Be careful of accounting identities

Yesterday Krugman had a post claiming that freer trade would not boost demand, and hence would not promote recovery from the current recession:

There are a lot of good things you can say about international trade. But it does not, repeat not, do anything to alleviate a shortage of overall demand. Yes, if you liberalize trade countries will export more. But they will also import more. If you’re worried about C+I+G+X-M, it’s a wash, because X and M rise equally.

Which makes this WaPo editorial on things Obama should be doing about jobs truly bizarre. Even if the proposed trade deals with Korea and Colombia were remotely big enough to bear mentioning in the context of the crisis “” which they aren’t “” they wouldn’t be job creation measures.

I am going to take issue with his argument, but first let me acknowledge the areas where we agree.  The Washington Post editorial was very weak.  And the two free trade agreements would not significantly impact the macroeconomy.  Nevertheless, I am worried that people will draw the wrong inferences from the argument that he makes.  I will argue that his analysis here can only be defended if you assume that fiscal stimulus is also ineffective.  Of course Krugman doesn’t think fiscal stimulus is ineffective.

Let’s start with the problems inherent in arguing from accounting identities.  We know in a closed economy with no government that C + I = C + S.  Income not consumed is saved.  And there are two kinds of goods, consumer goods and capital goods.  From these definitions we know that savings equals investment.

Some classical economists allegedly argued that a drop in spending couldn’t cause a recession, because a drop in spending meant an increase in saving.  And since savings equals investment, a drop in C will be offset by a rise in I.  I seem to recall that earlier in the year Krugman ridiculed that sort of argument based on accounting identities.  Yes, ex post savings equals ex post investment, but the attempt to save more may cause a reduction in both actual consumption and actual savings and investment.

I think there is a similar flaw in the argument that freer trade cannot boost AD.  But first let me nit-pick his wording.  The WaPo editorial argued that freer trade could create jobs.  Ditto for a lower minimum wage.  Those are generally regarded as supply-side policies.  It’s not clear they were arguing these policies would boost demand.  Nevertheless, I accept Krugman’s assumption that the main problem today is demand.  So let’s focus on that issue.

I going to reverse my normal procedure, by starting out with empirical evidence, and then asking why what is true in practice seems to conflict with our theories.

In the period around March and April 193o, there were a few “green shoots” in the economy.  The stock market recovered a significant chunk of the huge losses in 1929.  (I recall the Dow fell well below 200 during the famous crash, and got back up over 260 in April.  The 1929 peak had been 381.)  Then in May and June everything seemed to fall apart, and stocks crashed again.  So what happened in May and June?

The headline news stories during those months were the progress of Smoot-Hawley through Congress.  Each time it cleared a major legislative hurdle, the Dow fell sharply.  This pattern was obvious to those following the markets, and was frequently commented upon.  After it cleared Congress it went to Hoover.  The President received a petition from over 1000 economists pleading with him to veto the bill.  (A veto would not have been overridden.)  Over the weekend Hoover decided to sign the bill, and on Monday the Dow suffered its biggest single day drop of the entire year.

Yes, the stock market isn’t the economy, and this incident certainly doesn’t prove anything about the impact of Smoot-Hawley on aggregate demand.  But it will give us clues to the transmission mechanism.  Before explaining why stocks fell, let’s look at one additional piece of evidence.  Recall that standard theory says trade barriers are a supply shock.  And we all know that supply shocks reduce output and raise prices.  There is just one problem with the standard view; it doesn’t appear to be true.  The various commodity and wholesale price indices fell sharply at the same time that the stock market was plunging in May and June.  And commodities were a far larger part of the economy in the 1930s than they are today.  So Smoot-Hawley was almost certainly deflationary.

When I saw the real world impact of Smoot-Hawley, I realized that there was something wrong with the standard Keynesian view of protectionism.  Keynes wasn’t a protectionist, but he did argue that during a depression a country might benefit from tariffs that kept out cheap foreign goods.  According to this view, Smoot-Hawley should have been inflationary.  So what happened?

