The President laid out a series of policy measures in today’s speech which are, by textbook standards, entirely reasonable. And yet, many have been declared by the pundits to be DOA. I’ll leave the assessment of political feasibility to others, but the very fact that these specific measures [0] are so reasonable by textbook standards makes me wonder if we have in fact experienced technological regress in our politico-economic discourse. Maybe those shocks in RBC models are just the fact that so many individuals with influence never took an intermediate macro course, let alone an economics course [1] (I highly recommend Robert Hall and John Taylor’s Macroeconomics, or the later editions, by Hall and David Papell).
The Context: The Stimulus Package of 2009
To begin with it’s useful to review a little history, given the heated rhetoric that has been used in the past few years. From Chapter 5 of Lost Decades (forthcoming 9/19, W.W. Norton), written by me and Jeffry Frieden:
The standard macroeconomic view is that in recessionary conditions,
incremental government spending and tax cuts can stimulate
the economy. If the government spends an additional million dollars
to build a bridge, that spending directly adds to GDP. But that is only
the beginning: the spending on materials and labor is income to
suppliers, contractors, and workers, and some of this income will be
spent on consumer goods and services, which then further increases
GDP. This in turn becomes income for other workers, who similarly
increase their spending, again adding to GDP. This process suggests
a “fiscal multiplier,” typically associated with Keynesian macroeconomics,
such that every one-dollar increase in government spending
results in a greater-than-one-dollar increase in GDP. Especially
when the economy is stuck at ZIRP, so that private borrowing and
spending are particularly weak, the multiplier can be large…
Even before the new administration took office, its economic team
had been considering a fiscal stimulus. Romer’s evaluation indicated
that a $1.2 trillion package was needed. However, President-elect
Obama’s political advisers insisted that this was not feasible, and
the numbers were shaved. Once in office, President Obama proposed
a $675–$775 billion package to stimulate the economy.32
Intervention here: When I hear statements that “the stimulus failed”, I think it important to recall that between the election and inaugural day, business economists revised downward their estimates of GDP [2]. The proper metric is then to compare against the counterfactual as of the time of implementation, as described in here, here and here.
…
Some policymakers and economists believed that any government
intervention, even in so troubled an economy, was unjustified.
Congressman Ron Paul (R-Tex.) complained that “the US
government just won’t allow the correction the economy needs.” He
invoked the recession of 1921, which was deep but short, in Paul’s
view because the government permitted insolvent companies to fail.
“No one remembers that one,” he averred. “They’ll remember this
one, because it will last 15 years.”34 Paul’s view was reminiscent of
the position of the “liquidationists” of the early 1930s, who were led
by Treasury Secretary Andrew Mellon. Mellon’s advice to President
Herbert Hoover was typical: “Liquidate labor, liquidate stocks, liquidate
the farmers, liquidate real estate … purge the rottenness out of
the system.”35 The idea was almost moralistic: bad loans, bad debts,
bad businesses, and bad deals had to be exorcised before the economy
could right itself. Satisfying as such a scorched-earth policy
might be (at least to those not caught in its path), few serious economists
or policymakers ever considered it.36
Some theoretical objections to an active fiscal stimulus were
based on the view that government spending would inevitably be
wasteful, providing no real benefit to the population. Others emphasized
the expectation that increased government spending would
be counteracted by an offsetting reduction in private spending.
The budget deficits that would result from tax cuts or more government
spending would drive interest rates higher and reduce, or
“crowd out,” private investment and spending. The economists who
held these positions were generally hostile to traditional Keynesian
views, especially because of the Keynesian inclination to assume
that there were market failures that government could correct…
But among most economists, there was a general consensus on the
desirability of some form of active fiscal policy. In a February 2009
Wall Street Journal poll of economists, 68 percent said that the proposed
stimulus package was about the right size or too small. Only
31 percent said that it was too large. Most economists in the policy
and business circles viewed a stimulus package as something that
could soften the blows of a deep downturn and hasten the arrival of
a recovery.38
Well, we know about how much interest rates have risen in response to the stimulus package [3] [4]. And with that decline in long term interest rates, we know how much interest-rate-induced crowding out of investment has occurred: nil. Instead, the drama we seem to be witnessing now is a replay of 1937 [Krugman], when policymakers overly worried about inflation and deficits withdrew stimulus too early. Whether we will (re-)learn that lesson is the question of the moment.
The President’s Proposals: The Textbook Approach
In a standard aggregate demand/aggregate supply model [5] [5a] [6], output is demand determined in the short run, and supply determined in the long run. The President’s proposals focus, appropriately, on the short run, given the large output gap.
Figure 1: Log GDP (blue), potential GDP from CBO (gray), cubic trend in log GDP estimated over 1967-2011Q2 (gray-green), and forecasted GDP according to mean of August WSJ survey (red) based on July GDP release, all in Ch.2005$, SAAR. Source: BEA, CBO, WSJ, author’s calculations.
Using the CBO’s measure of potential, the lost output has been $2.8 trillion (Ch.2005$) through 2011Q2. Using the WSJ mean forecast, another $1.4 trillion will be lost by 2012Q4. The CBO-implied output gap as of 2011Q2 is 7.1% (log terms). Using a cubic in time trendline, the gap is still 3.4% (and then one has to believe that output was above potential 3.3% in 2007Q3).
Extended unemployment insurance, extension of the payroll tax holiday [CBPP], and infrastructure spending are all means by which aggregate demand can be sustained. To the extent that extended UI and payroll tax holiday benefit lower income/liquidity constrained individuals, the marginal propensity to consume is relatively high and hence the multiplier fairly large. Investment in infrastructure also makes a lot sense given the multiplier is fairly large for direct spending. The multiplier ranges are depicted in the below table (they are from a January 2010 report, so the dates relate to proposals implemented in 2010).
Source: CBO (2010).
I’ll note these ranges pertain to a situation where the Fed is assumed to tighten a year after implementation. To the extent that the Fed has committed to keeping the Fed funds rate close to zero to mid-2013, the relevant upper end of these ranges is probably higher.
Transfers from the Federal government to the states also has a high multiplier. Since the states are typically constrained in terms of borrowing, either due to constitutional or credit constraints, it makes sense for the Federal government to share revenues with them during times of fiscal stress.
The policies also make sense from a supply side perspective. Here I want to distinguish between faux supply siders, who think supply can only be augmented by reduced tax rates or reduced regulatory burdens, and those who have a broader interpretation of what factors determine supply. Clearly, investment in infrastructure increases the productive capacity of the economy. But some of the other measures also work on that margin. So when the President aims to retrain workers, or keep employed workers who otherwise might undergo long periods of joblessness, this is working on the supply side by enhancing human capital. (See these posts and conference proceedings regarding the impact of long term unemployment [7} [8] [9] [10]).
From the Public Finance Textbook
The employment tax credit is not straight out of the macro textbooks, but the rationale can be found in most standard public finance textbooks, specifically the analysis of an investment tax credit (ITC). An ITC reduces the cost of capital facing the firm, increasing the optimal capital stock. The cost per unit of additional investment is in principle low because only the marginal investment spending requires tax expenditures; there is no rent going to other units of capital already in place. The logic for employment is the same.
Summing Up
The President’s proposal will probably not have an enormous impact on GDP (in principle, it should have been larger, but I bow to political realities), although the estimates vary since the details are still coming out. Macroeconomic Advisers guessed about a percentage point acceleration, more than a week before the speech (they are to have a more specific estimate soon). Mark Zandi from Moody’s, with more details at hand, estimated 2 ppts acceleration relative to baseline, according to Bloomberg. What perhaps is of key importance is that these measures prevent the economy from falling below stall speed. [11]
Addendum: Replaying 2008?
On a slightly different note, I note that the opposition to adequately funding the new consumer financial protection bureau, and confirming the new head [12], seems to signal an intent to let the financial sector “self-regulate”. Well, that turned out well. Once again, from Chapter 8 of Lost Decades, our assessment of why the financial crisis was so devastating:
… Lawmakers disarmed financial regulators,
who in turn used few of the weapons left in their arsenals, allowing
financial institutions to develop new instruments that were largely
untested and wholly unsupervised. Financial institutions worked
madly to increase their profits in a low-interest-rate environment by
taking on ever riskier assets, insisting that they had mastered risks
they barely understood.
In our view, the argument that one can “just say no” to future bailouts, and eschew regulation represents a willful misreading of the lessons of history, or a deliberate attempt to obscure the buildup of contingent liabilities that will be paid by ordinary taxpayers, while spewing platitudes about free markets — in other words, and excuse for Akerlof-Romer type “looting” [13].
Update, 9/9, 7:30AM Pacific:
Estimates suggest a more substantial boost than early assessments (including mine). From Macroeconomic Advisers:
American Jobs Act: A Significant Boost to GDP and Employment
We estimate that the American Jobs Act (AJA), if enacted, would give a significant boost to GDP and employment over the near-term.
