GDP Forecasts
The WSJ August survey indicates a resumption of growth in Q3. What was perhaps a bit surprising was the bump up in the Q3 q/q SAAR growth from about 1 percent to 2.4 percent. The out-quarters were little changed. These forecasts imply the following trajectory for GDP.
Figure 1: Log GDP (blue), mean WSJ forecast (red), and trimmed high (maroon) and trimmed low (maroon) forecasts, based on 2009-10 growth rates. Trimming removed the top 4 and bottom 4 forecasts out of 51 responses. NBER defined recession dates shaded gray, assuming recession end is 2009Q2. Source: BEA advance 2009Q2 release, WSJ August survey, NBER, and author’s calculations.
I’ve included the top forecasts (based on 2009-10 growth) and bottom forecasts, trimmed (that is, leaving the top four and bottom four forecasts out).
Some of the consistently (and inexplicably) high forecasters have dropped out of the August survey (e.g., James Smith) [1], so I think this rebound is quite remarkable. Even so, GDP will reattain the 2008Q2 peak under the (trimmed) high between 2010Q2 and Q3, and will only reattain the peak under the WSJ mean forecast after 2010Q4.
Potential GDP and Output Gap Forecasts
Since potential GDP estimates in Chained 2005$ have not yet been published by CBO, it’s not possible to say what the output gap will be.
The IMF has recently presented an analysis of the future of potential GDP, in the US Article IV supporting analyses [pdf]. They work through how potential is likely to evolve, given the lower investment rate (due in part to constrained financing), as well as other factors (demographics, etc.), and compare against the OECD estimates (discussed in this post). Here’s their assessment of the potential growth and the output gap:
Excerpt from Figure 4. United States: Decomposing Potential Output Growth, in
Natalia Barrera, Marcello Estevão, and Geoffrey Keim, “U.S. Potential Growth in the Aftermath of the Crisis.” Source: IMF, “United States: Selected Issues,” IMF Country Report No. 09/229 (July 2009).
One notable feature of these calculations is that the estimated output gap is smaller than the CBO estimate, around -4 percentage points of GDP, rather than the CBO estimate of approximately -6 percentage points [2], and the OECD (annual) estimate of -4.9 [3]. Readers can compare to statistical approaches in earlier posts [4] [5].
Debt-to-GDP Forecasts
Given anxieties about the trajectory of the debt to GDP ratio, the IMF analysis using stochastic simulation and the historical correlation between the primary surplus, debt, and output gap is particularly salient. While the staff estimate predicts a debt to GDP ratio approaching 100 percent 2019, the econometrics suggest something much less alarming.
Figure 4 from Oya Celasun and Geoffrey Keim, “The U.S. Federal Debt Outlook: A Stochastic Simulation Approach.” Source: IMF, “United States: Selected Issues,” IMF Country Report No. 09/229 (July 2009).
Implied Multipliers
By the way, Professor Mulligan has asked if the CEA’s estimated effects of the ARRA imply fiscal policy multipliers of “20 or 50 or something like that”.
I don’t know how he got that number, but here is what I wrote in my August 1st post.:
Here’s a way to think about what the impact of ARRA has been on 2009Q2 growth. About $60 billion of stimulus funds had been expended by end-June, of which a large portion is in the form of tax rebates. The price deflator is about 10% higher in 2009Q2 than in 2005, so $60 billion translates to about 54.5 billion 2005$. This is a cumulative figure, while 2009Q2 SAAR GDP was 12892 billion, or 3223 billion Ch.2005$ at quarterly rate. If the multiplier is 0.5 (keeping in mind a large chunk of these funds are tax rebates), then growth was about 3 percentage points higher (q/q SAAR) than would have otherwise occurred; I think this is how Josh Bivens arrived at the conclusion that GDP growth would have been 3% lower in the absence of the ARRA.
Since Dr. Romer indicated that the amount spent (see footnote 4), based on Recovery.com and IRS data we are not privy to, was about $100 billion by end-June (rather than the $60 billion I used), then the multiplier required to get to the percentage point impacts on SAAR q/q GDP she indicated is even smaller than I assumed.
For previous installments in the multiplier debate, see: [6] [7] [8] [9] [10] [11] [12]
Technorati Tags: GDP, recession,
debt, potential GDP,
output gap,
multiplier
.
I’m not concerned so much with the debt itself, but with the interest. What are the interest to GDP and GDP growth ratios?
