And why countercyclical macro policy is still needed — pro Summers, contra Barro.
As we contemplate the future of macroeconomic policymaking (or not), it’s important to recall the stakes. We have avoided the disastrous outcome that would arisen had the plan for further massive tax cuts aimed a high income households and extensive spending cuts been implemented; however, output remains far below full employment levels. Since 2008Q1 through 2012Q3, the cumulative output loss has totaled $3.75 trillion (Ch.05$).
Figure 1: GDP (black), WSJ survey forecast GDP (red), and potential GDP (gray), in bn. Ch05$, SAAR. Source: BEA, 2012Q3 3rd release, WSJ January 2013 survey, and CBO (August 2012).
A further $0.94 trillion loss will be incurred by the end of 2013, should the economists surveyed by WSJ be proven correct on average. This makes clear that we are far from a complete recovery. However, it is important to recall that the Administration did propose additional measures that would, had we implemented them, have reduced the gap measurably. Consider where we would be had the President’s American Jobs Act been passed:
Figure 2: GDP (blue), WSJ survey forecast GDP (red), and implied by Macroeconomic Advisers’ estimates of the American Jobs Act (green squares), in bn. Ch05$, SAAR. Vertical dashed line at 2009Q1 (passage of ARRA, aka “the stimulus package”). NBER recession date shaded gray. Source: BEA, 2012Q3 3rd release, WSJ January 2013 survey, Macroeconomic Advisers (Sept. 2011), and NBER.
Macroeconomic Advisers did not incorporate into their estimates the impact from the incentive effects on employment from the payroll tax credit for new hires, so in some sense, the MA estimate understates the impact. (For contrast, consider the “Real American Jobs Act”, which was backed up by Heritage Foundation statistical analysis.)
This is an obvious point, but it bears repeating: the slow growth we have experienced thus far is partly the result of insufficient fiscal stimulus (as well as partly a result of the aftermath of a balance sheet/financial crisis, a la Reinhart and Rogoff).
I think the Macroeconomic Advisers estimate might understate how much we could have improved the situation for another reason: that is because recent evidence suggests the fiscal multiplier is larger than we thought on the basis of pre-crisis data. I’m thinking of this as I have just been writing a survey of fiscal multipliers, and one of the interesting results that has come out in recent research is the asymmetry in responses to fiscal policy conditional on the state of the economy. Auerbach and Gorodnichenko and Fazzari et al. have tackled this issue from the expansion/contraction distinction, as discussed in this post.
But I think that the most relevant finding is from Baum et al. (2012), regarding effects conditioned on the sign of the output gap:
Excerpt from Figure 2. Cumulative Fiscal Multipliers: Fiscal Expansion from Baum et al. (2012). Shaded bars indicate statistical nonsignificance.
The blue (red) bars pertain to cumulative multipliers four (eight) quarters out. The left hand side bars pertain to positive spending shocks, the right hand side bars pertain to negative revenue shocks. The bottom line is that positive fiscal shocks, particularly when the output gap is negative, are particularly expansionary.
Obviously, the prospects for further fiscal stimulus are not auspicious. Nonetheless, these results are important, if only they highlight the fact that previous estimates based on homogeneous responses to fiscal policy, might lead to misleading conclusions when applied in the current macroeconomic environment.
But more importantly, if expansionary fiscal policy is depends on the state of the economy, it would not be surprising to find contractionary fiscal policy to also be state-dependent. This is what Baum et al. also find:
Excerpt from Figure 3. Cumulative Fiscal Multipliers: Fiscal Contraction from Baum et al. (2012). Shaded bars indicate statistical nonsignificance.
The bars on the left pertain to negative spending shocks, on the right to positive revenue shocks.
Hence, we must be very careful about attempting too much fiscal consolidation right now when the output gap is so negative (further confirmation of this general principle is provided by the recent paper by Blanchard and Leigh (2013)). All something to keep in mind as certain groups demand sharp front-loaded spending cuts.
I anticipate hearing something about crowding out from the usual suspects; these fears have been abounded for the past four years. I will merely observe that if crowding out occurs via elevated interest rates, evidence is somewhat lacking for that thesis.
