The WSJ survey of forecasts has just come out [link]. One key finding is that the mean forecast has barely budged since March. In other words, unlike previous months, the perceived outlook has ceased deteriorating.
That being said, the dispersion of forecasts is pretty high, even q4/q4, ranging (-3.5%, 3.4%).
Figure 1: Histogram of 4q/4q growth rate of real GDP (in percent) from March WSJ survey. Source: WSJ April survey and author’s calculations.One is tempted to ask who is forecasting 3.4%. That would be James F. Smith, of Western Carolina State University and Parsec Financial Management. Dr. Smith has been extremely consistent in his forecasts for q4/q4 growth, having forecasted 3.4% in the December 2008, as well as in the January, February and March 2009 surveys (I didn’t go further back than December…). Note that once his forecast is removed, the distribution of the survey responses is approximately Normal (i.e., a Jarque Bera test can’t reject the null of a Normal, at the 43% msl). In addition, the mean growth rate drops to -1.48%.
I noted in the first paragraph that the mean forecast had ceased deteriorating. One can see this if one plots the March and April mean forecasts. The forecasted trajectory of GDP is essentially unchanged. One has to go back to the February forecast to see the detioration, as is shown in Figure 2.
Figure 2: Log real GDP (blue), April WSJ survey mean forecast (red), real GDP advance (teal), February WSJ mean forecast (pink), CBO potential GDP (black), all in log of Ch.2000$. Source: BEA GDP final and advance releases, WSJ, CBO, NBER and author’s calculations.In this sense, the statements by several individuals that the outlook has stopped deteriorating are consistent with forecasters’ views.   However, this is not the same as saying economic conditions have stabilized. In fact, the mean GDP forecast still indicates continued decline into 2009Q2. And of course, means by definition do not show the variance in forecasts.
Because of the aforementioned sensitivity to outliers (I’ll call it the James Smith problem), I’ve plotted in Figure 3 (log) real GDP, the mean WSJ forecast from the April survey, and trimmed high and low forecasts (that is, looking at the 6th highest and 6th lowest q4/q4 forecasts; thus I’ve dropped the top 5 and bottom 5, out of 54 forecasts).
Figure 3: Log real GDP (blue), April WSJ survey mean forecast (red), trimmed high and trimmed low forecasts (gray), CBO potential GDP (black), all in log of Ch.2000$. Source: BEA GDP final release, WSJ, CBO, NBER and author’s calculations.The mean forecast implies that the output gap will be -8% (in log terms) in 2009q4. If the optimists are right, then the output gap will only be -6%.
The survey was conducted between April 3-6. Thus, they came before the trade release for February. Since the trade balance was above consensus, conditional nowcasts of GDP have probably risen .
On the other hand, the OECD forecast cited in this post implies continued decline throughout 2009. I’m not certain why the OECD is so gloomy (or alternatively, why the US-based forecasters are so optimistic). Using the OECD forecast and the CBO potential, the output gap will be 10.9% (log terms) by 2010q4. Perhaps this is in part due to a more pessimistic assessment of potential GDP (eyeballing the “Output Gap” table in Appendix 1.2 of the March OECD Economic Outlook, it seems that the OECD’s estimate of potential is about 1.2% less CBO’s).
A final observation: given the substantial negative output gap under reasonable assumptions, it’s hard for me to be particularly worried about inflation in the current year, as evidenced in some fevered accounts (e.g., ). Given that 0% of respondents perceived inflation as the biggest threat to their forecast, I think I’m in good company. (Digression: in 2000-01, when I was following the Japanese economy on the CEA staff, I also heard worries about hyperinflation in the wake of rising debt-to-GDP ratios; so far we haven’t seen that outcome materialize).
Originally published at Econbrowser and reproduced here with the author’s permission.