“Some of the consistently (and inexplicably) high forecasters have dropped out of the August survey (e.g., James Smith) [1], so I think this rebound is quite remarkable”
I think it is remarkable, too. And mistaken.
“If the multiplier is 0.5”
From the sidelines, this whole multiplier debate seems murky, although I find your views the most understandable. It seems as if many people don’t want to admit that the stimulus is based on a number of intangibles. In other words, what the hell, this might work.
I don’t have any problem defending that idea, especially in that the Chicago Plan of 1933 posited that a stimulus could help reinforce QE. Also, I’m just a citizen. I can afford to be wrong.
But my take is that a multiplier is almost impossible to determine, it could vary depending upon the circumstances, and it’s based on some ideas we have about how we think markets and people behave that are not empirically nailed down, and might never be. Since that’s how a lot of economics appears to me, I don’t see the problem.
The rise in GDP growth from the first quarter to the second was the largest in almost a decade, and the second largest in the past quarter century. Private forecasters predicted or attributed the unusual behavior of real GDP to the Recovery Act. For example Macroeconomic Advisers in their Outlook Commentary on April 2, 2009 estimated that ARRA added 2.2% to GDP growth at an annual rate (p. 6). Economy.com estimated that ARRA added 3.0% to GDP growth at an annual rate in Prcis: U.S. Macro, July 2009 (p. 6). Goldman Sachs in US Daily: Fiscal Stimulus: A Little Less in Q2, A Little More Later, August 4, 2009 estimated that ARRA added 2.2% to GDP growth at an annual rate. etc. etc. On average private forecasting firms think that ARRA added about 2.4% to GDP growth at an annual rate, or 0.6% at a quarterly rate.
Given what is known about fiscal multipliers should this be surprising? According to recovery.org about $56 billion of the stimulus was spent by the end of June (I still haven’t seen the IRS data). Virtually all of that was in the second quarter. Nominal GDP was about $3.54 trillion in the second quarter. Thus the stimulus was about 1.6% of GDP. The fiscal multiplier thus comes to 0.6%/1.6% = 0.4. The initial projections of of the stimulus’ impact by Romer and Bernstein gave it an average fiscal multiplier of 1.2. This seemingly worse than expected performance is due to one simple fact: a large part of the stimulus has so far been in the form of tax cuts which have a low fiscal multiplier.
We should see performance in the quarters ahead improve as more direct government spending enters the mix. What I find exciting about this is that it is a real time experiment in the ability of fiscal policy to lift the economy in the midst of a liquidity trap. We are literally observing the Keynesian “break the glass in case of emergency” tool in action.
AAARGH.
More idiotic curve fitting.
Also…if you are going to waste all our time curve fitting (curve fitting is what cause this mess)..at least curve fit the right data. Why do econ geeks insist on only worrying about public debt to GDP. It is TOTAL debt to GDP that matters. Private debt load is what caused mess. Increasing public debt (by refusing to recognize private sector DEFAULTS…and instead papering them over with new public debt) will solve nothing.
It is the total debt to GDP…now surpassing 380% that matters. Debt service (Fed, State/Muni, Corporate, Mortgage, Consumer) is all paid out od the same economic profit stream. Worrying about only one segment is idiotic.
This is what economic “thought” in todays world has degenerated to. No one actually understands the economy. All economists do is plug parameters into some model..curve fit it to the past..and make some bogus conclusion.
The best analogy I can come up with is an engine tuner who works with an engine management system. He knows what happens when you richen the mixture. He knows what happens when you change the timing. He knows what happens when you adjust high and low end cam switchover points…he knows when happens when you do these things in combination. But he does not really know, mechanically, how the economy works. So when something is broken…he sits there at the dyno tweaking settings..because he does not really know how the engine works. That tune is a modern economist. Worthless as tits on a tree.
“Given what is known about fiscal multipliers should this be surprising?”
Or what we think we know. Most, if not all of the recent relatively good macro data could just as easily be put down to the affects of lower oil prices.
http://www.inflationdata.com/inflation/images/charts/Oil/Inflation_Adj_Oil_Prices_Chart.htm
In fact all recent recessions and stagflationary periods could be viewed as the result of spikes or extended periods of high oil prices.
Believing that we fixed this recession by design is a very tempting one from a confidence point of view, but underlying fundamentals may be the real cause of the seemingly quick turnaround.
