Downshifting and Reversion in Forecasts

My post on CEA’s forecast discussed the similarity between the Administration’s forecast and the Society of Professional Forecasters forecast. What can we learn from recent SPF forecasts, in the wake of decelerating growth.

In Figure 1 below, I’ve plotted the last three SPF forecasts and the closest vintage log real GDP.


Figure 1: Log real GDP and Society of Professional Forecasters’ forecasts for 15 May (purple *), 14 August (black x) and 13 November (red +), 2006. Source: BEA via St. Louis Fed ALFRED and SPF via Philadelphia Fed

What is interesting is that even as the level of GDP has been revised downward, predicted growth rates remain largely the same. Since it’s hard to distinguish growth rates in log levels, I also plot the annualized growth rates calculated using log differences. This yields the graph in Figure 2.


Figure 2: First-differenced (annualized) log real GDP, Society of Professional Forecasters’ forecasts for 15 May (purple *), 14 August (black x) and 13 November (red +), 2006, and author’s calculations. Source: BEA via St. Louis Fed ALFRED and SPF via Philadelphia Fed

What is true is that as each quarter’s data comes in with lower growth rates, the nearest-quarter forecast growth declines, but the out-quarters forecast converges back to slightly under 3 percent annualized growth (in log terms).

In other words, forecasters continue believe that whatever shocks perturb the economy, within about two quarters growth resumes at roughly pre-shock levels (although the level of GDP is permanently lower).

Technorati Tags: ,
, and


26 thoughts on “Downshifting and Reversion in Forecasts

  1. Anchoku

    Heheh. Thanks. That’s the funniest thing I’ve seen, yet, today. It makes me wonder if they should have been called The Society of Professional Fudgers.

  2. c thomson

    Is there any solid evidence that economic forecasting is even useful? Studies I have seen suggest that the consensus is better than individual forecasts but that naive extrapolation is better still. Just project what is happening and the professional forecasts are left in the dust over time.
    Surely economists would do themselves a favor if they stuck to explaining how we got where we are and let the future take care of itself. Economics is properly a branch of history with a mathematical overlay.

  3. kharris

    Odd, that implicit assumption that growth missed in the near term becomes a permanent fixture. If population growth is not pulled lower by a temporary slowing in output growth, then assuming growth is never made up is a very strong assumption about either productivity gains or labor market participation resulting from the period of slower growth.
    It looks like an admission that forecasters don’t understand productivity in the recent past.

  4. kharris

    c thomas,
    Economic history may be a branch of history, true enough. Relegating the rest of economics to that category misses most of its theoretic underpinnings and its explanatory power.

  5. menzie chinn

    anchoku: By this post, I didn’t mean to criticize the SPF. Rather, I hoped I was indicating that right now, forecasters do not appear to be forecasting a persistent slowdown, despite downward data revisions and hits to growth in q2 and q3.

    c thomson: I believe the current view is that forecasters do have information beyond last period’s output level, or what is captured in a simple ARIMA. See this statistical assessment of the ASA-NBER forecasts here.

    kharris: This pattern of forecasts is consistent with a belief that log real GDP is an integrated process of order 1. At a first cut, that is what GDP looks like, although short run dynamics and nonlinear asymmetries are often detected as well.

  6. Dick

    c thomson,
    Hey, what you trying to do? Put a bunch of government economists out of work. It is not about being right it is about being employed.

  7. spencer

    The major econometric model tend to converge on trend growth rates if they are just allowed to run. To get them to generate something significantly different you generally have to plug in some shock to the system.
    So all these results are saying is that the forecaster are not assuming some external factor like higher oil prices or inflation will significantly impact the economy.
    For all pratical putposes it is a “default” forecast.
    To understand forecasting you have to understand the risk – reward ratios for the forecaster. It is like being a portfolio manager. If you lose money doing what everyone else is doing you probably will not lose your client. But if you go out on a limb and are wrong you will lose clients like crazy. So most bank, wall street forecasters tend to change the consensus forecast by one or two tenths and claim it is independent thinking.

  8. Dick

    I have not worked directly with any government economists. I do work directly with tax agencies and auditors at the local, state, and federal level concerning corporate taxes and with congressional staff. My background is business not academics.
    My understanding of government economists comes from reading their writings both before their work in government and after. I do understand that they are often hindered by political mandates and concerns, but since most are chosen from the same general economic philosophy that supports the central planning preference of politicians I doubt that presents much of a problem. I have only been aware of a handful of classical economists in government service.
    Wayne Angell was good at the FED, not perfect, but good. I was really excited initially with Greenspan but he was quickly seduced by the dark side. Power has a way of warping principles. I was also disappointed when Glenn Hubbard left the Bush administration to go back to Columbia, but that was definitely best for him.
    I hope that helps you understand where I am coming from. I imagine I am somewhat of a minority posting here.

  9. menzie chinn

    spencer: I agree that in the long run, these models converge to whatever the trend growth rates of labor force, capital and TFP are. But it is remarkable that the convergence is within a couple of quarters. Given some serial correlation in the first log difference of output, I would expect slower reversion.

