One Times X is X

by Don Boudreaux on October 1, 2009

in Myths and Fallacies, Seen and Unseen, Stimulus, Taxes

Here’s Robert Barro and Charles Redlick reporting, in today’s Wall Street Journal, on their recent research that finds the alleged Keynesian multiplier to be largely mythical.

Closing paragraph:

The bottom line is this: The available empirical evidence does not support the idea that spending multipliers typically exceed one, and thus spending stimulus programs will likely raise GDP by less than the increase in government spending. Defense-spending multipliers exceeding one likely apply only at very high unemployment rates, and nondefense multipliers are probably smaller. However, there is empirical support for the proposition that tax rate reductions will increase real GDP.

Be Sociable, Share!

Comments

comments

78 comments    Share Share    Print    Email

{ 78 comments }

Justin P October 1, 2009 at 12:21 pm

This is one area of Keynesian mythology that has fascinates me. You can take the same data and put it into different models and come out huge differences in multipliers.
I think the multiplier from G will usually be less than one. My thinking is, Government spending is dominated by special interests, usually very small groups to begin with but with big pocket books. You have time lags to consider, inefficient government bureaucracies, and the calculation problem. As a result, the multiplier will be so low that fiscal spending just doesn’t give you enough bang for the buck.
I agree that tax rate reductions will produce a higher multiplier in the short run, but unless it is coupled with spending cuts, won’t give any long run benefits.

Anonymous October 1, 2009 at 12:36 pm

Wow. So we should cut taxes to zero and end government spending completely. I don’t know but it’s just kinda weird how the economy crashed after we massively cut taxes and how the stock market is now up 17% since the stimulus package was passed.

Oh and just a little follow up on sub-prime lending.

http://www.ritholtz.com/blog/wp-content/uploads/2009/06/cra-chartg1109.gif

Justin P October 1, 2009 at 12:41 pm

How would your world work? Jack the tax rates to 100% and increase Government spending to what, 100% of GDP?
Really what do you have against Capitalism?

Anonymous October 1, 2009 at 2:14 pm

Tax rate of maybe 40% for 300K to $2,000,000. After that 60, 70 80% whatever it takes. Then that excess will be re-invested in the company and not its CEOs and board members.

Government spending ideally closer to 30% of revenues.

Is that communism??? No it’s pragmatism.

Mcwop October 1, 2009 at 4:06 pm

And of that 30% of revenues 40 cents of each dollar will continue to go to defense spending. Yes great use of resources Muir. I will accept 30% of GDP to the government, but only if they cut military spending by 60-70%.

Justin P October 1, 2009 at 9:02 pm

Those are just details, they aren’t important.

Anonymous October 1, 2009 at 9:33 pm

Agreed!

Anonymous October 1, 2009 at 12:48 pm

Muirgeo,

Post hoc ergo propter hoc continues to be your guiding philosophy. That and using the stock market as your measuring stick of the economy’s health. Always the populist. Unemployment (or employment) might be better.

Anonymous October 1, 2009 at 2:16 pm

I’ll take consistent correlation without certainty of causation over consistent anti-correlation with unyielding certainty any day.

matt October 1, 2009 at 2:43 pm

So like, hmmm. I get it. So the correlation/causation thing goes like this:
You speak…. then people hear stupid things…. therefore, you must say stupid things…. damn! this is fun!

Anonymous October 1, 2009 at 3:33 pm

What is anti-correlation?

Guest October 1, 2009 at 4:42 pm

Semi-clever rhetoric?

Anonymous October 1, 2009 at 6:42 pm

I was wondering the same. I don’t know if he means negative correlation, or non-correlation. Most likely he doesn’t know what he means either.

Anonymous October 1, 2009 at 9:40 pm

That’s like when your hypothesis includes things like;

Cutting taxes is good for the economy…

Unions are bad for the economy…

Minimum wage is bad for the economy….

Free trade is good for the economy…

Deregulation is good for the economy…

Privatizing public goods and services are good for the economy…

Wealth cncentration is inconsequential…

and then when you do these things the economy looks like it does now. That’s anti-correlation.

Anonymous October 1, 2009 at 2:18 pm

To add to Russ, correlation != causation. The government also ramped up spending during the great depression and the 1970s and it neither helped the stock market, unemployment, nor GDP.

muirgeo, it’s also been widely established that sub-prime lending was backed and even encouraged by government as a way to push home ownership on specific segments of the population they felt should have homes. This was encouraged by political party hacks on both sides that felt if they got credit for this they’d get more votes for their party.

Of course if government magically did “all the right things” then we’d be in paradise, even if we ignore the fact that everybody’s idea of paradise is different.

Finally, few people here are arguing for either extreme. If there was a 100% tax rate I wouldn’t be able to chase my dreams. If there was a 0% tax rate there’d be nobody there to protect them.