Very interesting analysis Menzie. To put the projections into perspective, compare them to the results of Reinhart and Rogoff (Dec 2008). They simply look at a collection of previous banking crises (mostly G-5 and the 1997 Asian crisis) and consider the average macroeconomic behaviour that results.
They find:
– average increase in unemployment of 7 percentage points and over 4.8 years (trough to peak.)
– average drop in GDP of 9.3% over 1.9 years (peak to trough)
– cummulative increase in real public debt in the three years following the crisis of 86%
I’ve found that last figure to be the most startling one, and it is useful to contrast it with the IMF methodology underlying their fan chart. Their assumption is that US history repeats itself; based on a VAR analysis of US 1948-2008 data they find that
“Federal primary budget balances in the United States. have historically increased rapidly in response to higher debt and the deviation of real GDP from potential output.”
In effect, they’re arguing that there is no need to condition on the fact that there has been a banking crisis. Reinhart and Rogoff (the latter an IMF Chief Economist for many years) argue in a series of articles that banking crises appear to create unique dynamics (and another IMF paper by Stijn Claessens, M. Ayhan Kose and Marco E. Terrones 2008 tends to support that view.)
So I guess the open research question is whether the fiscal policy response to banking crisis debt shocks is different from its response to “typical” shockss; neither camp has anything to from say about that.
I think even the most severe skeptics of Keynesianism would agree that fiscal stimulus temporarily boosts GDP, when funded by foreign borrowing, or when funded by domestic borrowing and applied during a period of risk aversion.
However, skeptics of Keynesianism tend to see that boost as a temporary cushion, which has side affects that slow recovery. That is, the debate is over which is better in the longer run.
Funding fiscal stimulus with domestic borrowing when there is no particular reluctance to invest retards growth, as fund-raising for public spending competes with fund-raising for private investment, and drives up the cost of capital.
I’m not sure where Keynesians stand generally on the value of fiscal stimulus funded by money creation, which is what happens when the Fed purchases Treauries.
I’m simply dumbfounded at the WSJ consensus projection. It appears even more unrealistic and inaccurate than EVERY forecast that the WSJ consensus has made throughout this recession. Until now, each GDP forecast (which was way high when reality was subsequently reported) has been shifted lower with the bottom farther out and recovery at a modest pace.
This time, they’re ready to declare the recession’s bottom now and a 5% GDP increase in the next 5 quarters. Who in the real world believes we will see a 5% gain in GDP by the end of next year?
Slightly off topic, but also related to economic growth and extrem capitalism :
Bernie Madoff’s credit card statement is now available for the public :
http://www.scribd.com/doc/15027931/Bernies-Credit-Card-Statements
Let us spend 160.000 USD per month …
I like Tom’s assessment of Keynesianism and fiscal stimulus. I would add though that critics of Keynesianism would say that stimulus does not just slow long term recovery, it actually severely prevents it by adding to the deficit and raising borrowing costs. It provides further drugs to the drug addicted economy and makes it more and more difficult for it to return to a healthy, sustainable path based on savings and productivity.
Mark A. Sadowski: Thanks for the summary of estimates from the various investment bank research groups. That’s helpful information. It would be interesting to know if they were conditioning on the early estimates of the spend-rate for ARRA, or contemporaneous with 09Q2 release.
Don, the libertarian Democrat: There is empirical evidence on the multipliers, both from the old styles Cowles commission approach to identification, and the VAR approach; see this post for a discussion. It’s just a question of whether you believe the identifying assumptions.
Simon van Norden: I agree that in principle they should’ve conditioned on banking crises, but in the sample they have, there’s only one other (S&L). So what they have is a simulation based on an average of several “ordinary” adjustments and one entailing a banking crisis. Perhaps a Theil sort of mixed estimation could’ve been done (using priors from the cross country analyses)?
Looks like a “V” over at the WSJ. The comps for a panic are not the S&L crisis (I recall that as a systemic meltdown, but not an expensive mess), but rather ’29 and a smattering of financial crises in the 19th century. See Gorton.
Steve:
re: “The comps for a panic are not the S&L crisis (I recall that as a systemic meltdown, but not an expensive mess)….”
GAO 1996 estimated direct costs to taxpayers as just over $130 billion and total costs as around $500 billion. (see http://www.gao.gov/archive/1996/ai96123.pdf)
But I agree with you that it was not a panic.
Menzie,
“On the sample they have…”
Another example of econometricians making spurious conclusions based on inadequate data set. “But its all we had,” doesn’t cut it as an excuse.
