Small is irrelevant (in macro)

Paul Krugman and I occasionally disagree on the fine points of some macro issue, but almost never do I see him making a basic logical error.  Until today:

Dean Baker points us to Feyrer and Sacerdote, who use cross-state variation in stimulus spending per capita to estimate the employment effects of the stimulus. They find a clear positive effect: states that got more money per person did better on jobs. And as Baker points out, the national effects must have been larger, since some money spent in New Jersey presumably creates jobs in New York and vice versa.

One thing I might point out, by the way, is that this is something of a “Well, duh” result. Of course more federal spending in a given state or county creates more jobs. And the burden of proof should always have been on stimulus critics to explain why this doesn’t mean that stimulus spending creates jobs at the national level too. In normal times you can argue that the positive job effect of higher spending is washed out by higher interest rates “” that fiscal expansion will be offset by contraction on the part of the Fed. But with interest rates up against the zero lower bound, that argument doesn’t apply.

The argument about Fed policy in the second paragraph is flat out wrong, as I’ve pointed out 100s of times.  But even if it is right, this study tells us nothing about the macro effects of fiscal stimulus.  It’s a near perfect example of the fallacy of composition.  Every single anti-stimulus model would predict exactly the same finding at the micro level.  If the federal government builds a billion dollar military base in Fargo, North Dakota, I think all economists agree that the number of jobs increases—in Fargo, North Dakota.  Does the number of jobs increase at the national level?  Very possibly yes, but nothing in the Feyrer and Sacerdote study addresses that question.  Krugman, Dean Baker, Brad DeLong, etc, are very smart guys, I don’t know why they keep hyping this study.

Even worse, Krugman suggests that their instrumental variable analysis (which used state population) tells us that stimulus mattered.  Small states got more aid, and small states did better job-wise.  My preceding argument suggests that study is also irrelevant, but there’s another problem here.  Smaller states may have a different industry mix (say more commodity-oriented) that allowed them to better ride out the recession.  So it’s doubly irrelevant.

By the way, just a few weeks ago I criticized a Brad DeLong post on similar grounds.  If Paul Krugman would have read that post he would have avoided falling into the fallacy of composition.

Micro studies can’t tell us whether fiscal stimulus works.  Micro studies can’t tell us whether monetary stimulus works.  Micro studies can’t tell us whether RBC or sticky price models are better.  Micro studies can’t tell us anything about macro.  That’s why macro is a different field.


Tags:

 
 
 

49 Responses to “Small is irrelevant (in macro)”

  1. Gravatar of Will Ambrosini Will Ambrosini
    25. February 2011 at 10:05

    Micro studies can test assumptions of macro models…

  2. Gravatar of StatsGuy StatsGuy
    25. February 2011 at 10:13

    “Micro studies can’t tell us anything about macro.”

    Ok, isn’t that taking things a bit far? Micro studies tell us in a duopolist environment, oligopolist profits are likely. That has no macro implications? Micro behavioral studies tell us that people respond to price signals. That has no implications? Micro studies tell us that it’s hard for nominal wages to drop. No implications? Micro studies tell us about expectation formation and how people use recent trands to make decisions. Nada?

    Why incite the prisoners? Maybe something a bit less dramatic, like “Micro studies treat many things as exogneous that are really endogenous in a macro context.” Doesn’t quite have the same fire to it, I guess.

    BTW, I should point something else out – MACRO studies also treat things as exogenous that are endogenous in a POLITICAL context.

  3. Gravatar of Brian Brian
    25. February 2011 at 10:22

    “In normal times you can argue that the positive job effect of higher spending is washed out by higher interest rates “” that fiscal expansion will be offset by contraction on the part of the Fed. But with interest rates up against the zero lower bound, that argument doesn’t apply.”

    I’m confused by this part. In normal times higher spending is washed out by higher rates. But when the nominal rate is against the lower bound, the Fed can’t contract and rates can’t go up?

    And what’s real and what’s nominal in that excerpt? I suppose I should read Krugman more often to follow his arguments when he writes shorthand like this.

  4. Gravatar of ssumner ssumner
    25. February 2011 at 10:33

    Ambrosini, The only one that I can think of right off the bat is wage and price stickiness. And even then micro studies can’t easily refute RBC models (which allow for some types of price stickiness.)

    Maybe there are some examples that I am not aware of.

    Statsguy, You said;

    “Micro studies tell us in a duopolist environment, oligopolist profits are likely. That has no macro implications?”

    I’d like to see those studies, I doubt the effect is very strong. (Think Boeing and Airbus). But even if you are right, what sort of macro implications does that have?

    I agree with you about wage stickiness.

    What I was thinking about was cases where people say “Let’s do a micro study to see whether the RBC theory is correct, or whether stimulus matters.” Those kind of micro studies are looking in the wrong place.

