Evan Soltas
Aug 20, 2012

Are Things Looking Up?

Following yesterday's post on my new real-time forecasting model for key macroeconomic data, I wanted to discuss further its projection for the coming quarter -- which, contrary to what I wrote in the original post, is 1.9 percent quarterly annualized growth in real GDP. (There seems to have been a coding bug which I've since resolved.)But before that, a side note: I don't have access to Excel right now, so I can't share with you the latest progress on the model in any real detail -- but I will say that, with the lagged unemployment rate and initial claims, I expect to be able to come quite close to zero root mean square error. My new goal for the real-time unemployment forecast is to be able to call an increase, decrease or hold with 75 percent accuracy. I haven't started the inflation forecast yet, though I am optimistic about that. It seems to me now that quarterly change in real GDP is the most challenging to forecast, given weak autocorrelation and the wide variety of possible input variables. By comparison, autocorrelation is powerful in the unemployment rate and somewhat strong in inflation, and the inputs into the model are more obvious.

The model is "thinking" that last quarter's real GDP growth slowdown will fade in Q3 2012. Given how my model works, this means is that recent data implies, with striking uniformity, that things are indeed looking up.

Quarterly growth in industrial production, my model thinks, is the most powerful predictor of GDP -- and industrial production numbers in July suggest 3.0 percent real GDP growth. Payroll employment, the second most predictive, corresponds to 1.9 percent real GDP growth. Beyond that, the Leading Index's July measure implies 2.5 percent real GDP growth, the ISM survey in July would have you think GDP will grow at 2.1 percent, and the St. Louis measure of financial stress suggests clear sailing, with 2.1 percent real GDP growth. Least of all, S&P performance anticipates 3.4 percent growth. Tempering this optimism are real personal consumption expenditures, whose growth implies real GDP contraction of -0.3 percent, and initial claims for unemployment, which match with 0.1 percent real GDP growth.

The model data, broken down in this way, reveal clearly that the case for pessimism in the short run has weakened significantly over the last month or so. 1.9 percent quarterly annualized real GDP growth is nothing to get wild about, for sure, but it looks a lot better now than in June, when I saw increasing risk of recession starting in Q1 2013.

In comparison to the Philadelphia Fed's survey of professional forecasters -- you know, the people who do this thing for a living, as opposed to me and my Excel files -- my model is just to the optimistic side of the median forecast of 1.6 percent growth. That number represents a sharp downward revision of their projection from 2.5 percent.