Jayson Virissimo October 1, 2009 at 6:43 pm

“Finally, few people here are arguing for either extreme. If there was a 100% tax rate I wouldn’t be able to chase my dreams. If there was a 0% tax rate there’d be nobody there to protect them.”

Couldn’t you, IDK, use some of that extra money you saved from not having to pay taxes and pay someone to protect them?

Anonymous October 1, 2009 at 7:11 pm

What if somebody with more resources decides to use force against me? The idea of individual liberty is that there should be no one allowed to use that coercion.

In theory I could become part of a community to protect me, but isn’t that what government is?

Justin P October 1, 2009 at 9:04 pm

re: “If there was a 0% tax rate there’d be nobody there to protect them.”

I do believe there were a few years before the 16th Amendment, where there was still a Federal Government.

Anonymous October 1, 2009 at 9:32 pm

You’re thinking only income taxes. The government collected revenue via a host of other means, from poll taxes to duties to excise taxes. There were also other levels of government involved.

Anonymous October 1, 2009 at 9:45 pm

“If there was a 100% tax rate I wouldn’t be able to chase my dreams. If there was a 0% tax rate there’d be nobody there to protect them.”
hylarides

Great point! I am simply arguing for minor changes well within the bounds of anything this country has done before. And for that I’m called all kinds of names by people who generally are the ones arguing for something extreme and unproved and with a lot of “anti-correlation”.

Anonymous October 2, 2009 at 12:33 am

You can’t argue for minor changes within bounds that we’ve already done, as then nothing new would ever come about. Even if you did, what most people are arguing for is well within what’s been done. Things like taxation at about 10% of GDP, minimal government only doing the basics to protect individual rights.

During this time, places like Japan, Europe, and North America saw huge gains in the standard of living. People went from relatively brutal serfdom and backbreaking labour in farms to cities and their own places to live. Though looking back we may see the factories as sweatshops, 100 yeas from now they’ll probably read in disbelief at how we were stuck in office chairs breathing recirculated office air.

Anonymous October 1, 2009 at 1:00 pm

Russ, you probably should have linked to this:

http://www.textlog.de/4227.html

Nathan Scott October 1, 2009 at 4:43 pm

You do realize inflation is up like 10% since the stimulus passed? The stock market is not an inflation adjusted measure. Whenever the government prints money it invariably finds its way into stocks. This was seen with the recent 140 P/E ratio of the SnP 500, the highest in history.

Anonymous October 1, 2009 at 6:38 pm

It would be an improvement if people would simply realize that government is ALWAYS an expense.

Anonymous October 1, 2009 at 1:03 pm

I have two concerns -

1. So he looked at a macrohistory of the U.S. and generally found multipliers are 1. Fine. That’s not really counter to the Keynesian view. Keynesians say that fiscal stimulus is appropriate when interest rates are around zero, savings exceed planned investment, and there is deflation or extremely low inflation. Otherwise, Keynesians expect the impact of fiscal stimulus to be much less significant. So Barro looks at a huge range of U.S. history to test an average effect of a multiplier, with only one previous instance where the Keynesian arguments apply – and then concludes that he’s disproven the Keynesian argument? It just doesn’t make sense. That’s EXACTLY what a modern Keynesian would expect for most of 20th century macro history (it’s true – early Keynesians were more naive about what to expect from multipliers… my understanding is they were actually thinking about multipliers on the scale of 4 in the early days!).

2. It’s awfully conceited to write “the available empirical evidence does not support the idea that spending multipliers typically exceed one”, when all he’s really presenting is his own empirical evidence. The IMF recently published a sort of meta-analysis of (by my quick count) 108 different estimates of fiscal multipliers. Their conclusion? 1 to 1.5 for spending multipliers in a large economy. Exactly what most of the discipline is assuming.

So who do we trust? One economist that has staked his career on bringing Keynes into disrepute – or 108 different studies that certainly vary in their findings, but center on a 1 to 1.5 estimate that makes good sense given theory?

To take a line from the last post, these 108 economists and their colleagues are the “unsung heroes”. They shouldn’t be tossed out when we find one famous guy that gets a few inches in the Wall Street Journal.

Justin P October 1, 2009 at 1:05 pm

IMF reports also show multipliers in the negative range as well.

Anonymous October 1, 2009 at 1:16 pm

Right. But look at where they occassionally come up negative:1. Tax cuts, which makes sense for the same reason that Taylor recently pointed out the rebate checks don’t make much of an impact.2. Most of the negative impacts on the spending multipliers come from estimates for European economies in the 1960s. Europe was booming in the 1960s. No (modern) Keynesian in their right mind would expect a substantial spending multiplier then. How much relevance do you think those estimates really have for depressionary conditions now?, and 3. Often the negative estimates are for the “3 years out” impact estimate. This makes sense too. If you basically trade future growth for current growth, you’re likely to have positive multipliers in the first year and negative multipliers in later years. Note the cumulative multipliers on these cases – they’re almost always positive.Context matters a lot. Yes, multipliers can be negative. A caricature Keynesian may think all government spending is always and everywhere great – but real live Keynesians advocate it under specific conditions. The IMF estimates do present a range. But that range makes perfect sense given the expectations of Keynesian theory.

matt October 1, 2009 at 1:44 pm

Rebate don’t make “much of an impact” because they don’t change behavior. Tax cuts do.

