Greg Mankiw cites a study by Cogan, Cwik, Taylor, and Wieland to buttress his arguments that fiscal multipliers are small, especially when considering New Keynesian models. He also provides a startling graphic showing the dynamic multipliers from Romer-Bernstein versus the Taylor (1993) model, incorporating model consistent expectations; this graphic motivates Wieland et al. to remark:
We first show that the assumptions made by Romer and Bernstein about monetary
policy — essentially an interest rate peg for the Federal Reserve — are highly questionable
according to new Keynesian models. We therefore modify that assumption and look at the
impacts of a permanent increase in government purchases of goods and services in the
alternative model. According to the alternative model the impacts are much smaller than
those reported by Romer and Bernstein.
Cogan et al. use a New Keynesian dynamic stochastic general equilibrium (DSGE) model, specifically the Smets-Wouter model (Working Paper version of AER paper here).
In contrast, here’s some remarks based on results from the IMF’s New Keynesian DSGE (The Case for Global Fiscal Stimulus).
…the effect on U.S. GDP of
investment expenditures is 3.9 when there is global fiscal expansion and only 2.4 when the United States acts alone. Similarly, the effect on Japanese GDP of targeted transfers is 1.5 when there is global fiscal expansion and only 1.0 when Japan acts alone. Differences in multipliers across regions relate to the size of leakages in the different areas, including leakages into saving and imports. [italics added]
Here’s a graph summarizing what happens when monetary policy is held fixed for two years, and then is allowed to revert to a Taylor rule thereafter.
Figure 3 from The Case for Global Fiscal Stimulus. Assumes monetary accomodation in the two years of the fiscal stimulus, and stimulus equal to 1 ppt of GDP in year one, and 0.5 ppt of GDP in year two.
The obvious question — why the disjuncture? They’re both New Keynesian DSGEs? (documentation for the IMF model, GIMF, here).
While I don’t dispute the conclusion that in the Smets-Wouter model implies small fiscal multipliers when monetary policy follows more Taylor-rules, I do dispute the conclusion that this is a general conclusion relevant for all New Keynesian DSGEs. After all, in the IMF simulations, monetary policy reverts back to a Taylor rule after two years, and yet they obtain large multipliers.
Hence, I think New Keynesian DSGEs — like any other large models incorporating many equations — can yield differing results depending on the assumptions made, some of which might seem inconsequential or conventional; what I have in mind include those assumptions regarding the nature of asset markets. (The Smets-Wouter model has a completely standard assumption of complete markets, a generic one-period bond that all households can access. The IMF model incorporates an overlapping generations framework. The differences in the two approaches yield drastically different implications for fiscal policy, as has been discussed in the context of fiscal consolidation — see Chinn (2005).)
Now, I don’t believe that DSGE’s are useless for policy analysis, as Willem Buiter has argued. And I count myself a New Keynesian. But I also believe that just because a model has microfoundations, intertemporal optimization, model consistent expectations, and so forth, doesn’t mean it necessarily provides more plausible estimates. It matters what assumptions are made (is “complete asset markets” a good assumption, for instance? And if they were three years ago, are they still now? And I’ll bet the proportion of “rule of thumb consumers” or “liquidity constrained consumers” is probably higher now than three years ago…).
Link to previous installment of The Great Multiplier Debate
.Technorati Tags: multipliers, fiscal policy,
New Keynesian, monetary policy, intertemporally optimizing, microfoundations.
This is way to confusing to me.
I am against the DSGE. The way they solve their model is to assume and the end of time there is a steady state, and hence everything unravels and is defined from that point. It is completely ridiculous to assume that is the way people in the present make choices (It is also ridiculous to assume that an economy has a steady state when billions of functions and decisions could be non-linear).
If someone gets a job through the stimulus bill, he won’t say to himself, well gee, I’m going to be taxes tomorrow, thus, I won’t increase my consumption today. He will likely go nuts. But it will be all the more likely he won’t know his job was from the stimulus bill anyway. And if he is like 95% of the population, he won’t know even what a Stimulus bill does.
Surely any discussion of multipliers, esp. with implicit stochasticity, should include a reference to the Schartz-Metterklume approach?
Why would a DSGE model describing the current situation include a Taylor rule? The short-term interest rate is at zero… This also suggests that the amount of crowding out is likely to be overstated in these models because interest rates aren’t going to rise (all that much) in response to fiscal expansion.
Menzie-
From your 2005 Twin Deficits paper:
My contrasting view is that:
While I do understand that you and Dr. Hamilton are not financial market experts, I continue to be perplexed by your avoidence of the regulatory capture/derivative explosion issue. WHAT GIVES?
MarkS: I think I’ve been pretty critical of the deregulatory zeal and private sector fraud that took place under the Bush Administration: see [1], [2], [3]. That being said, I think derivatives are only part of the problem, as indicated in this post.
Sorry to infer that the Econobrowser has been avoiding derivatives and regulatory capture- I did find JDH’s comments at the Fed’s Jackson Hole Conference in 2007 “Borrowing Short and Lending Long” . In it he pointed fingers at many of the culprits.
Menzie- Thanks so much for providing the links to your previous comments on looting, fraud, and lax regulation. I am especially interested in the 1994 Akerlof/Romer “Looting” paper. I’ll try read it at my local library… Much Obliged!
Is this a parody? Do real, grown up, economists debate which model is better without ever referring to the real world? In most social sciences “a study” means you went into the field and gathered some data, then did a bit of straightforward statistical analysis on it to see if there was some generalization possible.
Something along the lines of giving vaccinations against measles correlates with a drop in cases, or better nutrition during childhood leads to taller adults.
Papers like this and those who then engage in debating the “results” are more like medieval scholastic scholars debating some theological point or other. Amazing.
If there is a 1.5 increase in GDP when the government spends money why isn’t there a 1.5 decrease in GDP when they confiscate the money to spend?