    Dick: I guess that we’re talking about slightly different sets of government economists. There are the policy-level political appointees (who I think you’re focusing on) subject to Senate confirmation; there are economists on the Congressional staffs; and then there are professional staff in the Federal Government, who sometimes spend their entire career in the civil service. These are the economists I was thinking of, when I think about the generation of the Troika’s forecast. In addition there are the staff economists at the Fed. In general I’ve been impressed with the dedication and intelligence of these individuals. Political appointees — such as the Wayne Angell’s of the world — have different forces being applied to them, and you may very well be justified in having low regard for them. But I think that it’s useful to distinguish between different sets of government economists when attributing motivations to them.

  10. calmo

    Fig 2 reminds me of normalized response curves for 2nd order systems where there is under/overshoot and a critical damping ratio that reaches the steady state value first.
    We expect the feedback (previous forecasts) to force a convergence of expectations as long as the system’s shocks are well-behaved (the system is stable) and the ‘turns’, non-disasterous.
    So here it looks like the perturbations (“shocks”) are defined/recognized to be disturbances that have a duration of 2 quarters. If we went back and looked at 2000/1 this convergence might stretch out somewhat.
    It suggests the shocks are unduly domesticated.

  11. Dick

    One other thing that may help you understand me. I preferred Glenn Hubbard to Ben Bernanke as FED Chair. I know that no matter who is FED Chair it would be difficult to change their failed methodologies, but I would rather have someone like Hubbard who recognizes the weaknesses of the theory, rather than someone like Bernenke who believes the theory and attributes any FED failures to a failure to use the same old methodology properly.

  12. John Thacker

    Professor Chinn:
    Does it make a difference to this post (or any recent one) that the Fed just did an upward revision for Q3, from 1.6% GDP growth to 2.2% GDP growth? (And inflation revised slightly downwards from 2.3% to 2.2% annualized.)
    See for example.
    I would imagine that that might slow somewhat talk of an impending recession. I’d be interested to see how plugging 2.2% instead of 1.6% in affects the models.

  13. kharris

    On what do you base your assessment of Fed “failures?” The variability of economic performance has fallen markedly over the past quarter century. Measured core inflation has trended lower for roughly that same period.
    On what do you base your assessment of Bernanke’s predilection for theory (or is it methodology?) as it exists, rather than some unspecified other theory? I can’t identify, from one five-year period to the next, a stable theory that is central to Fed policy. The Fed seems to devote more effort to learning and adjusting than most of its critics.
    I’m not saying the Fed is perfect. I am just unaware of an alternative way of doing business that we can reasonably expect to provide better results.

  14. Dick

    Thanks for the question. I cannot take credit for these words in bold, but they express my feeling.
    The Fed…has a model available to it that uses thousands of variables; it’s called the free market. Back in the Sixties, …[the free market] told the members of the board their monetary policy was full of … error, and they chose to ignore it. When they removed the promise to exchange dollars for gold, their folly was exposed as the Greenback’s purchasing power almost totally collapsed; over the next decade it did drop from 1/35 of an ounce of gold to 1/850.
    To this day, that august body has refused to admit that an outside force, the promise of gold conversion on demand, was the only thing that gave the dollar stability. They tried to shift from targeting a value for the dollar to using the economy, or some other “feel-good” indicators, and that process was far to vague to support purchasing power. It did, however, appear to address the needs of those running for election….
    The real question is whose interest is being served, why should anybody buy dollars? In that sense, …
    [currency] is like any other product; the public sets a value based upon the product’s utility. When the federal government put its needs ahead of the people’s, the value of its money vanished. You can pick any sort of product analogy to see this, because when a manufacturer takes quality out of a product to raise his margins, the customers wise up and go elsewhere. He certainly may continue to sell whatever it is, but at a far lower price, which reflects its reduced utility.
    Over the past quarter century the dollar has varied from around 1/250 to 1/750. As Nobel Laureate Robert Mundell has observed, this cannot be considered stability.
    There is a huge organized effort to prevent the FED from ever going back to a gold standard, headed mostly by those who make their living off of forecasting and projecting the value of the floating dollar in the academic and political sphere.
    Their demonization of gold has been very successful, but, sadly, the day will come when the dollar will experience the same fate as the Continental in the late 1770s – we were close in 1979. Hopefully, when the currency crashes the next time, we will find another Alexander Hamilton to pull us our of the mire.

  15. calmo

    But to Dick’s surprise, Ben showed up at the FOMC wearing a gold earing…pierced through the left nostril.
    “The board members will just have to get used to it” he is reported to have said in response to those who claimed they could no longer look him in the eye.