    Brian, Krugman meant that at the zero bound the Fed would like to be more expansionary than it is currently being, but just can’t. So if they do more fiscal stimulus it won’t be offset with higher rates. Of course we saw last November that Krugman was wrong, the Fed wasn’t being passive because there was nothing it could do, it was being passive because it wanted to be passive (until November 2010, that is, when it stopped being passive.)

  5. Gravatar of david david
    25. February 2011 at 10:52

    I’m slightly confused here… do you happen to have a post laying out your proposed link between NGDP to RGDP? Do you have a modified sticky-price NK model buried in there, or something? Or do your agents just form expectations based on NGDP (not entirely implausible?).

  6. Gravatar of MarkW MarkW
    25. February 2011 at 11:34

    Let’s try to phrase this a little bit differently. Let’s assume that the federal govt takes $1billion from each of the states. Then let’s assume that the feds turn around and give the full amount, all $50B to say, Kansas.

    Is there anyone who doubts that the job outlook in Kansas would go up compared to the other 49 states?

    The question then becomes, do the negative affects of taking $1B from the 50 states outweigh the benefits to Kansas?

    This is why the study in question is flawed. When the negative effects are evenly spread, and the positive effects are concentrated, it is no surpise that the areas receiving largess do better than other areas. Such a finding in no way validates the belief that govt spending can increase economic activity. It merely reaffirms the truism that when you take money from one group and give it to another, those receiving the money are better off.

  7. Gravatar of Benjamin Cole Benjamin Cole
    25. February 2011 at 11:59

    Interesting commentary–btw, small states have been doing well by the federal government for decades. Rural states may be small, but they have two Senators each. And the rural-agro-defense bloc is very powerful. I also call it the Red State Socialist Empire.

    If you look at Census or Tax Foundation figs, you will see some astounding stats. Kentucky, for example, gets back $1.50 for every dollar sent to DC, or about $4k per resident. Pre resident!

    Kentucky is home to Senate Minority Leader McConnell. and the new star of the Tea Party, Senator Rand Paul. Oh boy, do you suppose they really want to cut federal spending?

    An interesting idea would be to require each state get back roughly equal what they send to DC. Would result in balanced budgets, and no more interstate subsidization.

    As it is, there is a bloc of rural states, with 24 Senators, who have a strong incentive to argue for more and more federal spending.

  8. Gravatar of Doc Merlin Doc Merlin
    25. February 2011 at 12:08

    @david:
    There are lots of ways of doing it… you don’t even need sticky prices. Just having nominal assets that aren’t money but must be paid in money will do it, and will impose downside to deviations, even in the long term, (housing loans are usually 30 year loans) from expected per capita NGDP paths.

    Note 1: If this is are the mechanism that makes adverse NGDP shocks real, you still have an RBC like internal economy if NGDP paths are maintained.

    Note 2: Nominal contracts across currencies would still be able to cause adverse real shocks even with a per capita NGDP targeting regime.

  9. Gravatar of JPIrving JPIrving
    25. February 2011 at 12:23

    And at the end of the day, say someone comes up with a study showing massive effects from fiscal stimulus. Who cares? You still had to borrow to do it. Monetary stimulus is 1. more egalitarian because it is independent of geography within the USD zone 2. Doesn’t run up debt or carry a big risk of CB losses on purchased bonds, and is invulnerable to rent seeking (everyone owns U.S. gov paper).

  10. Gravatar of JimP JimP
    25. February 2011 at 12:45

    Yellen on communications

    http://www.bloomberg.com/news/2011-02-25/yellen-says-fed-could-use-communication-to-affect-easing-unemployment.html

    Talk can do a lot – she says. (Thats me talking of course).

    And she is right.

  11. Gravatar of JimP JimP
    25. February 2011 at 12:49

    begin quote from her
    “Such a shift in policy expectations would be associated with a lower trajectory for the unemployment rate,” Yellen said at the University of Chicago Booth School of Business’s annual U.S. Monetary Policy Forum. The shift would also cause “a somewhat higher path of core inflation,” Yellen said.
    end quote.

    Thats right Janet. But why put it in the subjunctive? Why not JUST DO IT.

  12. Gravatar of Bams Bams
    25. February 2011 at 12:50

    Bennie C,

    Like a good lib – always focused how those dollars are distributed, as opposed to how they are accumulated.

    As opposed to the moral equivalency you try to create between McConnell/Paul and your fellow travellers in the Democratic Party (who have been purchased by the public sector unions), do you think the real issue responsible for the balance of payments in some red states is actually a function of oh-so-compassionate progressive taxation system which extracts relatively fewer dollars on average from the disadvantaged souls with lower average incomes in many interior states vs. the high average incomes in coastal states where your and your fellow “reds” (in the historic sense) have your seat of power.