Forgive my obvious disdain, but it was the same type of analysis which brought us the following chestnuts:
“nominal house prices never fall”
“foreclosures are driven by unemployment”
“default risk correlates with FICO scores, so income documentation is superfluous”
“Fannie and Freddie are well capitalized under any reasonable recession scenario.”
I could go on and on, but I think I’ll just add:
“its extremely unlikely, given the past twenty (or 40) years of history, that debt-to-gdp will become a problem for the U.S.”
David Pearson: Well, I was not a party to any of those econometric conlcusions, nor did I ever cite them. The one I am familiar with — nominal house prices never fall — was demonstrably invalidated by just looking at Case-Shiller 10 city indices. By the way, if you didn’t notice: (1) the sample is 60 years, (2) the figure doesn’t say debt-to-GDP won’t be a problem, and (3) the report doesn’t say debt-to-GDP won’t be a problem. You are imputing a lot more than what anybody has said.
My point stands: drawing conclusions from inadequate data sets is at best sloppy and at worst dangerous. You use words such as “given the anxiety” and “much less alarming” that attach meaning to the debt-to-gdp forecasts you presented above. Does it rise to the conlusion, “it won’t be a problem?”. Maybe not, but your comments certainly head in that direction.
As for citing or not citing the above chestnuts, not you specifically, but the peers that you defend cited them continuously. None other than Ben Bernanke, as late as 2007, said on a TV interview that he “disagreed with the premise” that falling house prices might cause a recession — he simply didn’t think falling house prices was a plausible scenario. Where would he get such and idea? From some unquestioned economtric study perhaps? Why weren’t economists rolling with laughter that such a statement could be made by America’s most prominent economist?
Sorry. Got my tied up in ‘not’s. I meant to say S&L crisis was an expensive mess, but not a systemic failure, not the other way around.
Regarding debt buildup and Debt/ GDP the staff estimate of 100% of GDP looks reasonable after adding how much further fannie freddie Ginnie Citi Aig and the Fed itself are going to cost in federal assumption. And their mid range debt buildup ( and yours) unreasonably optimistic. You keep ignoring the loss assumption problem in your comment
Here’s a report from the ground here in NYC about how $360 million of the government spending stimulus is progressing. It’s only an anecdote, but a $360 million anecdote would be significant in many communities. For what it’s worth…
The Transit Workers Union contract has been up for renewal, and rather than go through negotiations the governor and politicians had the Metropolitan Transit Authority quickly kick it right over to a panel for binding arbitration.
For perspective, the MTA is so broke it just received a 9-digit bailout via transit fare hikes and a number of new taxes — including a regional payroll tax that applies to all workers in communities located as far as 80 miles away from the nearest NYC subway station or bus stop.
At the same time the average wage of TWU members is $64,000 (significantly higher than the median wage both in NYC and the US), they get full pensions at age 55, and have work rules such as let them get paid time-and-a-half overtime for sleeping at home, literally, and other ample benefits. It’s a politically influential union.
One of the members of the three-man arbitration panel was the head of the union himself — so let’s say that for the union it was a not hostile panel. (Some even think the politicians kicked the contract over to arbitration so quickly because “the fix was in“, but I don’t want to indulge in cynicism.)
Anyhow, the panel ordered an 11% wage increase over three years, $600 million. (With so little inflation in this economy, that’s 11% real.)
How is a broke transit agency supposed to pay a $600 million wage increase? The panel had a specific answer for that…
“The arbitration panel said the authority could use federal stimulus funds and money from its capital program to make up any shortfall in its operating budget” — NY Times.
The arbitrators specifically said that with stimulus money going to the MTA capital account the MTA could cut its own contribution to the capital account by like amount to pay for the wage hike.
The “next day” number being discussed by the budgeteers for the amount of the stimulus that will go to the wage hike is $360 million.
So the result of this $360 million of stimulus will be no new hiring … no construction or purchases to boost GDP … workers already getting above-market wages getting higher above-market wages … and, worst of all for the city’s fare-paying transit riders and taxpayers, the new permanently increased union wage level is paid for with a temporary shot of funding — so what will happen when the stimulus funding ends?
And, of coure, the cost of paying for all these benefits is added to the national debt.
The multiplier for this $360 million is hard for me to figure. And I must say that some of us around here are drawing little comfort from the oft-made claim that while a government spending stimulus may arrive very late, at least it is more effective and productive, due to the political direction of actual spending, than is a much more rapid tax cut stimulus or some such that let’s people spend money on their own however ineffectively they want.
