Transcripts of the monetary policymaking body of the Federal Reserve from 2002–2008.

The top panel of exhibit 5 summarizes our assumptions about the supply side of the economy. As indicated in line 1, we assume that potential output growth will edge down over the forecast period, from 2.7 percent in 2006 to 2.5 percent in 2008. This slowing primarily reflects our assumptions about trend hours growth (line 2), which steps down from about ¾ percent last year to ½ percent in 2008 because of a steepening downward trend in the labor force participation rate and a gradual slowing of population growth. Although we are comfortable with these assumptions—and, indeed, have not made any changes to them in this projection— we do see risks on both sides of our point estimates. For example, as shown in the middle left panel, many outside forecasters are basing their projections on a significantly higher estimate of potential output—in some cases above 3 percent per year. We suspect that these differences, at least in part, reflect different views about the underlying trend in the labor force participation rate.

The participation rate and our estimate of its trend are shown in the middle right panel. As indicated by the black line, labor force participation has risen about ½ percentage point since its trough in early 2005. Some forecasters appear to have taken this increase as a signal of faster labor force growth going forward. However, we see it as largely a cyclical response to steady employment growth and a tighter labor market, and we expect it to be reversed in the near future as the pace of hiring slows and the underlying demographic forces show through. In part, our view reflects the fact that the participation rate tends to rise above its trend when the unemployment rate is low. Periods in which the unemployment rate was below the staff’s estimate of the NAIRU are denoted by the yellow shaded areas in the chart, and the current gap between the actual participation rate and our estimate of the trend does not appear to be outsized relative to historical norms. In addition, as shown in the bottom left panel, the increase in the participation rate over the past couple of years has been fueled by a rise in the percentage of individuals who moved directly from out of the labor force to a job; this flow also exhibits noticeable sensitivity to labor market tightness. That said, some groups have behaved differently than our models would have predicted. On the one hand, the participation rate of older individuals, shown by the red line in the bottom right panel, has risen steadily for some time, presenting an upside risk to our forecast. On the other hand, participation among teenagers (the black line) has remained surprisingly low, and there are undoubtedly downside risks to our forecast of an upturn for this segment of the population.

Exhibit 6 describes another source of tension in the recent data that may have implications for our estimate of potential output. As shown in the top left panel, our standard Okun’s law simulation (the red line) suggests that the unemployment rate (the black line) fell more last year than would have been expected given our current estimate of real GDP growth in 2006. In the baseline forecast, we assume that an increase in the unemployment rate causes that gap to disappear gradually, an assumption that does not seem unreasonable given that the error in Okun’s law at the end of last year was within the bounds of historical experience. However, other interpretations are possible as well. One possibility is that current estimates of real GDP understate economic growth last year. One piece of evidence in support of this hypothesis is shown in the top right panel. We currently estimate that real gross domestic income rose 4 percent in 2006, about ¾ percentage point more than real GDP. As shown by the green line in the top left panel, if we replace real GDP growth with our estimate of real GDI growth over the past year and re-run the Okun’s law simulation, the actual unemployment rate in the fourth quarter lines up very closely with its simulated value.

An alternative interpretation of the recent error in Okun’s law is that potential output growth was weaker last year than we have assumed—perhaps because of a downshift in structural productivity growth. The middle and bottom panels address this possibility. As shown by the difference between actual productivity growth (the black line in the middle left panel) and a simulation from our standard model (the red line), labor productivity decelerated much more last year than the model would have expected. As shown in the panel to the right, a purely statistical model based on a Kalman filter would have responded to the incoming data since March of last year by cutting its estimate of structural productivity growth by a full percentage point. In contrast, because we place less weight on the recent data that have not yet been through an annual revision, we have reduced our own estimate by only 0.6 percentage point. The bottom panels provide a couple of reasons for our reluctance to lower our estimate of structural productivity growth as much as the statistical model would have lowered it. First, as shown on the left, labor productivity in the nonfinancial corporate sector was quite strong last year. In part, the better performance of productivity in this sector reflects the fact that its output is measured from the income side of the accounts and thus incorporates the difference between GDI and GDP noted above. In addition, this component omits some sectors that are notoriously difficult to measure. Second, as shown on the right, a measure of productivity that excludes the residential construction industry also held up fairly well last year, suggesting that much of the deceleration in nonfarm business productivity may be cyclical. As shown in the middle panel, all told we expect actual labor productivity growth to step back up to an annual rate of about 2½ percent by the middle of this year as businesses reduce the pace of hiring in lagged response to the slower rate of output growth in recent quarters.