    As a lib, you shouldn’t self-righteously preach about your superior level of compassion for the downtrodden, and then curse a symptom of the very system you set up so that you can feel good about yourself when you go to sleep at night.

  13. Gravatar of JimP JimP
    25. February 2011 at 12:50

    I will bet she is just warming us up. Lets hope so – shall we?

  14. Gravatar of Morgan Warstler Morgan Warstler
    25. February 2011 at 13:22

    Scott, the glaring disparity between you and DeKrugman will only increase…

    I told you they have sussed out where the arguments lead, and they do not like your trail, they have decided to beat back and go the other way.

    They are now officially the cheerleading team for economic failure, if things get better – sans more stimulus – they are cooked.

    Which is WHY you need to get more vocal:

    1. Goldman Sachs is screaming for stimulus because they are evil.

    2. Austerity where it produces productivity gains (public employees) is imperative.

    3. Do not worry, dear GOP, if things get harry – Ben will follow the Uncle Milty play book and relax Monetary Policy.

  15. Gravatar of StatsGuy StatsGuy
    25. February 2011 at 14:20

    Doc Merlin:

    “Note 2: Nominal contracts across currencies would still be able to cause adverse real shocks even with a per capita NGDP targeting regime.”

    I of course do agree with you, but you do realize that the above is only true if the distribution of wealth (or, at the very minimum, changes in the distribution of wealth), impact RGDP (growth).

    Either that, or we’re trying to trick everyone into thinking that society as a whole is richer as nominal asset values inflate (money illusion?), but that’s not very long term, eh?

  16. Gravatar of StatsGuy StatsGuy
    25. February 2011 at 14:30

    @ ssumner

    “I’d like to see those studies, I doubt the effect is very strong. (Think Boeing and Airbus). But even if you are right, what sort of macro implications does that have?”

    First, you picked two of the most regulated/government subsidized companies in the world. Even so, yes, there are implications on the structure of the macro-economy.

    For example Rotemberg/Woodford:

    “Oligopolistic Pricing and the Effects of Aggregate Demand on Economic Activity”

    “The authors construct a dynamic general equilibrium model in which the typical industry colludes by threatening to punish deviations from an implicitly agreed-on pricing path. They use methods similar to those of F. Kydland and E. Prescott (1982) to calibrate linearized versions of both their model and an analogous perfectly concerning model. The authors then compute the two models’ predictions concerning the economy’s responses to a change in military spending. The responses predicted by the oligopolistic model are closer to the empirical responses estimated with postwar U.S. data than the corresponding predictions of the competitive model.”

    http://ideas.repec.org/a/ucp/jpolec/v100y1992i6p1153-1207.html

    There are others.

    While on the topic of Boeing, Krugman had a cute post on contracting out/in-housing which is relevant to Williamson and past discussions of Coase – glad to see structure of the firm finally getting some play. The average optimal size of the firm is hugely relevant to macro issues (consider how self-employment impacts the choice to layoff vs. take a wage reduction).

    What better example than the oligopolistic banking sector. Surely _that_ has relevance to macroeconomic models?

  17. Gravatar of ssumner ssumner
    25. February 2011 at 15:06

    David, I assume that there is a connection between NGDP and RGDP due to sticky wages and prices–which is a pretty standard assumption. I suppose I put a bit more weight on sticky wages, but that difference isn’t key.

    MarkW, Exactly,

    Benjamin, Yes, they are net beneficiaries, but I think their current prosperity is due to the commodities boom.

    Doc Merlin, I’m not sure that nominal assets are enough, as that just creates sunk costs and gains, which should not affect behavior at the margin. But in a severe financial crisis like 2008, it probably is enough.

    JPIrving, I agree.

    JimP, Well at least she is saying sensible things. Now let’s see her do it.

    Morgan, You said;

    “1. Goldman Sachs is screaming for stimulus because they are evil.”

    No just foolish. I hope you are right about Bernanke–I think you are.

    Statsguy, I don’t agree on Doc Merlin’s point.

    You said;

    “First, you picked two of the most regulated/government subsidized companies in the world. Even so, yes, there are implications on the structure of the macro-economy.”

    They are subsidized because they aren’t particularly profitable, which undercuts your argument that duopolies are profitable.

    You said;

    “What better example than the oligopolistic banking sector. Surely _that_ has relevance to macroeconomic models?”

    What worse example, there are close to 5000 banks in the US. The industry is highly competitive (actually monopolisitic comp.)

    I don’t dispute that the US economy does not behave as a perfectly competitive economy would behave, but we don’t need studies to show that. But I think monopolistic competition is a far better model than oligopoly. Indeed I see very little value from oligopoly models.

  18. Gravatar of Morgan Warstler Morgan Warstler
    25. February 2011 at 15:25

    There is a BETTER chance there Scott, if you are beating the Three Step Drum.