- The various tax cuts aimed at raising workers’ after-tax income and encouraging hiring and investing, combined with the spending increases aimed at maintaining state & local employment and funding infrastructure modernization, would:
Boost the level of GDP by 1.3% by the end of 2012, and by 0.2% by the end of 2013.
- Raise nonfarm establishment employment by 1.3 million by the end of 2012 and 0.8 million by the end of 2013, relative to the baseline.
- The program works directly to raise employment through tax incentives and support to state & local governments for increasing hiring; it works indirectly through the positive boost to aggregate demand (and hence hiring) stimulated by the direct spending and the increase in household income resulting from lower employee payroll taxes and increased employment.
Because the package is some $100-$150 billion larger than the proposal widely reported in the press and that we wrote about two weeks ago, these effects are expected to be significantly larger than previously expected.
This simulation did not incorporate potential incentive effects on employment from the payroll tax credit for new hires.
Studies have found that such credits do incent increased hiring.
- However, it is difficult to know the employment base of the firms eligible to receive the credit, hence we could not form an estimate of the aggregate employment gain
- That we did not allow for these effects suggests some upside risks to these employment estimates presented here.
Because these initiatives are planned to expire by the end of 2012 — except for the infrastructure spending, which has a longer tail — the GDP and employment effects are expected to be temporary.
That is, these proposals will pull forward increases in GDP and employment, not permanently raise their level.
Nevertheless, there may be good reasons to want to implement such programs today, if the government can follow through on the commitment to trim deficits later:
There remains considerable slack in the economy and nearly all forecasts anticipate only a gradual decline in the unemployment rate over the next couple of years.
Given the elevated risk of recession the U.S. faces today, additional near-term stimulus reduces that risk.
Given the deleterious effects of long-term unemployment on an individual’s skills and long-term employment prospects, speeding a return to employment is both individually and socially beneficial.
- With monetary policy’s limited room to lower rates and stimulate demand, there is a role for counter-cyclical fiscal policy.
Ezra Klein summarizes the plan, and economists reactions:
The plan shows that Obama gets how dire the jobs situation is, writes Paul Krugman: “I was favorably surprised by the new Obama jobs plan, which is significantly bolder and better than I expected. It’s not nearly as bold as the plan I’d want in an ideal world. But if it actually became law, it would probably make a significant dent in unemployment…Some of its measures, which are specifically aimed at providing incentives for hiring, might produce relatively a large employment bang for the buck. As I said, it’s much bolder and better than I expected. President Obama’s hair may not be on fire, but it’s definitely smoking; clearly and gratifyingly, he does grasp how desperate the jobs situation is. But his plan isn’t likely to become law, thanks to Republican opposition. And it’s worth noting just how much that opposition has hardened over time, even as the plight of the unemployed has worsened.”
There’s some question as to whether payroll tax cuts are effective, reports Brad Plumer: “Chad Stone, an economist at the Center on Budget and Policy Priorities, points out that the payroll tax cut does tend to get money in the hands of lower-income workers, who are more likely to spend their savings, which means that the CBO tends to score it as relatively effective economic stimulus….Still, Stone adds, there are other tax credits that are better designed for providing stimulus — say, a refundable tax credit that more narrowly targets low-income workers…Former Reagan adviser Bruce Bartlett has argued that the measure doesn’t help the unemployed (who are most likely to spend any additional income), that it hits many higher-income Americans who don’t need the relief, and that many workers are more likely to pocket the savings than spend the money…There are two counterarguments here. One, even if workers are saving the money, that may not be so terrible.”
Obama’s setting up a win-win scenario for himself, writes Mark Schmitt: “Obama’s new approach, though, sets up, in theory, a different hypothetical win-win than the one we’ve been operating under for almost three years. One possibility is that Republicans have some qualms about a wholly obstructionist agenda, Congress passes some or most of the American Jobs Act, the economy improves (likely with some help from the Federal Reserve, international circumstances, and good fortune), and actual conditions get Obama out of the box he’s in. Failing that, if the White House and Democrats can keep their focus on the American Jobs Act (and if the left can avoid getting distracted by Obama’s wise concessions to reality, such as long-term reductions in Medicare spending), then Republican obstruction takes a new form. It’s not just blocking Obama, or his agenda–it’s blocking economic recovery, systematically, including ideas that Republicans have embraced in the past and will embrace again.”
The package is the right size, writes Jonathan Cohn: “The numbers I got from economists varied, but the rough consensus was that an investment in the neighborhood of $400 to $500 billion (including renewal of the existing payroll tax cut and unemployment insurance extension) would reduce unemployment by roughly a percentage point over the next year, relative to what it might otherwise be. If the current projections are right — always a big if — that’d still leave unemployment at close to 8 percent, which would be too high. But it’d be a whole lot better than 9 percent, which is where it’s stuck now. Would the American Jobs Act accomplish that?…Add it up and you get close to $450 billion, very much in the ballpark of what those economists had recommended.”
It’s too tax cut heavy, writes Jeff Madrick: “I do have some reservations about what the president said. One is the large majority of this plan is about tax cuts and tax credits, not real spending. We get much more bang for the buck with spending on infrastructure–with spending on unemployment insurance and sending more money to state and local government than we do for tax cuts for business. For the most part, we’ve had those and it hasn’t worked. But these tax cuts for business and workers will not have the pop per dollar that direct spending would have had–direct spending on infrastructure, on education and so on. For every dollar of direct spending, you get more GDP than you do for every dollar of tax cuts. So that was disappointing–that was the same old stuff. The other thing that was disappointing to me was that he mentioned that if we didn’t reform Social Security, it wouldn’t be here–that is totally not true.”
…
It won’t work, writes Douglas Holtz-Eakin: “The president has ducked entitlement reform and believes that government spending is growth, so there was no chance of spending offsets. That means that to provide the offsets would mean he had to come clean with the fact that this is just another tax and spend effort. So we got another speech with more promises and fundamentally mediocre substance. Indeed, even by late August ‘Macroeconomic Advisers’ was warning that a similar package would generate under 40,000 new jobs per month between now and the end of 2012. This package is ‘bigger’ and would like get scored differently, but the bottom line would be the same. Details aside, one knew in advance that the president wanted to spend nearly another half trillion dollars and not move the dial on unemployment. We didn’t know that it would be spend and promise to tax, but the shock value is small.”
AP via Chicago Sun Times, Globe and Mail and Izzo/WSJ RTE also summarize economic assessments.
Four comments:
(1) The textbook is wrong in the limit where marginal spending by the government is saved rather than spent by the recipients.
(2) The textbook is also wrong when the marginal dollar is spent on imported goods (e.g. oil at higher prices due to price inelasticity) rather than domestic production.
(3) The textbook is also wrong in the limit where corporate profit margins are high enough to siphon off all of the marginal spending as profit (not payrolls).
Therefore the multipliers are likely to be grossly overoptimistic and the result of stimulus may skew far too much toward commodity and financial-asset inflation rather than consumer balance sheet repair.
(4) I strongly disagree with your excuse-making comment “I think it important to recall that between the election and inaugural day, business economists revised downward their estimates of GDP [2]. The proper metric is then to compare against the counterfactual as of the time of implementation…”
The proper metric is whether the President and Congress have adapted their policies in realtime to proactively meet the nation’s challenges. “Business Economists” are irrelevant. It doesn’t matter if the government launched an undersized stimulus at first, their job was to realize it was undersized (a lot sooner than now, 3 years later!) and do something about it for the sake of the greater good.
It is nearly 4 years since the economic peak of 2007, 11 years since the 2000 peak, and many many more years since the lower half of the population saw its real earnings peak. When only 1-5% of the nation are doing better now than they were 10 years ago, that is not successful statesmanship. This administration, and the Bush administration which preceded it, are complete failures in terms of “promoting the general welfare”. Even the Clinton administration was an epic fail because the dot-com bubble prosperity was unsustainable and in many cases completely fraudulent.
We need to hold our government and its economic policymakers to a much higher standard of accountability that what you profer.
Not an economist, but your explanation of how the bridge is built, and that leads to GDP growth directly and indirectly is absolutely correct. You are able to game GDP with deficit spending. But the question that is never answered is “then what”. We have had nothing but deficit spending and stimulus for the last ten years . . . after the dot com bubble burst and the sept 11 attacks, the fear was deflation, so interest rates fell, tax cuts were enacted. And back then, the “virtuous” cycle did take hold in the form of a housing bubble. This time, when the bridge is done being built, the debt will be there, and all the people that were getting income from the bridge will end up at the same place they started . . . starving economically.
