Floyd Morris' Recession Predictor

This morning, on CNBC, Floyd Morris a writer for "The People's Daily," a/k/a The New York Times, predicted that the US economy is now, or is about to be in, a recession.

Morris predicts recessions with a simple model. When total deflated sales, year over year, by US new car dealers, falls 2% or more, a recession has begun or is imminent.

On its face, Morris' method is both interesting and somewhat sensible. But so many signals point the other way right now, not to mention so many analysts and economists with more training and experience in macroeconomics than Morris seems to have.

Now, on one hand, I respect empirically-driven conclusions. And this has an 'Occam's razor' sort of simplicity to it, which is attractive.

On the other hand, it just seems to be such an isolated indicator/predictor.

So, rather than criticize it, I would, instead, be interested in considering how and why it might be wrong this time?

In effect, to engage in Popperian analysis, and ask, 'what would fool or confuse the indicator?' Can I find a reason which would render the predictive power of the simple method suspect?

Here are some ideas:

1. A 2% real decline in annual sales is pretty small, as triggers go. It wouldn't take much in terms of consumer behaviors, not recession-driven, to attain this result, and give a false positive.

2. The last forecast recession was 5 years ago. Could pricing pressures in the auto sector since then have resulted in non-recession behaviors that simply reduced total revenues to dealers? Morris uses no unit volume data, only total deflated sales.

3. Is it possible that combinations of higher crude oil and gasoline prices, combined with better substitution of online services for local travel, have diminished total consumer demand for cars from new car dealers?

4. Has the growth of online car buying services, with which to negotiate prices with new car dealers, possibly lowered the price/car at new car dealers since 2000, when Morris' method last forecast a recession?

5. Morris showed a graph of his data, and the entire curve seemed to attenuate over the timeframe he displayed- something like 20 years. Is it possible that, due to pricing and competitive pressures, overall auto sales growth, year to year, is simply slipping anyway? And that the relative rate of sales decline is really what matters, not the absolute rate of sales growth/decline? Could dealer sales simply be falling secularly, and some periods fall harder, due to recessions? So that, with time, a 2% decline may no longer be relevant or significant as a recession signal?

6. Are there more or fewer new car dealers than in the past? Does this matter? Does Morris control for the number of new car dealers? Should he control for this variable?

7. Is there a change in the mix of car sales from dealers of different brands, foreign or domestic, higher- vs. lower-priced models? Is it possible that more features are now found on lower-priced, foreign-owned brands, so that new car unit sales could be healthy, but the dollar values of the value embedded in those cars is lower because of productivity and competitiveness in the sector?

Empirical analysis is good. But when one strips away so much information, as Morris has, and forecasts recession based upon such a simple measure, such as Morris' total sales by new car dealers, you have to wonder if there are some complicated factors working beneath the surface, that could confound this recession predictor.

Put another way, if I, as a quantitative marketing analyst, which I have been in a past life, tried to predict business or marketing outcomes with such a simple model, I'd probably be ridiculed for failing to take into account many, if not all, of the factors I have listed above.

I wanted to write about Morris' prediction, and method, because so much has been spoken and written recently about the state of the US economy, now and for the next 18 months. The reality, though, is probably that Morris was on CNBC because he provided a quick, provocative sound bite of opinion, and, thus, entertainment.

And, once again, we see a ubiquitous business "news" source eschew careful, analytical rigor for ratings and the shock value of inadeqately-designed or explained forecasting methods.