Thank you, Mr. Chairman. As usual, we have received a mixed bag of data since the January FOMC meeting. Over most of the past six months, when we said “mixed” data, we generally meant data that were either stronger than we had expected or that were a lot stronger than we had expected. By contrast, during this intermeeting period, we received figures on both sides of our January projection. As you know, on net, we read the incoming information on labor markets, production, and spending as suggesting that there was somewhat less upward momentum to activity than we had previously thought. Our largest disappointment was centered in the labor market. Private nonfarm payrolls were flat in February and have been increasing at a pace of only about 65,000 per month since the upturn in hiring began last September. That is well below the pace we had envisioned several months ago. To be sure, initial claims for unemployment insurance since the turn of the year have fluctuated in a range that in the past was associated with stronger employment gains than we have experienced to date. But those data likely reflect a decline in layoffs, and we have scant evidence, as yet, that new hiring has picked up materially. There have been a few hopeful signs in the recent data: Temporary-help employment has been trending higher since last spring, and the average workweek has moved up since the turn of the year. But those looked to us to be thin reeds on which to rest our earlier projection. As a consequence, we marked down considerably our projected improvement in employment over the first half of the year, and along with that lower employment path, we reduced the projected growth of hours worked.
If the pattern of the past couple of years had been repeated, this would have been yet another occasion for a weak labor market report to result in a further upward revision to productivity. But this time, somewhat softer readings on manufacturing production and on spending persuaded us to revise down our projection of the growth of real GDP by about as much as we revised hours, leaving our productivity projection about unchanged. Among the slightly softer indicators that influenced our thinking was manufacturing IP. Despite an increase of 0.8 percent in February, smaller gains in December and January took down our projected increase in factory output in the current quarter from 7 percent at an annual rate to 6 percent. Obviously, this is a revision best viewed as one of quantitative degree rather than one of qualitative substance. The anecdotal reports still seem to signal steady improvement in the manufacturing sector. We just don’t see quite the degree of vigor that earlier indicators had suggested.
We also have had, on net, some negative surprises in the spending data. Perhaps consistent with the slightly softer readings on IP, both exports and inventory investment have come in below our expectations. The weakness in inventory investment is somewhat surprising to us in light of the low carrying costs of inventories. We have heard reports that some firms have encountered difficulties acquiring needed materials and some discontent on the part of businesses with breakdowns in their just-in-time inventory arrangements. These difficulties may have held down inventory investment. But we don’t believe this is a phenomenon with sufficient breadth or magnitude to explain the meager additions that firms have made to their inventories in recent months. Instead, we think the subdued pace of stockbuilding likely reflects the same continued cautious behavior on the part of businesses that is evident in their hiring decisions.
The single largest source of downward revision to our GDP projection has been construction activity, where the recent data have come in below expectations for the residential, nonresidential, and state and local sectors. To be sure, we had anticipated some falloff in residential construction from the torrid pace set late last year. But the drop-off in starts in January, especially in the single-family sector, was larger than we had penciled in. Moreover, sales of new and existing homes also fell significantly, and as a consequence, the commissions on those sales, which are part of GDP, amplified the slowdown of spending in this sector. Homebuilding figures for February were released this morning, and those data show another drop in single- family starts—from an upward-revised figure of 1.55 million units in January to 1.49 million units in February. Nevertheless, permits have remained quite firm, and they point to some bounceback in construction in coming months. All told, these data might take a tenth off of our current-quarter GDP projection. Nonresidential construction activity also has come in below expectations. Construction of office buildings, commercial structures, and manufacturing facilities continues to contract. Although there are faint signs of improvement, vacancy rates remain high and rents soft. Meanwhile, although budget conditions are improving among state governments, funds are still very tight, and these entities have further trimmed their construction spending.
The only other notable area of greater-than-expected softness has been spending on light motor vehicles. Sales averaged about 16¼ million units in January and February, about ½ million units below our previous projection. At the same time, production of light vehicles has come in above our expectations. As a consequence, inventories appear a bit high to us. In the projection, we resolve that tension through a combination of production restraint and a pickup in sales that is induced in part by a sweetening of incentives.
It is important to note that not all of the news over the past seven weeks was negative. Excluding motor vehicles, consumer spending was actually a bit stronger than we had expected. And business outlays for equipment and software were noticeably stronger in both the fourth and first quarters than we had projected in January. High-tech spending has continued to increase rapidly, while spending on capital equipment outside of high-tech and transportation has posted a marked acceleration in recent months. As you know, equipment spending had appeared to us to have been unusually weak over the past year or so given our view of the key fundamentals—a low cost of capital, accelerating final sales, and ample increases in cash flow. That gap is now closing, and along with the continued boost from the partial-expensing provision, we are projecting that real E&S will maintain double- digit gains through the end of the year.
