Are Estimates for US GDP Growth Believable?
The recent disappointing re-estimates for US GDP growth in the first half of this year, which were released in July, generated considerable discussion this summer, especially in political circles. No less striking, however, were some less well publicized revisions in the numbers for 2009 and 2010, which have raised doubts in my mind about the accuracy of many of the other numbers released in the last two years. These concerns raise significant policy implications for decisions made in the last two years, and possibly for the decisions confronting economic policymakers now.
The revised estimates for US GDP released in July (along with the advanced estimate of 2011Q2 GDP) paint a quite different picture of the performance of the US economy since the current recovery began in mid 2009. Basically, the revised data show that there was no economic growth slowdown in the spring of 2010 as earlier estimated, and there was no acceleration of growth late last year. The revised estimate for 2011Q1 showed virtually no increase in GDP, and the advance estimate for 2011Q2 showed very weak (annualized) growth of only 1.3 percent—but still a surprising advance over the revised estimate for 2011Q1.
The very sluggish economic growth now estimated for the first half of 2011 (only eight-tenths of 1 percent at an annualized rate) has clearly contributed to recent pessimism about the strength of the current recovery and to the sell-off in the stock market. In my view, however, the recently released GDP estimates do not portray an accurate picture of the recent performance of the US economy and are seriously inconsistent with much other data concerning that performance
Before the revised GDP estimates were released in late July, the image we had of the performance of the US economy was that after a quite strong 5 percent growth 2009Q4, growth moderated to a still reasonably rapid rate of 3.7 percent in 2010Q1. Growth was then estimated to have slowed considerably to 1.7 percent in the spring of 2010. Other data about the performance of the US economy was broadly consistent with this pattern—growth slowed considerably in the spring of 2010 after the initial stage of recovery. The Federal Reserve’s consideration of additional quantitative easing was motivated by this perception that the economy was slowing, as well as the concern that core inflation was falling too low, and there was a risk of deflation.
The revised estimates for GDP growth now show that 2009Q4 was weaker than earlier reported, 3.8 percent versus 5.0 percent. I have no particular quarrel with this revision, although I note that it may be influenced by revised seasonal factors and from the revised treatment of oil import costs for a quarter when world oil prices were falling sharply. GDP growth for 2010Q2 was revised up substantially from 1.7 percent to 3.8 percent. Thus, there is now no estimated slowdown in the US economic growth in the spring of 2010. Insofar as the Fed’s decision to pursue QE2 was motivated by concerns about a serious slowdown, and if one believes the recent GDP revisions, then the Fed’s decision was not well rooted in the economy’s performance. Although I was not a fan of QE2, I do not believe that the Fed was fighting a phantom. Aside from the revised GDP estimates, most of the economic data (available at the time) suggest that there was a meaningful slowdown in US economic growth in the spring of last year. Perhaps growth was somewhat stronger than 1.7 percent, but not as strong as at the start of the year.
The estimates for 2010Q3 growth (when they were initially released in late October and revised the subsequent two months) showed a modest acceleration to 2.6 percent from the spring estimate of only 1.7 percent growth. This modest acceleration seems broadly consistent with other economic data. The newly revised estimates are virtually unchanged at 2.5 percent.
Before the revisions of this July, the “final” estimate for 2010Q4 (released in March) showed further modest acceleration to 3.1 percent annualized growth. (As I recall, this was somewhat lower than the “advanced” and “preliminary” estimates released in January and February.) The “final” estimate for 2011Q1 growth (released this June) showed a substantial slowdown to 1.9 percent growth. In view of other data about the performance of the US economy, these GDP growth estimates struck me as quite low. In particular, the employment data show an acceleration of (establishment) employment growth, a significant reduction in initial and continuing unemployment claims, and a substantial fall in the unemployment rate from 9.8 percent last October to 8.8 percent this March. Other economic data generally appear more consistent with the story told by the employment data than by the GDP estimates (before the July revisions). The bulk of the economic data since March point to a significant slowing of growth this spring.
The revised estimates released last month, now show even slower growth late last year and early this year than was previously reported. Specifically, growth in 2010Q4 was revised to 2.1 percent in 2010Q4, and 2011Q1 growth is estimated to have collapsed to 0.4 percent, before picking up modestly to 1.3 percent in the spring. The story told by these revised GDP estimates, however, conflicts strongly with what most other data show—and strikes me as not really believable. Pretty clearly, there was a slowdown of US growth in the spring of 2010 and an acceleration late last year and early this year before another significant slowdown this spring. All of the problems may not lie with the GDP estimates, but in my view they should not be taken at face value.
What is going wrong? I am not sure. The people at Bureau of Economic Analysis are skilled professionals and follow established procedures to estimating and revising estimates of GDP. Nevertheless, this is not the only occasion when I have concluded that the GDP estimates have gone awry, and the revisions to these estimates do not always seem to me to be an improvement relative to what other data may show.
One possibility is that the seasonal adjustment factors have been distorted, not only for GDP and the things that feed into the estimates of GDP, but also for other data series. The enormous movements in US economic data that occurred during the recent recession, especially in 2008Q4 and 2009Q1, must be playing havoc with the standard seasonal adjustment procedures.
If the problem is with seasonal adjustments that have artificially depressed estimates of GDP growth for the first quarter, then we may get back at least part of the estimated GDP growth that we have recently lost in later quarters. Any such boost in growth, however, would be primarily a statistical artifact. The solution in my view is to keep an eye on a broad range of economic data and not to be overly impressed with the GDP estimates.