The CBO released its Economic Outlook today. Here’s its projection of the output gap, under current law.
Source: CBO, The Economic Outlook, 7 January 2008.
This forecast differs from the output gap projections shown in my previous posts [1], [2], in that it conditions upon current law. The forecasts from the WSJ survey (and SPF) embody differing estimates of the size of the stimulus package.
Note that the projection implies that — in the absence of further action — the size of the output gap will be comparable to that of the early 1980’s. The output gap is more persistent than that of the 1980’s, implying a larger cumulative output loss. Hence, the projected recession will also be quite long, as the trough is projected for 2009Q3. The peak-to-trough period from 2007Q4 to 2009Q3 is seven quarters, and would constitute the longest recession since the Great Depression (1929Q3-1933Q1) (see NBER).
Update: 9:20pm Pacific Oops, I was scooped by Paul Krugman.
The Bush tax cut stimulous to the economy failed to close the output gap. Yet they continue to propose more of the same. Republicans never learn do they?
spencer: It would be of interest to know how big the output gap would have been in 2008Q2-Q3 if the stimulus package had not been implemented. And even more interesting to me would have been another counterfactual — what would have occurred if a properly structured stimulus package (less business tax incentives, more spending) [1], [2] had been implemented instead.
Menzie, I agree with everything you wrote above, but in your reference to the earlier piece a pocket full of m you show some point estimates of different multipliers and some simple algebra on how those can be derived. My question is, how confident are we in those numbers? Mark Zandi, lists the multiplier for food stamps as 1.73, two digits after the decimal implies a lot of precision. What is the standard error?
Does this projection by the CBO include any stimulus package/tax cuts by the Obama administration? I glanced through the report and don’t think it goes. Any thoughts as to whether Obama’s proposed stimulus would close the output gap any sooner?
MikeR: I don’t know the exact manner in which Zandi calculated the multiplier. There should certainly be a standard error. If the multiplier is calculated as say b/(1-b(1-t)) for instance, where b is the marginal propensity to consume, and t is the marginal tax rate, then the multiplier standard error is a nonlinear function of the standard errors surrounding the parameter estimates for b and t. Or it could be calculated as a reduced form regression, in which the standard error would pop out of the regression.
theo: No, as I mentioned, this is a “current law” projection, so follows that standard that CBO uses — current law as written, and discretionary spending rising with projected CPI. See this post for discussion and link to CBO’s explanation.
So you haven’t read Christina Romer’s paper, huh?
Buzzcut: If you mean Romer and Romer (2008), why yes I have read it. That’s one particular way to identify fiscal shocks. But I’m not sure what your query is meant to imply.
By the way, Romer and Romer have a related paper dealing with the “starve the beast” hypothesis. Recommended reading!
Tax cuts works, as long as they are accompanied by spending cuts.
Nice qualitative description of Harding’s actions in the ’20-’21 downturn, and how quickly things turned around:
“…Harding inherited Wilsons mess in particular, a postWorld War I depression that was almost as severe, from peak to trough, as the Great Contraction from 1929 to 1933 that FDR would later inherit. The estimated gross national product plunged 24 percent from $91.5 billion in 1920 to $69.6 billion in 1921. The number of unemployed people jumped from 2.1 million to 4.9 million…
“…Federal spending was cut from $6.3 billion in 1920 to $5 billion in 1921 and $3.2 billion in 1922. Federal taxes fell from $6.6 billion in 1920 to $5.5 billion in 1921 and $4 billion in 1922…
“…With Hardings tax and spending cuts and relatively non-interventionist economic policy, GNP rebounded to $74.1 billion in 1922. The number of unemployed fell to 2.8 million a reported 6.7 percent of the labor force in 1922…”
http://article.nationalreview.com/?q=MWI2OWUyOWE2NmZjMmQ2ZTg5YzIzZjczY2I2Mzg2N2Q=&w=MQ==
Rothbard stated that the ’20-’21 recession was the last one unfettered by Keynesianism. I had not seen the numbers on the turnaround, before. Looks mighty compelling to me, a ‘slash government spending and taxes’ approach.
Republicans never learn do they?
It is a well-known fact that Republicans have higher IQs than Democrats. Maybe not high enough, but higher than Democrats.
Tax cuts works, as long as they are accompanied by spending cuts.
does it bother you that CBO thinks we were above potential significantly thru most of the 50s and 60s? I think they overestimate NAIRU and thus the downside gap is underestimated. Doesn’t necessarily have anything to do with the medium term trend tho – it’ll take us a long time to get back to potential, even with sizable stimulus. The gap will be large & persistent in part because we have to move quite a bit of productive factors into new lines of work. Giving lots of tax breaks to money-losing firms probably isn’t going to speed that process up.