From my reading of Krugman and others, it is an uncontested fact that the overindebted consumer is the main driver of this economy. In my view, economists need to frame the need for stimulus as a way to get from point a to point b. Point a is the status quo, where we are right now. And people need jobs, in the long run we are all dead, and all that. And Stimulus is needed for that. But, the economists thinking that a virtuous cycle will develop out of this are not correct, things need to occur to correct what is wrong in our society . . .
house prices were raised too high, so a “do-over” needs to be enacted with across the board principle reductions, or at the very least, refinance everybody to a low rate on 15 year terms so principles are paid down much quicker. Every ounce of fraud that was perpetrated to blow that bubble needs to be prosecuted . . . the rule of law needs to be imposed or this country will be finished. Accounting rules need to be reinstated with prompt corrective action required at all banks . . . no more foreclosures drawing out for two years of rent free living. If somebody needs to be foreclosed on, foreclose on them, no more extend and pretend. If this crashes the banking system, so be it. They were the ones who wrote the bad loans. We should have nationalized the banking system in 2008 anyway, and better late than never.
Our elites have exported our manufacturing jobs with the promise that we would educated ourselves for more high-tech professions of the future, yet College education costs are rising much faster than the pace of inflation, and loans to unsophisticated 18 year old borrowers are borderline predatory. Some sort of means testing on the degree being sought or protection for the borrowers on these loans need to be enacted, rather than the current system of an almost debtor’s prison where the loans can never be discharged.
Our biggest government fiscal problem is the long run health care costs and health care inflation. But in the Hoopla over “obamacare,” overall health care inflation in the private and public sectors was little changed (i.e., nothing was fixed except that now poor people will be able to get overcharged for crappy healthcare as well). Numerous changes will need to be enacted on healthcare to bring this bloated medical industrial complex under control. In my family’s name, 15% of my income (most of which my employer pays) goes to health care premiums before I even see a doctor. My co-pays have gone up from $20 per visit to $30 per visit in just the last two years, and last year, while getting work done on my two year old’s AVM’s and Birthmarks, my deductible doubled and out of pocket maximum jumped in the middle of the year by 50%, costing me another $1500. Medicare costs are not the only costs that are outrageous.
Finally, something resembling an energy plan will need to be enacted, as the status quo is not sustainable. A combination of efficiency measures with offsetting increases in the gasoline tax to get cars that drive 60 to 70 to 100 miles per gallon while mimicing $3.00 gasoline at today’s usage rates of 25 miles per gallon would seem to do the trick.
The military industrial complex needs to be reigned in as well.
All of these ‘structural’ changes need to be enacted . . . and it is going to take time. Stimulus can bridge the gap between point a and point b, but Stimulus alone is not going to fix the problems outlined above, and because of the problems outlined above, won’t lead to a ‘virtuous’ cycle.
Menzie,
If you have a moment to help me with a question…
When I see estimates of the impact on employment that “the stimulus” (ARRA) has had, such as CBO’s estimate http://cbo.gov/ftpdocs/123xx/doc12385/08-24-ARRA.pdf I wonder: Are these after-the-fact analyses simply a reflection of models, just as the pre-stimulus projections were based on these models (the only difference being any difference in inputs), or based at least on some refined models based on actual data of observed immediate post-stimulus effects that are theorized to translate to employment, or what?
After all, actual employment data, at least on a macro level, doesn’t in itself tell us about magnitude of impact, since impact is the difference of the actual vs. the counterfactual of “no stimulus” rather than just before vs. after absolute levels. And if we just apply the same or similar model after as before, we’re just adjusting the counterfactual per the model to validate the model, which is circular logic.
So how is the methodology that now estimates that difference any more likely to be valid than would be models used to project the impact in the first place?
And when does VP Biden declare another “summer of recovery?”
http://www.brainyquote.com/quotes/quotes/a/alberteins133991.html
gee, republicans seem to be skeptical of policies they have favored before. gee, republicans who loved deficits in the 00s and 80s now use them as an excuse to dismantle programs they have hated for over 70 years. gee, i wonder why.
No Direct Incentives to Consumption – No Dice
The payroll tax cut will simply be used to pay down debt or saved, so the net effect is zero. With no direct tax incentives to stimulate consumption, Ron Paul may be right about 15 years. We have learned nothing from Keynes:
“If it is impracticable materially to increase investment, obviously there is no means of securing a higher level of employment except by increasing consumption. . . .I should readily concede that the wisest course is to advance on both fronts at once. Whilst aiming at a socially controlled rate of investment with a view to a progressive decline in the marginal efficiency of capital, I should support at the same time all sorts of policies for increasing the propensity to consume. For it is unlikely that full employment can be maintained, whatever we may do about investment, with the existing propensity to consume.”
The General Theory of Employment, Interest and Money, p. 325
If the President’s proposals are all so reasonable and so great, why not go much, much further, all the way and more? Why not spend 5 Trillion dollars (instead of 500 billion) and worry about “fiscal consolidation” (Bernanke dixit) later, sometime in the “future”? Because evidently, according to this post, somebody somewhere in some office or dungeon in Washington DC can allocate resources much, much better than 100 million households and tens of thousands of businesses. So, let’s go for 5 Trillion, or even more, because if 500 million help, much more will help 5 Trillion.
River’s has posted the most valid comment to date. A practical solution to a speculative problem.
Yet,the real issue has not been touched upon, that the economy is an invention of man. In such we speculate on what we control?!
Can we reframe the conversation?
First, Markets are not sentient. Prices do not change due to conflict or disaster, they change because some group of persons changes them.
Remove the veil of anonymity , state the largest traders and the losses or profits they expect from the change and the effect this will have on the average person, the economy.
This would be a start, now how about coming to the realization that we have finite resources and global control over the resources is needed.
Oh a better dream, remove the ability to make money from money.
I am starting to have Utopia visions here! Why not.
To paraphrase James Burke. The economy is at anytime, what we say it is. So say!
Year before last I attended a luncheon speech by the president of the Boston Fed.
After the speech he was asked if the US is on a path like the Japanese into a lost decade. He had spent several years in Japan in the 1990s and was well qualified to address that question. And he gave a very good answer explaining how the Japaneses made repeated policy mistakes at critical points. He finished by he was confident the US could avoid making similar policy mistakes.
Next month I’m going to hear him speak again and I know what I want to ask him. I will remind of his answer two years ago and ask that given the developments in Washington over the last two years if he is still confident that the US can avoid making similar policy mistakes. I sure his answer will be interesting.
Porkulus I was reasonable by textbook standards. How’d that work out?
Time to get a new textbook.
Comparing 1937 to now is no reasoning, in 1930 1937 few banks and many investments trusts funds engaged in pyramidal chains collapsed but the banks systemic risk was promptly remedied.The Glass-Steagall Act was established in 1933 and repealed in 1999.
The present plan is a text book of the measures undertaken by Hoover when addressing the productive sector of the economy railways and energy.It will create jobs and at last move away from the capital consuming economy,the financial industry.
Banking system remains to be addressed.Many banks remain unmanageable,VAR a mathematical ideology,capital increase scant, assets divestiture to raise adequate capital
has not been seen.Derivatives and macro risks are flourishing.Real counter party risks are on the rise.
Given the long-term decline in household savings rates (compared to GDP) and the long-term increase in household indebtedness (compared to GDP), why is it unreasonable to believe that output (and consequently, employment-to-population ratio) was above the financially sustainable trend?
Not only are we repeating the late 1930s, the other historical parallels are the Long Wave Trough depressions of the 1830s-40s, 1890s, and Japan after ’98-’99.
But those who are students of the Long Wave and Schumpeter already know this.
Therefore, the debates about this or that incremental monetary and fiscal policy prescription or stop-gap borrowing and spending boondoggle are largely moot in the face of an unprecedented “global” debt-deflationary slow-motion depression coinciding with population overshoot, peak global oil production, and falling net energy per capita and available net oil exports.
There can be no growth of real private economic activity per capita until, or if, the price of oil falls to the $20s-$30s and stays there for an extended period; consider the scale of global contraction required for this to occur.
In the meantime, deficits and unemployment/underemployment will continue to rise and stay high; price inflation will decelerate; stock and real estate prices, bank loans, and interest rates will fall; the yield curve will further flatten; and the avg. 10-yr. rate of real private GDP per capita will contract further.
Moreover, local and state public spending will be continually reduced, including salaries and benefits and pension payouts and benefits for retirees, including for professors at public universities.
The worst conditions of the debt-deflationary slow-motion depression are still ahead, and this time around the BRICs will be hit hard.
Anonymous 9:30PM: Let me go point by point: (1) I think your terminology implies you mean “transfers” rather than government spending on goods and services (in the NIPA lexicon), when you discuss private saving. When the transfers are directed to liquidity constrained individuals (as I stressed in the post), then the MPC is high, MPS low. (2) Yes, when the marginal propensity to import is high, the Keynesian open economy multiplier is lower than otherwise; but oil imports are declining, and are not the only component of total goods imports. (3) Departing from standard undergrad textbook to higher level textbooks, if investment follows an accelerator model, the policies will be expansionary; if a financial accelerator is relevant (Bernanke-Gertler) then better economic conditions relax the collateral constraint, inducing more investment.