To add to robertdfeinman’s comments, there is no indication in the summary that this model includes the low cost manufacturing of China and S Asia, China’s immense reserves of US dollars and the impact that has on the US economy, US and Euro speculation in housing funded by high risk loans, the impact of unregulated credit default swaps, and the impact of high risk loans to Eastern Europe. It is beyond belief that anyone could find useful conclusions with practical value in our current global situation from this model. It lacks fundamental raw material.
It is better to be 60% correct on a problem statement that is 100% on target, than to be 100% correct on some problem statement (like the one summarized up top) that is only 60% on target to the situation.
It seems to me we are in a tight spot as far as research on the multiplier goes. Models such as that used by Cogan (a formidable economist) et al. basically assume the results (unless you argue that they cover the range of likely possibilities). But then what of this: “Keeping interest rates constant for two years still does not seem very realistic and would likely result in an increase in inflation, but it is certainly more realistic than pegging the interest rates at zero forever, or even for four years.” (I, for one, see nothing unrealistic in pegging the rate for four years – look at Japan.) Thus, they assume the fed is able to raise the nominal interest rate after 2010. I doubt that we will be out of our current malaise by then.
As for empirical estimates, they suffer from a dirth of observations (situations like the current one in which we have a liquidity trap).
I think we need a reality check. These models are no better than models that try to predict weather or climate change. The bottom line is that the world is far too complex for economists to model and anyone who accepts the models as useful or accurate is being naive. What we need is to look at the real world and the verdict there is that government is a burden on the productive classes. If we want to end this mess the best way is to get government out of the way and to let the markets get rid of the malinvestments. That means much smaller government and much lower taxes.
I may be a bit dense in these matters, as I am a physicist, not an economist, but I can’t accept the idea of multipliers at all. What makes a group of players – such as the government – more valuable for the GDP than any other group? To use reductio ad absurdum it surely implies, that all spending should be by that priviledged group, to maximize the GDP. Nobody should be allowed to spend at a smaller than 1 multiplier, which surely means built in bias for communism.
Where do I go wrong?
But I also believe that just because a model has microfoundations, intertemporal optimization, model consistent expectations, and so forth, doesn’t mean it necessarily provides more plausible estimates.
In most sciences, we would then say then it isn’t even a model or theory, if it does not provide a useful prediction. Sort of like the “how many angels can dance on the head of a pin” debate.
What use is a model so strongly dependent upon an imaginary parameter with little empirical support that makes no valid estimates? At best it provides a false pretense of mathematical validity.
Menzie is very diligent about these matters and will undoubtedly step in with his own replies, but I guess I cannot resist replying to some of the comments above (which may not fully coincide with what Menzie will say anyway).
Robert D. Feinman,
We all know, or should, that economics is a very imperfect science. These models are in some sense constantly being tested against the ever-evolving reality that is the economy. Part of the problem is that we are experiencing dramatic shocks and events that are outside the ranges of data on which the models were estimated, with the old problem of the Lucas critique also looming over things, that people change their behavioral response functions in response to major changes in policy, and we are seeing major changes in policy along with these major shocks. That we are not able to adjudicate ahead of time which of the competing models (or other non-DSGE alternatives) will perform better under these circumstances is not surprising.
DickF,
Sorry, but this is elementary principles textbooks stuff. A tax increase only reduces economic activity to the extent that the people who have their taxes increased actually reduce their expenditures, and we do not expect them to do so in a 100% way. However, spending increases are 100% increases in spending, more than offsetting any tax increase.
Of course in the current situation, while taxes are to go up somewhat on some upper income groups, most people are getting tax cuts, which will reinforce the spending increases. Yes, this will be coming out of future generations, but hopefully the growth stimulated by all this will make it easier for future generations to handle the higher debt.
In any case, it is pretty much accepted that at least in the short run, spending changes have bigger impacts on aggregate demand than do equivalent tax changes, that the former have larger multipliers than the latter, to put this into the terminology of the current thread.
Vangel,
Yes, the economy is complex and hard to model very well, at least as bad as trying to do so for the weather or climate change, and I am not aware of any econometrician who has ever claimed otherwise. However, do you ask the meterologists and climate modelers to stop it?
Kuthi,
No, this is not about the government being a “privleged group.” The multiplier from private investment is just as big (or even bigger) than that from government spending. The problem is that private investment has collapsed, as have exports, which is what happens during bad recessions and depressions. That is why people fall back on the government to carry the ball in jump-starting the economy again.
Steve,
“Imaginary parameter”? You may be right that the empirical support for the value of the parameter is weak, but that does not make it “imaginary.” If you wish to argue that most chemistry and physics models are more accurate in their predictions than most economics models, you will not be revealing anything to anybody that has not been well known all along. Do you wish to shut down all econometric modeling in the way that Vangel seems to appear to want to shut down all climate modeling?
Oh, and for all the strutting physicists here, last time I checked there was a big fat zero amount of empirical evidence either for or against string “theory.” Does this mean that it is therefore neither a “model” or a “theory”? (And, indeed, there are some, such as John Horgan and Lee Smolin, who have rather pointedly raised such issues.)
Barkley,
I am not talking about the impact of tax cuts. If government spending increases GDP by 1.5% then government taking would reduce GDP by 1.5%. You cannot look at only one side of the transaction.
Now you may be correct that we will spread the 1.5% reduction over many years in the future or pass it on to posterity but the Piper must be paid at some point.
There is a saying, “You cannot have your cake and eat it too.”
P.S. You are correct that it is “elementary principles textbooks stuff.” This is exactly what I was taught in Macroeconomics. It is one reason that economic education is so dismal today. Most economics today looks at only part of the transaction and assumes that it is the whole, very poor logic. This is why 17th Century mercantilism has had a resurgence in education over the past 60 years.
DickF,
Sorry, but this sort of argument is only meaningful in a fully employed economy on its production possibilities frontier, a world of opportunity costs where government-purchase guns must come out of the resources that otherwise would have produced private-purchase butter (or whatever). That is probably why the estimates of spending multipliers from WW II cited by Robert Barro do not find them exceeding 1.0. But we are not at all in such a situation: there are plenty of unemployed construction workers sitting around ready to pick up those “ready” shovels that the stimpack will fund.