But nevermind that, how about this: “but real live Keynesians advocate it under specific conditions.”

How do they know under what conditions their goofy spending ideas should be implemented and how they know that those conditions are actually happening. (And then, for fun, tell me how you’re actually going to get the money spent in the “right” places at the “right” time.)

Anonymous October 1, 2009 at 1:50 pm

Note the tax cuts don’t fare as well empirically either, for a similar reason to the rebates, although obviously not the exact same reason – because as you say they don’t change behavior.RE: “How do they know under what conditions their… spending ideas should be implemented”Theory and empirical work on when stimulus is effective.RE: “and how they know that those conditions are actually happening”BEA has data on savings and private investment. The Fed has data on interest rates and inflation expectations. BLS has data on inflation rates. It’s not that hard to check all the boxes off.Your final parenthetical question is obviously tougher – that’s never a guarantee, and I should hope nobody acts like it is. That’s why we need an open political process, outspoken watchdogs, and a willingness to live with the prospect that of course it won’t always come out perfectly. Timely, targeted, and temporary – that’s been Brookings’s mantra and I think it’s a good one.

Anonymous October 1, 2009 at 6:46 pm

Rebates are not tax cuts.

The IMF paper that I think you are referring to specifically looked at recessions. The wide variance is best interpreted as a lack of evidence of a multiplier, not support for “real live Keynesian” beliefs that sometimes there is a >1 multiplier.

Anonymous October 1, 2009 at 7:47 pm

And I should have actually said what IMF paper I was refering to! It was this one: http://www.imf.org/external/pubs/ft/spn/2009/spn0911.pdf

It’s not just recessions – some of the smallest multipliers come up from European estimates during their economic miracle. Which is no surprise – they SHOULD be small during those booms. So if you whittle down the 108 estimates to those that were done on recessionary periods, the variance will shrink dramatically and the average impact estimate is going to increase too.

I’m not sure why you think a variance of results means non-support. We expect variance in these estimates. Nevertheless, it would have been nice to have more of a meta-analytical approach to be more specific about exactly how much variance is too much.

And while you’re making the point on taking estimates during recessions and not during recessions, it’s also important to note that the fact that stimulus is done during recessions introduces a major identification problem that will bias any estimate downwards. Now the job of a good empirical study is to properly identify the impact – but we should always keep in mind that there is a background downward bias to multiplier estimates (just like education researchers always keep in mind that there is a background upward bias on returns to education estimates).

Daniil Gorbatenko October 1, 2009 at 7:52 pm

Daniel, read the research paper on which the WSJ article is based.

http://www.nber.org/papers/w15369.pdf

These are the crucial lines:

“The empirical evidence on the response of real GDP and other economic aggregates to added government purchases and tax changes is thin. Particularly troubling in the existing literature is the basis for identification in isolating effects of changes in government purchases or
tax revenues on real economic activity.”

“It is hard to be optimistic about using the macroeconomic time series to isolate multipliers for non-defense government purchases. The first problem is that the variations in the non-defense variable are small compared to those in defense outlays. More importantly, many of
the changes in non-defense purchases are likely to be endogenous with respect to the determination of real GDP.”

And remember: science never progresses by consensus. One good argument/piece of evidence can and must refuse whatever amount of bad arguments/evidence. This is not to say that Barro is necessarily right. But your argument about 108 economists is not proper.

Daniil Gorbatenko October 1, 2009 at 7:54 pm

Sorry, I meant “rebuttle” not “refuse”.

Anonymous October 1, 2009 at 8:08 pm

Hi Daniil! I was searching for this paper when I was talking to you on Econlib, but I couldn’t find it.

There are definitely estimation problems – if there weren’t we would all agree by now! One important thing to remember is that given the context of multiplier estimation (ie – during a recession), the estimate is expected to have a downward bias unless some identification strategy is used. That’s always good to keep in mind – that holding all else equal we’re probably underestimating this (just like education economists all know in the back of their heads that they’re probably overestimating the returns to education).

Barro solves the endogeneity issue fantastically – by looking at defense spending. But by doing that he makes the multiplier much less meaningful. He solves the endogeneity problem but introduces a new problem of estimating a multiplier during non-depressionary conditions. What good is that multiplier??? Of course it will be lower – that’s why we don’t do fiscal stimulus when we’re not in a recession!!!