Figure 3: Ten year constant maturity yields on TIPS (blue) and real interest rate on ten year constant maturity Treasurys (calculated using ten year constant maturity Treasurys and expected ten year CPI inflation from the Survey of Professional Forecasters (red squares). NBER defined recession dates shaded gray. Source: St. Louis Fed FRED, Survey of Professional Forecasters, NBER, and author’s calculations.
How many more years into this centrally planned fiasco before we stop calling it “cyclical?”
W.C. Varones: Thank you for your typically insightful and comprehensively reasoned comment. You are nothing if not consistent.
Hi, Prof. Gorodnichenko is my adviser and I’m glad to see his research featured here. At Berkeley, a huge portion of the first and second year macro sequence is devoted to thinking about how to estimate various multipliers. Another paper worth mentioning here is the Romers’ estimates of tax multipliers around 3 (http://emlab.berkeley.edu/~cromer/RomerDraft307.pdf).
Question for you, Menzie: how much difference is there between the multiplier for a positive consumption spending “shock” and the multiplier for a positive investment spending “shock”?
In response to figure 1. I would like to give my view of it taken from my work. I have prepared a one-page explanation with a graph. Link…
https://docs.google.com/file/d/0BzqyF_-6xLVEdlhJTWg4UUF4RDQ/edit
The basic idea is that we are at a new low-level trend line for potential real GDP. As such we will eventually have a contraction that takes real GDP even lower compared to the official potential real GDP trend line if the new low-level line holds steady.
Greg Hill: In the Baum et al. paper, the US fiscal spending variable is government consumption and investment, so the effect is the same. Typically, in other papers, the multiplier for investment is larger.
I’ve been thinking about the issue for other reasons and would like to suggest one other reason why fiscal stimulus effects might be greater than expected: economic geography. The Republicans in the House who are blocking all stimulus measures except tax cuts are in primarily rural areas or smaller cities, or sometimes exurban areas. If stimulus money were equally distributed by Congressional district, the inner city districts where money was spent would see people walking, taking the bus, etc., to spend money. So the consumption multiplier would be high: the recipients in general would be doing something similar and leakages would be low. For the exurban or rural districts, the same money would have to pay for gasoline or diesel for the same purchases. So you have an automatic leakage of a substantial percentage to Saudi Arabia and Venezuela, and the consumption multiplier must ceteris paribus be lower. Any current stimulus is likely to be more urban-oriented, and so the implied estimates from previous equal distribution will overstate the leakage.
So far Central Banks balance sheets have expanded up to 18 Trillions USD, Everything remaining the same,the Federal Reserve will purchase monthly throughout 2013,85 billion USD worth of financial papers ,extending the central bank balance sheet to around 4 trillion USD (source Bloomberg Fed Debated QE End in 2013 Amid Concern Over Total Assets) that would more than match in amount the output gap of 3.5 Trillion USD as mentioned in introduction of this post. Still left more than 14 Trillions USD displayed among other Central Banks. The real output gaps have not been lifted up and Japan experience is or should haunt the strategists of the Bernouilli paradox.
Pr Ramey paper is still convincing as the empirical data show a switch from private investment to public investment followed by a very weak multiplier.This multiplier alone,does not carry much hope in Jobs creation.
Since year 2000 Central Banks are filling the output gap. The results are financial,this year again Banks and financials have outperformed the rest of the other sectors, when it comes to their capital gains gaps in the equities markets.
Central Banks have become the largest hedge funds, they play the markets,they do not have to worry about the Value at Risk and to their exit strategies but private individuals should.
As we may notice a security bears,three components the risk, the yield and the time to maturity.
Edward Lambert,thanks for the link to your study as it is proved to be usefuL
Thanks for the very timely round-up of recent research on fiscal multipliers. While I take comfort from that fact that minor changes in methodology produce similar conclusions, I have some doubts about the Baum et al. study. That study focuses on the output gap (which is unobserved) rather than GDP growth. In the authors’ words
“The reasons to employ the output gap instead of the GDP growth rate are manifold. The
output gap is the measure most commonly used to identify economic cycles, as it is seen not
only as reliable ex-post but also as a reliable real-time indicator for policy-makers. [p. 6]”
Instead, both of those statements are generally acknowledged to be false. The authors show the impact that fiscal policy would have if we knew the revised value of the OECD output gap. However, at best we have only forecast (and nowcast) output gap estimates when fiscal policy is decided. This makes it likely that the differences they find overstate the differences facing policymakers.