Thanks to Barkley Rosser for stepping in the breach. I admit I could not face answering some of the questions for the umpteenth time. In particular, I tire of the pseudo-intellectual nihilism of those who say this or that (or any particular) model is useless, without saying why they think the model is useless. For instance, Vangel asserts the models have no predictive power. Well, in fact they do, in the sense they have been calibrated to fit the actual real world data. The issue is that, as a consequence, they have about as much predictive and/or forecasting power as the standard old-fashioned structural macroeconometric models that Cogan et al. critique. And so we cannot differentiate on that basis (and in fact an atheoretical VARs might be even better, despite the fact that there is minimal theory involved in the construction/estimation of such models.)
don‘s point is more useful. The question is whether we can use our models during periods when we’ve hit the zero-interest bounds. Indeed, I’d say that it probably isn’t a good idea to use a flex-price RBC with perfectly working asset markets to model this type of situation — which is exactly the point I was trying to make with this post.
I must confess I don’t understand Kuthi‘s point. But I suspect that if he read a standard macro textbook, he could understand the concept of the multiplier, and then he could identify exactly what points in the standard model he agreed and disagreed with. But Barkley Rosser is exactly right that the argument regarding the multiplier typically presupposes some slack in the economy. The nature of that slack, and whether it can be eliminated depends on the source (misperceptions a la Lucas) or monopolistic competition (a la any number of models, including New Keynesian), or sticky prices, or combinations thereof.
Sorry for (another) off topic question. I wonder if this, the ramblings of another armchair economist, makes sense?:
Assume the baseline scenario is persistent high unemployment, and the central bank continue printing money ‘behind the scenes’. I don’t think it matters whether they simply load it into bank reserves, or use it to buy crud from banks. Sooner or later, after a period of deflation and sustained central bank profligacy, we might get positive inflation again.
But then Fisher’s debt deflation goes into reverse and inflation (and expectations thereof) will boost the economy. This growth will cause even more inflation, giving us hyperinflation.
In short, if we get 10% unemployment and 2% inflation, is it not certain that the ensuing lower employment cause even higher inflation?
(The expectations of hyper inflation would make things even more difficult.)
Therefore, the only way to get from here to low unemployment and low inflation is by pumping the freshly minted (virtual) money directly into spending people, via Keynesian stimulus?
the empirical support for the value of the parameter is weak, but that does not make it “imaginary.”
Well, it sure doesn’t make it “real”, does it?
Let me put it another way: if your model is heavily dependent upon unobservable parameters, it ceases to be a predictive model and becomes an exercise in “curve-fitting”.
Do you wish to shut down all econometric modeling
No, I’d like to see econometric modeling that doesn’t immediately fail when applied to the real economy. Unfortunately It’s important that major policy decisions are being made based on a flawed assumption of the existence of a significant GDP multiplier from government spending. When that’s taken away, it becomes obvious that you are gambling that tax revenue growth will outrun compounding interest, as expressed by:
but hopefully the growth stimulated by all this will make it easier for future generations to handle the higher debt.
with catastrophic consequences if you are wrong. What if the multiplier is only 0.6 after the embezzlement factor? Oops, sorry, grandchildren.
In particular, I tire of the pseudo-intellectual nihilism of those who say this or that (or any particular) model is useless, without saying why they think the model is useless.
Criticizing the mainstream neo-Keynesian view is not nihilism, it’s just uncomfortable when you realize that your macro textbook hasn’t worked out well in practice. The specific models I’m talking about that involve large GDP multiplier effects of government spending are useless because the only plausible stimulative mechanism, increasing the marginal propensity to spend, cannot mathematically be higher than a few percent. It’s absurd.
For instance, Vangel asserts the models have no predictive power. Well, in fact they do, in the sense they have been calibrated to fit the actual real world data.
Are you seriously saying that fitting your model to the data after the fact is the same as prediction? Given enough free parameters, you can fit any theory to any curve. Induction is the opposite of prediction.
Multipliers (for Dickf and others)
To be very precise, all popular economic models have multipliers. However, the classical model has a vertical aggregate supply, so the multiplier has no effect on output. Here, the multiplier occurs because there is a positive feedback between aggregate demand and the components of aggregate demand. If I print more money to buy ice cream, that is someone else’s
income that rises, and he spends some of it, and so on down the line until there is no new spending (it is all accounted for in savings).
Now, the reason to think that aggregate supply is flat right now, is well, as evident as it was when Keynes first began writing the General Theory.
Steve: I wasn’t objecting to criticism specifically directed at New Keynesian models. I was objecting to criticisms of essentially all models, in favor of “common sense”, “reality”, and other vacuous and un-implementable approaches. In particular, no alternatives were specified, so I think it right to characterize that as “nihilistic”.
By the way, your statement “the only plausible stimulative mechanism, increasing the marginal propensity to spend” is gibberish, and suggests that we are talking about different things.
I am sorry but I can’t see how people who missed the crisis are supposed to be credible. The bottom line is that these models are not any better than a coin flip in predicting future events and most of the people who talk them up are, in the words of Nassim Taleb, empty suits. These people think that they know much more than they actually do and as such offer dangerous advice.
We need to keep in mind that the Austrian School economists saw the crisis coming and explained why it was inevitable so if any economists should be listened to it is them. And for the life of me I cannot understand why people pretend that Keynes’, General Theory makes sense. My only guess is that they never actually read it carefully or critically.
I am sorry if I gave the impression that I am strutting about proudly. In fact, I am quite ignorant of basic textbook economics – as kindly pointed out by both professors. Also, I am only an interested layman, and won’t take up another Ph.D. just for the fun of it. Frankly, economics is not that fascinating, especialy if fundamental concepts such as multipliers cannot be described in a few well chosen words. Sorry to have been of obvious but unintentional offense. Think of my puzzlement and question about multipliers as over a nice dinner from one of your guests…
I was perhaps misled by the lack of sophistication in the general discussion, it gave me the impression that maybe a physicist could join in the fun.