RE: “This is not to say that Barro is necessarily right. But your argument about 108 economists is not proper.”

Of course it’s proper. My 108 economists argument was simply that Barro was wrong to say “The available empirical evidence does not support the idea that spending multipliers typically exceed one”. I didn’t say they are all right or even that the question is closed. I’m simply pointing out that Barro’s point about the “available empirical evidence” doesn’t really reflect that “available empirical evidence”. The available empirical evidence is not conclusive – it is mixed. But if you had to choose which side it leans towards, it clearly leans towards the idea that multipliers on spending in a recession are greater than one.

Daniil Gorbatenko October 3, 2009 at 9:10 pm

=Barro solves the endogeneity issue fantastically – by looking at defense spending. But by doing that he makes the multiplier much less meaningful. He solves the endogeneity problem but introduces a new problem of estimating a multiplier during non-depressionary conditions. What good is that multiplier??? Of course it will be lower – that’s why we don’t do fiscal stimulus when we’re not in a recession!!!=

But at least in the case of WWII there were depressionary consitions. The unemployment rate in 1940 was 9,4%. And if these were not depressionary, nowadays’ conditions also aren’t.

As for endogenity issue, the important line from the Barro’s paper is this:

“A common approach in the existing empirical literature, exemplified by Blanchard and Perotti (2002), is to include government purchases (typically, defense and non-defense combined) in a vector-auto-regression (VAR) system and then make identifying assumptions
concerning exogeneity and timing. Typically, the real-government-purchases variable is allowed to move first, so that the contemporaneous associations with real GDP and other macroeconomic
6 aggregates, including real personal consumer expenditure, are assumed to reflect causal influences from government purchases to the macro variables. This approach may be satisfactory for war-driven defense spending, but it seems problematic for other forms of
government purchases. More generally, it is perilous to treat an arbitrarily selected component of GDP as exogenous and then demonstrate a positive, “causal” effect of this component on
overall GDP.”

Of course, it is up to professional statisticians to opine whether Barro is right in this criticism of the aforementioned way VAR is applied.

Anonymous October 3, 2009 at 11:32 pm

RE: “But at least in the case of WWII there were depressionary consitions. The unemployment rate in 1940 was 9,4%. And if these were not depressionary, nowadays’ conditions also aren’t.”

In WWII the government enforced strict rationing of private consumption, precisely so they could direct resources to defense spending. Is it any wonder that Barro came up with an insignificant impact of government spending on private consumption?

It’s a bad paper Daniil.

Seth October 1, 2009 at 1:54 pm

Coefficients are fiction.

A business I work with was once addicted to managing by coefficients. That didn’t work so well for them. They eventually found out they have to produce value for their clients.

Anonymous October 1, 2009 at 2:10 pm

Hahaha – so are coefficients fiction or are coefficients not a panacea? You say both but only give evidence for one of those two contentions :)

Coefficients are just averages. The fact that they can be misused doesn’t mean that they have no use.

Seth October 1, 2009 at 2:44 pm

Most of the time, Pyrite.

You’re right, there might be some uses. Let’s say that a multiplier came out to be 5, 10 or 100 under some conditions and 1 and others. That might catch my attention. But, if that’s the case, then it’s very likely that that difference is already built into our intuitive senses.

But something in a small range of even -1 to 2 is all the same to me. Given the imperfections of the data, the data that’s not even available to added to the model, or is never considered to be added, and the dynamic complexity of the real world, the opportunity cost of doing anything based on those results is too high. I could get better bang for the buck on something else.

Ike Pigott October 1, 2009 at 3:22 pm

You got the puntcuation wrong.

The truth of Keynesian economics is that “One times X is x.”

Anonymous October 1, 2009 at 10:00 pm

Here’s a formulation more germane:

GDP = X + Y
X2 = X1 * 1
Y2 = Y1 – X1

Anonymous October 1, 2009 at 6:39 pm

The IMF beat them to the punch:

“The average fiscal multiplier across all 43 recession episodes is -1.5.”

“It is probably more appropriate to conclude that the fiscal multiplier is very small in open economies (and probably close to zero with a flexible exchange rate).”

http://en.scientificcommons.org/34265622

Anonymous October 1, 2009 at 6:51 pm

Cato publishes an interesting correlation in their policy handbook. It presents an empirical demand curve on a plot of tax rates (price) versus government spending (quantity). The implication is that government grows when taxes are cut, because people do not feel the costs of government spending. Therefore popular demand (or toleration) for government spending increases. The Cato conclusion is that the focus must be on cutting spending.

iamse7en October 1, 2009 at 7:48 pm

That’s an interesting point. It’s true that tax cuts don’t work in the long term if government doesn’t cut spending as well. Reagan believed you can’t wait for government to control spending, so you apply pressure by going ahead with tax cuts, but government still didn’t decrease the spending. What’s the solution? You can’t rely on government to cut spending. It’s an always-growing beast. We can’t just sit around and say, “cut spending!” Because even if we got someone like Ron Paul in office, Congress wouldn’t allow him to get rid of the Departments of HUD, Energy, Commerce, Education, and so on. There certainly is a flaw in the system somewhere, and I like Ron Paul’s immediate solution: Ending the Fed would help to prevent government from just printing and spending money whenever it wants. Back to commodities standard!

