Richard Posner displays his failure to understand the difference between expenditures (reported on Recovery.com) and tax rebates (not reported online, but have to be estimated).
Romer’s speech argues that the disbursements of stimulus funds through the end of the second quarter of this year (that is, through June 30) have had a big effect on economic output and employment. I said this was unlikely as a matter of theory, and that she had no persuasive evidence to back up her claim. And I raised the question of the ethical responsibilities of an academic who takes a government job and then makes a speech that although it deals with a subject that she had studied and written about as an academic is not a responsible academic analysis. My concern is enhanced by the statement of one of my critics that the Council of Economic Advisers, of which Romer is the chairman, has a sterling reputation for political neutrality and analytical rigor. Romer’s speech does not bode well for the preservation of that reputation. Another critic argues that since it was just a speech, intended therefore to be heard rather than read, Romer should be permitted to have rounded off her numbers, and thus to have rounded off $89 billion (this critic’s estimate of how much stimulus money had been disbursed by the end of the second quarter) to “more than $100 billion” (her language). This overlooks the fact that the speech was posted on the CEA’s website, and is replete with footnotes, which I doubt she read aloud.
In fact, while I am on the subject of the amount of stimulus money disbursed so far, $89 billion seems too high. The government’s official figure is $60 billion, and a recent estimate by msnbc.com is $58 billion. One of my fiercest critics estimates the figure at “about $60 billion,” without however remarking the discrepancy between “about 60 billion” and Romer’s “more than $100 billion. [emphasis added – mdc]
“One of my fiercest critics” apparently is me. Since Mr. Posner clearly did not follow the link provided in my previous post, let me quote from Donald Marron, who was CBO Acting Director, and a member of the CEA under George W. Bush (hence, I suspect, not a “left-wing” economist, as Mr. Posner has characterized me):
In her recent speech about the impact of the stimulus effort, Christina Romer, Chair of the President’s Council of Economic Advisers, noted that “as of the end of June, more than $100 billion had been spent.”
If you visit the government web site tracking the stimulus (Recovery.gov), however, it will tell you that the government had paid out only about $60 billion by July 3. (You can find this figure in the chart at the lower right hand corner of the home page.)
Why does Christi report a figure so much larger than the one reported on the official website? Because Recovery.gov isn’t tracking all of the budget effects of the stimulus.
Christi’s figure includes the $60 billion of spending reported on Recovery.gov plus an internal estimate, prepared by Treasury, of the tax reductions resulting from the stimulus effort through June 24. Those tax reductions are obviously a big deal, totaling $40 billion or slightly more through the end of June.
I used $60 billion, and IMF estimate, in my August 1st post, for want of a better guess. The $89 billion is yet again another estimate, from Mark Zandi (all of which I stated explicitly in my post). But from personal experience, I know two things: (1) the CEA has access to more information than those of us in the public have access to; and (2) the CEA would not publish and allow into a White House document (be it a speech or white paper or ERP) unverifiable numbers.
I think a whole bunch of former CEA economists, who have served in both Republican and Democratic administrations, will back me up on this specific point. Hence, on this issue, Mr. Posner simply does not know what he is talking about; I suggest he talk to somebody who has worked in the CEA.
So, to recap: $60 billion paid out on the spending side. $40 billion in tax rebates. No mystery, if one understands the numbers.
Technorati Tags: stimulus, Christine Romer,
annualize, Council of Economic Advisers,
Richard Posner,
multiplier
Thanks for the heads up Menzie. Posner’s ethical lapse this week has been “fun.” I left the following on his latest blog entry and thought I would leave it here as well for two reasons: 1) I doubt it will pass moderation 2) I just wanted to say how much I appreciate all the good work you do in addition to your serious academic work.
———————————————–
Posner wrote:
“In fact, while I am on the subject of the amount of stimulus money disbursed so far, $89 billion seems too high. The government’s official figure is $60 billion, and a recent estimate by msnbc.com is $58 billion. One of my fiercest critics estimates the figure at “about $60 billion,” without however remarking the discrepancy between “about 60 billion” and Romer’s “more than $100 billion.”
And I remind Judge Posner for the *third time* that Romer said:
“These numbers reflect outlays through July 3, 2009 from Recovery.gov website and internal calculations from the Department of Treasury through June 24, 2009.”
(Thats for the obvious reason that two dfferent government organizations are keeping track of expenditures and tax cuts.)