Barro’s argument seems to boil down to, the less government spending the better, even if this does cause a short run recession, and besides those are “Keynesian models that always predict that GDP expands when the government gets larger.”
Funny thing is, that’s exactly what the data above from Baum seems to be saying. They are showing multipliers of 1 or more for the US, even with a positive output gap. Though looking at the Baum paper, this only seems to be true of the data for the US and Japan.
But this isn’t generally true of Keynesian models. This shouldn’t be the case if you get crowding out when you have a positive output gap. And I believe Keynes himself advocated reducing government spending with a positive output gap.
reply to ppcm…
Could I ask for a brief comment on what part of my study proved useful?
Yawn… we’ve been running trillion dollar “stimulative” deficits for a few years now, along with the fed monetizing a couple trillion to no real effect. When will you guys give it up. “It would have been worse” is not an acceptable rebuttal because you simply do not know that.
Edward:
1. I don’t agree with your divergence from official potential GDP. I think potential GDP is much closer to what Menzie is showing on his graph above.
2. Your argument seems to be that declining labor share will reduce potential GDP. But there is no evidence that declining labor share has led to declining personal income. Instead, personal income has held steady as a percent of GDP, as increases in personal transfer payments have offset falling compensation of employees.
RPI
PCTR
Compensation & PCTR
In short, I think the fall you are seeing in labor share since 2009-2010 is coming in part from the jump in personal transfer receipts (from 12% to over 15% of GDP) which occurs at about the same time. Much of that should reverse as recovery proceeds.
The Jobs Act was to cost $447 bn. Your graph shows GDP $200 bn higher with it than without it, ie, you seem to be suggesting the Act would have destroyed half its value.
The core of the Act was a payroll tax holiday. Does that mean you think Democrats should not have voted to end the payroll tax holiday?
Your chart shows 2013 GDP about $400 bn higher than its previous peak in 2008.
I have to admit my sense is that GDP is in fact lower than it was in 2007.
Some stats (courtesy CR):
– Employment today is 4% below its previous peak
– Vehicle sales are 2 m lower (14 v 16 m) than in a decent year (about 15% down)
– new homes sales are 2/3 lower than the previous peak; about half of normal
– Industrial production is about 2% below the previous peak
– Imports are near previous peak
– LA ports imports 1/5th below previous peak
These statistics reflect the economy as I perceive it subjectively. The country is not as prosperous as it was in 2007.
GDP may be higher on paper, but it doesn’t correspond to the daily reality I see.
reply to acerimusdux…
Are you implying that capital income has not increased? here is a quick graph of real corporate profits after tax divided by real GDP…
https://research.stlouisfed.org/fred2/graph/?graph_id=104326&category_id=0
The tax transfers you are seeing must not be transfers from capital income to labor income… they must be labor income to labor income.
My study talks about the divergence between capital and labor income. and if capital share of income is increasing as a share of GDP, labor share of income must be decreasing, since as percentages they add up to 1 together.
My study seeks to show the effects of increasing capital share of income in a growth model of increasing real GDP.
So let’s if I understand the concept of a “multiplier” correctly.
In business, the multiplier would be the equivalent of return on invested funds, if I understand correctly.
If you spent your money on pure consumption, the multiplier would be zero. In economics terms, the multiplier is zero if crowding out is complete, ie, if, say increased government spending is fully offset by decreased private consumption. This would occur if incremental debt is not allowed.
If the multiplier is one, then GDP increases by as much as the stimulus, say, government spending. In this case, the stimulus pays for itself through higher than average growth rates in the payback years.
There appear two version of this as measured: gross and net. The gross version looks only at the stimulus. It does not consider the repayment of any associated debt. The estimates for the gross number vary all over the board. For Ilzetzki et al, these are initially estimately at 0.8 and 0.2, long term for developed and developing countries respectively.