Anonymous October 1, 2009 at 8:17 pm

You make good points. The Cato correlation suggest not just that tax cuts don’t result in spending cuts, but that it results in EVEN MORE government spending.

I agree that taxes should be cut, PERIOD. They are FAR too high and fundamentally immoral. And although there appears to be a demand curve, that doesn’t mean a slower spending growth won’t still follow higher taxes, perhaps a faster growth as taxes become more progressive. We just have to recognize the fact that tax cuts will accelerate government spending and influence in the economy.

But there is a limit to debt that the government can raise. We have to accept that the government should be allowed to reach that limit as the only possible constraint upon spending.

And though they may dabble with an inflation tax, that too is self limiting as the horrors of inflation set in and the central bank takes action.

Why is this a good long term solution? Because it FINALLY conditions the government to function without growth. It requires them to make choices regarding spending. It imposes a level of discipline that never hindered them before.

We wind up with European stagnation (and if you look at our decelerating GDP since WWII, it is apparent we are headed there). But that may lead to popular support for more rational fiscal policies. There is some glimmer of that in places like Ireland, Germany, and some Eastern European countries.

iamse7en October 1, 2009 at 9:23 pm

Couple of questions:

1) Do you have a link to that specific part in their handbook that relates to this issue? I’m interested in reading it.
2) What is the good long term solution? Are you referring to the cutting of spending? I certainly agree, and although there are official limits to debt that the government can raise, those are set by government and when they want to go past that limit, they can get that limit changed, as Geithner is lobbying to do so now.
3) What of Ron Paul’s solution to ending the Fed, which he believes is the #1 tool that government uses to spend, spend, spend?

Thanks.

Anonymous October 1, 2009 at 10:16 pm

I own a paper copy. They sell it, so I don’t think it is online. It is a terrific read covering almost every policy topic you can think of. I highly recommend buying it.

The long term solution I was referring to is the conditioning of the government and public to function without deficits or high taxes. To reach that end, we must always fight higher taxes. Absence of deficits is not a realistic long term solution, as long as it is at all possible for the government to borrow.

They will always raise the debt ceiling when they can. You have to realize that no way has been found that will stop debt growth for any significant period of time, particularly the debt ceiling which repeatedly gets raised. The only possible way, I’m convinced, is for the money to run out. That is, the interest they have to pay for new debt must swamp all other government spending. Only then will all non-debt repayment government spending be forced into some sort of discipline.

And if our political leaders are so crass as to allow overt default, I’m no so sure that would bad either. Default would immediately raise the risk of government bonds and severely hamper the government in raising further debt.

In short, spending truly is the problem, and the solution must be to cut spending. But I don’t see spending ever getting cut until there just isn’t any more money to spend.

Ending the central bank and privatizing the money and banking system in our country would be great step for stability, resilience, growth, and liberty. But merely placing it under control of Congress is unlikely to be much of an improvement.

Anonymous October 1, 2009 at 11:39 pm

http://krugman.blogs.nytimes.com/2009/10/01/multiplying-multipliers/

Another problem with Barro that I forgot about.

His solutions to the endogeneity problem (defense spending, etc.) make his estimates absolutely meaningless for the current situation. But people latch on to it because they like what he says about Keynes. Not very impressive.

Justin P October 2, 2009 at 4:20 am

Running to the Krug for help? That desperate?

How convenient that Krug picks the Romer estimate, I mean who would have thought he’d pick a multiplier greater than 1?
I mean come on, that’s like quoting Al Gore to prove AGW.

Anonymous October 2, 2009 at 10:56 am

No thoughts on the points he actually raises, huh?
I see. Fair enough.

matt October 1, 2009 at 2:54 pm

“Theory and empirical work on when stimulus is effective.”

“BEA has data on savings and private investment. The Fed has data on interest rates and inflation expectations. BLS has data on inflation rates. It’s not that hard to check all the boxes off.”

You should call and tell them that. Seems once they figure out they have all this data at their disposal, we’ll be problem free forever!

Anyway, all the data in the world cannot be relied upon. The system (e.g world economy) is too large and complex and dynamic to get the information you’d need to target whatever with spending. It’s a nice guide–all of their fun numbers–but those data probably aren’t accurate, could change in an instant, and even if not, you’d still have to deal with the parenthetical question I asked.

Hey, I think Kenyesians are super. Afterall, who doesn’t want prosperity forever without recessions? But as long as I’m asking for things that are not possible to achieve, I’m gonna come up with something more creative than that.