Again, sorry!
Kuthi: I merely assumed that as a physicist, you’d be able to see the logic in the multiplier model, and then see where you would disagree (I’m not saying that one has to agree with the Keynesian framework).
Suppose aggregate demand determines output; this presupposes slack. Otherwise, supply factors would come in. Assume an injection of one dollar of exogenously determined government spending. GDP rises by $1 since government purchases is a component of GDP. Anything produced yields income. Assuming no taxes, then then s proportion (0 less than s less than 1) of that $1 is saved, the rest consumed (c proportion). Consumption is part of GDP; hence GDP rises again, by $c. But consumption goods had to be made, so that becomes income for others, of which c of that $c is spent. The sum of the infinite geonmetric sequence:
1+c+c2+c3 = 1/(1-c)
See also this post.
Menzie: You are assuming that a complex, non-linear system like the economy can be modelled accurately but the evidence suggests otherwise. Le me repeat my previous statements again; the modellers missed the crisis and are only good at providing a narrative about the past. But anyone can look to the past and make up explanations.
Keynesian economists either imply or directly state that theirs is an empirical science. Most classical economic theories hold that economics is a behavioral or logical science. If Keynesian economists follow an empirical science then Keynesians should be able to use their theories to predict economic events.
Many of the classical economists of the day predicted the Great Depression. Keynes never predicted it. Irving Fisher never predicted it and, as a matter of fact, lost nearly everything he had investing basing it on his mercantilist beliefs.
Fast forward to our current Depression. Which Keynesian predicted the crash? I can list many classical economists who predicted it.
Keynesians fall into error because they equate equations with truth. Consider, just because a doctor can develop an equation for multipliers to calculate changes in blood pressure and can calibrate his constants to actual measurements of blood flows does not validate the practice of blood-letting.
Classicals criticize blood-letting for harming the patient, while Keynesians say classical criticism of blood-letting is invalid because they do not disprove their equations of blood flow and blood pressure.
As Barkley states most studies find that the multiplier is actually 1, in other words, not a multiplier at all. But Barkley uses the normal justification “things are different now.” In this case he cites unemployment, but there is no consideration that the Keynesian policies economy actually created the unemployment in the first place.
Debates of multipliers are debates on how many angels can dance on the head of a pin.
DickF: I am curious — what are the names of these many Classical economists that predicted this crash?
I must admit that I am much like a reformed alcololic or someone who has just quit smoking. Because I see how destructive my vice was (Keynesianism) I am strong evangelist against the vice. Sometimes I may over do it.
Vangel: I think this is a relevant aphorism: “Even a broken clock is right twice a day.”
What kind of specific complex nonlinear system are you talking about? So is it characterized by deterministic chaos? Is the characterization based upon an empirical test (BDS statistic, etc.). Is complexity of this variety implied necessarily by Austrian school models, and are they observationally differentiable from other types of models, say, incorporating Bayesian updating?
Menzie,
I also approach the problem of multipliers as an ex-scientist, and remain puzzled. I can follow the multiplier maths, but I would ask (a) what evidence there is that there is (involuntary) slack and, more importantly, (b) Dick F’s (not to mention Fama’s and Cochrane’s) question about why the government does not reduce output when they obtain their funding as much as they increase it by spending.
I am sorry if you are tired of discussing it, but you must appreciate that the multiplier idea is somewhat magical – presumably why it was such an innovation – and that a rigorous (ie anticipates and answers obvious questions) explanation will be needed to overcome the scepticism that it naturally arouses. Are there any textbooks that you would recommend for covering fiscal expansion and the multiplier especially convincingly?
Menzie wrote:
DickF: I am curious — what are the names of these many Classical economists that predicted this crash?
Let me just mention one because I sense a credential attack may be coming.
Frank Shostak
“What kind of specific complex nonlinear system are you talking about?”
It is called the real economy. The one that you can’t model because it is too complicated.
“So is it characterized by deterministic chaos? Is the characterization based upon an empirical test (BDS statistic, etc.). Is complexity of this variety implied necessarily by Austrian school models, and are they observationally differentiable from other types of models, say, incorporating Bayesian updating?”
I believe that it is clear that many on this board accept the mainstream assumption that uncertainty can be modelled mathematically, and that modellers should reach the same conclusion regarding a particular events probability, if they are provided with the same information. That suggests that the problem with modelling is not the process but the information.
But the Austrian School, which saw the crisis coming, points out that each individual has a unique set of preferences, a unique information set and a unique way of processing that information. This means that even if all economic actors were provided with the same information, their estimation about the importance of that information and their method of evaluating it will lead to different conclusions and different behaviour. There is no mathematical method and tool set that will allow economists create models that will make accurate predictions. The Austrians do much better because they ignore the mathematical models and stick to logic and sound theory. Given the financial and monetary environment they found it very easy to explain that a very large crisis was on its way and that the only hope was to have the Fed temporarily push that crisis into the future.
I would like to add more but I have a few chores to run at this time. We can take this up later.
Menzie,
I was somewhat pulled into an empircist discussion so I need to make sure that readers understand.
Most classical economists do no focus on predicting events. Classical economists describe the probable actions people will take when they are faced with certain events. Because classical economics is a behavioral or logical science it understands that conditions change based on behavioral reactions to conditional changes. For example if the government plans to increase taxes people will begin to reduce their spending before any governmental action is taken. Similarly when the government begins to talk of decreasing taxes then people will increase their consumption and production based on the anticipated change. There are not equations that tell a behavioral economist whether a person will choose a hamburger or pizza for lunch.
The burden of proof falls on those who claim to be empirical scientists. If their equations and predictions do not give answers then they are proven wrong. This is why it is illustrative when classicals see behavior bringing on a crash when Keynesians (Keynes himself) miss predicting the event.
Rebel Economist,
1) The unemployment rate is rising, now at 8.1%, with most of that recent increase being due to people being laid off, not quitting their jobs. I would say that most of them are “involuntarily unemployed.” I even know some personally. Don’t you?