So, for $1 of government spending, 80 cents shows up in increased GDP over time (cumulatively). And then you have deduct the cost of debt to obtain the net impact on GDP. Cumulative multiplier effect seems to max after 4 years, so at 4% interest, that’s about 17% cost of capital total.
Thus, one dollar of government spending borrowed from aboard would increase GDP by 63 cents, or a capital “loss” of about 3/8th of the funds borrowed.
Could I believe that? Yeah, maybe.
No doubt Slugs will have a comment.
Steven:
I think your perception is basically correct. The real GDP number Menzie is using adjusts for inflation, but not population growth. Per capita it’s still lower than 2007. So overall output is still up partly due to population growth. In addition, some of the increased production has gone into inventories, which are above the previous peak.
The biggest positive contributor seems to be that exports are up, by over $500B on an annualized basis from their low (imports are up only 1/4 as much from the prior peak). But what are we exporting, if production hasn’t fully recovered?
At any rate, it looks to me like businesses think we are in the early stages of a recovery, and are counting on consumer spending to have more upside here (hence the increases in inventories and business investment). It will be interesting to see how that plays out.
Endogeneity, thy name is macro.
The Baum et al paper doesn’t prove anything. It’s only identifying a correlation of spending increases coupled with recessions in which recessions cause the spending increases because of social welfare programs. I suspect that the model is picking out two features of the data: 1) As Ed Lazear has tried to teach us (although some refuse to learn), deep post war recessions, other than Obama’s non-recovery, have tended to be followed by very strong recoveries; and 2) Countries like the UK and France have tended to raise spending more than the US during the downturns.
Thus, the Baum et al model looks at the ratio of subsequent growth versus social welfare spending in the U.S. during the regime with a large output gap and sees a “multiplier” because the recovery from the deep recession was strong. However, social welfare spending went up faster in the UK and France for an equivalent recovery in the large output gap regime, and so their ratio, “the multiplier” looks lower. But spending didn’t cause the recoveries. Rather the recessions before the recoveries caused the spending. The “multipliers” estimated in the paper aren’t real.
Barro is of course right in his Project Syndicate article. But we need to listen to him on the multiplier issue too. Barro has emphasized the need to find exogenous spending increases to test the value of the multiplier, which is why he has focused on military spending. We need natural experiments.
Come to think of it–didn’t we just have one? The Obama stimulus program was a large exogenous fiscal policy action done when interest rates were near the zero bound and the output gap was very large. What happened to the recovery?
Move along–nothing to see here.
reply to acerimusdux…
Sorry about my graph above. I mixed nominal and real. It was a long ride home, because I realized my mistake as I was leaving work.
Here is the corrected graph with …
https://research.stlouisfed.org/fred2/graph/?graph_id=104326&category_id=0
You still see that after tax profits (capital income) has been increasing. This implies that labor income as a share of GDP has been declining.
It is important to understand what potential GDP is… I have seen it confused with productive capacity (100% capacity utilization).
I would clarify like this. Potential GDP refers to a state where the economy is functioning at the natural rate of unemployment. One must use the natural rate at the center of the business cycle. My work shows that this rate was 5.8% since 1967. My equations now show it has increased to 9.4% since 2010. That approximated rate will change as more data comes in.
But unemployment is now below that rate. Thus, my equations show that the current real GDP is actually above the current potential real GDP.
Edward,
No, corporate profits are up relative to wages or personal income, but really only over the last decade. In the graph you posted, you made one error there, in that the CPATAX series is not inflation adjusted, it’s a nominal measure not a real one. What you wanted was this. Still a sharp recent rise. But look also at this. There, corporate profits are up relative to personal income, but the ratio is only back to about where it was around 1947-1951. And unemployment was low then; it actually jumped to over 7% in the 1949 recession, but then recovered quickly to below 3%. So we are not really in any uncharted territory here.
Most importantly though, you will see on any of these graphs that profits tend to rise early in a recovery, but wages increase in the later stages. Look at wages to profits for example. If we have an extended recovery, continuing years after profits have peaked, then wages will recover, as you can see they did in the 1990s under Clinton, or the 1960s under Lyndon Johnson.