Brooks: The CBO figures are based upon previous modeling experience, derived from a variety of sources (see the appendix to the last CBO assessment of ARRA for a primer). What can then be done is to simulate a baseline, and the baseline-plus-stimulus, as discussed in the links after note [2] in the text.
Paul: Complete paying down of debt by households assumes some pretty odd intertemporal optimizing behavior, even if all agents are not liquidity constrained. I would say for hand-to-mouth consumers (or those on UI), that sort of behavior is highly unlikely. Might be true of millionaires or Todd Henderson households, but that’s not who the measures are aimed at.
Manfred: Obviously you can’t do math, or need to acquaint yourself with the AD/AS framework. We want to move closer to potential GDP. The gap is not $5 trillion, by anybody’s accounting.
W.C. Varones: I wish you’d read a textbook, even Barro or Williamson. Then you’d have an internally consistent framework in mind. I can only imagine what algebra textbook you used in school.
Manfred might like the platinum coin deal a little better. Instead of some petty functionary doling out conjured-out-of-thin-air fiat currency to faith based initiatives and diabetic supply companies we could mint a $10,000 denominated platinum coin for every adult American. Recession over? I’m not entirely sure but I think it costs less than $5T. And would probably be most stimulative!
There might be some initial inflation in the waterbed sector, and Disney vacation packages. But manageable, I think.
Regards!
Menzie,
Thanks, but I am still left with my question.
Perhaps I can be clearer with an analogy.
Suppose Weight Watchers had a weight-loss program for Joe, and they have a model that predicts the impact of Joe doing the program for the next month. Joe weighs 200 pounds, and the prediction is 190 at end of the month. That’s because the model calculates an effect of 10 pounds of weight-loss for Joe.
One month later, Joe still weighs 200 pounds. Weight Watchers tells Joe that my model says that the program still made him 10 pounds less than he would have been without the program (he would have been 210 without the program), because the model says that the program has an effect of 10 pounds.
In that case Weight Watchers is engaged in circular logic. They are saying that the model shows that the program worked (that he would have been 210 without the program) because the model says so.
Is CBO’s analysis or anyone else’s analysis on the effect of ARRA any different than the above? Is there a reason to consider such analyses credible estimates of effects rather than just a reflection of the assumptions of the original model?
Menzie:
And obviously, your math is so good and so precise (deserving of a Fields Medal, I am sure) that you have the US economy pinned down to the very last cent, wow.
You see, prefer Hayek when he wrote:
“The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design.”
Manfred: If my math had been really good, I would’ve become a theorist or a theoretical econometrician. All I’m asking for is grade school math — the ability to compare two numbers, figure out the gap, and determine up to an order of magnitude, what is necessary to close the gap by a certain proportion. $5 trillion doesn’t fit into even the extreme multiplier-skeptics calculations, so I could only infer an inability to subtract and divide, or unacquaintance with AD/AS.
So go back, and divide cumulative changes by point-in-time amounts if you want, and whatever mathematical gyrations you wish, but don’t expect us to take your resulting figures seriously.
What a great information packed post. I nominate Menzie Chinn for the president’s council on economic advisors!
River, you are spot on. The middle class needs the housing debt/valuation problem addressed, now! We have two entire generations on hold and learning austerity. These lessons are resulting in much reduced spending and have caused a steady spiral in a downward direction since 2007. For the boomers, the spending will never return, for their children, it is unlikely they will ever reach the level of their parents.
Professor Chinn, thanks for the point-by-point reply.
Regarding point (1), I do not think the stimulus policy changes can be targeted nearly as well as the policymakers seem to think. Payroll taxes, being a percentage of income, will provide greater rewards to those earning above $50,000/year (individually… up to ~$220K/year household where the FICA tax cuts off). A large portion of these folks have huge debt burdens and are not low-MPS. So I worry that more will be saved and less will be spent than expected. I would be interested in knowing whether detailed modeling of the expected effects has been done, and if so with what methods.
Regarding (2), to the extent that marginal dollars are spent and not saved, it seems reasonable to think the increase in consumption will flow through in high percentage to gasoline and fuel oil consumption. Prices in that sector are very inelastic and the marginal demand can easily drive a sharp increase in oil prices… effectively taxing everyone (not just those granted the extra marginal dollar). Marginal oil production is unlikely to come from domestic sources, so this is a trade deficit issue which impacts domestic savings rates and the ability to repair balance sheets.
With regard to non-oil imports, marginal consumption by consumers with low propensity to save may be on other imported goods (iPods, shoes, clothing) whereas non-marginal consumption may be more domestic, and I’m concerned that our modeling is inadequate to capture this.
Regarding (3) I lack your formal education in this area, but with my real-world eyes I don’t see corporations reinvesting their profits in new capacity, in a world where capacity utilization remains near historic lows. The St. Louis Fed National Economic Trends reports show clearly that around 10-15% of every dollar of GDP is harvested as corporate profit. Given fixed costs, the percentage harvested from the marginal consumption dollar is likely much higher. So one model I think economists need to consider is where a federal stimulus dollar actually reaches a high-MPC, low-income taxpayer as intended, but within 2-3 transactions the majority of that dollar winds up sitting idle on a corporate balance sheet. From there, given the MPS of corporations at present, it will most likely be invested in a money-market fund, the proceeds of which go to banks who end up placing the funds with the Fed as excess reserves. (One can imagine many similar chains of non-productive or even anti-productive savings, especially in our current environment of high financial fraud.)
I would add that on other sites, business-savvy commenters have made it quite clear that the $4K hiring incentive is likely to lead to a lot of additional layoffs in many positions, as corporations choose to recycle staff (and harvest the $4K bonus as profit) rather than retain experience. If relatively ordinary people can identify multiple ways to game the new policies into unintended consequences, within just an hour of Obama’s speech, I worry what the more sophisticated financial predators will do to the other portions of the new policies.
Addendum: If these Presidential spending proposals are indeed to be paid for through offsetting revenue enhancements, the net effects could well be more positive than I am fearing. For instance, if the payroll tax cut is funded by closing corporate tax loopholes, that offsets the impact of having the marginal spending get siphoned into corporate profits. But the chances of these proposals actually being offset by sensible revenue increases seem remote.
Menzie, Menzie:
My mathematical gyrations? It is *you*, it seems to me, who thinks that you have the whole thing figured out, “up to an order of magnitude” – whatever that means.
Since you do not seem to believe Hayek, let me quote from today’s Wall Street Journal:
“If President Obama’s economic policies [or one could say, if Menzie’s post] have had a signature flaw, it is the conceit that by pulling this or that policy lever, by spending more on this program or cutting that tax for a year, Washington can manipulate the $15 trillion U.S. economy to grow.”
Or it says later in the editorial:
“We’ve had the biggest Keynesian stimulus in decades. The new argument that the 2009 stimulus wasn’t big enough isn’t what we heard then. Americans were told it would create 3.5 million new jobs and unemployment would stay below 8% and be falling by 2011. It is now 9.1%. But this stimulus we are told will make all the difference.”
According to you Menzie, and your exhaustive calculations, “up to an order of magnitude”, 450 billion will make this difference to close “the GDP Gap”; that is, you, and the CBO, and Macroeconomic Advisors, and Mark Zandi, and whoever else, know to the cent (but “up to an order of magnitude”) the workings of a 15 trillion dollar economy; you know to the cent (up to an order of magnitude) the millions and millions of decisions that 100 millions households and tens of thousands of businesses will make, the expectations they will form (especially because of the temporary nature of many of the proposals) and the actions they will take.
Gimme a break, Menzie.
If you want to really make a move, announce you will stop buying foreign oil immediantly. This would collapse the US trade deficit as we take from the reserves to fill in the gap, but the key thing is, it will force business to replace the oil with non-oil products before we run out of oil.
Something is going have to be tried at some point, to deal with the global inbalances. I suspected they would sit on their arses hoping that things return to “normal”. Now is when things get interesting.
. . . why is it unreasonable to believe that output (and consequently, employment-to-population ratio) was above the financially sustainable trend?
Michael Cain, it was, in fact, by about a cumulative order of exponential magnitude of debt-money supply to industrial production since the 80s, the secondary peak of US domestic crude oil production and the onset of deindustrialization and financialization of the US economy.
Now we have a differential of debt-money supply and debt/asset values to industrial production and wages equivalent to 35-40 years’ worth of long-term trend industrial production and wages.
IOW, at the trend rate of post-’00 nominal GDP, the US economy will not grow through at least the early to mid-’30s in real terms per capita without a dramatic write-down of private debt/assets (and bank loans) of 50%. In effect, a halving of debt/asset values would result in a devastating banking system collapse and debilitating negative wealth effect, which has to happen, and will, to clear the decks.
