First Reading on (part of) Q1 GDP

Here’s a compilation of e-forecasting’s January GDP estimate, Macroeconomic Adviser’s December GDP estimate and forecast for 2009Q1. E-forecasting’s estimate is that January real GDP was declining at an 11.8% at annual rates.


snapshot.gif

Figure 1: GDP (light blue bars), GDP from e-forecasting, 2/20 (blue), and from Macroeconomic Advisers 2/13 (red); Macroeconomic Advisers’ forecast for 2009Q1 (red line with *), in Ch.2000$ SAAR. NBER defined recession dates shaded gray (assuming recession has not ended by 2009M03). Source: BEA, GDP release of 30 January, e-forecasting 2/20 and Macroeconomic Advisers [xls] 2/13 release.

Clearly, in order to hit the mean WSJ forecast of -4.6% SAAR growth (see discussion here), the rate of m/m decline must decelerate. As Macroeconomic Advisers notes:

…Our latest tracking forecast of a 4.6% annualized decline of GDP in the first quarter requires only 0.2% declines per month during the first quarter. …

The Macroeconomic Advisers forecast available online predates some key releases, including the industrial production release discussed here.

40 thoughts on “First Reading on (part of) Q1 GDP

  1. GK

    The steeper the plunge, the sooner in time the trough occurs. That is just the basic mathematics of it.
    So while the plunge is scary, the speed ensures that the destination is arrived at sooner. It is just not possible for GDP to contract this fast, and lose 500K+ jobs a month, without the recession trough happening by Q3.

  2. GK

    Even Nouriel Roubini says that US real GDP will contract by as much as 5% peak-to-trough. If Q109 ends up at -10%, and Q209 is also -5% (after Q408 was -5%) quarterly, then that -5% total GDP correction has finished by the end of Q209.
    Steeper fall = sooner thud of hitting bottom.

  3. Mike Laird

    GK, think a bit more about events in Europe and Asia, and then merge those trends into an interacting system with the US economy. This one cannot be analyzed effectively by looking only at US history and trends. It requires a system view. Lots of luck.

  4. JD

    GK, you assume that the recession trough is fixed. By your reasoning, the federal government should simply suspend all transactions, employment and production for 2.6 weeks (0.05%)….no GDP at all. Infinate speed and then begin recovery. Do you really think that this would provide the reallocation necessary to begin recovery???

  5. GK

    JD,
    Don’t be silly.
    I do believe that the global economy, as well as the US economy as a subset of it, always reverts back to the general trendline. We can debate where the trendline is, but estimations won’t vary beyond a certain point.
    The further the economy and markets get from the trendline, the less time they can sustain a level that far from the trendline. Each successive quarter of extreme distortion becomes exponentially more improbable.

  6. JD

    GK
    One might argue that this very trendline is effected by the magnitude of the distortions and the rate of change. Your arguement that the “the speed ensures that the destination is arrived at sooner” seems to suggest the sooner the better. I argue that the rapidity of this decline since Q4 2008 has impacted your trendline. Also, “Each successive quarter of extreme distortion” further reduces this trendline. Last, improbable events happen. Statistical models are tools that cannot adequately account for ‘black swan’ events.

  7. Rajesh

    GK,
    I disagree. The trendline is not fixed. Depending on whether we make investments in human and fixed capital, the trendline can move up or down. If the fall away from the trendline causes people to stop investing or to attempt market intervention such as protectionist trade barriers, then the economy can stay in a non-optimal path for a long time. This is why it is possible to have a Great Depression. Government policy does matter.

  8. wcw

    The betting money in general is indeed always on the trendline. The betting money tends to get thrown pretty hard on occasion, though. Take 1873 — please.

  9. GK

    JD,
    Time will tell if the trendline has moved. It did not move even through the Great Depression, so I am nowhere near willing to entertain that it has moved in this instance.
    Rajesh,
    The trendline of World GDP did not adjust even during the last GD. Go Google Ray Kurzweil’s the Law of Accelerating Returns, or Robin Hanson’s article on singularity economics.

  10. GK

    Brad Delong and Arnold Kling have also done a lot of work on long-term (200-500 year) economic growth trendlines.
    Remember that Kondratieff Waves, and the winters that are part of them, essentially amount to ‘three steps forward, one step back, repeat’.