Donald Marron, relevant to this fact said the following:
“In her recent speech about the impact of the stimulus effort, Christina Romer, Chair of the Presidents Council of Economic Advisers, noted that as of the end of June, more than $100 billion had been spent.
If you visit the government web site tracking the stimulus (Recovery.gov), however, it will tell you that the government had paid out only about $60 billion by July 3. (You can find this figure in the chart at the lower right hand corner of the home page.)
Why does Christi report a figure so much larger than the one reported on the official website? Because Recovery.gov isnt tracking all of the budget effects of the stimulus.
Christis figure includes the $60 billion of spending reported on Recovery.gov plus an internal estimate, prepared by Treasury, of the tax reductions resulting from the stimulus effort through June 24. Those tax reductions are obviously a big deal, totaling $40 billion or slightly more through the end of June.”
http://dmarron.com/2009/08/14/tracking-the-stimulus/
And Menzie Chinn chimes in with this article:
“One of my fiercest critics” apparently is me. Since Mr. Posner clearly did not follow the link provided in my previous post, let me quote from Donald Marron, who was CBO Acting Director, and a member of the CEA under George W. Bush (hence, I suspect, not a “left-wing” economist, as Mr. Posner has characterized me)”
“I used $60 billion, and IMF estimate, in my August 1st post, for want of a better guess. The $89 billion is yet again another estimate, from Mark Zandi (all of which I stated explicitly in my post). But from personal experience, I know two things: (1) the CEA has access to more information than those of us in the public have access to; and (2) the CEA would not publish and allow into a White House document (be it a speech or white paper or ERP) unverifiable numbers.
I think a whole bunch of former CEA economists, who have served in both Republican and Democratic administrations, will back me up on this specific point. Hence, on this issue, Mr. Posner simply does not know what he is talking about; I suggest he talk to somebody who has worked in the CEA.
So, to recap: $60 billion paid out on the spending side. $40 billion in tax rebates. No mystery, if one understands the numbers.”
https://econbrowser.com/archives/2009/08/richard_posner.html
And I won’t bother quoting Paul Krugman, but when he heard about the whole bruhaha (he’s been on the road and hence without access to the internets) he just had to chip in:
http://krugman.blogs.nytimes.com/2009/08/21/sour-sixteen/
So at this point economists Krugman, DeLong, Thoma, Chinn, Wolfers (as well as Waldman and Duy to a lesser extent) have all piled on Posner in the negative. In my opinion Posner needs to quite now before he falls even further behind. (Or he could just keep going, and dismiss all of them, as he does anyone who criticizes him, as “left wing” and see how much of the entire public economic intellectual universe takes notice of his magnificent ignorance.)
P.S. The lesson in all of this is the following. If you’re a public intellectual who accuses other public intellectuals of ethical lapses you had better make sure you’re not living in a giant elaborate hypocritical megalomaniacal glass palace (which is hard to do in Posner’s case).
Mark A. Sadowski: Thanks for the compliment and for the recap.
Here’s an update. It did post (I was a litle surprised).
Robert Waldmann officially joined the fray today mostly by analyzing where Posner is coming from.
As a result I decided to revisit Posner’s latest article and add the following.
————————————–
Another day and now Robert Waldmann has officially joined the fray (more on that in a bit).
First a few comments on the substance of this article:
Posner writes:
“The most serious problem with Romer’s speech is evidence. She thinks she has shown that the economy lost 485,000 fewer jobs in the second quarter as a result of the stimulus. There is no evidence for that, because she makes no effort to adjust for other developments in the economy that affected employment, including other parts of the government’s recovery program. I don’t criticize her or anyone for the absence of evidence concerning the stimulus program’s early effects. As I said in my book, when the government attacks a depression with several different programs, it is very difficult and maybe impossible to disentangle the causal efficacy of each one. I also said, and I have repeated this ad nauseam without my critics noticing, that it was right for the government to try a variety of measures for arresting the economic decline, including the stimulus, even though the result would be that the relative effectiveness of the different measures might be impossible to determine.”
The relevant Romer passage is the following:
“Of course, identifying the effects of the Recovery Act from the behavior of just a few data points is inherently difficult. We dont observe what would have happened in the absence of the fiscal stimulus. One way to try to add rigor to the analysis of the behavior of key indicators is to do a more formal econometric forecasting exercise.
There are various ways to do such an exercise, but let me discuss the results of a typical one. We forecast the usual behavior of GDP and employment jointly, using data from 1990 to 2007. We then forecast GDP growth and average job loss in the second quarter of 2009 using actual data up through the first quarter of the year.