My answer relates to my earlier discussion of China’s weak yuan policy.  Trade is not a zero sum game.  Smoot-Hawley, along with likely foreign retaliation, had a negative impact on the world economy.  It is very likely that the (unobservable) Wicksellian equilibrium interest rate fell all over the world as the result of these trade barriers.  And I believe the fall in the stock market was a symptom of this decline.  Keynes would have called it a decline in “confidence.”

In this paper:

Gibson’s Paradox and the Gold Standard ES9SYÏÏ

Barsky and Summers provided a model capable of explaining the deflationary consequences of trade wars under a gold standard (although they didn’t use their model for that purpose.)  In their model they assume the price of gold is fixed.  Because the supply of new gold from mines is fairly steady, short term fluctuations in the price level primarily reflect fluctuations in the demand for gold.  And these are caused by fluctuations in the interest rate, which is the opportunity cost of holding gold.  Indeed their paper provided a neat explanation of the so-called Gibson Paradox, the tendency of prices and interest rates to rise and fall together under the gold standard.

So if there was a technological revolution that made the real economy boom, or if the boom was created by free market policies, then investment would soar, credit demand would increase, and interest rates would rise.  As the opportunity cost of gold rose, demand for gold would fall, and price levels would rise.

Smoot-Hawley did the opposite.  It reduced investment all over the world.  Interest rates fell, the opportunity cost of holding gold fell, and the demand for gold rose.  This caused deflation, which made the Depression even worse.

Many of my commenters have an intuition that real shocks and a decline in NGDP are related, or “entangled” as I like to put it.  Their intuition is not correct for a properly managed fiat money regime.  The housing bust did lower the demand for credit, it did lower the risk free interest rate, but it should not have lowered NGDP if monetary policy was properly managed.

But their intuition is correct for the classical gold standard.  Some economists naively assume that the MV=PY equation proves growth is deflationary.  And yet under the classical gold standard cyclical growth tended to be inflationary.  Prices tended to be a bit higher in booms than depressions, even when the trend was down (as in the late 1800s.)  Growth is only deflationary if monetary policy is handled properly, i.e. if the Fed stabilizes NGDP, or its growth rate.  (This of course relates to Selgin’s argument for the productivity norm.)

Krugman has argued that fiscal expansion will raise demand, and by implication NGDP.  To those who argue that it will merely crowd out private spending, Krugman notes that this “crowding put” would only occur if the Fed tightened monetary policy, and (earlier in the year) he suggested that Bernanke wouldn’t do that in a deep recession.   Let’s assume Krugman was correct.  Then he is essentially arguing that monetary policy would be fairly passive in response to fiscal stimulus, in much the same way it was under a gold standard.  I have my doubts about that assumption, but let’s say he’s right.  In that case an expansionary fiscal policy can cause NGDP to rise for any given money supply.  Velocity will rise.  And that is precisely the argument I am making about trade.  Freer trade throughout the world boosts business confidence.  It boosts stock prices, often dramatically.  Stocks rose significantly after it was apparent that NAFTA would pass.  Higher stock prices and stronger business confidence raise the Wicksellian equilibrium real interest rate.  And this raises aggregate demand if the Fed holds nominal rates fixed.

In the end I agree with Krugman’s conclusion, the WaPo story was weak.  So why nit-pick his argument?  It’s fun.  .  . I mean it’s important that people not get the wrong idea about trade barriers.  The Keynesian model suggests that mercantilism can help an economy mired in a Depression.  Smoot-Hawley was an almost perfect test.  You may not care about the stock market, you may (wrongly) think the stock market only affects fat cats.  But the deflationary impact of Smoot-Hawley on goods prices was very ominous.  It was a very bad sign for the real economy, for working Americans.  It meant lower production was on the way, along with fewer jobs.  And the markets were right.  And Hoover and Keynes were wrong.

If Kim Jong Il conquered China tomorrow, and closed it off from the world economy, the US trade deficit would shrink.  But so would our economy.



21 Responses to “Be careful of accounting identities”

  1. Gravatar of JimP JimP
    11. December 2009 at 14:34

    And hopefully Martin Wolf and Robert Aliber will read this post.