  11. Paul

    GK says “the economy always reverts to the trendline.” Hmm, for about 1 million years the trendline was basically zero growth. For the recent 250 year abberation it has been 2-3%. So which trendline are we headed to Mr Chesterton?

  12. GK

    “Trendlines get broken. Remember the trendline of housing prices?”
    Bull. People who know about the basics of home valuation quickly saw that home prices were way too far above the trendline. I have been saying that since 2006.
    So if anything, the housing correction PROVES the trendline primacy.
    I don’t think you know what a trendline is.

  13. Anonymous

    JD, Rajesh, GK: Can you clarify what you mean by “trendline”? Do you mean some sort of time series property, if so which one (series and property)? Or do you mean it to stand in for some sort of theoretically justified value, i.e. what sort of economic exchange or production should we expect to see at a certain horizon out, given where we are today?

  14. Anonymous

    JD: You say that “Statistical models are tools that cannot adequately account for ‘black swan’ events.”

    That is not correct. For one, you can always assign whatever probability you want to the extreme events in a nonparametric framework, if you are so inclined, and, say, simulate your portfolio or economic variable behavior from that distribution.

    If you want to go parametrically, there is extreme value theory models for the tails of distributions. The multivariate versions are somewhat unintuitive (for a nonmathematician like me, at least), but it’s there all the same. So you can specify asymmetric comovements, no sweat.

    How you estimate or calibrate such models to your data (or judgment) is a whole other question. But it’s not the fault of the model. It’s a question of the modeler!

  15. DickF

    GK wrote:
    The steeper the plunge, the sooner in time the trough occurs. That is just the basic mathematics of it.
    GK,
    Your assumption here is that this is a normal business cycle downturn and that there is a natural bottom at which the economy will rebound. What if this downturn is caused by government action and the government doesn’t reverse its destructive policies? Additionally, assuming there is a natural business cycle bottom, what if government economic mistakes, or a government grab for power and control, have pushed us past the natural bottom and do not allow a recovery?
    Currently there are a lot of things that are sitting under the surface of our declining economy that will slap down a turn away from the market decline toward recovery.
    Let me cite three examples:
    The Price of Gold – Gold closed at over $1,000/oz. on Friday. This is a strong indicator of a loss of confidence in the dollar, a portent of impending inflation including another uncontrollable rise in the price of oil if there is even a small increase in demand due to the beginning of recovery. And don’t look for help here from the FED. They are in full Keynesian liquidity pumping mode.
    The Stimulus Price Tag – Contrary to what many in congress seem to think there is no free lunch. At some point all this massive “stimulus” must be paid for. If we borrow we are only postponing the inevitable. There are really only two alternatives to allow us to pay this unprecedented debt, massive tax increases or massive inflation. Either one will kick a recovery in the teeth.
    Nationalization of Business – If a government bureaucrat could run a business better than those currently running them they would not be government bureaucrats but rich industrialists. It is a serious delusion to believe that government bureaucrats, mostly lawyers, can run business better than professionals who have devoted most of their lives to learning the business. Just as the Zimbabwe example proves you cannot take farms away from productive farmers and give them to political cronies and get the same level of production.

  16. john smith

    sorry, is there anybody that can say why a recovery ahead ? It might simply not happen at all. Its a possibilty.

  17. calmo

    O john! You prolly won’t get a chance to read this on account of kickin the bucket…a real possibility that I’m at my wits end to arrest and find your recovery button.
    About “trend” and baselines and references and Chained 2000$…and the registry of GDP over the past few years on what now appear to be sky-high property valuations or possibly merely transitory property devaluations.
    I am yearning for another metric than Ch2000$, bothered by the sudden disappearance of $Ts in property values and what appears to be grossly inflated GDP values.
    Was the $650k starter bungalow in 05 San Diego, that is now worth $150k, raising doubts about the real height of the boom? Should the registry of all RI be spread out over the period of its amortization or just plunked down at the instant of the transaction?

  18. oops

    calmo- interesting. how do they account for homes that were 500k and are now 400K? is that a revision? has it happened? is this something new that has not been done before?