The baseline forecast implies further substantial job loss in the second quarter. Indeed, the implied average monthly decline is nearly 600,000 jobs. What you see is that actual job loss (the dark blue bar) came in substantially lower.
These calculations imply that employment is now about 485,000 jobs above what it otherwise would have been during the second quarter of 2009. This number is very similar to Mark Zandis estimate that stimulus added roughly half a million jobs over the second quarter, relative to what otherwise would have occurred.
I do, however, want to be very cautious. The approach we used is one of a number of sensible ways of predicting what would have happened in the absence of stimulus. Other methods could lead to somewhat different estimates of the jobs impact of the program in its first full quarter of operation. But the clear implication is, the program is working.”
The related footnotes say the following:
“12 These forecasts are based on a vector autoregression using the logarithms of real GDP (in billions of chained 2005 dollars) and employment (in thousands, in the final month of the quarter) estimated over the period 1990:Q1-2007:Q4. There are four lags, and the estimates are used to make projections beginning in 2009:Q2. Changes in the specification, such as using fewer lags and extending the sample through 2009:Q1, generally lead to projections of even slower recoveries of GDP and employment growth.
13 Moodys Economy.com. Prcis: U.S. Macro, July 2009, p. 6.”
Posner has a point in that Romer never addressed the other elements in the government’s recovery program. But this was not really an oversight for the simple reason that the timing makes them practically irrelevant to the econometric results for the second quarter. What are the other elements of the government’s recovery program? There are essentially only two elements: 1) Monetary Policy and 2) TARP.
With respect to monetary policy the Federal Reserve’s main policy instrument, the federal funds rate has remained where it has been since December 17 at 0%-0.25%. Any effect that it might be having should already be perceptible in the first quarter employment figures. The Federal Reserve has also engaged in some unconventional policies nearly all of which are reflected in its balance sheet. The Federal Reserve’s balance sheet surged by some $1.5 trillion in the last third of 2008, but since then it has contracted by $200 billion.
With respect to TARP by the end of the first quarter some $520 billion had been disbursed. Between then and the end of the second quarter another $100 billion went out but $70 billion was returned. Thus of the net disbursements by the end of the second quarter ($550 billion) approximately 95% had been disbursed by the end of the first quarter.
Romer probably should have brought up these facts formally. But I suspect that her audience was sufficiently well informed to be aware of these facts. Of the governments three largest recovery programs the only one that experienced a major change in terms of implementation was the Recovery Act.
It is also tempting to argue that since only an eighth of the Recovery Act had been spent by then of the second quarter that a similar proportion of its effect should be felt so far. But a discretionary fiscal stimulus is qualitatively different in its expected effects on the economy than TARP or the monetary policy measures taken. That is because its effects are proportionate to its flow rate and not its stock amount. The Recovery Act will be spent at a peak rate of $100 billion through the end of fiscal year 2010. Thus as of the second quarter it was already being spent at that rate. Nevertheless, due primarily to rising multiplier effects and changing composition its effects should continue to increase at least through the end of fiscal year 2010.
Naturally additional econometric analysis could be done taking into account the governments other recovery programs. But for the reasons i have just stated it is unlikely to show anything substantially different from the analysis that Romer and Mark Zandi have performed.
Now on to other points. Posner also writes the following:
“At this writing, it seems that economic growth is about to restart, yet more than $700 billion of the stimulus money remains to be spent. If economic growth turns out to be rapid, the effect of stimulus spending, on top of our huge deficits, may be to create significant inflationary pressures.”
The proper context for this worry is monetary policy. The primary concern here is will it be possible to spend the vast majority of the stimulus before a zero interest rate policy (ZIRP) is no longer desirable. Fortunately this has already been analyzed by several public and private economists. The best analysis on this outlook I’ve seen was by Glenn Rudebusch of the Federal Reserve Bank of San Francisco. Based on FOMC economic projections made earlier this year (which were excessively optimistic) he noted the following:
“The estimated Taylor rule can also be used in conjunction with economic forecasts to provide a rough benchmark for calibrating the appropriate stance of monetary policy going forward. The dashed lines in Figure 1 show the latest forecasts for unemployment and inflation provided by FOMC participantsthe Federal Reserve presidents and governors. (The dashed lines are quarterly linear interpolations of the median forecasts in FOMC, 2009.) Like many private forecasters, FOMC participants foresee persistently high unemployment and low inflation as the most likely outcome over the next few years. The recommended future policy setting of the funds rate based on the estimated historical policy rule and these economic forecasts is given as the dashed line in Figure 2. This dashed line shows that, in order to deliver a degree of future monetary stimulus that is consistent with its past behavior, the FOMC would have to reduce the funds rate to -5% by the end of this yearwell below its lower bound of zero. Alternative specifications of empirical Taylor rules, described in Rudebusch (2006), also generally recommend a negative funds rate.