    Because they are dreaming of a trade war with China.

  2. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    11. December 2009 at 16:36

    Wow, Gibson’s Paradox! I thought I was the only person in western civilization who knew about that. When was the last time you saw anyone ekse reference it, Scott?

  3. Gravatar of ssumner ssumner
    11. December 2009 at 17:13

    JimP Thanks. I agree that a trade war is a crazy idea. Too many fires to put out at once!

    Patrick, I actually published a paper on the Gibson Paradox way back when. I’d probably start doing posts on my old papers if I had time.

  4. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    11. December 2009 at 18:27

    So, I take it you don’t remember anyone referencing Gibson’s Paradox lately either, even though it was a famous historical controversy that, iirc, baffled Keynes and Wicksell, but not Irving Fisher.

    I came across it back in about 1980 in a text book used in an undergrad Money and Banking course at UCLA, and found it amusing that the author didn’t understand what it signified. As elsewhere in the book he/she wrote that, ‘interest rates can be thought of as the price of money’.

    As I don’t need to tell you, the key to understanding it isn’t a paradox at all, is that interest rates ARE NOT ‘the price of money’; they are the price of borrowing money, not buying it.

    Might make a good post someday for you to explain it.

  5. Gravatar of OGT OGT
    11. December 2009 at 20:12

    The US had a long history of applying tariffs in recessions before Smoot-Hawley. Generally the policy had been seen as relatively successful, the main difference was that by 1930 we had a huge current account surplus something like .5% of global GDP.

    So, the confidence thing seems like a bit of bank shot. True investors have a better idea about the long term implications of global trade, and global cooperation. But, I’d be surprised if the Chinese government announced greater export subsidies wouldn’t send the Shanghai Stock Exchange soaring. Or the same on Wall Street. And, yet, as we know, the trade distorting effects of taxes and subsidies are largely the same.

  6. Gravatar of ssumner ssumner
    12. December 2009 at 06:43

    patrick, I think the reason people no longer talk much about the Gibson Paradox is that it sort of broke down after we went off gold. Instead you get a “Fisher effect” The tendency of interest rates to be correlated with the rate of inflation. Under the Gibson paradox interest rates are crrelated with the price level. Yes, maybe I’ll do a post sometime. It helps explain what’s wrong with a gold standard.

    OGT, I don’t see my argument as a bank shot at all. We know that Smott-Hawley caused a crash in the stock markets in 1930. There are few stock market crashes in American history where the causality is more obvious. Of course there are many reasons why tariffs might reduce stock prices, and there are many reasons why the effect might be time-varying. For instance it is possible that retaliation was more likely in 1930. In my study of the depression I made the argument that Smoot-Hawley made international monetary cooperation less likely, and that increased deflationary ecpectations. I haven’t studied earleir tariffs, so I can’t say what their effect was. Again I do recall that NAFTA had a positive effect on the stock market, however.

    Regarding your second argument, I have an open mind. But I’d point out that almost all countries in the world start from a position of some trade barriers. If you have something like a flat 10% tariff on imports, and then add export subsidies, the subsidies will actually move you closer to a non-distortionary position. A better test of your argument would be as follows:

    Say Hong Kong or Singapore (which are tariff-free) raise their income tax rate and start using the money to subsidize exports. Would that raise their stock markets? I am not so sure.

  7. Gravatar of Bill Stepp Bill Stepp
    12. December 2009 at 09:06

    Y does not equal C + I + G + (X – M).
    When I go to the store and buy something, I’m exporting money and importing goods. When I sell my labor to an employer, I’m exporting labor and importing money.
    Also, the State is a criminal organization that steals to obtain resources, then showers the stolen loot on secondary criminals, so G is waste consumption, and T is theft.

    The correct equation is:
    Y = C + I – (G + T), where C + I = private production, and
    G + T = government depradation.