Menzie,
I’m thinking of, “The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks,” showing, “that “exogenous” tax cuts–that is, tax cuts not intended to offset the business cycle–have a large positive effect on gross domestic product. Specifically, a tax cut of 1% of GDP will raise GDP by about 3%.”
” Specifically, a tax cut of 1% of GDP will raise GDP by about 3%.”
Indeed. The only reason America is incapable of implementing this is that government spending must never go down, and public sector pensions must never be trimmed.
Other countries are doing this, and will thus attract corporations, capital, and eventually knowledge workers away from America.
Buzzcut: As I said, that’s one way of identifying shocks. In addition, the astute reader will note that all types of tax cuts (marginal rates, lump sum) are treated equally. Finally, I must confess I am more sympathetic to spending increases than further tax cuts. So, for me, the relevance of the paper is minimal.
MikeB: The characteristics that you mention don’t trouble me particularly. Other aspects revolving around the difficulty of estimating potential do (see some discussion here). My understanding of how CBO calculates potential is from this document. It is built upon a production function, rather than NAIRU, framework.
Great conversation, I love reading this blog. http://www.economixt.com
jg
I recently read Benjamin Anderson’s description of the precipitous 1920-21 depression & his description consistent with yours of the government’s response. He also mentions it as the last time the price system was allowed to work before Keynesianism caught everyone’s imagination. It should be more widely known & studied.
I’ve got to conjecture that Keynesianism had been so successful, not because of any compelling logice or evidence, but because it gives politicians cover to do the mischief they want to do anyway. Like the Catholic Church for the European monarchies of medieval times.
Speaking of evidence, look at the utter failure of heavy doses of Keynesianism in the Great Depression & modern Japan.
jg and bryce:
That’s an odd way to read the 20s. I’ve read Austrian followers characterizations of that decade as saying it was just one big massive credit bubble that was the underlying root cause of the Great Depression. If you look at a chart of total debt that certainly seems to be the case. They would say that it was avoided in the same way that Greenspan “avoided” a large recession after the tech collapse.
Whether or not you blame the Fed for that (they of course do) I think that large tax cuts during times where there aren’t many viable long term investments leads to bubble conditions.
Hello Prof Chinn,
In the discussions around the CBO’s projected output gap and US potential output, I never read discussions of how the current crisis could affect estimations of future potential output. Unless I am mistaken, CBO still uses a “growth model” to project potential output. However, from what I’ve read of it, this approach doesn’t allow for current financial and economic conditions to lastingly affect projected US potential GDP growth. Might the CBO be significantly overestimating future potential output, and thus the also the output gap?
mikkel:
I am not an economic historian, but my understanding of the 1920’s is that it started in a “negative bubble”. Asset and equity prices were probabably too cheap. Of course, you are correct that leverage caused things to overshoot on the upside.
Menzie:
My point about the standard error is that statistical measurement of phenomena in social sciences is difficult. Although simple keynesian models show a deterministic relationship between fiscal stimulus and GDP, the truth is, rational agents can not be assumed to act in predictable ways. For example, knowing that Democrats will spend us out of recession, why wouldn’t the recession end today, like a self fulfilling prophecy?
Finally, I must confess I am more sympathetic to spending increases than further tax cuts.
Just stating the obvious, huh?
Admission is the first step to recovery. 😉
Regarding the Romers’ lumping of all types of tax cuts, well, that’s just an opportunity for others to do more research, right?
Interesting, your admission of dissinterest in this paper. It is pretty hot in the economics blogosphere. Mankiw loves it. So does Kling.
Some would take your admitted disinterest as evidence that you are only interested in research that confirms your pre-existing liberal bias.
@ Buzzcut
So interesting to hear you cast the author as a libertarian. I suppose that makes you a conservationist? Aren’t broad stroke labels, cryptic references, invisible leapfrog thinking and innuendo helpful? Not.
One important note in the Romer paper is the following:
“We also find suggestive evidence that tax increases to reduce an inherited budget deficit do not have the large output costs associated with other exogenous tax increases.”
I would also note that even though the paper is somewhat supportive of the idea tax cuts that are financed through deficits have a positive effect on growth, Keynes never said that they would not. All other things created equal, government investment is more effective as a stimulus than tax rate reductions. The Romer paper did not address that as it was not its subject.
