Mar

12

The recent forecasts by the former chair of the Fed about the likelihood of a recession are a case study in how not to forecast. Alan Greenspan believes there's a 1 in 3 chance of a recession this year. There are 3 cornerstones to his theory because "we are in the sixth year of a recovery." First, short term interest rates are above long term, and according to Fed research, such divergences have been good at predicting recession. Second, we are in the sixth year of a recovery and there is a tendency for recessions to become more likely as we enter the "well-known 10 year cycle." Third, profit margins are declining, and Greenspan believes this will lead to a decline in capital spending that will lead to the recession. The problem with these forecasts is the same thing that's wrong with most over-determined forecasts. There are so few observations that it are guaranteed that one indicator or another will provide a perfect forecast of the recessions in retrospect. For example, since 1945, there have been only 12 recessions. The hazard rate for a recession ending given the number of months of the previous expansion is as follows:

   Number of          Number of           Chances of
   months of          observations            failure
   expansion           remaining           next 12 months %
    12                         11                        10
    24                         10                        11
    36                          9                         33
    48                          6                         17
     60                       05                          0
    72                        04                        20
    84                        03                        33
    96                        02                        50
  108                        00                      100

The results show that a fairly constant hazard rate of 20% throughout 84 months, with all of the remaining three expansions dying in the next 24 months. However, because there were just three observations, there is nothing but randomness in such an estimate because estimates based on three observations are completely unreliable. They have at least three times the variability of an estimate based on 25 observations.

The same flaw would apply to an attempt to relate the recessions to movements in interest rates. With only 11 observations to predict and approximately six turning points in the indicator that are doing the predicting, the chances of finding a pattern when none exists are close to one, when one considers the number of forecasting indicators, measuring in the thousands, that were implicitly considered before coming up with the divergence between short term and long term as the magic bullet.

I have previously discussed a similar error in using the short term — long term yield differential to forecast stock prices, showing that just a small error in the reporting was enough to change the conclusions, and the completely useless forecasts that such indicators gave for stock prices. (They were bearish throughout 2006 for example.)

A much more fruitful way of thinking about business cycles is the way Milton Friedman thinks about them. "I've always questioned whether there is really such a thing as a business cycle. What you have is an economy that is subject to shocks from time to time."

Such shocks according to Friedman are caused by the Fed with mistaken monetary management. Part and parcel of these mistakes is reliance on over determined, multiple comparisoned indicators, similar to but even more useless than the Super Bowl indicator for stock prices.

Q: So the business cycle is becoming less important?

Milton Friedman: No. I've always questioned whether there is such a thing, really, as a business cycle. What you have is an economy which is subject to shocks from time to time. And a shock comes along which knocks the economy down, and then it recovers. But the idea that there are regular intervals, regular size, I think that is not supported by the–I have no doubt whatsoever that to a large extent past recessions were produced by mistaken monetary management; that they were not natural in the economy, that they did not have to occur; but you had a situation in which the monetary authorities–this is particularly after the Federal Reserve was established–followed a policy of tending sort of a stop-go policy. They were late in reacting to changes in the economy, and when they acted, they acted too strongly.

And even the Fed has learned from experience. And I believe that the performance of the Fed under Mr. Greenspan has been better than any prior chairman. You may know personally I'm in favor of abolishing the Fed.

Peter Robinson: Yes, I know. I know. I'm going to get to that.

Milton Friedman: I would rather substitute a computer for it.

Peter Robinson: The business cycle is one of those phrases that is almost incantatory, people say it over and over again, and so I assume there must be a business cycle. Now I hear you saying that you're not sure that such a thing even exists.

Milton Friedman: If you want to see a real cycle, think of a seasonal cycle.

Peter Robinson: Right.

Milton Friedman: That's an event. It's warm in the summer, it's cold in the winter, you grow food in the summer, you don't grow it in the winter.

Peter Robinson: Predictable, regularity.

Milton Friedman: So you have a real cycle predictor, that would be a real honest-to-God cycle.

Peter Robinson: All right.

Milton Friedman: Now the image of the business cycle that people have had is that there are reactions in the economy of a similar kind which tend to run with reasonably regular frequency.

Now for one time, to give an example, one favored theory at a time was the so-called sun spot theory. There are spots on the sun which have a physical cycle, 10-year cycle roughly. And that does affect the fertility of crops on the earth. It affects the growing conditions. And Stanley [Jevins], an English economist in the 19th Century correlated, the movements in agricultural output with the movements in the sun, and argued that that was a cause of a business cycle. That would be a real honest-to-God cycle.

Peter Robinson: Right.

Milton Friedman: And people had been searching for some other mechanism. But what I think is really going on is a very different thing. I think that you have a reaction mechanism in the economy. The image I've always had is, think of a piece of wood up here with an elastic band glued onto the bottom of it. And every now and then something comes along that plucks it down. For example, you get the shock of the oil embargo.

Peter Robinson: You're talking about the '70s.

Milton Friedman: In the '70s. That knocks it down, and that creates a recession. And then there is a reaction mechanism within the economy, which you can understand, that it takes time for what happens then to have its full effect. Some things react immediately. Some things react later. And that reaction mechanism means that we'll take a reasonably predictable amount of time for the economy to react back to there, and get back up to that board up there which defines its long term path.

And similarly there might be something that pulls it up. All of a sudden, you've got a war in which in order to finance it, they print a lot of money, it causes inflation, that produces a temporary boom, and then you react down to it.

So the image I have is of an economy which is subject to shocks from time to time, and which reacts to those shocks in a rather predictable manner, with a predictable reaction mechanism in it.

Steve Ellison adds:

For sales forecasting, APICS teaches a decomposition method:

Demand = Base + Trend + Seasonality + Random Error

One can develop forecasts by evaluating the seasonal pattern and longer-term trend to estimate underlying base demand.

More sophisticated sales forecasting methods take substitution into account. For example, if a technology company introduces a new product, usually there is an old product for which sales will drop as customers switch to the new product.

Substitution can occur across time, too. Many sales promotions succeed only in rearranging the timing of demand, but fail to increase base demand. Customers who were thinking of buying in the future rush to buy at low prices. With much future demand already satisfied, sales fall far below normal when the promotion ends (possibly at about the time the marketing department receives bonuses for increasing sales). Such a case is a microeconomic example of Dr. Friedman's elastic band being stretched up and then reacting down.


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search