The shaded area in Figure 2 is the difference between the current zero-constrained level of the funds rate and the level recommended by the policy rule. It represents a monetary policy funds rate shortfall, that is, the desired amount of monetary policy stimulus from a lower funds rate that is unavailable because nominal interest rates can’t go below zero. This policy shortfall is sizable. Indeed, the Fed has been able to ease the funds rate only about half as much as the policy rule recommends. It is also persistent. According to the historical policy rule and FOMC economic forecasts, the funds rate should be near its zero lower bound not just for the next six or nine months, but for several years. The policy shortfall persists even though the economy is expected to start to grow later this year. Given the severe depth of the current recession, it will require several years of strong economic growth before most of the slack in the economy is eliminated and the recommended funds rate turns positive.”
http://www.frbsf.org/publications/economics/letter/2009/el2009-17.html
So, in brief, ZIRP will be necessary for several years. Thus it is extremely unlikely that spending the Recovery Act while ZIRP is in effect will be at all a problem.
Finally, as someone who is very familiar with Dr. Romer’s research I absolutely disagree with the following statement by Mr. Indyk:
“The analysis she presented to The Economic Club of Washington was quite different from what she would have said had she remained a private academic and not an agent of the current administration.”
There is really no contrast at all between Romer’s academic work and her speech. I am not alone in thinking her speech was a continuation of her life’s work. Robert Waldmann said this and several other things today:
http://rjwaldmann.blogspot.com/2009/08/richard-posner-there-is-no-need-for.html
As someone who agrees with neither Posner nor yourself, I find his arguments relatively well presented and cohesive, and yours ruined by an immature snippiness.
Just one more point about Posner’s apparently thinking. He seems to imply that fierceness of opposition has something to do with size of estimate. More than once, he points out that some figure which he thinks supports his criticism of Romer comes from one of his “fierce” critics. Perhaps in his world, the extend of one’s opposition could alter the size of one’s estimate, but we should worry if it does. If he had offered estimates from his most fastidious critics, or from critics with the greatest analytic resources or the best record for accuracy, we should be impressed. We should not are one whit about the emotional state of his critics.
The fact that Posner apparently doesn’t understand the data he is offering in questioning Romer’s analysis is, of course, the problem most deadly to Posner’s argument, but how good can an argument be when it relies on distractions like “fierce”?
It is, to my mind, somewhat shameful to go from disagreement over the interpretation of data to questioning of ethics, as Posner has done, in what now appears to be a fairly casual manner. Posner has written about the temptation to slide into slapdash analysis when one becomes a celebrity pundit. He made the particular point that if one becomes known as a public commenter in one’s area of expertise, invitations are often extended to comment outside of one’s area of expertise, and that is the path to saying stupid things in the guise of expertise. What, exactly, does Posner think now separates him from his earlier target of disdain?
Tom: Thank you for your literary criticism. However, I would appreciate any substantive comment on the nature of the data analysis. In your mind, did or did not Mr. Posner make an error in mathematics, and interpretation of statistics? If there is a problem with the math I have presented, please explicate.
One question on the data analysis, how do the time periods line up? Your calculation of a 2 percentage point increase in GDP for the second quarter, is based on spending of $89 billion of the stimulus funds. The total stimulus bill was estimated to be $787 billion, so there is something like $675 to nearly $700 billion left to spend in the next couple of years.
The question is this, how much bang does the remainder buy? Is that about 2 percentage points per quarter, if you assume that the remainder is spent at approximately the same rate, or is something bigger? Looking at the WSJ forecast survey, 2 percentage points would account for almost all of the forecast GDP growth in 2010.
I’m curious, do you dispute Posner’s characterization of you as a “left-wing” economist?
DDearmont: See this CBO document for a partial answer to your question.
pwyll: Relative to economists in the law-and-economics area, I’m sure I am on the left. Relative to the total economist population, I doubt I’m on the left.
Is there a terminology discrepancy here? Stimulus expenditures vs tax-cuts. If we’re talking about the time-to-market for GVT expenditure it’s wrong to include tax changes as those are instantly reflected in withholding numbers.