Buzzcut: Thank you for your insights into my assessment of research. For clarification, I think that the Romer and Romer line of research is important, and relevant. I do think it has limitations insofar as its relevance to the current debate — some of those limitations in terms of relevance having to do with the points I made (no differentiation between types of tax cuts, lack of relevance to spending impulses). I will also note that (i) the Romers are at UC Berkeley, where I received my PhD; (ii) David Romer wrote the textbook I used to teach macroeconomics.
Patrick: Typically the impact on potential GDP is incorporated in an ad hoc fashion; as investment falls, or trend total factor productivity falls, then projected potential GDP falls. I don’t know how these effects have been incorporated into these projections.
MikeR: Well, if you want to go down that route, trust me, you don’t need a fancy model to get indeterminacy or multiple equilibria/unstable equilibrium. Taking your starting point, I wouldn’t call that a standard error problem, except insofar as the distributions for the relevant variables would not be stationary and ergodic. In that case, calculating standard errors wouldn’t make sense at all.
I would call it a confidence problem. One can have no confidence in the estimates of the multipliers you quote, but you use them to support your preference for government handouts because, as you admit, you prefer normative economics. Again this is why you work in “public affairs” rather than economics.
Any research that disagrees with your social goals (the roemer quote), you call irrelevant.
MikeR: I did not say the Romer and Romer (not Roemer) work was irrelevant. Please do not mis-represent my statements.
If indeed we cannot have any confidence in the estimates of the multipliers, I’d be curious in what estimates you have confidence in…Given your previous statement regarding the failure of the predictable human behavior, it must be “none”.
Menzie,
Thank you so much for pointing out and for correcting my mistake, the authors with whom you disagree are Romer and Romer, not roemer.
In statististics, confidence refers to the variablity in a point estimate. If you say that goverment spending has a multiplier of 1.7 while tax cuts a multiplier of 0.5, you might conclude that government spending works better than tax cuts. But I would not be so niave. I would prefer to know the confidence intereval.
Among the liberal “public affairs” courses at Berkely, I am sure they have plenty on statistics. If not, there is a good book on the subject by Arthur Goldberger or you can start here:
http://en.wikipedia.org/wiki/Confidence_interval
MikeR: My PhD is from the Economics Department at Berkeley. Art Goldberger is a professor emeritus at University of Wisconsin-Madison’s Economics Department where I have a joint appointment. I know what a confidence interval is. In fact I taught the concept here. None of what you had said pertains to the situation where the time series are not ergodic, which is what you implied in your statement.
Menzie, have you seen this? It is extrapolation of that paper, to be sure. But there is at least reason to believe that spending and rebates are not as efficacious as across the board cuts in marginal tax rates.
Zero X Owner, I don’t claim the Romers as libertarians. Quite the contrary. With Mrs. Romer now an advisor to Obama, I’d call her a Democrat. But that makes her paper all the more interesting to me.
Menzie-
Do you have any knowledge of how the CBO calculates the affect of mortgage credit failure on GDP? Option-ARMs and ALT-A mortgages will reset at almost the same rate in 2010-2011 as subprime mortgages did in 2007-2008. While the underwriting quality of Option-ARMs and ALT-A mortgages is better than SubPrimes, wouldn’t the stress produced by a 2-year recession elevate the default risk to near the SubPrime rate of default?
I am far less sanguine than the CBO. Household savings will have to increase substantially to reduce the more than 350% of GDP Total Credit Market Debt accumulated by the US economy. Recovery also implies that banks, the government, and businesses accept losses on their balance sheets to reduce the debt burden, else we suffer decades-long stagnation like Japan.
Thanks for posting the link to the “Budget and Economic Outlook”, it made for interesting reading.
Buzzcut: Sorry, don’t usually give much credence to op-eds in Forbes. I understand the arguments, but not convinced.
MarkS: No, I’m afraid you’ll have to scour the CBO website, or contact somebody at CBO, to get an answer to your question.
Sorry, don’t usually give much credence to op-eds in Forbes.
Weak reply. Come on, it is written by this guy.
He’s an academic economist, not Steve Forbes!
Buzzcut: OK, I give. First, his “extrapolation” of R&R’s fiscal ineffectiveness proposition to spending doesn’t make sense if the recession is long — which is suggested by the figure above, and in my subsequent post.
Second, the “perversity” of the tax credit is on the supply-side. On the demand side, it’s not clear it’s perverse, especially if it goes to credit constrained households that have a high MPC. Critiques of the last rebate (in 2008) are somewhat mistaken since they cite an impact number.
Finally, with respect to public finance theorems, well, there are lots of theorems in economics. Ricardian equivalence is one… and if one believes in that, for sure there is no reason to ever undertake government spending (or tax cuts) for stabilization purposes.