    Reach out to the WSJ and tell them to expect your OpEd.

    Scott, they are baddies, you need to hit them, to make the stronger argument:

    http://washingtonexaminer.com/blogs/beltway-confidential/2011/02/check-out-krugman-column-i-mean-goldman-sachs-advocacy

    Ben doing QE to ensure there’s no more fiscal stimulus, is a way to proves he’s a good conservative.

  19. Gravatar of Benjamin Cole Benjamin Cole
    25. February 2011 at 16:07

    Bams-

    Actually, I am not a liberal. I would prefer to limit the federal share of GDP to 16 percent, by sunsetting the USDA, VA, Education, Labor and Commerce, and cutting defense by 75 percent.

    However, the budgetary success of rural states is longstanding, and a major drain on the Treasury. I see no reason why some states should subsidize other states not for a few years, or decades, but for generations.

    This has created a huge rural subsidy economy, where everything is subsidized, as infrastructure is subsidized–the roads, water systems, power systems, telephone systems, postal service, airports, train stops, etc etc etc. Add to that the huge ethanol sop, and crop subsidies, and you gEt an enfeebled, knock-kneed, weakling rural economy that cannot survive without constant federal infusions of cash.

    Scott Sumner–Whyen Kentucky gets $4k net per resident from the feds, I wonder if a commodities boom (read in part, “ethanol sop”) is really that important. It is filip, to be sure.

  20. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    25. February 2011 at 17:17

    ‘…cutting defense by 75 percent….’

    You mean reducing our 1.5 million man uniformed military to under 400,000? Be a bit of a problem if there was a war, no?

  21. Gravatar of Benjamin Cole Benjamin Cole
    25. February 2011 at 18:39

    Patrick S-

    If there was a war, yes, we might mobilize. However, a force of 400,000 men, including fleets of hunter-killer subs and “ballistic” subs would surely be enough to defend our shores. Just who do you predict would invade? By what route?

    Exactly which war since WWII do you think was necessary? Vietnam? Iraq? Afghanistan? Would we not be several trillion dollars ahead if US Presidents, for idiosyncratic reasons, did not have a tool to expensively entangle us in foreign conflicts?

    Are you aware that our Founding Fathers detested, loathed and nearly outlawed standing armies, which is the reason for the Constitutional language regarding “right to bear arms and form militias”? That George Mason (the far right-wing university is named after him) refused to sign the Constitution as it did not explicitly ban standing armies?

    That immediately after the Revolutionary War (during which Redcoats and Hessians held entire US cities, such as New York) the Congress completely demobilized the Continental Army? That during Washington’s first term as President, he did not have an army of 500 men?

    And that after WWII, the right wing led a near-total demobilization of our Armed Forces, so much so we did have the troops to commit to South Korea? (A war Eisenhower refused to prosecute, btw). Some right-wingers declared that the advent of nuclear weapons meant we no longer needed a navy.

    The idea of permanent mobilization started in the Cold War, a fancy of the “iron triangle” of the federal bureaucracy, defense interest groups, and Congressional leaders with defense-heavy districts. In this regard, the federal defense establishment is little different from the farm establishment, the HUD establishment, or the any other number of federal bureaucracies which have to justify their existences.

    BTW, I forgot to add HUD to the list of federal agencies that could be profitably sunsetted.

  22. Gravatar of StatsGuy StatsGuy
    25. February 2011 at 19:43

    ssumner: “But I think monopolistic competition is a far better model than oligopoly.”

    Not a battle I care to fight (though woodford might)… first, I agree with you in most cases, second, it’s nearly impossible to separate oligopoly and regulation since nearly every industry that is a potential oligopoly is regulated in some way.

    I suppose I’m just trying to figure out whether you are picking a fight, or you really believe micro studies are irrelevant to macro.

    “I don’t agree on Doc Merlin’s point.”

    Heh, of course you don’t! If nominal shocks propagated to the real economy through contracts even when NGDP was targeted successfully, this would mean that a purely distributional effect could impact RGDP (the share of NGDP growth that is real). That’s some serious blasphemy, man.

    Defending it requires that you break a lot of the clean assumptions in arrow/debreu. In the short term, things like mortality, bankruptcy/survival thresholds, or behavioral factors can get you there. In the long term, after an economy is structurally adjusted to a new wealth/consumption distribution, you need bigger arguments – things like impact of inequality on property rights and political institutions, relationship between wealth and culture, externalities due to insufficient education of poor, subpar investment in public goods, government capture by the rich leading to state-aided monopolies, predatory taxation… All the things developmental economists talk about.

    “Distribution” is still taboo in economics, as Alan Blinder recently discovered. Those who utter it get stuck with the unpopular crowd (like those “developmental” economists, snicker snicker) and don’t get to hang with the cool kids. Da Boyz.