But that won’t happen if policy makers think we are already near to potential GDP, and thus cut off fiscal or monetary policy prematurely. So mainly, I would like to see the Fed keep a foot on the accelerator. And even when inflation does begin to occur, I’d prefer it be dealt with first through spending cuts (as I think that can help extend a recovery).
Rick Stryker: You write:
Perhaps I am not understanding what Baum et al. write on page 8 and in the appendix that for the US; they state it seems pretty unequivocally that the fiscal (spending) variable is government consumption and investment. Hence, I don’t understand your point about rapid increases in social welfare spending. In point of fact, those are included in net tax revenue, and those multipliers are not statistically significant. I suggest you re-read the paper. Geez.
@Simon van Norden:
A little offtrack: I would like to read your opinion on the IMF/Blanchard analysis of Growth Forecast Errors and Fiscal Multipliers. It is based on structural balances, which are based on output gaps. Could the problems with output gap estimations somehow influence the results they get?
Thank you
Edward Lambert:
“The basic idea is that we are at a new low-level trend line for potential real GDP. As such we will eventually have a contraction that takes real GDP even lower compared to the official potential real GDP trend line if the new low-level line holds steady.”
Statistically, what evidence is there that your new low-level trend line will hold steady?
Menzie,
You have not addressed my point, which was about endogeneity. I did not say “rapid increases in social welfare spending” as you asserted but rather that increases in spending motivated by social welfare programs went up faster in welfare states such as France and the U.K. I am not talking about transfer payments and I’m fully aware that the paper uses government consumption and investment.
The reality is that variations in government spending are too small to infer anything about the multiplier. Government spending usually just goes up. But I’m suggesting that it could go up faster in welfare states whose governments already purchase a larger percentage of GDP than does the U.S. during recessions, particularly during severe recessions and their aftermath. That, coupled with differences in real GDP growth during robust recoveries could explain why the U.S. looks like it has a multiplier while France and the U.K. don’t. Thus, my point is about differences in rates of growth of spending relative to difference in rates of growth of GDP growth during a robust recovery.
The key point here is that this spending is endogenous. In the U.S. for example, state and local government spending dominates non-defense spending. That spending is highly dependent on revenues which in turn depend on the state of the economy.
Barro focuses on defense spending alone, because it’s most plausibly exogenous to the state of the economy, and on WWII, because that’s when variations in defense spending are large enough to measure anything. That’s the only natural experiment we have.
Because the spending is endogenous, Baum et al haven’t proven anything. I was trying to explain how they might have come up with the result that there is a significant multiplier for the US but not for the UK or France. The fact that the results are so different for these countries should be a clue that something might be wrong.
On that topic here’s a question for you. Paul Krugman has been thundering against what he calls the “Austerians,” blaming UK austerity programs for the countries macro outcome. But if you really believe this paper, you’d have to say he’s wrong. Are you willing to say that? Are or you cherry picking the results? If you really think this paper’s results are real, how do you account for the multiplier differences between the US, the UK, and France?
reply to simon van norden…
As far as support for a new potential trend line holding steady, my analysis uses the constraints of UT staying positive in the equation…
UT^2 = unemployment – capacity utilization + (0.78 * labor share, 2005=100)
UT is total unused “available” factor capacity. (non-negative)
According to the equation, capacity utilization is now capped around 80% and wouldn’t go higher than that. There is a lower limit for unemployment too. These constraints effectively “maintain” potential real GDP at a lower level.
My work does show that the natural rate of unemployment has increased. I commented above that this means that potential GDP has shifted down too. This is because potential GDP corresponds to a natural rate of unemployment.
We saw a guest post here by Laurent Ferrara and Valerie Mignon a few weeks ago that showed statistically that the natural rate of unemployment has risen 2.7%. Link…
https://econbrowser.com/archives/2012/12/guest_contribut_27.html#more
In regards to this guest post and my work…
Once one accepts that the natural rate of unemployment has in fact risen to a new “stable” level and isn’t showing signs of returning to its former level, then in my opinion, one should say that potential real GDP has fallen to a new “stable” level.
