September 2017 WSJ Forecasts

The new WSJ forecasts were released last Friday and it looks as if the recent data has caused the forecasters to be pessimistic about the short-term, but optimistic about the long-term. Forecasts for inflation and unemployment through 2018 worsened (unemployment ticked up, while inflation decreased), but consensus predictions for for both variables in 2019 improved. In addition, only the last two quarters of 2017 GDP growth were revised downward, and all subsequent quarters and annual projections rose. 

These general macroeconomic indicator forecasts were somewhat at odds with the changes in specific indicators like the ten-year bond rates and crude oil prices. Bond rates were all revised downward, despite increases of the expected federal funds rate in 2019. Crude oil prices are still expected to rise slowly over the next two years, but only reaching the low 52 dollar mark, instead of 53 or 54 from a couple of months ago. Despite the lackluster September employment report, payroll forecasts for next year rose by over 10,000 to 16,080.

These numbers suggest that the recent data implies that the economy is sliding a little below the long-run growth path. As I pointed out while discussing the recent payroll report, long-run time-series dynamics seem to be dominating current forecasting (as opposed to structural modeling and forecasting). I believe these recent round of forecasts supports that idea, because this pattern of revision is consistent with the behavior we observe. To see what I mean look at Crude Oil Price forecasts:

The graph above shows forecasts at different points in time (light to dark indicates old to new). All we see are level shifts (the intercept) holding the dynamics (the slope) the same. That suggest the new data are not changing anything about the fundamentals, which would alter the trajectory, but instead only reveal changes in the starting point of a more or less unchanged dynamic system. 

But is that good news or bad news? The good news: there really isn't any bad fundamental news. The bad news: models based on dynamic systems are correct on average, but since they are essentially data driven, it makes forecasters appear to be agreeing with each other. So the recent drop in forecast uncertainty (defined as the standard deviation amongst forecasters), does not necessarily indicate that we know a lot about where the economy is heading.