    The thing I find funny, though, is that the intellectual implications of many “austrian” ideas sync closely with the implications of many folks in developmental economics, even though the policy recos are often diametrically opposed.

  23. Gravatar of TGGP TGGP
    25. February 2011 at 22:30

    “Goldman Sachs is screaming for stimulus because they are evil.”
    I’d like to hear some elaboration on that theory.

  24. Gravatar of Lorenzo from Oz Lorenzo from Oz
    26. February 2011 at 02:31

    There is something in modern economics which seems to make people not want to see monetary effects. A case from economic history: it is fairly obvious that the flood of silver from the Americas from 1550, building on the fivefold increase in production from central European silver minds from c.1460, was a central feature driving global trade patterns until about 1830. It priced European goods out of Asian markets (except where there were no competitors) and Asian goods into European markets, as about a third of the total silver mined went into buying Asian goods. (Given silver was the dominant medium of exchange and the Asian economy was much larger than the European one, the price implications are surely obvious: Smith discusses it in The Wealth of Nations [Vol. I, Bk. I, Ch. XI, Pt. III].) Yet one gets perfectly serious economic historians talking about the “strange reluctance” of Asians to buy European goods, or even the “inferiority” of European goods. Or studies of global trade which do not even consider this factor.

    Money is how one makes and accepts offers, the basic signaling device of commerce, the markers for transactions. It is not some mere impotent ‘ether’ between goods and services, for it connects all monetised goods and services with each other. It is what you use to stay in the game, to go up and down in the game, how you keep score. What happens to and with your markers therefore matters.

    I wonder if (since the language you use affects how you think) whether the language of ‘nominal’ and ‘real’ does some damage here, as if goods and services are all there is to the “real” economy so what happens to mere markers, the “merely nominal”, does not matter.

  25. Gravatar of Lorenzo from Oz Lorenzo from Oz
    26. February 2011 at 02:34

    Or, to put it another way: imagine an economy without money. (I don’t mean without notes and coins, I mean without money at all, just goods and services.) Then try and claim that money does not matter for the “real” economy.

  26. Gravatar of Lorenzo from Oz Lorenzo from Oz
    26. February 2011 at 02:37

    Bugger, I managed to post this to the wrong post. Oh well, you get the point.

  27. Gravatar of Scott Sumner Scott Sumner
    26. February 2011 at 06:29

    Morgan, You said;

    “Ben doing QE to ensure there’s no more fiscal stimulus, is a way to proves he’s a good conservative.”

    He needs to convince Plosser, Fisher, et al.

    Benjamin, I’m not sure Kentucky is booming–it is the states with even smaller populations (Dakotas, Wyoming, Alaska, etc) that benefit most from commodities. Kentucky has a decent population.

    Statsguy, You said;

    “I suppose I’m just trying to figure out whether you are picking a fight, or you really believe micro studies are irrelevant to macro.”

    OK, I’ll amend that to “rarely have any relevance to macro.” I was thinking about studies that looked at particular regions, particular industries, or particular professions. If you define “micro” as studies of whether price stickiness is widespread across the economy, or whether market power is widespread across the economy, then I’ll grant you that they may have some significance, though I doubt they tell us much.

    All the distribution effects that you describe aren’t going to have important business cycle effects. You said;

    “”Distribution” is still taboo in economics, as Alan Blinder recently discovered.”

    There is nothing taboo about distribution. 75% of economists vote Democratic, and they almost all favor income distribution. The taboo was free trade. Blinder did not present a single good argument for protectionism.

    Indeed he wasn’t really making a macro argument at all, so even if right his example doesn’t show that distribution matters at the macro level.

    Doc Merlin was implying that a change in income distribution might have some affect on aggregate expenditure. But that wouldn’t matter with NGDP targeting–if the poor bought less the rich would buy more, and vice versa. In practice, income distribution changes far too slowly to affect the business cycle. The income distribution argument is basically a reallocation argument. It’s just not empirically important.

    Lorenzo, Yes, I think there is exactly that bias against nominal. Even the term sounds unimportant.

  28. Gravatar of LibertyDefender LibertyDefender
    26. February 2011 at 06:51

    Am I–a layman–the only one who bristles when Krugman speaks only of federal spending, without accounting for such facts as

    1. The spending is financed entirely by borrowing; and
    2. Too much of the federal spending is for consumption?

    We can argue over the definition of consumption, but much of the Stimulus spending went to states that spent it on retaining public sector jobs, along with their unsustainable pension and medical insurance obligations.

    Dambisa Moyo, in her book “How the West was Lost,” stresses that federal borrowing to support consumption is ruinous. Her prescription is (inter alia) infrastructure investment, and R&D of productivity-enhancing technology improvements.