  19. JDH

    Calmo and oops: The value of the new construction itself is included in GDP, but the sales price of the land (which is most of that $650 K figure you mention) is not. The sale of an existing home from one person to another is not counted in GDP.

  20. calmo

    And new construction is less than 20% of the housing market from which MEW was withdrawn to fuel consumption expenditures, yes? 70% of GDP coming from that bin, yes? Until recently…so this nagging feeling does not go away just because the BEA manages to somehow back the land value out of the property transactions. Of course OER, the proxy for housing costs nags me too…I could be too sensitive for Economics, but unlike john, I have staying power.

  21. MikeR

    An -11.8% number would be truly devastating… more than a 3 sigma event. Let us hope that does not happen.
    I have found individual GDP forecasts tend to be too extreme, given the low GDP volatility over the past 50 years. Maybe this time it is different.

  22. calmo

    So rather than stomach the 3 sigma event, downsize the previous GDP records and recognize that it really was a rather empty air-filled bag, not a super-sized decade of 3(?)% GDP growth….something everyone can digest.

  23. Sanjay

    GK,
    Trendlines are only useful under a constant structure. Structural changes break the trendlines. In econometric time series analysis, Chow test is used to determine the structural breakdown. If we use trendlines without considering the structural breakdowns, Iraq, Egypt, Greece and Italy would be the most prosperous nations. Black Swan events are structural events and have the potential to break the trendline.

  24. GK

    Sanjay,
    No. Even the Great Depression and the Long Depression (late 19th century) did not deviate the global GDP growth trendlines.
    Again, go read the work of DeLong, Kling, Hanson, and Kurzweil.

  25. GK

    DickF,
    Yes, but we have survivied such Government meddling before. This country survived both the GD and the Carter years.
    The economy always finds a way to adapt around government meddling. Some individuals are permanently damaged, while other reap a windfall simply by being in the right place and the right time, but the economy adapts.
    1) Inflation? Isn’t the next article from JDH worried about deflation?
    2) The stimulus, while bad, will be pared back, if not now, then after the 2010 elections. People are assembling in protest as we speak.
    3) The sectors that matter the most are not being nationalized.
    You are falling for the same ‘this time it’s different’ that the exuberati fell into at the top of both the dot-com bubble and the real-estate bubble.
    ‘This time it is different’ is wrong, and an indicator of turning points, both at the top AND the bottom.

  26. calmo

    “exuberati”, my goodnessiti! [along with “surviving the GD and the Carter years”…flabbergasticiti!!…”You are falling for…” seriously GK, what are you smoking? Tell the person typing for you that you need a break.]
    Sanjay,
    Yes.

  27. DickF

    GK wrote:
    You are falling for the same ‘this time it’s different’ that the exuberati fell into at the top of both the dot-com bubble and the real-estate bubble.
    No GK, this time it is not different. Recall that the Great Depression lasted over 10 years, through Hoover and Roosevelt, and did not end until FDR died.
    We may recover, as did the Soviet Union after communism fell, but what happens if this US version of National Socialism does not fall? What if there is not another Ronald Reagan to confront the communists with their own failings and defeat them with economic prosperity? What country will expose the failings of the US system? I suspect that only China could fill that role, but knowing what I know about China I would not say they will bring us another Ronald Reagan moment.

  28. MattYoung

    I love this debate with Rajesh, GK and DickF.
    My question is. Did Has the financial community reconciled balance sheets yet? Is the future decline in production calculated in? Have we decided, once and for all, that this is the Mini Depression? Is the stock market done with its major declines?
    My favorite theory about that is that we have been doing the dead cat bounce since Nov 2008, but I could be wrong.
    Second question, Now that the federals are adjusting to the new realities, is their some natural brake on nationalization and demonetization of the private sector?
    I would think so, in the bond market. Ultimately, I speculate, that the existence of the bond market proves its it has an essential function, to crowd government interest payments against increased government nationalization.
    My second speculation follows from the first, if financial markets are balancing the books in reasonable time then they must be accounting for the risk of excessive federal nationalization.
    Do I have too much faith in the financial markets?