  16. kharris

    The Fed is not in charge of the external value of the dollar. In fact, no central bank has much hope of regulating the external value of its currency through any means that would avoid whiplash to the domestic economy, given that the federal government – not the Fed – has chosen not to link the dollar to gold.
    In fact, most if not all or what you have laid at the Fed’s feet is in fact a set of choices made by our elected representatives. It may not be to your liking, but I have not voted in a single election in which the gold standard was an issue, and I’ve voted in a bunch by now.
    I can’t except the first quote you offer. Models are simplifications of reality. The “free market” is not a model. It’s a market. The “free market” has also never been observed in nature. Or in captivity. It is a fabulism, a thing of myth. It may have existed once, but there is no evidence it has. The first time a big guy figured out he could take what he wanted, markets acquired an element of coersion.

  17. Economic Investigations

    News of the World #15

    A bit of everything with something on the side… Remembering Friedman Free to Choose, available online in full! The Friedman Magic, Alex Tabarrok remembers how Friedman used a small dynamic model to make a good point about active stabilization. Cl…

  18. Dick

    I agree with you that the primary responsibility for the currency falls at the feet of congress and the President, through the Treasury, but they have abdicated their responsibility. The congress in the Full Employment and Balanced Growth Act of 1978 gave the FED a duel mandate, “promote full employment … and reasonable price stability.” The methodology to be used by the FED has not been mandated.
    As far as I know there is no law that would prevent the FED from reestablishing a gold or silver certificate and reinstating the gold standard to “promote … reasonable price stability.” There is sufficient gold in Fort Knox to allow a gold dollar if the FED established a proper gold exchange value, then properly managed the currency, but I agree that this should properly be accomplished by mandate of either congress or the President.
    There is an alternate method the FED could follow to establish a gold exchange value for the dollar by using the purchase or sale of bonds to manage the money supply, keeping the price of gold within a defined range. If this was announced publicly, speculators would do most of the hedging to drive the currency to the defined price, then the FED would simply manage it. Of course this runs counter to the pride of those on the FOMC because they love to use their econometric games to manage the currency. Even though a gold anchor would do a better job it would not be nearly as much fun for them, not to mention the loss of face time in the news media.
    You are correct that there has been no election based on a gold standard. There are a number of reasons for this but to mention a few.
    The election of Ronald Reagan in 1980 was primarily because of double-digit inflation, unemployment, and interest rates. In a sense this was an election based on Nixon’s ending the gold standard, and the fact that the presidents who followed him didnt have either the intelligence, or the guts, to reverse his error.
    Another significant reason that gold is not an issue is that politicians love the principle described in Keynes 1922 pamphlet, “The Inflation of Currency as a Method of Taxation.” Most politicians do not want gold to stabilize the currency because it would prevent them from increasing taxes through inflation without having to vote for it. It is interesting that economists throughout history have understood this nature of government and politicians to debase the currency, but for some reason our society is totally ignorant. Sadly, the 1970s proved that we will need to endure an absolute economic meltdown before people will relearn this vital lesson.
    There are other reasons but this post is too long already.

  19. menzie chinn

    Dick: I highly recommend to you Barry Eichengreen’s Golden Fetters: The Gold Standard and the Great Depression, 1919-1939, Oxford University Press, 1992.

  20. Dick

    I found a paper by Eichengreen dated October 2001, Counterfactual Histories of the Great Depression, that appears to outline what is in his book. It seems a very serious omission when he ignores Hoover’s increase of taxes in the teeth of the depression, but perhaps this is in his book. It was the Hoover tax increases that got FDR elected, but after the election FDR continued the same failed tax policies. Eichengreen also ignores the Smoot-Hawley tariffs that precipitated the worst trade crisis in the 20th Century.
    One thing Eichengreen does get right is the government intervention with wages and prices distorting the market system and preventing the necessary market correction. The only thing is that he does not connect the dots that this intervention was the primary reason the depression persisted for over a decade.
    Something that has to be addressed when talking about the Great Depression is what distinguishes it from other earlier depressions. Eichengreen asserts a loss of confidence in the gold standard when in fact the loss of confidence was in the government, not the gold standard. This was primarily due to the fact that the world monetary system suspended the gold standard during WWI. The primary difference between the Great Depression and earlier depressios was the government took charge at a time immediately after it reneged on its promise of convertibility (this was similar to the loss of confidence caused by the monetary suspension during the Civil War).
    Seeing that Eichengreen comes from a demand side perspective does give some insights into why he makes his assumptions.

  21. Dick

    The Labor Department said labor costs in the nonfarm business sector — a closely-watched measure of how much employees are paid to produce a single unit of output — rose at an annualized rate of 2.3% in the third quarter, considerably less than an earlier 3.8% estimate. In the second quarter, labor costs actually fell 2.4%, reversing an earlier reading of a 5.4% advance, the department said.
    How can anyone work with statistics like this? In private business if your forecasts went from showing a 5.4% advance to a 2.4% decline you would be fired – or your business would go bankrupt, but for the government such revisions are the norm.
    Historically government data becomes very close to actual because the government changes their forecasts to reflect actual, but at the time of the forecasts they are often abysmal.
    Is it any wonder that central planning makes such a mess of an economy, that the FED takes us on such a monetary roller coaster ride?

Comments are closed.