yields in Japan have retreated to about 89 basis points.
The top panel on the next page graphs the euro-denominated three-month cash deposit rate and the rate three, six, and nine months forward since mid-April. With weak sentiment and downbeat data, the market expected that the ECB would ease policy, which it did on June 6, cutting rates by 25 basis points. That is not marked on this graph, I should point out. The forward rates suggest that the market expected further easing, though less, through much of July. However, as the global bond market selloff accelerated, euro area forward rates trended higher and spiked on July 31 after the U.S. GDP report. The nine-month forward rate has eased off a bit since. The dollar has risen modestly against most currencies since June, consistent with the higher yields and the more upbeat view of the economic outlook. The dollar–yen exchange rate, shown in the middle right panel, continued to trade in the 117 to 120 range that it has been trapped in all year. The Ministry of Finance (MOF) continued to intervene, buying $19 billion in the period and raising its total intervention for the year to $75½ billion. In general, the yen has been depreciating or stable against most currencies. The bottom panel graphs the Australian dollar, the Canadian dollar, the euro, the pound sterling, and the U.S. dollar against the yen since January 1. The first three currencies have risen against the yen while sterling and the U.S. dollar have moved little. The U.S. dollar–yen relationship looks rather like a straight line since January 1 but perhaps mostly because of the frequent intervention by the MOF. The days Japan intervened in dollars are marked with small boxes on the red line.
Turning to page 6, let me say a word on reserves. The Desk has faced an unexpected decline in the growth of currency, as shown in the top panel on the page. That slowdown is due mostly to reflows from overseas for reasons that are not fully clear. Regardless of the reasons, the slowdown and the resulting overall pattern in autonomous factors have led us to shrink the System Open Market Account (SOMA) slightly by reducing our holdings of long-term RPs and essentially ceasing outright purchases for the time being. This is the second instance in the past year when we have had to contend with an unexpected need to temporarily contract SOMA. We have used the flexibility inherent in the long-term RP book to do that. But we have found that a book of RPs with a maturity of twenty-eight days at times does not give us enough flexibility to adjust the balance sheet on very short notice. So in order to give us more flexibility, I plan to shift the long-term RP book from a typical maturity of twenty-eight days to fourteen days at our regular Thursday operations. The change is purely a technical measure and does not require a change in the authorization, but I did want to advise the Committee before we implement the change. We plan to announce this modification to the primary dealers later this month and manage the shift in late September. Again, the main benefit of this change would be to provide the Desk with greater flexibility in addressing the demand for banking system reserves over the two-week length of the maintenance period. We will, however, continue to arrange operations with an original maturity of up to a maximum of sixty- five business days when reserve needs are expected to persist for relatively longer periods of time. That is often the case around the year-end, for example, when the total volume of temporary operations typically peaks in response to seasonal growth in currency.
Mr. Chairman, there were no foreign exchange operations. I will need approval of the domestic operations. If possible, after I take any questions on my regular report, I would like to make a few comments about Ginnie Maes.