  29. GK

    calmo,
    Your words are that of a person who has been intellectually overwhelmed.
    Rajesh,
    The GD actually had 4 horrendous years, followed by a 2-year weak recovery, followed by another 2-year recession, and then a final 2-year recovery to get back to the trendline. So the 10-year GD was actually broken into these 4+2+2+2 phases.
    No GK, this time it is not different.
    Indeed. So recovery always happens. Usually when the pessimism bubble is at its highest, and it is heresy to suggest that a recovery can EVER happen (just like it was heresy in 2006 to say that housing prices can go down).

  30. MikeR

    Following up on my earlier post, if my numbers are correct, qoq GDP at an annual rate has a mean of 3.3% and std dev of 4% since 1947. If 2009 Q1 GDP came in at -11.8%, that would be a -3.74 std dev event with a probablility of 0.01% assuming a normal distribution.
    Either this time it is different or GDP growth is not normal.

  31. kharris

    MikeR,
    GDP growth is very likely not normal, even if the Q1 annualized pace of GDP contraction is not -11.8%. Note, though, that was just one estimate, and just for January. A pace of decline in Q1 half that of the e-forecasting estimate for January would still be extraordinary, on the heels of what is shaping up to be a -5% drop in Q4.
    We need to get away from this mechanical notion of how the economy works. Arguing that this or that “must” happen in some quarter because of one guy’s trendline (I’m sure we could all fit lines that tell other stories) from way back when is silly.

  32. MikeR

    kharris,
    Thanks and I agree absolutely. Economies are complex systems more similar to an ecosystem than an engine. While we “prime the pump” of an engine and use oil to “keep the wheels turning” we don’t talk of “stimulus” for a rain forest. We can visualize linear and quadratic models, but can we can only begin to grasp fluid dynamics or the placebo effect in drug efficacy trials.

  33. aaron

    Even engines are more like ecosystems than we tend to think. I believe that’s a good part of why fuel economy has been in decline while fuel prices were rising. Cars and traffic a far more complicated than “slow is efficient”.

  34. Menzie Chinn

    MikeR: In a fashion similar to annual versus quarterly data changes, one can’t translate the m/m annualized changes to q/q annualized changes. As you might be able to see from the graph, the e-forecasting series (as well as the Macroeconomic Advisers series) are more variable than than the quarterly official series. Surprisingly, you get pretty close to the same answer doing the calculations with the actual data.

    Annualized mean growth (all in log terms) for the e-forecasting series from 1954 to 2009M02 is 0.0314, standard deviation 0.0522. This means the m/m annualized change in February of 12.5% (in log terms, 11.8% in exact terms) is 3.05 standard deviations from mean. However, a Jarque-Bera statistic rejects Normality at p-value of 0.00000. The kurtosis statistic is 6.33 (kurtosis for a standard Normal would be 3). Hence, the distribution is fat tailed. Still I suspect the p-value is pretty small.

  35. Robbie

    “JD: You say that “Statistical models are tools that cannot adequately account for ‘black swan’ events.”
    That is not correct. For one, you can always assign whatever probability you want to the extreme events in a nonparametric framework, if you are so inclined, and, say, simulate your portfolio or economic variable behavior from that distribution.
    If you want to go parametrically, there is extreme value theory models for the tails of distributions. The multivariate versions are somewhat unintuitive (for a nonmathematician like me, at least), but it’s there all the same. So you can specify asymmetric comovements, no sweat.
    How you estimate or calibrate such models to your data (or judgment) is a whole other question. But it’s not the fault of the model. It’s a question of the modeler!”
    Do you believe the claptrap you just wrote?
    First off; A Black Swan, besides being the new buzz-word, is a philosophical thinking game. Black Swans deal with the logical problem of induction; meaning, that the future is not based upon past results.
    You even state that the modeler is flawed and not the model and that judgements are needed. So why not just predict things with astrology? It uses data and judgement also.
    The fact of the matter is; you can not predict the future no matter what medium you use, math, astrology, darts, ect…

  36. aaron

    It does seem a lot like the 70s problem. When oil tried to take a big share of loose money, everything went to crap. The decline in fuel consumption and transit also looks very similar. Question is can avoid the mistakes of the past, or will we continue to make things worse and end up just like the 70s.

  37. larry

    I have a general question Do the CBO estimates of unemployment peaking at between 9 and 10% seem too low now?

Comments are closed.