Historically, these secular debt cycles and resolutions have coincided with wars every 54-60 (and now 70 years owing to the increased lifespan and associated demographic effects) years or so going all the way back to the 17th century (and perhaps longer), periods during which there were banking and currency crises, financial panics, bank runs and failures, social unrest, sovereign defaults, revolutions, and all manner of systemic crises.
The post-’00 avg. trend rate of real US GDP has been cut in half (1.6% today vs. 3.3% long term) since the secular peak in ’00, not unlike what occurred in Japan by the late ’90s and early ’00s. The US economy has produced ~$2.3 trillion less (15-16% less growth) than would have occurred had the long-term growth trend continued after ’00.
By the end of the decade or early to mid-’20s, the US post-’00 avg. trend rate of real GDP will have fallen to less than 1% (real private GDP per capita will be negative), reducing real growth from ’00 by 35-40% vs. what otherwise would have occurred had growth continued at the long-term trend rate.
Therefore, 35-40% should be the implied real GDP growth gap vs. trend by decade’s end or slightly thereafter; that is, the cumulative loss of real GDP growth from the secular peak to ’20 will be $7-8 trillion, which implies cumulative US gov’t deficits totaling 200% or more of GDP by the early ’20s (Japan-like levels of today) just to keep nominal GDP from contracting along the way.
The implied growth gap projects a grim U-6 (total unemployed/underemployed) rate in the 30-35% area by decade’s end (doubling from here). However, the dramatic deceleration in the rate of growth of the labor force will result in the “official” U rate being much lower than it otherwise would be were the labor force to have grown at the rate of the past 50-100 years.
Thus, structural deficits and the implied secular growth gap are so large cumulatively as a share of GDP today that there is no wiggle room virtually at all for incremental gov’t borrowing and spending apart from attempting to sustain various income support spending for food stamps, unemployment, Social Security, Medicare/Medicaid, SSI, etc.
Why no multiplier for tariffs, the most obvious way to reduce immediate overcapacity issues and reallocate demand towards domestic production?
Manfred Your first mistake was in reading the WSJ op-ed page. It will make you dumber. Time for some corrections to that nonsense. So let’s begin:
1. We’ve had the biggest Keynesian stimulus in decades.
We’ve also had the deepest recession since the 1930s coupled with a hobbled Federal Reserve due to ZIRP. The stimulus was not near large enough to do the job.
2. The new argument that the 2009 stimulus wasn’t big enough isn’t what we heard then.
That’s because the people over at WSJ live in a Murdoch/Fox News echo chamber. Go back and check Menzie’s posts from early 2009. Go check Mark Thoma’s posts from early 2009. Go check Krugman’s rantings and ravings as to the inadequacy of the stimulus. There was a poll among economists and almost two-thirds thought the stimulus was too small. Read Romer’s paper…and read between the lines. Her own numbers called for a much larger stimulus. So this claim by the WSJ is either ignorance or outright lying. Take your pick.
3. Americans were told it would create 3.5 million new jobs
It came pretty close, but even Romer’s own numbers showed that this would not close the gap.
4. unemployment would stay below 8% and be falling by 2011.
That report was written in Nov/Dec 2008…long before anyone knew just how bad things really were.
5. …it is the conceit that by pulling this or that policy lever, by spending more on this program or cutting that tax for a year, Washington can manipulate the $15 trillion U.S. economy to grow.
More evidence that Paul Gigot and the WSJ editorial board are clueless as ever and are in bad need of an undergraduate macro course. No one is talking about fine tuning the economy. We know that in the absence of an effective monetary policy we have to rely upon fiscal policy. We’ve known this for 75 years. The federal government has lots of policy tools for aggregate demand management. What the federal government does not have are a lot of tools for aggregate supply side growth. The government can easily close the output gap to potential GDP; the government has almost no ability to affect the long run growth rate of potential GDP itself. Strangely, the wonder brains over at WSJ have it completely backwards. They’re always jabbering about supply side policies, blah, blah, blah. They’re the ones who are always looking for phantom supply side levers.
@ Menzie,
You rolled out the policy options paper again…so I feel obliged to prod you on the analysis again. I went to the St Louis Fed web site and took data sets for nominal GDP (seasonally adjusted annual rate, quarterly), nominal consumption (SAAR, quarterly), total government expenditures (SAAR, quarterly), and net exports (SA, quarterly, had to generate my own annualized data from that data set). I plugged these data into the identity GDP = C + G + I + net exports and solved for I (to get nominal AR data set). Then normalized everything as % of GDP. Plotting it I get…something I cannot paste into this comment box…crap.
Anyway, looking at the stats, the calculated “I” correlates most strongly (-0.947 corr. coef.) with “G” and using LINEST, has a slope of -1.35. Put another way, a dollar of increased spending on “G” leads to ca. negative $0.74 spending on “I”.
I realize this isn’t a fancy model of the variables…it’s just what happened in the US economy between Q1 of 1960 and Q2 of 2011. Increases in “G” mostly came at the expense of “I” and decreases in “G” tended to show up as increased “I”. So once again, if we add -$0.74 to mean of the numbers in the first two columns of the policy options table you cited…most of those ideas look to be possibly harmful to the economy, not beneficial.
BTW, before you ask, I resorted to calculating “I” because I could not find data sets for all parameters in the identity that would sum to the GDP data set. Since I am interested in the effect of “G” on investment and since I had to calculate annual investment rate either this way or from the data sets on total investment that are available at the St Louis Fed web site, I opted for calculating it this way. Let me know if you would like to see the spreadsheet.
One thing that grates on people is having economists tell them that opposition to the most reasonable policies of their betters is due to ignorance. If only the people were more educated they would not oppose the reasonable policies that are being promoted by the Harvard educated elite that run the country.
One thing that I admire about James Hamilton is that he does not talk down to people. I can disagree with him without feeling like he is sneering at me for my opinions.
Admittedly a lot of the criticisms of past policies are based on economic ignorance. But perhaps there is some cherry picking going on in the selection of economists and textbooks.
For the lighter side of this discussion: http://1.bp.blogspot.com/-_bgPeJsFLqE/Tl_LguQLW9I/AAAAAAAABPo/Ti6e7cMMc0I/s1600/nonsequitur090111.gif
Brooks,
Exactly! Answer the question, Claire!
JFK,
Good stuff. You can share and link spreadsheets via Google Docs.
JFK Setting aside the rather obvious econometric weaknesses in your simple spreadsheet analysis (e.g., estimating nonstationry flow variables in levels rather than log differences; no weighting for heteroskedasticity; all values are contemporaneous and no lagged or ARMA terms; etc.), the only thing that you “proved” is government spending is countercyclical to investment spending. And that’s exactly how things are supposed to work. Would you rather have both private investment and government spending move in the same direction?
What you’re trying to make is an extremely crude “crowding out” argument. Now that’s a tough case to make even if you’re armed with some fancy pants econometric tools, but a crowding out argument is laughable when corporations are sitting on $2T in idle cash, banks are charging customers for holding large deposits, there’s a huge output gap, unemployment is at 9.1%, and the Fed is dealing with ZIRP, and the 10yr closes at 1.91%.
Though macro economically it is true that in recessionary conditions, incremental government spending and tax cuts can stimulate the economy but these assumption are true in ideal conditions which is not the case with US. Watching the political and economic crisis going there, Obama admin would have to do much more than spending.Prediction about the double dip recession is very relevant in present situation
colonelmoore: Please be specific about instances where I have sneered at policy recommendations you have made. Now, I will admit to being somewhat exasperated about the inability of people to subtract and divide, but that is hardly driven by an elite education; I learned those mathematical operations in grade school, and I expect people who are going to comment on policy issues here to be able to do simple math. At this level, math is not a matter of opinion (we’re not doing boolean geometry!). Or do you disagree?
@ 2slugbaits
As to making an “extremely crude crowding out argument” I plead guilty. As to the econometric weaknesses, again, guilty as charged.
However, Menzie’s article lamented the lack of understanding of even basic macroeconomic models and Menzie’s response to my earlier inquiry indicates to me that the policy options were evaluated on a ceteris paribus basis. Well, the “extremely crude” model based on a classic macroeconomic identity indicates that this is not an extremely safe presumption.
Still, I read this blog to learn. So, can you point me at better constructed model that looks at the effect of an added dollar of I (or C or net export) so I can make a fair comparison of a policy that relies on increased government spending versus increased consumption, increased investment, or increased net exports?
The government can influence the economy in many more ways that just creating debt and using that money to spur demand. I would like to see what the other options are so I can make a fair comparison.