Taken together, the incoming data led us to revise down the projected growth of real GDP by a little more than ½ percentage point at an annual rate in the first half of this year, after a bit larger revision in the fourth quarter of last year. All else being equal, multiplier effects would have argued for a lower growth rate in the second half of this year as well. However, the substantial rally that has occurred in the bond markets in recent weeks led us to mark down our projected path for long-term interest rates, and those lower rates are expected to provide a considerable cushion to spending going forward. Still, we do have on net a lower level of real GDP, and hence a higher output gap, throughout the projection period compared with our previous forecast. Whereas in January, the output gap was nearly eliminated by late next year, in the present forecast that gap remains at nearly ¾ percent of potential GDP at the end of 2005.
With more slack in resource utilization, you might have expected to see somewhat lower inflation in this projection. That effect, however, was offset by several countervailing considerations. First, the incoming data on core inflation were a bit above our expectations in recent months. Moreover, oil prices have moved up sharply and are expected to stay higher for longer than in our January forecast. And finally, the prices of non-oil imports have increased more rapidly of late than we had anticipated earlier. The bottom line is that we still expect core PCE prices to increase about 1 percent this year and next.
Stepping back from the details of the forecast, we continue to believe that the data and the anecdotes are consistent with an economy that is expanding fast enough to take up slack in the economy’s utilization of resources. We expect fiscal stimulus, accommodative monetary policy, and ongoing rapid gains in structural productivity to sustain a brisk expansion through the end of next year. Coming as it has on the heels of an extended period of mostly upside surprises, the events of the past seven weeks, including last week’s bombing in Spain, have clearly restored a sense of two-way risk to the outlook. Just how large are those risks? Well, that’s a question that we have been working to answer more precisely for you.
As you know, we included confidence intervals around some key variables in our forecast as a standard feature for the first time in this Greenbook. Over the past few years, we have shifted somewhat the emphasis of our forecast presentation away from a single narrative focused on explaining the baseline projection toward inclusion of more material highlighting some of the chief risks surrounding the economic outlook. The presentation of confidence intervals for the projection seemed to us a logical extension of that effort. We have calculated those confidence intervals both using the real-time track record of the staff forecast and using stochastic simulations of our large-scale econometric model. We thought several purposes could be served by including these measures in the Greenbook.
One very important purpose is to illustrate how much uncertainty surrounds the staff baseline projection. For this purpose, the confidence intervals calculated using the actual forecast record of the staff projection would seem the most informative measures. Those intervals are, indeed, wide. The 70 percent confidence interval around our projection of a 5¼ percent unemployment rate in the fourth quarter of 2005 runs from 4¼ percent to 6 percent. Similarly, the 70 percent confidence interval around our forecast of a 1 percent increase in core PCE prices in 2005 runs from about zero to 2 percent. And it is important to remember that this implies that there is a 30 percent chance that the actual outcome will lie outside these bounds. The width of these bands does provide me with one small consolation. When asked in the past how confident I have been about various aspects of our projection, I believe the record will show that I have almost consistently responded “not very.” It looks as though I’ll be pretty safe sticking to that answer. [Laughter]
We also included in the table and as fan charts confidence intervals generated by stochastic simulation of FRB/US. The reason that we have added this second approach to the mix is that it allows us to undertake calculations that cannot be performed with the Greenbook forecast errors. For example, we can decompose sources of forecast uncertainty in the way that we did recently in a pre-FOMC briefing for the Board, breaking out inflation uncertainty into uncertainty about supply, demand, and the unexplained equation residual. This approach also provides us with an ability to compute certain conditional probabilities—for example, probabilities of deflation under different monetary policies. We have provided some of this material to the Committee in the recent past, and we anticipate doing more of the same in the future.
We also will continue to plot in the fan charts the alternative scenarios that we present in the Greenbook. Like the confidence intervals displayed in those charts, the alternative simulations are generated using our large-scale model. A number of you have remarked on occasion that, after reading the descriptions of the alternative scenarios, you were surprised to see that all we got on GDP growth or the unemployment rate were a few lousy tenths and, in most scenarios, considerably less than that on the inflation rate. I have noted in response that a drawback to the alternative scenarios is that they do not account for all of the covariation in the shocks to which the economy is subjected. The confidence intervals generated by stochastic simulations do just that. By placing the alternative scenarios in the fan charts, we hope to provide you with a better perspective on just how much of the probability space we are exploring with those alternatives.
Finally, the confidence intervals may provide a context for the Committee to consider the uncertainty surrounding the projections that you include in the Monetary Policy Report to the Congress. As you know, those forecasts are presented as ranges and central tendencies. But, of course, ranges and central tendencies are measures of forecast consensus, not measures of forecast uncertainty. Based on our past forecast errors, the 70 percent confidence interval surrounding our forecast of the unemployment rate in the fourth quarter of this year is about 1¼ percentage points wide; in contrast, the width of the FOMC’s forecast range of the unemployment rate in the fourth quarter of this year was just ¼ percentage point wide. So, while there is considerable consensus on the Committee about the expected unemployment rate late this year, there is probably much greater uncertainty associated with each of your individual forecasts.
Needless to say, advertising our ignorance in such an explicit fashion gave us some pause. But given the increasing prevalence of product liability lawsuits, we thought some appropriate warning on the Greenbook forecast was warranted. Perhaps the confidence intervals should be viewed as something like a warning label that reads, “Do not operate large economies while under the influence of a baseline projection alone.” Karen will now continue our presentation.