Another piece of evidence of “slack” regards capital stock utilization: it has fallen from over 80% to 72%, although I think the voluntary vs involuntary issue is not answerable on this. But, in any case, there is plenty of capital stock and plenty of laid off workers around who would like to work. Plenty of slack.
b) This has been answered. When the government orders infrastructure to be built and provides the money for it (borrowed), and unemployed workers and capital stock are employed to do it (slack, you know, what we have right now), then output goes up. The “magic multiplier” effect happens when those people who are now getting money they did not previously have go out and spend some of it to buy things they would not have, thereby leading to more output increases.
A change in taxes only brings about a change in output if people experiencing it change their spending (or investing). In effect, any impact on demand from a tax change is a pure multiplier effect, whereas a change in spending by the government leads to a direct change in output followed by a multiplier effect.
This is why the textbooks describe the so-called “balanced budget multiplier” as equaling one. An equal change of spending and taxes in the same direction lead to the change in output due to the spending change (either up or down), while the pure multiplier effect from the tax change simply offsets the multiplier effect from the spending change, leaving the spending change still there and changing output (again, assuming some slack), hence, multiplier equals one.
This discussion can be found in most macroeconomic principles textbooks. I am not going to get into recommending any in particular.
DickF,
I did not say “most studies find the multiplier is actually one,” and this is not true. I did say that Robert Barro found this (actually for it to equal 0.8) for WW II. But WW II is an exceptional case of the economy being extremely fully employed, arguably more than fully employed, with Rosie the Riveter going to work from the kitchen, and prices being kept in check by price controls and rationing, the latter also forcibly keeping people from spending their extra earnings, and thus forcibly keeping the multiplier lower than it would be outside of full-blown wartime, quite aside from the absolute lack of any slack (and the US economy normally has some slack in peacetime).
There certainly has been a vigorous debate and many studies about the size of the multiplier, but as this has involved the study of actual data from real economies, it has not been similar to debating how many unobserved angels can dance on the head of a pin. Most careful observers are fully aware that multipliers can vary over time and also vary across different kinds of spending or tax changes at a point in time (with short-run multipliers for tax changes generally lower than for spending changes). Such variety is indeed seen in the initial posting by Menzie and in the debates between the competing models.
BTW, the garden variety “common sense” and back of the envelope guesses about spending multipliers in a mildly slack economy generally have run around 1.5 to 2.0 or so in the past.
I had never heard of Frank Shostak before. I do not think he would qualify as a “classical economist.” He is an Austrian of the Misesian branch, an adjunct scholar with the Mises Institute, supposedly a “former economics professor,” although I could never determine where, and also “chief economist” of M.F. Global, which claims falsely to be the world’s largest exchange-related derivatives trader. They were just sued for fraud and deception today for 30 million UK pounds by one of their former traders.
Of course, that has nothing to do with Shostak. I did look at some of his past postings at the Mises Institute. Back in April, 2003 he forecast a crash in the housing market for 2004. This seems to be about the quality of the forecasts by him in general. Mises was forecasting a crash from 1924 on. When it arrived, of course he claimed credit. Then in 1931, when the UK went off gold, he forecast that the pound would be “worthless” within a week. Oooooops! (Hayek has been reported, with some controversy, to have forecast a stock market crash in a newsletter in Feb. 1929).
Anyway, it looks like Shostak’s record of forecasting is about as good as Mises’s, his hero, sometimes not too bad, sometimes awful. But we get all these rantings about how great the Austrians are at forecasting (obviously this is relevant for Vangel also). And, the Austrians do not like the real classical economists like Milton Friedman. Him a commie.
Oh, and in some of his most recent posts, Shostak has been forecasting that the stimulus will not work, because when the government tries to build all that new infrastructure, there will not be any savings there, given all the low interest rates and so on. Except, of course, that as part of the recent decline, the US savings rate has risen from near zero to about 5%. Oooooooops! But, of course, we are not supposed to pay any attention to the incorrect forecasts or statements about mere data by such wise gurus as one finds hanging out at the Mises Institute.
Vangel,
Well, I am a hardcore complexity guy, as either Menzie or Jim Hamilton can tell you. I am even responsible for defending publicly Hayek as an early important figure in this (see my paper on “How Complex are the Austrians,” available on my website). But the sorts of claims you make about handwaving forecasts based on no models are not all that impressive.
I am also one of those who was forecasting a crash, but only really loudly began doing so in late 2006, including in an address at the Paris meeting of the Society for Nonlinear Dynamics and Econometrics in March, 2007 (Jim Hamilton was a plenary speaker at the next meeting of that society a year later in San Francisco). I did not predict the timing, however, although I pretty much laid out how it would go, housing bubble crash undermining housing-related derivatives, leading to broader financial crisis and collapse through all the overly entangled and opaque linkages.
I will simply close by noting that Menzie’s question was not silly. There are lots of ways to model macroeconomies and financial markets using “nonlinear complexity” methods, as there are a variety of those. Some have been doing better in this situation than others, but none are perfect. However, they are certainly not any worse than Frank Shostak in 2003 calling for a crash of the housing bubble for 2004. Duh.
DickF,
Your latest post showed up after I wrote a long post that has not yet appeared (I presume it will and hope so, as I spent quite a bit of time composing it). I will simply note that your supposed “classical economist” who is not one, Frank Shostak, seems to like to make all kinds of forecasts. However, his track record is not at all impressive, as I noted in my long post, still not up yet.
Thanks, Menzie.
I found two papers on this multiplier business in the local economy perspective from http://www.uaex.edu/other_areas/publications/pdf/fscdd-6.pdf and http://research.stlouisfed.org/publications/review/91/01/Consumer_Jan_Feb1991.pdf from the St. Louise Fed, on Google. My first reaction is shock (and disbelief) over the assumptions going into the derivation: the assumption of continuity and the existence of a derivative, when it is obvious that the granularity of spending and even the reactions to increased income from additional sales grossly invalidates those assumptions. I do think that there is place for mathematics and especially for statistics in macroeconomics, but let us not pretend that meaningful results will arise from such misuse of calculus.