With regards to a “crowding out” argument being laughable in light of corporations sitting on a large pile of cash…I think you missed my point. The classic identity says that GDP = C + G + I + net exports. The only things that promotes growth in GDP are a) a bigger population working at current productivity, b) increased productivity with the current population, or c) some combination of increasing population and productivity. I don’t see population making a big contribution to GDP for the next several decades. Of the four parameters on the right hand side of the identity, the only things that promote productivity are the investment component of G (not the transfer payments or the straight consumption components) and I.
Assume an extremely simple model wherein a dollar increase in G is bought entirely at the expense of a dollar decrease in I. Assume for the moment the the humans running government are exactly as clever as human in industry at choosing investments that promote productivity. Remind ourselves that we are all good Keynesians now and that we increased G to increase consumptive demand…and it is very hard to come to the conclusion that the negative impact on productivity of a dollar taken from I was adequately compensated for by the less than one dollar investment component of the dollar increase in G.
Perhaps there is some fancy pants econometric model that will explain all of this to me…but somebody has to point me to a link where I can access it before I can gain that enlightenment.
JFK Part of the problem is that you’re taking an identity too literally. In equilibrium investment equals saving, hence the identity I=S, which can be derived from the old Y=C+I+G+NX. Unfortunately the economy is not always in equilibrium. If demand for saving exceeds the demand for private investment, then something’s gotta give. Normally that would be the interest rate. A lower interest rate encourages businesses to invest and brings I and S back into equilibrium. Except that today the interest rate is at zero and businesses still don’t want to invest. You would actually need a negative interest rate in order to induce businesses to invest enough to make I=S. Since the Fed is already at a zero rate, the Fed doesn’t have a lot of other tools at its disposal. It could try to lower the long term interest rate, but Gov. Perry thinks that’s treasonous. Or the Fed could try to encourage some inflation, except three clueless Fed Heads are just convinced hyperinflation is right around the corner. So we’re basically left with fiscal policy.
If you’re completely new to economics, then I would suggest surfing YouTube for some videos on IS-LM analysis. It’s old school and your freshwater/real business cycle friends won’t want to be associated with you, but for all its problems the IS-LM model tells you pretty much what you need to know about today’s economic problem. Mark Thoma also has his entire semester’s macro/money&banking class covered in video, so that’s an option.
As to productivity. Those are all issues on the supply side. Long run you cannot ignore productivity, which is why it’s so strange that “supply side” Republicans want to cut education, cut research, cut infrastructure, restrict immigration and get rid of just about everything that actually increases productivity. In any event, supply side issues are not today’s problem. Today’s problem is too much supply relative to demand. We need to worry about aggregate demand policies in the face of a liquidity trap, and the old Keynes-Hicks IS-LM model is pretty good for understanding that kind of problem. Forgetting Keynes, Samuelson and Hicks is what Krugman meant by the “Dark Age” of macroeconomics.
Menzie, would you please indulge me by devoting a post to dealing with some of the following questions?
1. There is a large difference in the stimulative value of policies. (a) Half the stimulus is payroll tax cuts. While these deliver stimulus quickly, a more than half will go to businesses or to families above the median, who have a strong tendency to save and/or pay down debt, rather than consume. Even many families below the median are likely to pay down debt. Therefore, the true stimulative value of this is more like $75B, making the full package worth, maybe, $300B. (b) isn’t it pretty clear that the lower of the CBO estimates of stimulative effect of the tax cuts was probably too generous?
2. Preferences for veterans have little stimulative value, since they will encourage employers to hire veterans rather than other, but not affect the net propensity to hire.
3. To close the output gap, you say we need stimulus equivalent to about 7% of GDP annually. Assuming that stimulus achieved a multiplier of 1.5, we would need a stimulus of $650B, which (because of tax receipts) would have a net cost to government at all levels of about 75% of that, i.e., $490B, and a cost to the federal government of about $550B. Or, to turn it around, a stimulus of $300B (which is the maximum true stimulative value of the Obama proposal) would multiply to a maximum of $330B (roughly 2.4% of GDP) spread over two years, or about 1.2% of the 7% needed.
4. But all of this is using super-optimistic assumptions. In particular, Republicans are certain to demand offsetting cuts, the incidence of some of which will be in the period of the stimulus, reducing the power of the stimulus. Even assuming that they aren’t deliberately trying to crash the economy, that means that we’re probably down to 1% of GDP or less of stimulus, on top of predictions of 1-2% GDP growth (less if Europe crashes). In other words, enough to keep us out of recession, but nowhere near enough to spark a recovery.
5. In replaying 2008, it would be well to recall that the actual stimulus was quite small, that most of what was called stimulus was non-stimulative tax cuts.
6. Finally, let’s concede that allowing the banks to fail in a disorderly fashion is probably not be good for the economy. However, I did some simple calculations, and figured that the mortgage crisis could have be resolved for about $300B if we had forced the banks, the investors, the mortgagees, and the government to share the pain. The cost seems to have been much higher, and the banks seem to have profited (though, as BAC shows, we may still be in extend-and-pretend mode). So: isn’t it important, from the standpoint of gaining the political will to proceed, to make sure that stimulus is directed to people who are hurting, rather than allowing the perception, if not the reality, of huge sums flowing to the superwealthy, and pennies to the rest of us? How much do you think it would have cost to solve the mortgage crisis equitably without crashing the banks, foreclosing on millions, and having the taxpayers burdened with the full cost?
@ @slugbaits,
Thanks for the directions on where to find more discussion on this kind of modeling.
As for taking the identity too literally because of a base presumption of equilibrium and the classic model equating I = S, I have a few comments/questions to put back to you.
First off, I don’t think my modeling approach requires equilibrium because I took a snapshot of the system every quarter for 50 years and looked at how 200 instances of “G” correlated to 200 instances of calculated “I”.
Second, the “I” in my modeling is not presumed to be equal to “S” rather, the “I” in my modeling is a stand in for investment that allows me to complete the identity for each snapshot I took.
Does that make my “I” more than a little suspect with regards to the classic model…of course it does. However data sets for GDP, C, G, I, and NX that sum to the identity (while taking disequilibrium into account) aren’t readily available, so a crude measuring stick that isn’t NIST traceable (yes, I am engineer) is better than nothing.
Using that crude measuring stick I come up with the following:
1) “G” is most highly correlated with “I” (-0.947).
2) “G” is not so highly correlated with anything else (w/GDP = 0.64; w/C = 0.66; and w/NX = -0.52).
3) “GDP”, “G”, AND “C” all trend upward to the right (all in nominal dollars and both “G” and “C” as a % of GDP)
If the system being in disequilibrium were a real impediment to looking at the world this way…I would have expected a far weaker correlation between the 200 snapshots of “G” and the 200 snapshots of calculated “I”. After all, there have been a number of major governmental policy changes, technological changes, demographic changes, etc. over the last five decades and I am using an admittedly crude model…yet it appears that increases in “G” have a negative impact on “I”.
As well, there is no a priori constraint that dictates that “G” correlate with any of the parameters in the identity. My brief reading up on IS/LM modeling (thank you again) indicates that “G” is considered exogenous to these models.
So, going back to the whole reason I brought this up in the first place. The CBO evaluated some policy options that look at the projected change on GDP for the marginal dollar of increased “G” but assume that “I” will remain unchanged. This does not appear to be a safe assumption. For a fair evaluation, the benefits projected to accrue from any of these policy changes should be scaled back by some factor that includes the estimated negative impact on “I”…otherwise, there is good reason to believe that the result with regards to change in GDP will come in below the projected amount.
Next, you mentioned lagging data sets in your original comments. I am not sure that matters if I am taking a snapshot of the system “in real time” but am game to try it when I get the chance. Before I do that, would you guess i would be better off correlating absolute levels lagged however many quarters or quarter to quarter changes lagged however many quarters?
Next, from the data sets I looked at (courtesy of the St Louis Fed), C and G (as a % of GDP and in nominal dollars) are at all time highs. While I understand that everyone thinks that unemployment would come down if we could close the GDP gap…I still have to ask why are we so certain of that?
If we are producing the most we ever have and consuming the most we ever have and government is spending the most it ever has and we still have ca. 9% unemployment…maybe it isn’t the GDP gap? Put another way, in your third to last sentence you state, “Today’s problem is too much supply relative to demand.” If supply is GDP, the sum of C + G as a % of GDP is the highest it has ever been…and this sum relative to GDP has been trending higher for, well, ca. 1975 to now.
BTW, I don’t have anything to suggest in place of current dogma (i.e. lower unemployment by upping demand to close the GDP gap)…I am just wondering what other schools of thought are regarding this issue.
Thanks
Professor Chinn,
I was not referring to my policy recommendations and not even to the responses to other commenters. I was referring to the general tenor of the main blog posts.
colonelmoore: Please then identify a specific passage in this post that constitutes sneering. Is it the citation of the article that notes most Congressmembers do not have any training in economics? If statement of fact constitutes sneering, we are in bad shape. Or maybe you are right, and I should expect my MD to give high credence to my alternative view of medicine based on the four humors — after all, I took a biology class in college!