Anonymous October 1, 2009 at 2:58 pm

RE: “You should call and tell them that.”

Huh? People are already pointing to this data and saying these things. That’s why it’s there, matt. This isn’t news.

RE: “Seems once they figure out they have all this data at their disposal, we’ll be problem free forever!”

Why do you say that? That makes no sense.

RE: “Anyway, all the data in the world cannot be relied upon. The system (e.g world economy) is too large and complex and dynamic to get the information you’d need to target whatever with spending.”

You’re mismatching my arguments. Nobody said that we have adequate data to target spending. I went into detail about how that can be problematic. We can use data to recognize that we’re experiencing depressionary conditions where fiscal stimulus is conceivable – that’s all I’ve claimed.

RE: “Hey, I think Kenyesians are super. Afterall, who doesn’t want prosperity forever without recessions?”

You’re obviously confused about what Keynesians claim.

matt October 1, 2009 at 4:02 pm

“Why do you say that? That makes no sense.”

It was sarcasm. Point is, “they” had the “data” last year, in 2003, in 1998 and 1959 and so on. Like you said, it isn’t news. So why now, is it a matter of “checking the boxes”? If the data were so useful to policymaking, one might think we’d have far fewer “problems”.

“We can use data to recognize that we’re experiencing depressionary conditions where fiscal stimulus is conceivable – that’s all I’ve claimed.”

Conceivable? I can’t argue that. I can conceive a lot of things, no matter what the data… If you mean where fiscal stimulus should be applied, that’s a bit different, but I needn’t yet comment on that because all we’re doing is conceiving…

And true, you never claimed to be your definition of a Keynesian. But neither did I say you did. (When I said “you” I meant “one…” not you you). Anyway, all I claimed was that Keynesians will never have the data they need to target when or where to spend, nor the ability to actually spend when and where it would be “necessary” even with the data.

“You’re obviously confused about what Keynesians claim.”

How could one not be confused after listening to a Keynesian?

Anonymous October 1, 2009 at 5:14 pm

:) I understand it was sarcasm – that’s my point. The sarcasm doesn’t even make sense. Sarcasm has to have a kernel of truth or else it’s a non sequitor. The only people I know of that promoted anything close to a problem-free view of the world were the EMH people.

matt October 1, 2009 at 5:32 pm

Oh, sorry. It was a non sequitor. (And so is my “sorry”. For there was no kernal of truth to my apology.)

matt October 1, 2009 at 5:35 pm

And weren’t you told that your nitpicking is boring? Tsk tsk.

Anonymous October 1, 2009 at 6:23 pm

:) says the guy that started this sidebar after an original comment that wasn’t nitpicking – it was outright, up front disagreement.

Russ is funny about that. When I’ve got an issue with what he says it’s nitpicking. When he has an issue with what someone else says it’s an important disagreement. Tom”ay”to/tom”ah”to. What would be boring is a Cafe Hayek where everybody just nods their heads, or where the only disagreements are minor skirmishes within the generally agreed upon libertarian framework. THAT’S nitpicking and boring.

Anonymous October 1, 2009 at 8:02 pm

That is not the one I was thinking of. I recommend you look at an older IMF paper:

http://en.scientificcommons.org/34265622

“they SHOULD be small during those booms”

Which is why you should just look at recessions. The paper I quote you above did just that.

“I’m not sure why you think a variance of results means non-support.”

?!!!! That is ALWAYS what it means. Variance is a measure of uncertainty which is a reflection of lack of evidence. If there was something other than uncertainty to be said, then the underlying data would be modeled differently. There would both be a different model and maybe a lower variance.

“introduces a major identification problem that will bias any estimate downwards”

Look at the paper I gave you. The variance is still high, but the approach is sound–or at least as sound as retrospective economic analysis can be.

Perhaps the bigger problem is that people hinge their opinions on any of these studies. The keynesiac theory is fundamentally flawed and the whole notion of government spending for economic stimulus makes no sense.

Anonymous October 1, 2009 at 8:23 pm

Re: “Which is why you should just look at recessions. The paper I quote you above did just that.”

Yes, but the point is the paper I cited (which looks at over twice as many estimates as yours) got positive results even though it included extremely low results from non-recessionary years. I’m glad you posted this. It demonstrates there’s wide variation in estimates and the research is not in agreement (as Barro seems to think it is). But it doesn’t refute the more recent IMF paper either.

RE: “That is ALWAYS what it means.”

Yes but there’s going to be sampling error in any empirical investigation. When there is variance in estimates you don’t say “oh no – no effect”. You determine if the variance is great enough to reject the effect. And then you ask yourself “well did I reject this because there is really no effect, or did I reject this because this 2002 IMF paper was only able to look at 43 observations”.