  8. Gravatar of 123 123
    12. December 2009 at 14:19

    It is not really right to mention Paul “Smoot Hawley” Krugman and Kim Jong “Che” Il in the same post, as Krugman is seeking the optimal sixth best policy, whereas Il is more concerned with finding the optimal thirteenth best policy.

  9. Gravatar of ssumner ssumner
    12. December 2009 at 20:03

    Bill, OK, but what about exports?

    123, Yes, that was unfair. 🙂

  10. Gravatar of Doc Merlin Doc Merlin
    13. December 2009 at 01:31

    Good posting, Scott.

    I agree, tariffs, would trash the US economy at this point, and free trade would help it a lot. Again, these legal/policy factors have a huge impact that is hard to predict quantitatively. Since 2006, this congress has done everything horribly.

    1. They didn’t extend the bush tax cuts which would have helped during the recession.
    2. They scared a lot of domestic producers with cap and trade.
    3. They have been talking about tarrifs, included some protectionist things in the stimulus, and raised some tarrifs on chinese goods.
    4. They built a stimulus package, but one which most of the spending wouldn’t go into effect for several years. (Even if you are Keynsian, what is the logic of that?)
    5. They haven’t voted for the free trade package with South Korea.

  11. Gravatar of Bill Stepp Bill Stepp
    13. December 2009 at 05:54


    What about exports? As Don Boudreaux has pointed out many times (at Cafe Hayek), exports are what someone pays to import something.

    When you export something, you are selling goods or services (ultimately services, as Rothbard pointed out in Man, Economy, and State) and receiving an import, a good or service.
    Rothbard also pointed out that international trade theory is basically a big fallacy. National barriers are imaginary lines, and don’t affect the fundamentals of trade. They do effect the empirical terms of trade, but so do local taxes, to name one class of state crime.

    For example, Jeff Bezos left the People’s Republic of New York to start in Washington because of the former’s absurdly high tax robbery rates. He got better terms of trade in Washington. Fewer commies out there, and less of an entitlement culture of crookeaucracy.
    Let the Soviets in City Hall think about that. Starting with Comrade Bloomberg.

  12. Gravatar of Scott Sumner Scott Sumner
    13. December 2009 at 08:10

    Doc and Bill, I agree.

  13. Gravatar of Michael Pettis Michael Pettis
    14. December 2009 at 23:47


    I know I am coming in late, but I have to agree with what I think OGT is saying. It seems to me that the impact of trade contraction depends to at least some extent on whether a country is running a large trade surplus or deficit. One of the reasons Hoover had difficulty vetoing SH was because in the past raising tariffs seemed to have been expansionary, so it seemed a logical policy in 1930 to deal with the contraction in global demand. I would argue however that because before WW1 the US mostly ran trade deficits, and sometimes very large ones, the impact of trade contraction provided at least a short term increase in net domestic demand.

    Once it began running large trade surpluses, however, the same strategy would have the opposite impact since with a contraction in trade the excess of domestic production over domestic consumption had to be reconciled domestically without recourse to foreign net demand. In other words SH was a bad idea for surplus countries in both the long and short term if it invited retaliation, but not necessarily bad for deficit countries in the short term.

  14. Gravatar of Pacer Pacer
    14. December 2009 at 23:53

    I’d like to see some studies about the velocity of money in the hands of different countries. For example, high savings and investment rates mean that Japanese earnings are less apt to be re-circulated into world trade than are U.S. incomes to its consumption-oriented consumers.

    Then there is the effect of credit. When China runs a trade surplus, it invests that surplus in hard infrastructure and additional production capacity–each of which tend to lower the cost of goods (i.e. deflation). Latin American trade surpluses are more likely to be consumed in the form of end-user commodities and finished goods, which is inflationary.