The implications of this forecast for the labor market are shown in the top panels of exhibit 7. In particular, gains in nonfarm payroll employment—shown by the black line in the top left panel—are projected to slow to about 60,000 per month by the second half of this year. This pace is somewhat below our estimate of trend employment growth—the red line. As a result, the unemployment rate—shown in the top right panel—drifts up to just under 5 percent, our estimate of the current level of the NAIRU. To help gauge whether the estimated gap between the unemployment rate and the NAIRU is sending an appropriate signal about the degree of tightness in the labor market, I have included some other measures of slack in the remaining panels of this exhibit. As shown in the middle left panel, the job openings rate from the BLS’s JOLTS (Job Openings and Labor Turnover Survey) rose over the second half of last year to its highest level since early 2001. Because the job openings rate has such a short history, its equilibrium level is difficult to estimate. However, we can learn something by combining the openings rate and the unemployment rate to form the Beveridge curve shown to the right. The curve is estimated using data from the first quarter of 2001 through the fourth quarter of 2006, with the openings rate on the vertical axis and the unemployment rate on the horizontal axis. The relationship between job openings and unemployment appears to have been fairly stable in recent years, which suggests that the NAIRU has not changed materially over that period. Moreover, the latest data point is in the far upper left portion of the graph—the segment of the curve indicative of a tight labor market. Two other margins of slack are shown in the lower two panels. The bottom left shows the percentage of employed persons working part time because of slack work at their firm or because they couldn’t find a full-time job. This measure has moved down over the past couple of years and is currently below its average level in the second half of 1996— an earlier period when we thought that labor markets were roughly in equilibrium. Also, as shown to the right, the capacity utilization rate in manufacturing remains a little above its long-run average level.

Exhibit 8 presents the inflation outlook. Despite our view that labor and product markets are tight, other influences on our inflation projection have been more favorable than we were expecting at the time of the last Greenbook. Perhaps most notably, the recent data on core consumer prices—shown in the top left panel—have been lower than expected. Core PCE prices were about unchanged in November and, based on the latest CPI reading, we expect an increase of only 0.2 percent in December. As a result, as shown in the second column of the table, we have marked down our estimate of core PCE inflation in the fourth quarter by ½ percentage point, to an annual rate of 2.1 percent. As shown in the top right panel, the lower path of oil prices led us to revise down our projection of consumer energy prices. These lower prices directly pull down our forecast for total PCE prices; they also imply somewhat smaller indirect effects from energy costs on core prices over the forecast period. As shown in the middle left panel, a higher exchange value of the dollar in this forecast led us to reduce the projected path of core nonfuel import prices. The combination of these various influences led us to shave our projection for core PCE prices—line 4 of the middle right panel—by 0.1 percentage point in both 2007 and 2008, to 2.2 percent and 2.0 percent respectively. As before, the slight downward trajectory to core inflation reflects our projections of waning indirect effects of the earlier increases in energy and other commodity prices, declining relative import prices, and a deceleration in shelter costs.

In light of my earlier discussion of the risks to our assumptions about potential output growth, the bottom panels present one alternative simulation from the Greenbook, in which we assume both a higher trend for the labor force participation rate and slower growth in structural productivity. Specifically, we calibrated the simulation so that overall potential output growth was essentially the same as in the baseline forecast, but with structural productivity growth ½ percentage point weaker and trend hours growth ½ percentage point stronger. As shown in the bottom left panel, this change to the composition of potential output growth has little effect on aggregate activity and the unemployment rate. However, the implications for inflation—the middle panel—are noticeable, with core PCE inflation moving up toward 2½ percent next year because of the effects of lower structural productivity growth on trend unit labor costs. Joe will now continue our presentation.

Keyboard shortcuts

j previous speech k next speech