I’m not sure if anyone knows what level the “natural rate” of unemployment is currently at. But I think the current data makes it clear enough that anyone who had a number higher than 8% in the pool has already lost on their bet. Better luck next time.
The (in)famous Barro paper has a lot of well-known problems; e.g., he used annual NIPA data rather than quarterly, he did not account for monetary policy reactions, improbable symmetric consumer responses, etc. But one problem with his study that hasn’t gotten a lot of attention is his misinterpretation of what defense spending on line 23 of Table 1.1.6 actually means. Line 23 is based on internal DoD financial accounting data that captures Defense Finance Accounting System (DFAS) expenditures and receipts. In other words, “expenditures” is equivalent to “disbursements.” For the laymen (and apparently for many academics) “disbursements” is a synonym for “spending.” And this is probably good enough for most non-defense expenditures. But if you want to talk about defense “spending” as fiscal stimulus, then you should be looking at “budget execution” rather than disbursements. Budget execution basically means obligating a contract with a vendor. Obligations are what stimulate economic activity, not disbursements. Disbursements don’t occur until the final product is delivered and basically reimburse the vendor for costs incurred during the production process. It’s at the point of budget execution that vendors line-up bank loans, start buying materials and parts, start hiring, etc. For the private sector and even for most government spending outside DoD (and even O&M funding within DoD) equating disbursements with economic stimulus is close enough. But it’s not good enough when it comes to buying spare parts, buying new weapon systems, and military construction projects. For example, some event might prompt Congress to build more aircraft carriers, so Congress authorizes XXX gazillion dollars ($XXXGZ). It might take up to five years to execute that $XXXGZ (it varies by funding type). And the law allows for another five years before disbursements, and it’s disbursements that show up in NIPA Table 1.1.6, Line 23. So the actual economic stimulus might be up to five years after Congress authorizes the funding, and the NIPA tables won’t reflect those expenditures until ten years after Congress authorized the spending and five after the economic stimulus event. This is a big problem with Barro’s analysis. It’s also a problem with some of Valerie Ramey’s studies, although to her credit she does try to ameliorate some of this with her news event approach.
Interestingly, as part of ARRA the White House posted budget executions rather than disbursements. That was the right thing to do despite the fact that it created all kinds of consternation with some conservatives who insisted that the website should post disbursements to its ARRA website tracker.
Rick Stryker: Gee, you will forgive me for mistaking your exact words “However, social welfare spending went up faster in the UK and France for an equivalent recovery” as “rapid increases”. Oh, well. Please nitpick away.
Now, regarding Barro, wasn’t there something called “rationing” during WWII? Or is this a “no Lucas Critique” zone?
With respect to the small multipliers for UK and France, I’ll merely observe that those are the countries where the no threshold null is rejected at 10% and 5% levels (rather than the 1% level for the US, etc.)
Finally, I still don’t understand your point about endogeneity; Baum et al. use a Blanchard-Perotti type of structural VAR, and adjust the net revenue variable for the endogeneity you describe. So what exactly is your alternative (aside from Barro’s approach which I have a suspicion is sensitive to sample, since he never starts a sample post-1955)?
Steven Kopits: Please, please, please, read an intermediate macro textbook. Barro, or Williamson, if you like. At least then we would have a common point of reference. (And no, I don’t think of a multiplier as a RoR.)
Simon van Norden: Point well taken; I’m more comfortable with the idea that there is heterogeneity in effects across growth rates, but we know that those are also revised — maybe less than output gaps.
Menzie,
The Blanchard and Perotti identification procedure assumes that feedback from output to government spending is zero and that output affects taxes only automatically through features built into the tax code. In plain English for people who might be reading this comment, they are assuming that government spending can affect output but output does not affect the level of government spending. And that assumption is built into Baum et al.
But that’s just the point I’m raising. It’s hard to believe, for example, that state and local government spending is not affected by tax revenues which are in turn dependent on the level of economic activity. Baum et al, by using the Blanchard and Perotti procedure only partially adjust for endogeneity and assume away the really important question as to the extent to which the level of government spending is affected by changes in output.