RE: “The variance is still high, but the approach is sound”

It’s not clear to me what their identification strategy is for dealing with the downward bias. Could you explain it? I’m not finding it.

Note at the end they also characterize when fiscal policy does make sense – when there is excess capacity and accompanying monetary expansion. They don’t come out and say it, but that sounds to me like they’re thinking of something like a liquidity trap. Keynesians (at least modern ones) don’t say that fiscal stimulus is always necessary. They say it’s necessary under these conditions – so even these select 43 studies don’t look exclusively at recessions that Keynesians would suggest using fiscal policy responses for.

One final note – the major factor that they say shrinks multipliers is having an open economy (ie – your stimulus leaks to people you import from). No Keyensian would dispute that. That’s why a coordinated global response is being pushed as so important.

Justin P October 1, 2009 at 9:14 pm

Keynes is a theory in search of data.
Remember your General Theory, it’s full of wonderful quotes like, “If we had the Data, it would show…”
It’s a great line, kinda like “we’ve saved or created X jobs.” How can you say it’s wrong? They will just come back to say they don’t have enough data yet.
Large G multipliers can never be disproven because they will say they don’t have enough data.

Anonymous October 1, 2009 at 8:46 pm

“When there is variance in estimates you don’t say “oh no – no effect”. You determine if the variance is great enough to reject the effect.”

You’ve got to be kidding me. Where did you learn statistics? What you should ask is, ‘Is there evidence to reject the null hypothesis’–not ‘Can I reject an alternative hypothesis’. You should see that there isn’t evidence for your Ha, then you should stop claiming that there is.

What you are doing is creating a self-selecting process for your hypothesis. That is the worst kind of bias.

Anonymous October 1, 2009 at 9:58 pm

“Keynes is a theory in search of data.”

Beautifully said. Keynes knew what he wanted to accomplish (the subject of another recent blog post). So he created a theory to justify it. The theory was absurd where it wasn’t vague, but it was just what the progressive troops wanted. Ever since the keynesiacs have been on a fruitless search for empirical justification–all the time claiming success.

Anonymous October 1, 2009 at 10:39 pm

Yes, I stated that a little too casually. But I certainly stated it more accurately than you when you said “Variance is a measure of uncertainty which is a reflection of lack of evidence” or “The wide variance is best interpreted as a lack of evidence of a multiplier”.

RE: “You should see that there isn’t evidence for your Ha, then you should stop claiming that there is.”

The point is there are LOTS of estimates that have rejected the null. And variations in these estimates – contrary to your position – does not indicate a lack of evidence.

RE: “What you are doing is creating a self-selecting process for your hypothesis.”

How?!?! I’m the one admiting that some have come up significant and some haven’t. You’re the one claiming “there isn’t evidence for your Ha”.

Anonymous October 1, 2009 at 10:46 pm

“But I certainly stated it more accurately than you when you said “Variance is a measure of uncertainty which is a reflection of lack of evidence” or “The wide variance is best interpreted as a lack of evidence of a multiplier”.”

What I said is precisely true. You’re not gaining any points with me as you continue to further reveal an unbridgeable chasm in your understanding of statistics.

I can only recommend you pick up any beginning statistics text, start with chapter 1, and then go back and read my post.

Anonymous October 1, 2009 at 11:05 pm

Wow – getting snarky. Look, statistics puts food on my table (or is it “food on my family?” I always forget), so no need to lecture. We clearly have converged on a definition we agree on (your rejecting the null clarification). I admit I wasn’t speaking very precisely before and you’ve supplied a great substitute for what I said.

But there is NOTHING in that (accurate) definition you gave that you can twist into justifying “the wide variance is best interpreted as a lack of evidence of a multiplier” after the fact.

Wide is relative, vikingvista. You don’t just look at a variance you consider wide and say there is a lack of evidence for a multiplier, much less dismiss decades of statistically significant empirical findings with “you should see that there isn’t evidence for your Ha”.

Anonymous October 1, 2009 at 11:23 pm

“statistics puts food on my table”

That’s the best thing I’ve heard since muir said he was pursuing a degree in cognitive dissonance.

I hope you are not telling the truth. Of course I stand by what I wrote–I have to. It not only isn’t controversial, it is both true and basic. I don’t know how to make you see how mistaken you are. Perhaps you will read as I told you. But I cannot here give you a starter course in statistical theory. The understanding gap is too wide for a meaningful discussion.

Anonymous October 1, 2009 at 11:26 pm

OK. The vikingvista position is simultanuously “wide variance means there is no evidence” and “wide variance can be failure to reject the null if it is wide enough”.

Let me know when you figure out how to make those two things consistent.

Anonymous October 1, 2009 at 11:28 pm

Jesus, DK. That doesn’t even make sense. I think you’ve become infected by muir.