    When Krugman says (I’m paraphrasing slightly) “Well duh, of course more federal spending in a given state or county creates more jobs,” it strikes me as aggressively ignorant re the targets of the federal spending. For example, one part of the $787 Billion Stimulus Bill was to subsidize COBRA health insurance premiums of the unemployed. In other words, subsidized unemployment. As even Krugman understands, if you subsidize something (e.g., unemployment), you get more of it.

    How much of that $787 Billion Stimulus package was spent on inefficient public sector employees? More federal spending automatically means more jobs? I’d laugh it if weren’t so frustratingly tragic.

    Perhaps more to the point of this blog post, though: I’ve seen Dambisa Moyo speak twice re her book “How the West was Lost,” with its discussion of the choices and consequences of macro policies on labor, capital, and productivity. In each talk, the first question invariably is “but what about corporate executive pay?”–a micro issue. Argh. Dr. Moyo politely answers the questions (usu. by pointing out that bondholders should police executive pay, but government meddling through bond guarantees removes the incentive for bondholders to engage in such policing), and leaves the micro-whiners with the impression that their questions are relevant.

  29. Gravatar of TheMoneyIllusion » Random links TheMoneyIllusion » Random links
    26. February 2011 at 10:27

    […] 1: Yesterday I did a post showing how all sorts of prominent people on the left recently engaged in the fallacy of […]

  30. Gravatar of Doc Merlin Doc Merlin
    26. February 2011 at 16:42

    @Stats Guy
    “I of course do agree with you, but you do realize that the above is only true if the distribution of wealth (or, at the very minimum, changes in the distribution of wealth), impact RGDP (growth).”

    I don’t see why forced distributional changes away from the optimum couldn’t hurt RGDP growth. Its something conservatives have been saying for years.

    In addition, you don’t need even distributional changes to cause the damage, only for everyone to realize they are poorer than they thought they were.
    Here is a thought experiment:
    Everyone in society owns a 100k worth of debt. The payer on that debt defaults for everyone simultaneously, suddenly everyone is poorer than they thought they were, and the recalculation costs directly would hurt RGDP.

    @Scott
    “Doc Merlin, I’m not sure that nominal assets are enough, as that just creates sunk costs and gains, which should not affect behavior at the margin. But in a severe financial crisis like 2008, it probably is enough.”

    Depends on the distribution of the changes, a small change for everyone, and everyone would just tighten their belts and continue to pay their debts (which makes the money behave stickily), however if the changes were severe for some and small for others, it would still be felt strongly RGDP.

    Really this is yet another reason why we need to get rid of the fed, or at-least tie its hands. It prints too much when it shouldn’t which causes the need for excess tightness later, and doesn’t print enough when it should, which causes the need for excess printing later, thus leading to more instability!

  31. Gravatar of Doc Merlin Doc Merlin
    26. February 2011 at 16:49

    @Scott:
    “Doc Merlin was implying that a change in income distribution might have some affect on aggregate expenditure. But that wouldn’t matter with NGDP targeting-if the poor bought less the rich would buy more, and vice versa. In practice, income distribution changes far too slowly to affect the business cycle. The income distribution argument is basically a reallocation argument. It’s just not empirically important.”

    Even with per capita NGDP targeting, you could still get nominal shocks across currencies, however you wouldn’t find nearly as many big ones within a currency, so /most/ of the problem is taken away. You can also have policy shocks from bad regulatory policy which are real shocks but look kind of like nominal ones. (Amity Shays has been basically pointing at the GD as an example of that.)

  32. Gravatar of Scott Sumner Scott Sumner
    26. February 2011 at 18:40

    Liberty defender, I think that people like Krugman are so convinced they are right, that they overlook the logical fallacy in their analysis of stimulus.

    Doc Merlin, You said;

    “Depends on the distribution of the changes, a small change for everyone, and everyone would just tighten their belts and continue to pay their debts”

    I puzzled, by this, the issue is not who pays what, but how much people want to work. It’s not clear that debts make people want to work less.

    Amity Shlaes wrote a good book, but it’s more a cultural history than an economic history. There’s no modeling of NGDP

  33. Gravatar of Mark A. Sadowski Mark A. Sadowski
    26. February 2011 at 20:13

    Scott wrote:
    “Micro studies can’t tell us anything about macro.”

    Um, yeah. But some of the big geniuses evidently don’t get that quite yet.

  34. Gravatar of StatsGuy StatsGuy
    27. February 2011 at 09:42

    @Doc Merlin

    “Everyone in society owns a 100k worth of debt. The payer on that debt defaults for everyone simultaneously, suddenly everyone is poorer than they thought they were, and the recalculation costs directly would hurt RGDP.”

    No, not at all – if I owe a 100k debt to you, and I default for no reason other than I decide to default, you’re 100k poorer than you thought you were, and I’m 100k richer.