OTOH, it’s just an anecdote, and politics in New York State has got to be entirely different than everywhere else in the US, so this sort of thing can’t be any kind of systematic concern, I’d guess. It can’t be a problem in the big picture or the politicians who drew up the stimulus spending program would have voiced more concern over it. Probably not even worth mentioning.
Re: Brant W. The problem is the economists think the model their using is a Range Rover but in reality it’s a Yugo.
Menzie,
Thanks for the link. Yes. That’s why I quoted you saying “If…”. I was trying to say that the assumptions are key. In other words, there might be no easy way to determine that one set of assumptions is conclusively better than another. I wasn’t doubting that there were reasonable estimates and evidence, since I said that I liked yours, but that it might not be easy to agree on them generally. But, I still feel, like Shiller, I suppose, that there are behavioral assumptions underlying the assumptions, that are very hard to prove. I’m fine with that, by the way. As Wittgenstein said:
“What I hold fast to is not one proposition but a nest of propositions.”
By the way, since I’m a philosopher, linguist, and novelist, I really appreciated the response.
Take care,
Don
@Drewfuss,
Certainly the oil price spike last year played a role in overturning an already wobbly economy. The best discussions of that role I’ve come across have actually been at this site. See this by JDH for example:
https://econbrowser.com/archives/2009/04/oil_shocks_and_1.html
However if you are implying that oil prices have played a role in the unusual performance of the GDP in the second quarter I will have to strongly disagree with you. They have more than doubled from their lows in December. In my opinion, if anything, oil prices acted as a drag on the economy in the second quarter.
@Menzie,
Alas, I am getting the private forecasting numbers from Christina Romer (see page 13 and footnote):
http://www.whitehouse.gov/assets/documents/DCEconClub.pdf
Thus I do not have the answer to your question (and to buy these reports, as you well know, requires a lot of cash).
P.S. A few of your research papers were required reading a course on international finance and development I took as part of my doctoral program. I have visited this blog many times in the past but only recently felt compelled to comment. I consider it among the top three or four econ sites.
For another perspective on the current rosy economic projections, read this by Doug Kass at The Street:
http://www.thestreet.com/story/10576300/2/kass-experts-agree-recession-is-over.html
…sounds like something I would have written. Bottom line: These are the same economists who didn’t see the recession coming or acknowledge it was here. Why believe them now??
If I write an article and get it published in the WSJ, but my chart shows a projected flat-line recovery, or worse, a double inflection down, then would you no doubt find an economic model to justify MY curve? Probably not.
The WSJ projection is wishful thinking. That’s it. Self-serving myth propagated by models and data (as your readers point out) is based on times unlike the present.
Discontinuities in history, just as in data, have a nasty way of making asses of those who make assumptions based on past history.
In the words of Yves, wring, rinse, and repeat.
Jim,
The problem is deeper than that. There are some ‘big picture’ issues with modeling in general that most economists fail to grasp. Worse yet, some economists and analysts go graso them, but use what they know be flawed models, because ‘they have nothing else’…and to admist the models are fatally flawed would basically be an admission that their entire career, their entire body of modern economic ‘knowledge’ and methodology, has been a complete and useless waste.
#1 and most important: The reflexive nature of the use of models. They only work, long term, if you are the only one in the universe using them. When others also use them, they lead to a boom bust cycle.
#2: They are self reinforcing. The behavior the use of models encourages by participants (who believe in, or at the very least act on what the models tell them) causes the models predictions to come true. The models says adding debt to the economy to invest in X,Y,Z will work out well because it has in the past. Everyone piles on…and because everyone is using borrowed money, the asset price is bid up….way beyond its fundamental value. Cap rates in CRE are a perfect and current example.
#3: The very success of the models causes the users to forget about fundamentals, causing the self reinforcing nature to become even more exaggerated. This leads to the build up of imbalances that are unsustainable. We are currenlty in the middle of a monumental effort to now sustain the unsustainable. That no one recognizes that the debt load is unsustainable is clear, because we are trying to solve the problem of too much debt…with more debt.
Menzie:
“I agree that in principle they should’ve conditioned on banking crises, but in the sample they have, there’s only one other (S&L).”
Um…so panel estimation is out of the question? Didn’t you and I go into international economics partly because we thought looking at the experience of other countries was very informative? ‘specially when studying rare events? like crises?
Just asking….