But I don’t get how anyone can prove that the stimulus has led to any increase in jobs saved or created. If cash that went into the US Treasury market would have wound up going into something else, what would that effect have been on GDP.
If you visit the government web site tracking the stimulus (Recovery.gov), however, it will tell you that the government had paid out only about $60 billion by July 3. (You can find this figure in the chart at the lower right hand corner of the home page.)
Nancy: Yes, exactly, that’s what we all agree upon. But that figure does not include the tax rebates portion of the stimulus disbursements. That is the entire point of this post — pointing out that he missed that error after it was explicitly pointed out to him.
Nobody made any math errors, and you know that. The alleged math argument is really one of semantics: does the verb “spend” apply to tax relief? I think it can only if relief is in the form of a tax refund. In other words, assuming Romer was including tax relief in her “more than $100 billion” line, she misspoke when she used the word “spent”.
The more important argument is whether Romer had grounds to credit the rapid deceleration of the economic contraction in the 2nd quarter to the stimulus. I agree with Posner that Romer was, apparently with political motives, claiming to know something unknowable.
In your analysis, your math is correct, but your models are simple and arbitrary. You don’t explain why you think that an estimate of an extra 50 cents of consumption from every stimulus dollar is “low”, and you offer as the other end of your range the impossibility of every single stimulus penny being a penny that was spent on consumption that would otherwise have been saved.
PS Stepping back and looking at the bigger picture, arguments over the scale of GDP boost relative to the scale of stimulus miss the point: production is measured not only by quantity, but also by quality. And that quality is ultimately judged by the market. In other words, there is no point temporarily boosting GDP by increasing production of things for which there is no sustainable demand. This is where Posner is very wrong in arguing that this stimulus is poor because it’s low on public works.
I don’t think any serious person disputes that increasing deficit spending during periods of risk aversion boosts consumption and lifts GDP, to some extent. The folly of Keynesian economics is its belief that such boosts can initiate a growth cycle. What such boosts actually do is cushion the economy by sopping up extra capacity, while prolonging misinvestment and delaying the adjustment to new conditions needed to initiate a growth cycle.
Tom: You can’t be serious. Even if you didn’t accept the standard definition of the tax cuts as part of the stimulus, the math of dividing by annualized GDP is wrong, and there is no way to square that.
Incidentally, assume you have never heard of the term “tax expenditures”. I didn’t make up that term.
@Tom,
You wrote:
“What such boosts actually do is cushion the economy by sopping up extra capacity, while prolonging misinvestment and delaying the adjustment to new conditions needed to initiate a growth cycle.”
Based on my simple analysis, it’s my observation that the rate of growth in potential GDP is extremely dependent on the output gap. That is to say, growth in GDP per hour worked (productivity) seems faster when the economy is closer to “full employment.” I suspect that the cause of this is, in the case of full employment, the lack of the relative importance of of “hysteresis.”
In other words, in a liquidity trap, stimulus can minimize the loss of human capital, which is a very important determinant of future potential output.
Menzie – I’m serious. I believe you that there is a term “tax expenditure”, but that’s specialist jargon. The definition of “spend” means to disburse or use up, so canceling previously planned taxation is not “spending”. Certainly it is part of the stimulus.
Mark – Your models might well tell you otherwise, but productivity growth tends higher when and where employment is lower. Stimulus avoids short-term declines in employment and wages at the cost of long-term growth, which is why I call it “cushion”.
Menzie: I agree with Posner that comparing stimulus expenditures to 2nd quarter GDP is “half right”, for the reasons he gives.
@Tom,
I suspect you’re thinking about short run changes, as for example happened last quarter. The productivity data can often be very noisy.
If you look at the CBO’s series on potential real GDP what you will observe is much faster GDP growth during periods when unemployment was low relative to NAIRU (the Korean War era and 1964-1973 springs to mind). And when unemployment is high relative to NAIRU the opposite holds true. That’s not proof of course, but the pattern is rather striking.
My view on this is partly influenced by an old book (1957) called The American Economy by Harvard economist Alvin H. Hansen. The section of the book I am thinking of is aptly titled “The Case for High-Pressure Economics.”
Mark – No, I mean historic statistics on productivity growth and employment over the past several decades, in the US and elsewhere. I agree that productivity is difficult to measure and so the value of statistics is somewhat dubious. Potential GDP is theoretical so one could get any result for that, depending on one’s theories.