    If the REASON I default is because there was a massive supply shock, well, that’s different.

    However, even if there is no supply shock in the economy, and suddenly half the population defaults on its debt because it feels like staging a debt revolt, I guarantee you there will be a massive economic crash. You have to think about why – the reasons lead one down one of the many theoretical paths…

    Maybe the demands of the new wealthy (aka, less indebted) are different, and so the economy must recalculate to serve their needs. In this case, an NGDP target may help restore the existing wealth distribution and hence preserve the current distributional structure, and thus limit the cost of recalculation.

    Maybe there’s a massive de-risking, as people start to think being a creditor isn’t as safe, and so money velocity overall shrinks (cutting the credit supply), and this creates a demand collapse and a nominal price implosion.

    Maybe there’s a “Minsky moment” as everyone realizes that the _reason_ that people are defaulting isn’t because they are keeping their own future wealth, but because they never had the future wealth with which to pay their creditors, and so (as you say) everyone realizes they are poorer.

    There are other explanations.

    Which theory, or variant, you believe tends to determine your belief in the efficacy of monetary policy (or an NGDP target). Is the shock a repricing of risk, or is it a sudden recognition of future supply constraints (of which malinvestment is one cause)?

    The strongest plank in the pop-austrian argument is that the recent crisis was a recognition of future supply constraints, which means we should conserve more now to preserve future consumption. Oil, land, water, metals, etc. Forcing consumption now will only impoverish future selves/generations.

    Of course, if that really IS true, then every attempt by the Fed to restore the real trajectory will be met by arbitrage – inflation in commodities and wealth redistribution instead of a real pickup in aggregate consumption. Pop austrians DO believe this, but they believe that money illusion will cause many people to fail to recognize the truth, and thus continue consuming until there is a truly massive crash.

    This is the locus of the real argument – ** Will or will not technology keep up with population growth and resource depletion/contamination? **

    Scott is a technological optimist, as he’s shown in many posts. He rejects, for instance, the argument that living standards for median income families has held flat (or possibly declined). This is consistent with his belief that the current problem is 80% an aggregate demand/price level problem.

    I don’t pretend to know the answer here, but it’s one reason why I haven’t completely bought into ssumner’s arguments, or completely rejected the pop austrian position.

  35. Gravatar of Mindestlöhne, Gewerkschaften und die Neoklassik « Aus dem Hollerbusch Mindestlöhne, Gewerkschaften und die Neoklassik « Aus dem Hollerbusch
    27. February 2011 at 13:39

    […] den Artikel hat mich ein Link von Scott Sumner geführt, der selbst zuvor einen Artikel über den angewandten Trugschluss der Verallgemeinerung geschrieben hat. Kurz zusammengefasst: Datenpunkte, die darauf hinweisen, dass es einzelnen wegen einer Maßnahme […]

  36. Gravatar of Morgan Warstler Morgan Warstler
    27. February 2011 at 16:39

    “”Ben doing QE to ensure there’s no more fiscal stimulus, is a way to proves he’s a good conservative.”

    He needs to convince Plosser, Fisher, et al.”

    Scott, so you mean to tell me if Ben said for his fed oppositon,

    “ok give me more QE, support it out loud – and I’ll announce that the Fed categorically neutralizes Fiscal Stimulus, so even trying it, even discussing it, is a fools errand.”

    You don’t think Plosser etc. would go along?

  37. Gravatar of Scott Sumner Scott Sumner
    28. February 2011 at 06:13

    Statsguy, You said;

    “Scott is a technological optimist, as he’s shown in many posts. He rejects, for instance, the argument that living standards for median income families has held flat (or possibly declined). This is consistent with his belief that the current problem is 80% an aggregate demand/price level problem.”

    Ironically, I’ve been posting in support of Tyler Cowen’s argument that technological progress has slowed since 1973.

    I don’t agree with your default hypothesis. I don’t think a major default would cause a recession. It’s wrong to think of borrowers as “the poor.” Borrowers tend to be affluent.

    Morgan, Plosser would not go along.

  38. Gravatar of Lies, damn lies, and Paul Krugman « Internet Scofflaw Lies, damn lies, and Paul Krugman « Internet Scofflaw
    28. February 2011 at 20:58

    […] Scott Sumner points out that Paul Krugman’s latest column makes a basic logical error. Krugman notes a study that finds that stimulus spending leads to more jobs where the money is spent, but then makes the logical leap to the proposition that it leads to more jobs overall. Since stimulus spending is highly uneven (indeed, it is allocated exclusively by political considerations), the one hardly implies the other. […]

  39. Gravatar of Doc Merlin Doc Merlin
    2. March 2011 at 07:16

    @Scott
    ‘”Depends on the distribution of the changes, a small change for everyone, and everyone would just tighten their belts and continue to pay their debts”

    I puzzled, by this, the issue is not who pays what, but how much people want to work. It’s not clear that debts make people want to work less.’