I’m not sure what your point is on France and the UK. Mine was that the estimated multipliers are substantially lower than 1. If you really believe that, then Krugman has been blowing smoke with his Austerian rants and you should call him out on it.
Also, you are mistating my point yet again by referring to “rapid increases.” It’s not nitpicking. To restate, I was referring to relative differences in growth rates of endogenous spending between the US and the UK and France as a possible explanation for the differences in multiplier estimates.
I was not attempting to defend Barro’s conclusions by mentioning him (although I do think he’s right.) Rather, I was offering him as an example of someone who tries to take the endogeneity problem seriously.
Rick Stryker: Mebbe. Looking at actual (BEA reported data), q/q growth rates of state and local government consumption regressed on q/q GDP growth (Ch.05$) yields slope coefficient of 0.21, not significant at 10% msl.
Obama’s Global Warming junk science being debunked more and more by the day :
http://online.wsj.com/article/SB10001424127887323485704578258172660564886.html?mod=WSJ_MostPopular_US
The sad thing is, this prevents us from finding real energy alternatives. Most people have lost interest in the scam, but Menzie is one of the last holdouts among the true believers.
reply to acerimusdux…
“…anyone who had a number higher than 8%…”
My study shows two rates of natural unemployment. They represent different economic states. One rate (around 9%) matches up with potential GDP. Another (around 7%) matches up with 0% available factor capacity.
If you are talking about how low unemployment can go, that is the lower of the two rates. If you are talking about the rate matched to potential GDP above and below which the business cycle moves, that would be the higher rate.
Clearly you are referring to the lower rate. As such, my approximation of 7% is still below 8%.
In this brief, the rates are distinguished in graphs #4 and #5…
https://docs.google.com/file/d/0BzqyF_-6xLVEWUdJb2wxamJibW8/edit
The y-intercepts are the points of reference for the natural rates of unemployment. According to the study, the higher recent rates will hold steady if labor share of income does not rise enough.
JanesK:
I have not yet read the Blanchard and Leigh study. My understanding is that they regress actual forecast errors on forecast changes in structural deficits, arguing that there should be a correlation of zero if their model, particularly their estimates of the multiplier, is correct. Instead, they find that those countries where structural deficits were forecast to shrink the most tended to grow less than forecast. I agree that this is what should be expected if the fiscal multipliers that were used to make the forecast were too low.
To the extent that structural deficits are mismeasured, I would normally expect that to make it harder to find a relationship between “noisy” forecasts of structural deficits and economic performance. Weakening their results requires, I think, a link between growth forecast errors on the one hand and the measurement errors in their structural deficit forecasts on the other.
But let’s consider the case where the IMF simply overestimated potential output in many countries. We would expect that such overestimates would (on average) lead to overly optimistic forecasts of growth. They would also cause structural deficits to be underestimated. Over time, as more is learned about the original (2008) level of potential output, we might see that overly optimistic growth forecasts tended to come from countries where estimates of structural deficits were subsequently revised upwards. However, I don’t think that is what Blanchard and Leigh have done.
Rick Stryker I’m not sure what your point is on France and the UK. Mine was that the estimated multipliers are substantially lower than 1. If you really believe that, then Krugman has been blowing smoke with his Austerian rants and you should call him out on it.
First, the signs are correct for the statistically significant multipliers, which is more than can be said for Barro’s approach. Second, even if you don’t believe an expansionary fiscal policy in a negative output gap environment, that does not mean France and the UK should pursue a contractionary policy. Contractionary fiscal policies still make things worse even if you don’t believe expansionary policies would make things better. Third, you might want to look at the threshold values for France and the UK as well as the percent of time those countries were operating in a low output regime. That alone probably explains most of the small multipliers.
I was not attempting to defend Barro’s conclusions by mentioning him (although I do think he’s right.) Rather, I was offering him as an example of someone who tries to take the endogeneity problem seriously.
So ignoring a monetary policy response is your idea of “taking the endogeniety problem seriously”?
Edward, the rate associated with potential GDP, above and below which the business cycle moves, is currently around 4.5%.
And when we move below that rate, you’ll know, because we’ll start to see some inflation, and especially rising wages.