Sunday, May 11, 2014

Groundhog Day

Torsten Slok once again makes a beautiful graph, of the kind I posted at the bottom of Punditonomics, reminding us of the foibles of forecasting.  (I would have connected each projection to the actual value at the time it was made, but should make my own graphs if I want to criticize.)

Torsten views the result as an indication of failure for the Fed's models. I think the message is deeper, and tells us a lot more about the macroeconomic situation.

Every serious forecast looked like this -- Fed, yes, but also CBO, private forecasters, and the term structure of forward rates. Everyone has expected bounce-back growth and rise in interest rates to start next year, for the last 6 years. And every year it has not happened. Welcome to the slump. Every year, Sonny and Cher wake us up, and it's still cold, and it's still grey. But we keep expecting spring tomorrow.

Whether the corrosive effects of government microeconomic and regulatory policy, or a failure of those (unprintable adjectives) Republicans to just vote enough wasted-spending Keynesian stimulus, or a failure of the Fed to buy another $3 trillion of bonds, the question of the day really should be why we have this slump -- which, let us be honest, no serious forecaster expected.

(I add the "serious forecaster" qualification on purpose. I don't want to hear randomly mined quotes from bloviating prognosticators who got lucky once, and don't offer a methodology or a track record for their forecasts.)


  1. What on Earth are they using for forecasting? A Smets-Wouters model??

    1. I haven't spoken to a private forecaster who uses a DSGE model.

    2. I know of at least two private sector firms which use DSGE models for macro forecasting. But sure, it's atypical.

    3. They're using FRB/US:

      But of course no serious forecaster uses a general equilibrium model without conditioning information, so there is of course a lot more that goes into generating an actual forecast than simply one model.

  2. I know that the private forecasters use a mean reverting interest rate model. It is shockingly rare to see VAR's use in private sector forecasts.

  3. Nice post. I think the question of why is interesting, noting that it may be the economy, not our forecasting models that have been "broken" recently. Here's one take from last September's press conference (p 14

    "CHAIRMAN BERNANKE. Well, you’re certainly right that we have been over- optimistic about out-year growth. There are a number of reasons for that. One reason for it, though, is that it appears—and I talked about this in a speech last year—it appears that, as part of the aftermath of the financial crisis, that, at least temporarily, the potential growth rate of the economy has been slowed, perhaps because new businesses are not being formed at the same rate, innovation may not be translated into new technologies at the same rate, investment is slower, et cetera. So it appears, again, that the potential rate of growth of the economy has been slowed somewhat, at least temporarily, by the recession and the financial crisis, and you can see that in the slower productivity figures.

    Now, we have—you know, we haven’t anticipated that slowdown in productivity, and that’s one of the main reasons why we haven’t anticipated the relatively slow growth. Now, it’s important to recognize though that what monetary policy affects is not the potential rate of growth, long-run rate of growth, but rather the cyclical part, the deviation of output and employment from its normal level. And in predicting the amount of slack in the economy, so to speak, we’ve done a little better. Our predictions of unemployment, for example, have been better than our predictions of growth. And, in particular, one thing that’s been quite striking is that unemployment—we were too pessimistic on unemployment for this year. Unemployment has fallen faster than we anticipated. So, in that respect, we were too pessimistic rather than too optimistic."

  4. It's not just the Fed. The Australian Treasury could make far better forecasts if they were to incorporate last year's forecast error in their current forecasts.

    Here's a little chart I made of their forecasting performance. Note the 3 per cent of GDP downgrade in budget forecasts between the Mid-year budget update (MYEFO) in 2012 and 2013. During that period, the then government abandoned their commitment to return the budget to surplus. It makes one wonder how much the forecasts were informed by political commitments...

  5. For five years running, the Fed has shot high on growth forecasts, and higher even on inflation forecasts.

    In 2008, the FOMC board was told by Fed staffers to expect 2 percent annual increases in unit labor costs for next five years. That was baked into the forecasts (this was in the transcripts of the FOMC meetings, which finally became public).

    Instead, we see unit labor costs have deflated since 2008. In fact, unit labor cats are falling for last six months, as measured and are below the level at the start of the Great Bust.

    How do you get inflation with falling unit labor costs?

  6. Unfortunately, Torsten's figure contains a couple of misleading pieces. First, he uses the upper band of the central tendency for the FOMC forecast. This will have an upward bias and therefore make the FOMC projection look worse than it actually was. Second, he plots an “actual” Q4/Q4 growth rate for 2014 when we currently only have the advance release for 2014q1. Once these two errors are fixed the forecast still looks bad in 2010 and 2011, but overall the picture is not as bad as Torsten suggests.

  7. "This Time is Different"
    Caveats - R-R was a descriptive history really, and several steps between it and coming up with workable macro model assumptions; huge variation in individual historical events - yet, after reading that, I remained deeply skeptical of any forecast that looked like those above - I won't make a too strong personal claim - skepticism is itself several steps away from calling a turn (or lack of one) - but seems that better understanding financial system crisis is key

  8. "no serious forecaster expected."

    I hope you see the irony in the fact that all of the people you consider to be "serious" turned out to be wrong. They might have been "serious" and put on long faces and stroked their chins thoughtfully but they were WRONG. In my profession, if you are wrong about big stuff you get fired. If you are really wrong, you get sued. And if you are really, really wrong, you get disbarred. But then my profession believes in professionalism and personal responsibility.

    As far as blame for the slump goes: there is no point in arguing over who gets how much; there is more than enough for everyone.

    1. Only if there is someone who can reliably do better. Which there is not, accent on the reliably. In many businesses of course they chase the guy who got lucky yesterday.

    2. "they chase the guy who got lucky yesterday."

      That seems to be an issue mainly in finance where it is a systemic problem.

      The fact that no one can reliably do better (Krugman might disagree) is really proof that all of the models are broken. The test of any theory is its ability to make correct predictions and by that test all of the macro economic theories are wrong.

  9. Hi John,

    I also like Torsten's stuff and I liked that chart, at first glance. There are discrepancies, but one can replicate it by using the high end of the central tendency for 2010 and 2011. Then use the high end of the full Range for 2012 and 2013. Revert to the midpoint of the central tendency for the March forecast. Given the title of the chart, it was an unfortunate mistake.



Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.