I happen to hold that morality is at the root of economics – that is why societies with relatively ethical populations can live well even in the absence of natural resources while tribalized corrupt societies will always stay near the subsistence level – but apart from the ability to print money the government is exactly like any other entity or group taken randomly from the population, so its spending pattern cannot by definition be different from that of any other economic entity.
Now you may say, that since the rest of the population is not spending, the government should spend, but it begs the question why the rest is not spending (maybe they can’t find profitable things to invest in as prices loose their indicator status) and why is government spending inherently profitable? Infrastructure? A bridge to nowhere just destroys natural resources. If it made economic sense, a private group may have done it already. I only can think of altruistic spending by the government as helpful now, i.e. housing the homeless and feeding the hungry and financed by straight printing of new money as the sudden deleveraging of debt sucks money out of circulation. The rest is not possible as the people in government at best are only as good and only as clever as people in the general population.
I hope I am not trying your patience with my long rant,
Sincerely,
Andy Kuthi
“Well, I am a hardcore complexity guy, as either Menzie or Jim Hamilton can tell you. I am even responsible for defending publicly Hayek as an early important figure in this (see my paper on “How Complex are the Austrians,” available on my website). But the sorts of claims you make about handwaving forecasts based on no models are not all that impressive.”
First, I did read your paper, How Complex are the Austrians some time ago. I just looked at my notes and found several critical points that I had written. From what I can tell you have not read enough of von Mises. In the paper you wrote, “von Mises would never abandon the equilibrium perspective, which he defended strongly.” But you do not seem to acknowledge that von Mises claimed that equilibrium was an artificial construct that did not represent the real economy. I refer you to Section 3, Chapter 16, of von Mises’ great book, Theory and History. He ends that section by writing, “However, economists must not confuse this thymological notion of equilibrium with the use of the imaginary construction of a static economy. The only service that this imaginary construction renders is to set off in sharp relief the ceaseless striving of living and acting men after the best possible improvement of their conditions. There is for the unaffected scientific observer nothing objectionable in his description of disequilibrium. It is only the passionate pro-socialist zeal of mathematical pseudo-economists that transforms a purely
analytical tool of logical economics into an Utopian image of the good and most desirable state of affairs.”
And I do not consider the Austrian predictions very impressive because anyone with a clear mind and a knowledge of economics should have seen the crisis developing. And no, I do not expect anyone to get the exact timing right because it is impossible to determine when reality will intervene in the real economy and force a revaluation and rethinking on the part of the players.
“I am also one of those who was forecasting a crash, but only really loudly began doing so in late 2006, including in an address at the Paris meeting of the Society for Nonlinear Dynamics and Econometrics in March, 2007 (Jim Hamilton was a plenary speaker at the next meeting of that society a year later in San Francisco). I did not predict the timing, however, although I pretty much laid out how it would go, housing bubble crash undermining housing-related derivatives, leading to broader financial crisis and collapse through all the overly entangled and opaque linkages.”
As I pointed out, we can only get the timing right by accident because there is no tool that would permit us to predict the exact event that would trigger the correction.
“I will simply close by noting that Menzie’s question was not silly. There are lots of ways to model macroeconomies and financial markets using “nonlinear complexity” methods, as there are a variety of those. Some have been doing better in this situation than others, but none are perfect. However, they are certainly not any worse than Frank Shostak in 2003 calling for a crash of the housing bubble for 2004. Duh.”
While I disagree with the way that Shostak handles and interprets some of the data I do not think that he was suggesting that the housing bubble would burst in 2004. The way I read him, he simply points out to weaknesses in the system that make it vulnerable to external factors that will trigger an economic event. Shostak was consistent by pointing out that the economic ‘miracle’ was caused by monetary expansion on the part of the Fed and the financial system. He was clearly correct but if he expected to get the exact time of the correction right he is way too overconfident in his abilities to predict something that is not really predictable.
My approach on this topic is similar to that taken by Nassim Taleb who points out that most smart people in the financial industry and academia are empty suits because they don’t really understand what is necessary and overestimate their ability to get things right.
Two further remarks here.
1) Wise observers of mathematical or econometric models do not generally just blindly believe in them. Indeed, one sees careful examination of how and why they differ in their predictions. Menzie was doing that here in this post, and it should be kept in mind that in the crucial apparent differences between the two models, the one that is looking less realistic is the one that also generates the lower multipliers.
2) For all those folks out there who think that multipliers are imaginary or never greater than one, I would suggest you think carefully about the current downturn. Lord Kahn first cooked up the multiplier to explain how downturns can go down so hard, that a decline will multiply itself. So we see now. The US slows down and stops buying Chinese imports, who then stops buying Taiwanese imports, and so on. Some of the most important multipliers are those involving foreign trade, and they are working now with a vengeance to turn what was initially a more localized downturn into a full-blown global recession.
Barkley,
I don’t think just being laid off makes a worker involuntarily redundant; I think that it is also necessary to decline the jobs that are available, including, say, picking crops. There has been some proportion of economic activity in the USA that was never really sustainable (eg in finance) and I cannot see those workers accepting their next best opportunity, for a while at least. I am closely acquainted with at least one such worker. I dare say similar arguments apply to industrial capacity (eg in the Detroit car industry). It is just not economically viable to operate in the USA and may as well be scrapped.
Your answer to part b fails to explain why those who pay for the stimulus do not cut their expenditure by the same amount as those who receive it. I do not see why it makes a difference whether the money is extracted by taxation or borrowing. Either way, they have less to spend.
Rebel Economist,
If one declines some available job, then one is most clearly not involuntarily redundant. I think you did not mean to say what you did. But, if what you wanted to say was to support the new classical idea that there is no such thing as involuntary unemployment (or redundancy), then you must assume or believe that all laid off workers are being offered jobs. Are they? Even if there might be other jobs, they may be too far away or hard to find for the laid off workers to find them, and this gets harder the higher is the unemployment rate and the worse off the economy. Do not just assert that all those laid off workers can go pick crops or whatever. Some cannot, really.