Prof. Chinn, do you agree with Bruce’s answer to Michael Cain’s question: “. . . why is it unreasonable to believe that output (and consequently, employment-to-population ratio) was above the financially sustainable trend?”
Anyone who can help:
I asked above:
When I see estimates of the impact on employment that “the stimulus” (ARRA) has had, such as CBO’s estimate http://cbo.gov/ftpdocs/123xx/doc12385/08-24-ARRA.pdf I wonder: Are these after-the-fact analyses simply a reflection of models, just as the pre-stimulus projections were based on these models (the only difference being any difference in inputs), or based at least on some refined models based on actual data of observed immediate post-stimulus effects that are theorized to translate to employment, or what?
After all, actual employment data, at least on a macro level, doesn’t in itself tell us about magnitude of impact, since impact is the difference of the actual vs. the counterfactual of “no stimulus” rather than just before vs. after absolute levels. And if we just apply the same or similar model after as before, we’re just adjusting the counterfactual per the model to validate the model, which is circular logic.
So how is the methodology that now estimates that difference any more likely to be valid than would be models used to project the impact in the first place?
Menzie replied (September 9, 2011 10:09 AM) but I’m hoping someone can provide a nutshell conceptual answer to my question.
I followed up with the following:
Thanks, but I am still left with my question.
Perhaps I can be clearer with an analogy.
Suppose Weight Watchers had a weight-loss program for Joe, and they have a model that predicts the impact of Joe doing the program for the next month. Joe weighs 200 pounds, and the prediction is 190 at end of the month. That’s because the model calculates an effect of 10 pounds of weight-loss for Joe.
One month later, Joe still weighs 200 pounds. Weight Watchers tells Joe that my model says that the program still made him 10 pounds less than he would have been without the program (he would have been 210 without the program), because the model says that the program has an effect of 10 pounds.
In that case Weight Watchers is engaged in circular logic. They are saying that the model shows that the program worked (that he would have been 210 without the program) because the model says so.
Is CBO’s analysis or anyone else’s analysis on the effect of ARRA any different than the above? Is there a reason to consider such analyses credible estimates of effects rather than just a reflection of the assumptions of the original model?
Can anyone please answer my question? It’s not rhetorical or an attempt to be snarky. I really hope to get the answer.
JFK That’s a lot of stuff to cover in a comment section of someone else’s blog, but I’ll make a couple of points. First, you’re really saying that an increase in G causes a reduction in I. That really invites a causality test; e.g., something like Granger Causality. Second, G might be exogenous in a structural model, but today we would expect some kind of lagged feedback effect between variables. The issue of endogeneity and exogeneity is fuzzy, so you would want to use something like a vector autoregression (VAR)or structural VAR model to capture this ambiguity. Third, you really need to work with first differences of log terms rather than actual levels. One of the reasons you’re getting results with incredibly high R-square values is because the levels data is not stationary. You’re getting a spurious regression. At this point Menzie is reading this and banging his head on the table because of all the handwaving I’m doing, but if you want to pursue this kind of thing a good understandable undergraduate level book on time series analysis is “Applied Econometric Time Series” by Walter Enders. Our host JDH has a book on time series analysis, but quite frankly I wouldn’t recommend it as an intro book. It’s a good reference and I keep a copy on my desk at work, but it’s more at a graduate level. (Sorry JDH),
Finally, keep in mind that G is supposed to move countercyclically with I. And ordinarily (i.e., when we’re not in a liquidity trap) the Fed will actively frustrate the stimulative effects of an increase in G by raising the interest rate, which cools down I. When the economy is operating at something like equilibrium there is a crowding out effect; it’s just not a problem today. This means that a lot of the historical data isn’t relevant for today’s problem. This is a big issue with some of the DSGE models that Menzie references.
@ 2slugbaits,
Thanks for the guidance. I will give both of those books a gander.
Let’s see:
The budget deficit fell from 5.5% of GDP in 1936 to 2.5% in 1937 and the budget was virtually balanced in fiscal year 1938, with a deficit of just $89 million.
So in two years about 5.5% of GDP was withdrawn from the economy. Compare this against the projections through 2012 – which show little change in the deficit versus GDP. Note also that as measured by Deficit to GDP the fiscal stimulus over the last 3 years has been larger than in any year prior to 1942.
I don’t see the parallel to 1937. The “Stimulus Package” is not the same as fiscal stimulus.
Brooks,
Often in writing about economics, you’ll run into the notion of a “counterfactual” – what would have happened under circumstances other than the ones we got. It is one of the elements to economic discussion that drive some folks crazy. It is, however, unavoidable in most lines of economic thinking to employ a counterfactual, because there is no other way of thinking about alternatives. You can’t run a control group along side the experimental group.
So in answer to what I think I understand your question to be – all the after-the-fact estimates of the impact of ARRA, or any other event in economics you care to name, are based on modeling. Sometimes, the model is far removed from reality, based on ingrained belief. Sometimes, the model is constructed without bias, but still wrong by a mile. Sometimes, the model can get close. The only way we have of knowing which is which is to either develop the skills to assess the model, or find economists who are unfettered by political alliance and dogma. Neither is easy.
Kharris,
Thanks for trying to answer my question, but you haven’t really addressed my question.
I’m certainly aware of the concept of the counterfactual and of the usage of models for projection under alternative scenarios.
I’m asking whether the “post” models that claim to measure the impact of ARRA are any/much more than a reflection of the same assumptions used in models to project its impact. See my Weight Watchers example. If my assumption both “pre” and “post” is that the program causes one to end up 10 pounds less at the end of the program, I am just rigging the counterfactual to fit the ASSUMED impact that my model has incorporated all along, not really estimating the impact on some empirical or other basis.
Kharris,
Thanks for trying to answer my question, but you haven’t really addressed my question.
I’m certainly aware of the concept of the counterfactual and of the usage of models for projection under alternative scenarios.
I’m asking whether the “post” models that claim to measure the impact of ARRA are any/much more than a reflection of the same assumptions used in models to project its impact. See my Weight Watchers example. If my assumption both “pre” and “post” is that the program causes one to end up 10 pounds less at the end of the program, I am just rigging the counterfactual to fit the ASSUMED impact that my model has incorporated all along, not really estimating the impact on some empirical or other basis.
Brooks Tagging along with what kharris said, as a practical matter there really aren’t a lot of alternative models out there. Unless you tweak it a lot, a DSGE model probably isn’t very helpful because it wasn’t designed or parameterized with the intention of modeling an economy in a liquidity trap with a weak central bank. Austrian economic models are big on theories about how an economy gets into a slump (if you can get past all the mumbo jumbo about “concertina effects,” “roundabout,” etc.), but they really don’t have a theory of recovery. Austrian models are all about predicting doom right around the corner. Sometimes they’re right…but so is a stopped clock. And Real Business Cycle models are constrained by their theory to be silent about short-run output deviations due to weak aggregate demand. Don’t expect a model to forecast recovery from a recession if that model starts off by assuming that 9.1% unemployment reflects an optimal labor/leisure trade-off. So really you have to fall back on some kind of workhorse Keynesian model. But how do you get the Keynesian theory crammed into an econometric model? Some of those econometric models are big and structural in nature (e.g., Macro Advisors), while others emphasize a less structural approach and focus on impulse/response analysis. Of course, those are also contaminated with irrelevant historical episodes just like the DSGE models, but they are more forgiving in application.
At the end of the day you look for the soundness of the theory and the soundness & reasonableness of the econometric assumptions. Finally, if different models give similar answers, then that should give you some comfort. Awhile ago Menzie had a post comparing various forecasts with what actually happened (e.g., CEA VAR analysis, MacroAdvisors, Moody’s, CBO, etc.). All the models told a similar story even though the econometric approaches were different.
2slugbaits,
Thank you, too, for taking time for my question.
But again, unless I’m missing something here, my question remains unaddressed.
You and Kharris seem to be discussing the question of degree of validity of models for estimating the effect of some variable such as “stimulus” (either before or after introduction of that variable).
My question is: When a model is being used supposedly to estimate in retrospect what the effect of the variable was — as “evidence” of the size of the effect — is the result of that model anything more than a reflection of the very assumption (in the model) that the variable has that size effect (per a given magnitude of input of the variable)?
It’s like Weight Watchers (in my example) telling a client: “Well, our model shows that the program caused you to weigh 10 pounds less than you would have weighed without the program. We know that because the model says you would have ended up weighing 210 pounds without the program, rather than the 200 you actually ended up weighing.”
Well, if that model “estimate” of the counterfactual of 210 pounds is based on the model’s assumption of a 10 pound effect, then the model is not really indicating any sort of evidence of the effect, but rather just reflecting the model’s assumption.