    I meant distribution in the stastical time series sense, not a cross sectional sense. I should have been more clear.

  40. Gravatar of Doc Merlin Doc Merlin
    2. March 2011 at 07:17

    bah, that should be statistical not stastical.

  41. Gravatar of TheMoneyIllusion » Paul Krugman: Ignorant, and proud of it TheMoneyIllusion » Paul Krugman: Ignorant, and proud of it
    9. March 2011 at 09:11

    […] get out a bit more he might make fewer embarrassing errors,  like this one, where he forgot the fallacy of composition, something taught in EC101.  I guess none of his liberal friends have the nerve to point out these […]

  42. Gravatar of orionorbit orionorbit
    12. March 2011 at 09:18

    This post felt weird when I read it and it took me a while to figure what was wrong with it, but thanks to your response to deLong’s post, I think I got it.

    You and Krugman have a misunderstanding on what social science does. We do not advance knowledge by finding a hypothesis (such as “fiscal stimulus works”) but by refuting it, i.e. finding evidence that it does not work. In your response to deLong, you kind of got it the wrong way, by taking as a starting point that the hypothesis is that stimulus doesn’t work and then showing that infective stimulus can still record significant differences in cross sectional data, but this was never the question.

    Krugman starts to refute the hypothesis “fiscal stimulus works” and correctly thinks that finding a cross section of states where smaller states had higher federal funding and spending but did not do significantly better in jobs, would be enough evidence against the hypothesis. He fails to find that evidence, ergo he is glad as he can continue operating under the working assumption that stimulus works; he even says so in his post.

    Now of course you make a good point that if the positive effects of spending in small states would be offset by negative effects in other, it would simply be a transfer payment, but I would say that there isn’t much evidence in favour of this one. IF you find that evidence then yes, there is fallacy of composition, but you seem already convinced.

    But on the other hand, the two main ways you can look for evidence against the hypothesis “stimulus spending works”, are to either find a cross section like the one above and find no relationship, or to find the relationship there but that the effects of spending in one state are offset in another. I can’t think of another way, perhaps you have? Or perhaps you have evidence that the country-wide effects aren’t there? If so I’d like to read about it but until then I would say that Krugman and deLong are quite right to be happy about this study and also could just as well be right about the conservative blogs; even though I do read Mankiw and Taylor every now and then.

  43. Gravatar of Doc Merlin Doc Merlin
    12. March 2011 at 10:44

    @orionorbit

    NONSENSE!

    We require some amount of reason for our social science positions as well, or at least some razor to rule of thumb positive beliefs. Otherwise we end up with the “I believe in a teacup behind Saturn” sort of nonsense.

  44. Gravatar of Doc Merlin Doc Merlin
    12. March 2011 at 10:45

    I should add, using regression analysis, its REALLY easy to generate null results, by picking a model in such a way as it does.

  45. Gravatar of ssumner ssumner
    16. March 2011 at 16:00

    orionorbit, You’ve misunderstood what is going on here. You have two hypotheses. Hypothesis A predicts X. Hypothesis B predicts X. A study shows evidence of X. Krugman and DeLong say this is evidence in favor of A and against B. But it’s not. That’s just logic 101

    You are making it way too complicated.

    You can do cross-sectional studies to test fiscal stimulus, but they must be between more than one country–not within one country. Otherwise you have the fallacy of composition.

  46. Gravatar of TheMoneyIllusion » Progressive wishful thinking TheMoneyIllusion » Progressive wishful thinking
    26. March 2011 at 07:39

    […] have a number of posts showing that Krugman, DeLong, Dean Baker, etc, all falsely claimed that evidence of regional growth […]

  47. Gravatar of Conley & Dupor is not the first empirical ARRA study | Lobster Stuffed With Tacos Conley & Dupor is not the first empirical ARRA study | Lobster Stuffed With Tacos
    16. May 2011 at 19:03

    […] Sumner critiques here. So Conley & Dupor is now the second empirical ARRA paper that 1) exploits regional […]

  48. Gravatar of TheMoneyIllusion » The Keynesian bubble TheMoneyIllusion » The Keynesian bubble
    16. January 2012 at 08:55

    […] all the anti-Keynesian models that I am aware of would predict the same result.  But people like Paul Krugman and Brad DeLong latched on to this as evidence in favor of the Keynesian model.  How can it be evidence in favor, […]

  49. Gravatar of TheMoneyIllusion » DeLong on the fiscal multiplier TheMoneyIllusion » DeLong on the fiscal multiplier
    18. February 2013 at 18:37

    […] within the euro.  They will grossly overestimate the aggregate fiscal multiplier.  This includes studies cited approvingly by Krugman and […]

Leave a Reply