If the money was just going to sit in banks or other financial institutions and not get loaned to a private interest who wishes to do some spending due to low expectations of future prospects (lack of “animal spirits”), then lending to the government to spend may indeed increase the economy.
Menzie Chinn & Barkley Rosser,
Your patience is exemplary.
Barkley,
Thank you for the well thought out and extensive post.
I do need to correct one thing. Milton Friedman was a Keynesian not a classical economist.
I apologize. Only the past few days have I read any of your writings other than here. I did enjoy them especially those comparing Austrian economics to other disciplines though I would have liked more depth of economic analysis.
Your admission that you did not know Frank Shostak is telling of your limited knowledge of Austrians. I do not know if you are familiar with Roger Garrison but I recommend him. Shostak is a little too monetarist; that taints his Austrian credentials for me.
To the question at hand. One of Hayek’s strongest criticisms against Keynes was that his theories only operated at the extremes and he did not allow for incremental changes in conditions and responses by economic actors. This is especially true concerning unemployment. Considering only full employment and significant unemployment leaves out most economic activity because neither of these conditions is common.
There can be temporary changes in economic conditions with a Keynesian money-based approach but as I have stated before it is actually only borrowing from Peter to pay Paul. Unless you analyze totally economic events surrounding a condition you will be mislead.
Money solutions are illusions because they do not change the foundation of wealth and that is the supply of goods. Money in itself will not feed a child or put shelter over your head. I realize that the Keynesian assumption is that the consumption of goods will stimulate production but that gets the cart before the horse and actually feed inefficiencies such as inflation and destructive malinvestment of capital. These things are seldom taken into account by the theorists of multipliers.
I simply do not understand why economists today hold to questionable indirect effects on economies when the direct approach is not questionable, as is clearly demonstrated by those economists in the Adam Smith tradition. The favorite mysticisms of today, stimulation by increases in the money supply or manipulation of interest rates, only distort economic signals such as price and are rife with unintended consequences.
robertdfeinman writes
“Is this a parody? Do real, grown up, economists debate which model is better without ever referring to the real world? In most social sciences “a study” means you went into the field and gathered some data, then did a bit of straightforward statistical analysis on it to see if there was some generalization possible.”
This is why it is so difficult for economists to be understood. Economics is a much more difficult science than, say, physics. You have got to take into account market expectations about the future. The approach outlined by robertdfeinman suffers from the Lucas critique / Goodhart law / not to mention the post hoc ergo propter hoc fallacy.
The apple doesn’t have to worry about the expected duration of its fall, when it falls. Wolves don’t worry about the rabbits’ reproductive rates, when they decimate them. But consumers and producers must form expectations about the future value of prices and other economic variables. And so must economists. The “animal spirits” exhibited by humans, so aptly described by Keynes and some Keynesians, is a situation where expectations are unstable, shifting from optimism to pessimism at the slightest change of information, making humans look like animals in a panic.
In a nutshell, there is no “straightforward statistical analysis” available here. Is it a surprise, then, if some of the most remarkable advances in statistics have come from econometrics?
robertdfeinman would be well advised to take an introductory course in economics. Econ 101 would do.
d4winds,
If there is one thing that really annoys me about the Krugman types it is their condescending attitude to people who question them. It almost turns me against them, before my scientific training rescues me and I realise that it is what they say that matters rather than how they say it. Such a massive amount of resources are potentially going to be put behind their view that they should have the patience of job if that is what it takes to justify it.
Barkley,
You are right; I mixed up the sentence. I meant of course that I suspect that much of the unemployment is really voluntary. I cannot say what the employment situation is like in the USA, but I can assure you that most unemployed people here in the UK could find work – eg as a care assistant – if they really wanted it. The fact is, that, if they receive sufficient income without taking such an undesirable job, they won’t take it. As it is, such jobs get done by immigrants.
There is no such thing as money sitting in banks unused. It is always lent out, even if only to the state – ie to the central bank if held as banknotes in the vaults or as reserves.
The Administration seems to finally be seeing the writing on the wall.
Geithner: Higher Taxes Hurt Economic Growth
March 12, 2009 01:22 PM ET | James Pethokoukis
Here is an illuminating exchange today between Sen. Mike Crapo and Treasury Secretary Tim Geithner:
SEN. MIKE CRAPO, R-Idaho: You say that the tax increases will only happen when the economy has recovered. I understand that a lot of economists are saying we are going to be recovered by 2011. Frankly I think there are economists who are saying maybe our recovery will not be so strong by then. My question to you is, are these tax increases contingent on a recovery or are they going to happen regardless of what happens in 2011?
SEC. GEITHNER: Senator, I think it is a very important question. I think, again, we need to lay out an ambitious path for bringing those deficits down, commit to achieving that with a mix of measures on the resource side and the spending side to do the best possible job of leaving our economy stronger, and thats what the Presidents budget tries to do. Now of course we are going to have to watch how the economy evolves and I want to underscore that one of the mistakes governments have made over time in dealing with economic crises is putting the brakes on too quickly or in ways that hurt growth just as its starting to take hold. We just want to be careful not to do that.
http://www.usnews.com/blogs/capital-commerce/2009/03/12/geithner-higher-taxes-hurt-economic-growth.html
DickF,
It is only Austrians who label Milton Friedman a “Keynesian,” and only a subset of them at that. Most Austrians do not label themselves as “classical.” To the extent that word gets used it is more likely to be as “true neoclassical” in the sense of “Mengerian,” after Carl Menger, the generally accepted founder of the Austrian School in 1870, who was one of the people who introduced marginal analysis at that time (Jevons in Britain, Walras in France), and who thus get labeled as “neoclassical,” although most current Austrians would not say that they are “neoclassical” in a current sense.
I believe that Friedman himself considered himself to be a “classical,” although he was influenced by Keynes and always insisted that Keynes be taught at Chicago, if for no other reason than for students to be able to criticize his work intelligently.