Either I’m missing something or nobody responding so far is understanding what I’m asking.
2slugbaits,
Thank you, too, for taking time for my question.
But again, unless I’m missing something here, my question remains unaddressed.
You and Kharris seem to be discussing the question of degree of validity of models for estimating the effect of some variable such as “stimulus” (either before or after introduction of that variable).
My question is: When a model is being used supposedly to estimate in retrospect what the effect of the variable was — as “evidence” of the size of the effect — is the result of that model anything more than a reflection of the very assumption (in the model) that the variable has that size effect (per a given magnitude of input of the variable)?
It’s like Weight Watchers (in my example) telling a client: “Well, our model shows that the program caused you to weigh 10 pounds less than you would have weighed without the program. We know that because the model says you would have ended up weighing 210 pounds without the program, rather than the 200 you actually ended up weighing.”
Well, if that model “estimate” of the counterfactual of 210 pounds is based on the model’s assumption of a 10 pound effect, then the model is not really indicating any sort of evidence of the effect, but rather just reflecting the model’s assumption.
Either I’m missing something or nobody responding so far is understanding what I’m asking.
Perhaps my question will be clearer if I represent my Weight Watchers example in dialogue form – a sort of Abbot & Costello “Who’s on First?” but with the potential circular logic I’m asking about.
Here’s the dialogue…
Costello, Weight Watchers client: If I do your program for one month, how much will I weigh at the end of the month?
Abbot, Weight Watchers manager: Well, you weigh 200 pounds now, and our model projects that if you don’t take the program you’ll still be 200 pounds in one month, but if you do take the program you’ll end up 190 pounds, because the model is projecting that the impact of the program will be that you’ll be 10 pounds less if you do the program than if you don’t.
[Costello does the program for one month, then speaks again with Costello]
Costello: Well, how much do I weigh now?
Abbot: 200 pounds.
Costello: So does that mean the program had no effect?
Abbot: No, it doesn’t mean that. In fact, our model shows that our program caused you to be 10 pounds less than you would have been otherwise. Remember our model? Well, our model says that, without the program, you would now have weighed 210 pounds. So you see, the program had a 10 pound effect, just as we expected.
Costello: But how do you know it had that effect?
Abbot: Because the model says so.
Costello: But how does the model calculate that effect?
Abbot: Because the model says you would have weighed 210, but you only weigh 200. 210 – 200 = 10, so that’s a difference of 10 pounds. Thus, we can see that you are 10 pounds lighter than you would have been without our program.
Costello: But how does the calculate that 210?
Abbot: Because the model factors in an effect of 10 pounds, and you weigh 200. 200 + 10 = 210.
Costello: But how does it get that 10?
Abbot: I told you, it says you would have weighed 210 without the program, but you took the program and you only weigh 200. So, 10 pound difference.
Brooks,
I understood the question, but as slugger notes, models are what they are. In a sense, their results become invalid when you change the assumption on an ad hoc basis. The alternative is to run a sort of sensitivity test, varying assumptions progressively to see what effect that has. Then, you can talk about results under given sets of assumptions. That’s as close as it gets.
Understand, though, that the assumptions at work aren’t all found in the inputs. Some are found in the structure of a given model. You can build a model which, by its nature, doesn’t expect much response from fiscal policy. That tends to be true of models built to examine the effect of things other than fiscal policy. Yes, that’s a bug. A perfect model would take everything into account, but perfect models don’t exist. Often, bad conclusions are reached because the wrong model is used to examine the issue at hand.
That’s where judgement comes in. A good researcher will recognize that a particular model is not appropriate to the issue at hand, and go look for another model.
Even then, the structural assumptions built into the model are not neutral. Again, judgement is needed to come to anything approaching a valid conclusion. When you run into economic debates that seem like a less stupid version of political debate, those tend to be debates about assumptions. In the end, assumptions are just about impossible to get around in the absence of good will, extraordinary honesty and skill. You choose the best models available for the issue at hand, look for consistency among results, try to find reasonable explanations for inconsistencies.
It’s a boatload of work, but a few people do it. More common, and fairly useful, is a sort of “thought experiment modeling” in which smart people (one hopes) think through the implications of what they know. Thought experiments have the advantage of being unbounded by explicit assumption, so more easily get around the problem you have raised. New GDP data change what we know, so anybody who hasn’t, for instance, lined up the new GDP data against ARRA thinking usefully about what ARRA accomplished. Do GDP revisions serve to bolster or harm the case for ARRA helping growth? People who don’t bother asking that question don’t want to allow their pre-baked view of things to be upset.
Kharris,
Thanks again, but again, either I’m missing something or my question remains unaddressed.
In my Weight Watchers example, what’s happening is simply that the model assumes that the program has an effect of 10 pounds, so it will always just adjust the counterfactual to “confirm” that the “results” were indeed that effect of 10 pounds, but that’s not really any kind of measurement of results; it’s simply a reflection of the same assumption the model had all along. The model will always rig the counterfactual to fit the ASSUMED effect.
I’m asking if that’s all that these models estimating “results” of ARRA are doing, and if not, how not (how are they more reflective of likely actual results rather than essentially just a repetition of the original assumptions or something close to that)?
If anyone can answer (or even address) that question I’d appreciate it.
Brooks Try taking your Weight Watchers analogy a little further. There is a theory that is based on well accepted science that you must expend more calories than you consume in order to lose weight. Just buying the product but then helping yourself to a second donut does not invalidate the theory or the forecast if you don’t end up losing weight. Forecasts are always conditional. Always. Always. Always. The model can also be tested against a large population. Other factors like age and genetic history can be cranked into a fancier model. The key is that a diet plan has to be coherent. Same with economic models. If their really isn’t any internal consistency in the model, then there’s a problem. And econometric models are based on empirial evidence through time (time series approaches) and across different economies (panel approaches). That doesn’t mean the model will always predict things exactly, but it’s better than a guess. A lot of what we’re seeing from conservatives is a knownothing approach that says all models are bad so I’m entitled to believe any crackpot theory that comes along. Supply side economics is a case in point.
2slugbaits,
Again thanks for trying, but again you seem to be misunderstanding my question.
I’m not suggesting that a model is invalid because it failed to predict accurately, nor that a model must predict a total outcome (e.g., one’s weight) accurately in order to accurately predict the impact of one factor (Weight Watchers program). I realize there are numerous factors in a total outcome, and the actual values of those factors (and their effects) can be different than expected and can cause a different outcome even if the model was correct about the effect of the variable in question. I get all that. And I’m not one of those idiots who say that lack of decrease in unemployment proves or even implies that ARRA didn’t have any effect of reducing unemployment vs. the counterfactual of not having done ARRA.
None of that has anything to do with my question.
And your response does not seem to address my question at all.
Unfortunately, I don’t know any way of expressing my question that could make it any clearer than what I’ve already said.
But again, I appreciate your attempt to help me with this question.
I wish someone could just actually address it.
some basic math
Assume stimulus was 800 Bn (800 e9) and 3.5 million jobs (3.5e6); simple math shows about 228,500 dollars per job, which sounds like a lot; however, I don’t know how long those jobs lasted – if they are three year jobs, that’s not to bad.
Stimulus/us population, about 300e6 yields 2,700 dolars per capita, or about 11 grand for my family; we are fortunate – over 10 years or so, that is certainly something we could afford.
(actually, probably 3X that – 2X, because 40% or so of households don’t pay income tax,which I assume is where dept service comes from, and another fold reflecting that we are doing ok) still, a manageable amount.
I think this discussion, as so much about obama, is driven largely by the astonishing, almost un believable ineptitude of the obama whitehouse in getting out their message.
What would Roger Ailes or Lee Atwater have done with “porculus” as the limbaughites call it?
Every single *ffing week, they would have had 10 families from each state, saved by stimulus.
Or maybe a big counter, like the debt counter, ticking off the number of jobs…
Or maybe a running TV monitor, every hour, another family saved by stimulus money.
Instead, obama put togehter this astonishingly slick, complex website; when I checked it, I found that in my area, there were like 10 jobs that had been created (I forget the exact number, but it was sort of indistinguishable from zero)
You can see the ineptness of obama most clearly in healthcare; it is inconstestable that a huge majority of americans favor community based insurance fees [aka, no pre exisiting conditions;everyone pays the same rate, based solely on age]; this is in obama care, yet most people don’t know it !!!
That obama says stupid things like desimone deserves his pay doesn’t help either.
That “liberal” websites like thinkprogress or commondreams devote almost all their energy to attacking wackos like M Bachmann doesn’t help either; I guess if you are a lazy web journalist, it is a lot easier to write about Ms Bachmann then to actually do some work on obama initives
The point is, if obama had sold this properly – let people actually see the millions of americans working on porculus money, instead of letting limbaugh make jokes, most people would be a lot more supportive.