This time from Eric Chaney and Anna Grimaldi:
Updated with fresh data coming from March polls, our survey-based model -- call it the MP model -- is unequivocally negative. It says that manufacturing posted an initial contraction in Q1 (-0.9%Q) and, mind the steps, is likely to post another one in Q2 (-0.5%Q). This might well be true and, as a matter of fact, our general rule is to trust the MP model rather than hard data, as long as only the first month of the quarter is available. However, we take the MP model Q1 estimate with a grain of salt, because manufacturing production actually jumped 1.1% in January. Although susceptible to revisions, this strong entry point implies that, in order to validate our model's prediction, production would have to drop by 1.5% in February. Although not impossible, we think that such an outcome is at odds with the quasi-stability of opinion on current output in the February surveys. At this stage, we think that a fair compromise is to assume that actual manufacturing GDP growth in the first quarter dipped into negative territory, but by less than indicated by our MP model -- say a 0.5% contraction. Note that the important point here is that, even with this discretionary amendment, manufacturing production is contracting for two quarters in a row, unless business conditions change radically in the coming weeks.
A manufacturing recession does not necessarily imply an overall recession. When the Asian crisis hit Europe, manufacturing contraction lost 0.5%, in Q4 1998; this did not prevent GDP growth from growing a solid 0.3% rate. Our early GDP indicator, which is based not only on surveys but also on interest rates and US cyclical indicators, is nevertheless highly sensitive to manufacturing production, which it takes from the MP model itself. Then, if fed the raw results of the latter, our GDP model would print 0.0% in the first quarter, followed by 0.1% in the second. Alternatively, fed with our compromise manufacturing production estimate, it would say 0.1% in the first quarter and 0.0% in the second. Given the very relative precision of econometric models, these differences are totally insignificant. If there is one message to take away from quantitative tools, it is that euro-area GDP growth came to a standstill in January and did not show any sign of taking off since then. Of course, zero growth implies higher unemployment and higher government deficits, but that is another story.
Nonetheless, our quantitative tools do not yet indicate that GDP is going to contract in the second quarter, as indicated in our economic forecasts. One of the assumptions we used to build our full-fledged GDP forecast (see "Recession Alert", February 24, 2003, by the European economic team), namely that crude oil prices would shoot up to $40/bbl in March, did not materialise. Instead, oil quotes declined enough to bring the 30-day average to the vicinity of $30. For this reason alone, we could find a justification for a forecast change on the upside, for the first time since 2000. However, we are not yet ready to take this step. Gloomy we are, and gloomy we are likely to stay for a while, even though market conditions are slightly better than they were three weeks ago. Put simply, the volatility caused by inconsistently reliable news from the front in Iraq makes us very cautious about the markets' judgement.
Source: Morgan Stanley Global Economic Forum