I know Roger Garrison personally and respect his work considerably.
Rebel Economist,
I have not checked, but I would lay odds that there are many banks right now that have higher than usual excess reserves. And it is in exactly such times as these that we would expect to see such a phenomenon, the period in which Say’s Law does not do so well in explaining things, a major contraction of the economy from the demand side.
Since I’m the one that started the discussion about the applicability of models in the first place…
Pat Toche:
Ad hominem attacks are usually the tactic of those with a weak argument. You don’t know my background, or how much economics I’ve studied. You make two claims which are also just axioms and not verifiable. First, is the claim that people will modify their behavior as a result of predictions, thus making predictions impossible.
People may modify there behavior on basis of expectations or events, but this doesn’t mean these can’t be adjusted for. In fact this is the entire basis of advertising and public relations.
Second, is your claim that econometrics has been responsible for some advances in statistics. I’ll concede that they have been responsible for some very complex mathematics, but whether this leads to useful results is another matter. In fact I’ve analyzed much of this from the point of view of signal processing science and don’t find anything exceptional.
You can read my discussion on the topic here:
http://robertdfeinman.com/society/econometrics.html
Barkely Rosser:
If models fail under unusual circumstances then they shouldn’t be relied upon, especially by those making policy decisions. You can claim this isn’t the fault of the models, but those misusing them, but what was the purpose of the paper that started this discussion, if not to influence policy? In fact it was issued by an organization with a pronounced ideological bias, thus putting its scientific objectivity in doubt from the start.
It’s lack of connection to real world events is, to me, further evidence that it is nothing but propaganda disguised as pseudo-science. We have seen far too much of this in the past few decades, most of it coming from rightwing institutions funded by big money interests, like the Koch brothers, Scaife, Olin and the rest of the cabal.
As to string “theory”. This is a misuse of the word theory. Even though its backers call it string theory, it is more like string conjecture at this point. We physicists usually reserve “theory” for reasonably well tested mathematical constructs, or at least ones where the tests are possible, if difficult. One of the predictions of General Relativity was only confirmed within the past decade. It required high precision devices in orbit and wasn’t technologically feasible before.
So string theory is in an early stage of development and may have over-enthusiastic proponents, but it is still mostly conjecture. Believe me those working on it would like nothing better than to see some actual evidence to support it.
Finally to a theme I’ve raised elsewhere, which has to do with the “multiplier”. We don’t know what course of action will produce the most bang for the buck, that’s what all the rival claims are about. That’s because economics looks at aggregate effects like GDP. What policy makers should be looking at, instead, is the moral aspect of their actions.
Giving money to those in distress has a known outcome – less suffering. This means that we should be expanding social services like unemployment and health insurance, food stamps and similar programs. It means that the government should be putting people to work, paying them directly if necessary.
Indirect measures such as tax breaks to business or fiddling with the marginal tax rates or giving more bailout money to one sector or another get justified by arguments about their macro effect, totally ignoring what is happening to real people. This is why economics is held in such low regard. Money has a higher focus that people.
Robert D. Feinman,
I shall only respond to your remarks about multipliers and the models of them, such as those Menzie posted about. I would say that one of the bottom lines to this is that his post verifies what has been much discussed by many, that there are many competing models and estimates of the values of various multipliers for different periods and for different kinds of spending or transfers or tax changes.
What I would note out of this is that while the actual numbers vary considerably across these studies, by and large there is a good deal of agreement about the relative degrees of strength associated with some of these different multipliers across studies, such as that infrastructure investment will probably generate higher multipliers than some transfers or tax cuts, at least in the long run. Those arguing for stronger effects from tax cuts do so on the basis of longer run estimates.
So, I would say there is some use to all of this, although clearly wise policymakers will recognize as do wise econometricians, that one must take any specific forecasts or estimates with appropriately large grains of salt.
I meant to say “in the short run” before I said “in the long run” the second time.
Barkely,
I do not dispute that banks are holding large reserves at present; I just do not see why reserves have a greater bearing on real economic activity than any other form of debt. In fact, at the moment, reserves are effectively just an intermediate step in bank lending that provides the banks with liquidity and credit insurance – ie the banks lend to the Fed and the Fed does some of the lending that the banks would have done (eg by buying commercial paper).
Money (or reserves) represents wealth (real resources owed to the holder by the rest of society). I suppose that the problem of understanding fiscal stimulus is to explain why, when the government takes wealth from one group and gives it to another group to induce them to work, it does not simply create the same circumstances (ie unemployment) in the group it takes from.
A Note on Choosing between Macroeconomic Models –
Since the econometric models developed by competing schools of economic theory all seem to confirm the particular theory being tested, it’s hard to believe this is the best method of choosing between theories.
One alternative is to take a close look at the theory’s premises. According to Cogen, et al, the New Keynesian (New Classical?) model they use to estimate the (low) government spending multiplier is superior to the Romer/Bernstein moldel because it’s built on microfoundations, assumes rational expectations, and takes account of agent responses to policy changes.
To paraphrase, the model assumes that households and firms make decisions on the basis of the same (rational) view of how the economy will unfold over time.
Two problems immediately come to mind. In the first place, households and firms hold many different and conflicting views regarding the future course of prices, interest rates, taxes, etc. To cite but one example, many large firms subscribe to forecasting services that use Old Keynesian models to forecast macroeconomic variables. In addition, the very idea of rational expectations rules out the conflicting views of the “bulls” and the “bears.”
Secondly, the notion of “rational expectations” assumes a stable underlying structure about which it’s possible to form such expectations. Assuming such stability proved disasterous for Long-Term Capital Management and its Nobel prize-winning creators of option pricing theory, and for the latest band of financial engineers with their securitized mortgages and the like.
Did the model used by Cogen, et al, forecast the current crisis? Is the current crisis even consistent with the notion of “rational expectations”?
If the answers are “no,” then one is hard-pressed to accept its results regarding the size of the gov’t. spending mulitiplier.