    Of course, the trends today aren’t sustainable. The U.S. has moved forward so much consumption–through borrowing–and neglected investment in future productivity, that it’s a gravy train with limited tracks. Perhaps this ends in a bailout of the U.S. and a reversal of fortunes that will eventually lead to a more investment/savings-oriented U.S. subsidizing consumption in other countries. Won’t be such a nice time for Americans, but in today’s interconnected world there just doesn’t seem to be any way around generational warfare…

  15. Gravatar of Scott Sumner Scott Sumner
    15. December 2009 at 07:18

    Michael, I’d like to see the studies that show previous trade restrictions helped the US economy. I am not saying they are wrong, but I had a vague memory of articles suggesting that the US stock market never much liked protectionism. Perhaps my memory is wrong. Nevertheless, both the stock market and US economics professors strongly opposed Smoot-Hawley, so whatever the previous experience, there was both academic and “real world” belief that SH would have negative consequences, and they were right. The question is why.

    In other posts I have discussed a general problem with the “ceteris paribus” assumptions in all these policy counterfactuals. If monetary policy is doing its job, and targeting AD to grow at the desired rate, then no policy that has negative efficiency effects can help the economy, for the simple reason that it won’t boost expected AD, and it will worsen AS. This applies to both fiscal stimulus and protectionism. So any sort of second best policy like fiscal stimulus or protectionism requires some sort of assumption about inefficient monetary policy, so that the second best policy somehow makes NGDP rise. That is of course possible, especially if you are in a liquidity trap. But it is also far from certain, even in a liquidity trap. In that case one looks for real world evidence. Obviously SH didn’t work, and your surplus country explanation is plausible. But that is why I’d like to know more about the earlier examples. I recall the previous tariff was under Harding. And of course the economy did well under Harding. But that raises the issue of ceteris paribus. Harding also had very pro growth policies like big tax cuts and effective monetary policy after the steep deflation he inherited from Wilson.

    My own view is that it is dangerous to adopt policies that one knows are bad from an AS perspective, on the hope that it will be a backdoor way of getting NGDP higher, i.e. a way of circumventing monetary policy. The Fed clearly is looking at variables like real growth and inflation when deciding when to adopt its exit strategy, so Krugman’s assumption that we can hold Fed policy fixed in any counterfactual is very dubious. I’d rather hold NGDP fixed, and think about the policy’s impact on RGDP under those circumstances.

    Pacer, I think you are mixing up velocity and trade flows, which are two unrelated issues. Also, trade surpluses don’t mean a country is saving and investing a lot. I believe that both values have fallen sharply in Japan, it just means saving is greater than investment.

  16. Gravatar of Smoot-Hawley Revisited – Economics – Smoot-Hawley Revisited - Economics -
    26. December 2009 at 14:36

    […] were not a main reason for the severity of the Great Depression.  In an interesting blog post, economic historian Scott Sumner calls this conclusion into question:In the period around March and April 1930, there were a few “green shoots” in the economy. The […]

  17. Gravatar of Colin Colin
    26. December 2009 at 16:14

    The connection between the stock market’s performance and momentum for passage of Smoot-Hawley was covered fairly extensively by Jude Wanniski in his book The Way the World Works.

  18. Gravatar of jean jean
    26. December 2009 at 17:14

    Greg Mankiw reads your blog:

  19. Gravatar of ssumner ssumner
    3. January 2010 at 11:58

    Colin, Yes, I read his book, I agree with him about 1930, but not 1929.

    Thanks jean.

  20. Gravatar of Korbin Korbin
    6. March 2010 at 19:14

    This has been a very interesting article. It is interesting to note that there is no consensus when it comes to the effect of the Smoot-Hawley tarrif. From the U.S. Statistical Abstract, the real rate of tariff back in 1929 was approximately 13.5% and in 1933 it was effectively 19.8% where duty free imports effected 63% of total imports. From 1821 to 1900 the US tariff rates hovered around 29.7% and hit a high of 57.3% overshadowing the Smoot-Hawley rate.

  21. Gravatar of ssumner ssumner
    6. March 2010 at 19:44

    Korbin, I beleive the biggest problem might have been political (something I didn’t talk about in this post.) Smoot-Hawley poisoned the atmosphere for international monetary cooperation, which was the only hope for staving off world-wide deflation. Once central banks began acting competitively, not cooperatively, it was virtually impossible to arrest deflation under the gold standard.

Leave a Reply