Anonymous October 2, 2009 at 12:48 am

Other posts here and episodes of econtalk have gone into all of your points.
I can summerize some of the arguements:

“Cutting taxes is good for the economy…”

By increasing the incentive for people to take risks, work, and reduce the need of the government the economy tends to grow more over the long term.

“Unions are bad for the economy…”

By driving up wages and removing competition for other workers and protecting their own bad workers, it hurts companies when they try to compete in the marketplace.

“Minimum wage is bad for the economy…”

By “discriminating” against those with lower skills who’s only ability to compete with others is on price, we prevent them from being productive members of society. Many people must start off at lower salaries as they learn and become more productive. Many industries have that start at below minimum wage.

“Free trade is good for the economy…”

By allowing me to import cheaper goods from abroad, it will allow me to use those cheaper goods to produce things that are cheaper and therefor more competitive. If I have to spend less money on cheap tires, I can spend more on other things. This means more people benefit from my spending.

“Deregulation is good for the economy…”

Well this one is more complex, generally speaking regulation impacts the ability to innovate by stifling risk and removing experimentation.

“Privatizing public goods and services are good for the economy…”

This depends on what you’re privatizing. But if the government allows more efficient operators to provide goods, then everybody benefits. There’s be more organizations providing competition to give us those goods, thus providing a dynamic market for them.

“Wealth cncentration is inconsequential…”

This is true so long as the wealth is fairly acquired. It may sound unfair, but the fact is that there has to be wealth concentration. If everybody had equal wealth, NOBODY would have enough to develop complex and expensive to design things such as a computer. Nobody would have enough to build a large building or a railroad. Apple needs that capital to invest and design it. Oil companies need that capital to build huge pipelines.

Justin P October 2, 2009 at 4:06 am

Then you need to be more specific. Muir was talking about income taxes, to which Russ and you, replied.

Justin P October 2, 2009 at 4:11 am

Why do progs love theories with no data, then claim anyone that doesn’t agree with them suffers from cognitive dissonance? It’s maddening!

Anonymous October 2, 2009 at 5:15 am

Data in economics are not likely to ever be completely satisfying. What maddens me is theories with no basis in reality.

Anonymous October 2, 2009 at 6:42 am

Great post, Hylarides.Remember however, that it was a reply to a ducktor who believes that only government can determine how much liberty a man needs to be free, and that the only way to end corruption in government is to give the government even more power to heap favors on some corporations at the expense of others. He is the guy who rails against corporate welfare, and how eeevil it is that GE is such a large corporation, but favors corporate welfare for GE, because they make photo-voltaic cells that His Holiness: The Divine Prophet Algore I has determined are crucial to the survival of Mother Gaia.In short, and I know you’ve already deduced this: Yasafi isn’t the brightest CFL in the box.

Justin P October 2, 2009 at 7:14 am

The Data is corrupt from the start. I mean look where we get our unemployment data, yeah no conflict of interest there!

That leads to an important point, the models calibrated with faulty data are going to be faulty…who would have thought! Could this be one reason why the MSE (mainstream economists) didn’t see anything coming?

Anonymous October 2, 2009 at 7:25 am

“Could this be one reason why the MSE (mainstream economists) didn’t see anything coming?”

Yes.

But even worse, they don’t incorporate all of what we can and do know even without collecting any data. And even worse than that, they actually permit conclusions from data that contradict those known facts.

And then you read Krugman, and you understand why. It is because they have no interest in understanding reality. They just want to achieve their ends, and don’t think reality should be an impediment.

Anonymous October 2, 2009 at 12:09 pm

Which is why he needs to be educated! :-)

Anonymous October 4, 2009 at 12:31 am

I should take that back – I shouldn’t say whether it’s a bad paper or not. But I fail to see how Barro thinks it illuminates any of the questions we’re dealing with now. His endogeneity solution makes his answers irrelevant, and the singular case of WWII is bad because of the private consumption controls. I’m sure it’s a very well done paper – I don’t think it provides any insight into what kind of multipliers we can expect from the current stimulus package.

And I should also point out – I didn’t dream up that criticism of him myself (the one about rationing). People have brought that up to Barro himself in conferences and have been frustrated that he hasn’t seemed to have done much to address that objection. There’s not much point to professional conferences if you don’t take criticism you get there into account.

Daniil Gorbatenko October 4, 2009 at 10:00 am

Daniel, I don’t buy the argument about rationing. For me, it proves Barro’s point. Because the government did not ration consumption for want of it. It did it because it could not have more of both.

And if you are willing to say that WWII is not an illustration of successful Keynesian policies what is left? Japan of the 1990s?

But regardless of that, the most important point from Barro’s paper for me is that it is currently extremely difficult to measure the non-defense multipliers. And that all the supposed evidence is unreliable. I don’t know what to make of that argument. But if it is correct my view that the stimulus does not pass the “first do no harm” test holds.

Previous post:

Next post: