Evidence of Reverse Causality?

In the comment section of an earlier post here, there has been some interesting discussion of real estate prices, so I thought a new post would be in order.  Last October when I first began arguing the reverse causality view (tight money–>>expectations for falling NGDP–>>falling asset prices–>>worsening financial crisis) I actually didn’t have much evidence.  I viewed the fall in NGDP as a Fed error of omission, and I saw circumstantial evidence that the debt crisis was spreading out of the subprimes and Alt-As into other sectors.  But it was mostly just common sense—I figured that rapidly falling NGDP could not be good for debtors, and banks were in a very fragile state even before NGDP began falling.

The 5 graphs in the post a few days ago showed how movements in real interest rates, inflation expectations, exchange rates, stock prices and commodity prices all pointed to a contractionary monetary shock in the late summer of 2008.  I forgot industrial production, which also broke sharply lower beginning in August 2008.  Now I’d like to consider a seventh variable, housing prices.

I had never thought much about this variable, as I had seen graphs showing fairly continuous declines for several years.  But I recently came upon data that caused me to sharply revise my views.  The graph attached here from the NAR suggests median home prices were about the same in January and August 2008 (roughly) $200,000, and then fell to about $172, 000 in January.  Their website just listed February 2009 at $165,400, down nearly 17% if I am not mistaken.

A commenter named Thruth showed me that this was too good to be true.   The NAR data is not seasonally adjusted, is not repeat sales, and underweights more expensive homes.  The Case-Shiller data uses repeat sales, is seasonally adjusted, and represents average prices.  But the C-S data has its own problems if one wants to understand the forces driving the banking crisis.  Foreclosures have been concentrated in lower and middle income areas like California’s “Inland Empire”, not the expensive homes near the coast.  So C-S may overweight those pricey homes if we are looking for evidence of housing distress affecting the banking industry.  Wealthy people who see their home values decline are less likely to default on mortgages.

More importantly, C-S overweights the 4 states at the epicenter of the subprime crisis; California, Nevada, Arizona, and Florida.  The overall C-S index of 20 cities declined by 5.9% between March and August 2008, and then by 9.8% over the most recent 5 months for which data is available (August to January 2009.)  But when I looked at individual cities I noticed a striking pattern.  Here are the 13 cities not in the 4 bubble states.  The first number is the percentage change in the five months up to August 2008 (which is roughly when I believe money starting becoming too tight), and the second number is the change over the next five months:

Atlanta:         -2.64%     -9.68%

Boston:          +0.28%    -5.20%

Charlotte:      -2.00%     -6.33%

Chicago:        -1.80%    -11.79%

Cleveland      +0.76%     -4.33%

Dallas            +0.55%     -5.71%

Denver           +0.70%     -5.09%

Detroit           -4.41%     -15.19%

Minneapolis   -2.70%     -14.33%

New York       -1.72%      -6.40%

Portland         -3.64%      -8.50%

Seattle            -3.70%     -9.80%

Washington    -6.55%     -9.79%

Average          -2.06%     -8.63%

Notice that the slowdown after August is even more striking in the non-bubble states.  And even this list underweights the center of the country.  With the obvious exception of Detroit, it looks to me like prices were fairly stable in Middle America during much of 2008, and only began falling sharply after August.  And judging by the NAR data, I expect another sharp drop in the C-S index when February is announced.

Now think about the implications of my “reverse causality” view.  I agreed that the initial recession was linked to the subprime bubble bursting (and also the oil shock.)  I also suggested that the steep Fed rate cuts in January helped keep NGDP growing in the first half of 2008.  But I argued that Fed policy was modestly too contractionary in the late summer of 2008, and then far too contractionary in the fall.  What would one expect from this pattern?

1.  The initial housing decline should have been concentrated in the key bubble states, although because there were some excesses everywhere, smaller declines might occur in some non-bubble states.

2.  After August 2008, the problem should no longer have been concentrated in the subprime states.  Monetary policy affects the entire country, not just one region.  Thus one would have expected an acceleration in the rate of decline house prices everywhere, but the change might have been most noticeable in the non-bubble states, where prices had stabilized somewhat after the January 2008 easing by the Fed.  Notice the decline in the 13 city average in the first period (-2.0%) is far below that of the 20 city average (-5.9%), so you can imagine how rapid the decline must have been in the seven bubble cities.  After August 2008, however, the 13 city decline is pretty close to the 20 city average.

And once again, I believe that if I had nationwide data this pattern would be even more dramatic.  The NAR data is certainly flawed, but it is picking up a striking change that the C-S data disguises.  The true figure is probably somewhere in between.

To summarize, we now have seven variables which showed striking shifts after August 2008.  In every case their behavior was what one would expect after a severe monetary contraction.  Admittedly five of those seven variables responded as they would to any adverse AD shock.  But two of the variables (rising real interest rates and a sharply falling euro) specifically pointed to tight money.

Why did people miss this pattern?  Perhaps it was because we are used to thinking in terms of nominal interest rates or economic aggregates.  But real interest rates are a much better indicator when inflation expectations are falling fast.  And monetary aggregates can be misleading when money demand is increasing.  Increases in money demand can have just as contractionary an impact as lower money supply—but it is much harder to recognize.  My very first post after the intro started off with a quote from David Hume in 1752.  Even back then, Hume recognized the danger of an increased demand for base money:

If the coin be locked up in chests, it is the same thing with regard to prices, as if it were annihilated

Because Fed policy was neither aggressive enough nor forward-looking enough, these problems intensified.  In the second half of 2008 and early 2009 a rapidly increasing part of the financial crisis represented reverse causation; falling NGDP causing more loan defaults, pushing an already weakened banking system over the edge.  Because we were so terrified that we might overshoot inflation by one or two percent, we now face trillions in extra budget deficits, a ballooning national debt, and talk of nationalizing the banking system.  The President is now the de facto CEO of GM.

A few weeks ago on 60 Minutes, Ben Bernanke finally admitted what went wrong.  When asked why each bailout seemed to be followed by more requests for government money, Bernanke responded that the economic situation had gotten much worse than when the initial estimates were made.  And why did that happen?

Update (4/2/09)  I just did the numbers for the seven bubble cities.  The rate of decline increased from 11.42% to 14.34%.  But this is misleading as the decline actually slowed in the huge Southern California markets (LA/San Diego) which comprise about half the total in terms of population.  I believe that a population weighted average would only show about a 1.7% acceleration.  If I am right that the rapid acceleration in the non-bubble cities represents the impact of falling NGDP (rather than simply bubble correction) then that force should have hit all markets, including the bubble states.  So what looks like a continuous decline in the bubble states hides a much more interesting pattern; a sharp correction in 2007 and early 2008 reflected a necessary market adjustment, followed by a further and unnecessarily steep decline in late 2008 and early 2009, which represented a failure of monetary policy.

When people ask me if I am just trying to reflate the bubble, I say that the initial housing correction was necessary, but that we are now overshooting.  Preventing this overshooting through aggressive monetary policy would not increase the “moral hazard” problem as long as we targeted 5% NGDP growth.  Those borrowers who survived the initial subprime crash but are now being pushed into default by falling NGDP, should not be lumped in with those who took out the really bad mortgages.  The more recent group of defaulters probably expected NGDP would keep growing (of course I mean subconciously, as in no sudden depression) and it is not their fault that we had a sudden and severe failure of monetary policy, a failure of the Fed to set policy at a level expected to hit their own announced targets.

[P.S.  I am still working on the 1932 post.  But I needed to get a new post up so newcomers don’t keep running into my silly piece on Krugman.  I welcome suggestions on this topic, as I expect to be on bloggingheads TV soon, and would like to discuss this issue.  If you are interested in researching the data, I suggest first looking at Thruth’s excellent comments toward the end of this post.]



39 Responses to “Evidence of Reverse Causality?”

  1. Gravatar of TGGP TGGP
    1. April 2009 at 19:44

    If people want to jump directly to the comments where Thruth linked to the data, they’re here:

  2. Gravatar of anon1 anon1
    2. April 2009 at 03:10

    For clarity, could you just briefly illustrate the difference between your reverse causality and the accepted wisdom causality? Thanks.

  3. Gravatar of ssumner ssumner
    2. April 2009 at 03:42

    Anon1, Good question. The conventional view is that causality runs from the bad subprime loans (and other Alt As and adjustable rates, etc) to high default rates, to rapidly falling house prices to banking crisis to economic recession. I think that view does have some merit for the first year of the crisis. But even though the last 6 or 8 months look like more of the same, I think the causation has switched. I see the latter period as monetary policy failures causing falling NGDP, causing falling asset prices (houses, stocks, commodities etc.) causing bank balance sheets to worsen much more than the initial subprime crisis. I call this “reverse causation” because the conventional view sees the banking crisis causing falling AD, and I reverse that causation. (Of course there was some of each.)
    My specific complaint about the Fed is that in the 3rd quarter they made a modest but forgivable error of not easing policy when expectations turned lower, but in the 4th quarter, right from early October, the policy errors were unforgivable in two ways. Expectations about NGDP had become so bearish that an aggressive policy was needed (zero rates and QE even in early October) and they committed a huge mistake in paying interest on reserves (a highly deflationary policy.)
    BTW, I added a update of two paragraphs just as you were in the process of commenting.

  4. Gravatar of Cogar Cogar
    2. April 2009 at 04:43

    Are you saying that if there’s an asset price bubble, we should inflate the rest of the economy (NGDP) to normalize its price?

    Sort of like unscrewing a lightbulb by holding it steady and turning the room.

  5. Gravatar of lxm lxm
    2. April 2009 at 05:30

    I have trouble understanding how a failure of monetary policy could cause or correct what was essentially massive fraud in the finance industry. See this link for an example: http://www.ritholtz.com/blog/2009/04/aig-before-credit-default-swaps-there-was-reinsurance/#comments

    Nor do I understand how a failure of monetary policy could cause or correct massive over capacity in industries such as the automobile industry.

  6. Gravatar of Alex Golubev Alex Golubev
    2. April 2009 at 05:33

    ARGH! Bad subprime loans? there are no bad loans, just like guns don’t shoot people. Subprime debacle was contained until July ’07 when the rating agencies dramatically increased forecast defaults on ARM loans. That cutoff a major portion of the shadow banking system (TIGHT MONEY) and created the initial shock. The banks also reacted to increased default rates across their book and started raising rates and tightening up underwriting (TIGHT MONEY). Tight credit in turns limits borrow ability to conduct business like it’s 1999… i mean 2005. There is a FEEDBACK mechanism on the downside just like there’s one on the upside. chicken or the egg? butterfly flaps it’s wings and if all the domino’s align correctly it will create a black hole. Fed’s by far not the biggest player controlling the tightness of money. The securitization market created a ton of liquidity outside of Fed’s control and when that disappeared one can hardly blame fed for not supporting the shadow banking system… only in hindsight.

  7. Gravatar of Carl Futia Carl Futia
    2. April 2009 at 05:41

    Did the Fed make a mistake in the fourth quarter? Ex post the answer seems obvious – it did. But I think it is less clear that its ex ante policy was defective.

    The difficulty with targeting expectations is that expectations are (much!) more volatile than the underlying economic aggregates. So the problem of separating signal from noise is correspondingly more difficult. It seems like you want to argue that the elasticity of policy with respect to observed expectations should have an absolute value of 1. But I doubt that this could be a property of any policy that is ex ante optimal in a world of uncertainty.

  8. Gravatar of Thruth Thruth
    2. April 2009 at 09:14

    Carl Futia,
    >The difficulty with targeting expectations is that
    >expectations are (much!) more volatile than the underlying
    >economic aggregates. So the problem of separating signal
    >from noise is correspondingly more difficult.

    I don’t know about you, but I see a lot of signal here:
    FRED Graph comparing 5 year bond, 5 year TIPS
    (focus on the difference, a market proxy of inflation expectations)

    (Scott: If this link doesn’t work, could you please repair it. Also, thanks for the nice comments.)

  9. Gravatar of Carl Futia Carl Futia
    2. April 2009 at 09:39


    Yes, after the fact the signal is clear. But I notice that the entire deflation “signal” lasted about 10 weeks from start to finish. And I recall that at the time there was debate among experts about the exact significance of this movement in the interest rate differential.

    How should the Fed have acted differently than it did in response to this signal ? (I think you can be confident that the board members and staff were well aware of this data at the time.) There was uncertainty then about what it meant – experts debated this question. At what point should the Fed have done more than it in fact was doing? How much more? When expectations turned inflationary again should it have begun unwinding its actions?

    My point is that it is a lot harder to predict the future than it is to predict the past. And any sensible policy would not be responsive to temporary changes in expectations. The difficulties lie in specifying IN ADVANCE the policy rule’s sensitivity to the targeted time series.

  10. Gravatar of Scott Sumner Scott Sumner
    2. April 2009 at 09:57

    TGGP, Thanks.

    Cogar, No!! Let me emphasize that I do not want the Fed to worry about asset bubbles, I want them to target NGDP as they were doing (very roughly) before the crisis. They only gave up this policy 6 months ago. If we had NGDP targeting the peaks of bubbles would probably we lower and the troughs higher. That would be a nice side effect. But even if I am wrong, at least the bubble wouldn’t wreck non-bubble parts of the economy like manufacturing. Again, I certainly oppose any attempt by the Fed to bail out housing–but let’s not make the housing bust even worse than it otherwise would be.

    lxm, The Fed cannot correct for massive fraud, nor should it try to. Nor should it correct for massive overcapacity in the auto industry. BTW, by overcapacity I hope you are not referring to the current gap between capacity and output, as that number is meaningless during a recession. My point is that we wouldn’t be in a recession at all, if the Fed had done it’s job.

    Alex, I am afraid I disagree. The Fed has a monopoly on the medium of account, the base. They can create hyperinflation or deflation anytime they wish, or anything in between (although obviously they may under or overshoot.) But they have all the power over nominal aggregates. Real aggregates? Only somewhat and only in the short run.

    Carl, It’s not just me, the distinguished Princeton economist Lars Svensson also says target the forecast. It makes no sense to set policy where you think it will fail.
    Not only are expectations not far more volatile than the underlying aggregates, they are far less volatile. Look at the consensus forecast or real growth and inflation. Then add them together to get the implied nominal growth forecast. That consensus is much more stable than the underlying NGDP series. Under my proposed NGDP futures contract targeting scheme there would be absolutely zero fluctuations in NGDP expectations–so the ratio of actual to expected fluctuations would approach infinity. The internal Fed forecast (which Svensson prefers), is also more stable than the underlying series. And the Fed even failed by that criteria–they let their internal forecast fall far below the policy objective.

  11. Gravatar of Thruth Thruth
    2. April 2009 at 11:10

    Useful link:
    Bailout timeline
    (I haven’t verified all of those dates and notice the money market guarantee program is missing)

    Carl Futia said
    >How should the Fed have acted differently than it did in
    >response to this signal ? (I think you can be confident that
    >the board members and staff were well aware of this data at
    >the time.) There was uncertainty then about what it meant –
    >experts debated this question.

    Actually, it’s pretty clear it was panic stations at Treasury and the Fed within a day or two of letting Lehman go. The TARP legislation was put together around Sep 20 with both Treasury and Fed endorsement. The money market guarantee was set up Sep 19. The FDIC insurance limits were increased in October.

    Perhaps the more pertinent question is why did Fed decide the best response was legislative rather than monetary? And why is qualitative/quantitative easing suddenly ok now but not then?

  12. Gravatar of ssumner ssumner
    2. April 2009 at 13:46

    Carl and Thruth, You both make good points, but I’d like to offer a slighter broader perspective to the time line question. Throughout my blog I’ve always said that monetary policy was probably too tight in August and September, but at least defensible. For several weeks after Lehman the consensus was that we were looking at a slowdown and a pretty bad financial crisis. But things changed really fast in early October. It seem like there was an absolute flood of really negative forward-looking news from all sorts of fronts. Most famously the stock market crash of the first 10 days (more than half the decline for the whole year.) But also, the severe commodity price crash. Very bad news about world trade, and manufacturing. Worsening financial crisis. Virtually every forecast was suddenly for deflation and severe recession. Because Bentley is a small school I went to Harvard a few weeks later and had lunch with Greg Mankiw and Larry Ball. I was all prepared to argue that the Fed didn’t realize how bad things were. They both said something like “you can be sure they do know, they just not sure what to do about it.” This caught me off guard, because I had assumed the Fed’s slow response on easing was a sign they didn’t think the drop in nominal spending was as bad as I thought. I wish I had been ready for that answer, I would have spent more time on ideas like a penalty rate on excess reserves. I think Carl’s point is defensible for a few weeks after Lehman. In my view the reason stocks didn’t immediately crash is investors thought (hoped?) the Fed would act decisively enough to prevent a severe recession. It crashed when they saw Paulson and Bernanke bogged down in the bailout debate, and no activity on the monetary policy front (where bank hoarding was increasing.) I feel very confident about applying my argument to October. September, less so. Sometimes my rhetoric gets a bit carried away because I feel so passionate about this issue. So my view is the the August/September policy was a bit too tight, but excusable. But October, November, December, January, February, March, and right now, not excusable.
    One small technical point. The interest rate spread is misleading because a definitional change around Dec. 1 suddenly meant an inversion was impossible. (New indexed bonds cannot lose value.) But the sudden rise in early October was real. It’s the quickest change in the several year forward outlook I’ve ever seen (similar to 1929 and 1937.) One reason deflation expectations fell this year (even with a proper index), is that the actual deflation was sharp and short last fall. Now we are looking at near zero inflation for several years.
    BTW, The blog is 2 months old today, and we just crossed the 1000 comment point. Thanks everyone for all the comments.

  13. Gravatar of Alex Golubev Alex Golubev
    2. April 2009 at 14:57

    I guess i misunderstood you in sayign that the Fed was responsible for this. If we’re gonna blame the Fed starting in October, that’s fine. I would hardly call them responsible though considering that this has been years in the making and required a lot of things to go wrong first (flood of really negative news from EVERYWHERE). I can see how you feel so pationate about it considering your specialization and it was quite a mistake on Fed’s part and seems very obvious and blunt to you. But a whole lot of professionals in a wide range of professions made misalculated decisions to get us to this point first. I was in the heart of it from the very beginning (subprime securitizations) and i had a front seat and watched with WAMU management fail in astonishing fashion starting in early ’06 along with rating agencies, national associatons of realtors, brokers, lenders. stock analysts and all kinds of experts on CNBC. Anyways. My point is that it’s a self-reinforcing process and you can’t hold the last guy responsible just as much as you can’t hold the first guy responsible. That’s too simplistic. If we want to talk about the “evildoers” that caused this, then we’re practicing a witchhunt ritual. It would me more productive to discuss the similarities in assymetrical incentive structures for all the decision makers and goods/evils of leverage. Private decision makers are either misinformed or get paid in call options. Politicians mostly care about getting reelected and not anything long range. it would particularly be dangerous to end the good times! That’s the cause. very unsatisfying cause it’s hard to point a finger at it.

  14. Gravatar of ssumner ssumner
    2. April 2009 at 17:41

    Alex, It comes down to what you mean by “responsible for this”

    1. If you mean the very mild recession that began in December 2007 and was still mild in mid-2008, then no I don’t think the Fed was responsible for this. If you mean the initial subprime crisis. No.

    2. If you mean by this the very, very severe recession that began in late 2008, then I do believe the Fed is responsible for this. And also responsible for making the banking crisis (which was pretty bad to begin with) even much worse.

    This is the point I have trouble conveying, because most people see one smooth continuous crisis. And I don’t blame them.

    Lots and lots of other people are trying to establish blame for the original subprime crisis, I am focused on monetary policy failures, which I think is an under-reported story.

  15. Gravatar of TGGP TGGP
    2. April 2009 at 19:20

    Steve Keen argues in his post The Roving Cavaliers of Credit that the Fed (like other central banks) does not have control, but rather that the money stock is endogenous and they are pushed into accommodating it after the fact. He also claims we don’t live under a fiat-money system.

  16. Gravatar of Bill Woolsey Bill Woolsey
    3. April 2009 at 02:19

    I have seen Keene’s article and I think it is wrong.

    It is true that if the central bank targets the interest rate, then, at first pass, the amount of base money it creates is “determined” by the interest rate target it chooses. By paying interest on reserves, we now have experience showing that a central bank can impact both, though not in a way that is relevant to monetary policy.

    Even if the central bank targetted base money and let all interest rates vary, then it would still be true that in dynamic equilibrium, you would expect both base money and deposits of various sorts to grow together. The “principles” level money multiplier analysis isn’t quite what should be expected–an increae in reserves and then large loans and then progressively smaller loans as the money multiplier works itself out.

    I would think that the natural expectation is that changes in policy should show government policy changing first and then the monetary aggregates following. However, with “lending” by holding short term securities being a possibility, the adjustment may not appear in even weekly data.

    Finally, the notion that bank money is greater than credit by government money is extremely odd. Not all lending comes from banks. Not all bank lending is funded by the creation of money.

    Frankly, I had trouble getting beyond that very basic error.

  17. Gravatar of Jon Jon
    3. April 2009 at 06:28

    Keen is wrong: his conclusions do not necessarily follow from the data. In particular, contrast his analysis to Hayek’s: “forced savings” drives GDP growth above the natural rate; sustained growth above trend requires an increasing rate of “forced savings” over time.

    That narrative matches Keen’s data exactly. As the boom progresses, the Fed must pump the base faster to sustain their growth target.

    He has this strange notion that in Fractional Reserve Banking “Total Debt is Less than the amount of money”. Which is true only if you phrase “total” as “net”, but in his charts he goofs. The debt numbers he cites are not “nets” but “totals”. So when he graphs the theoretical FR ratio as 0.9, he actually means “10” to make it apples-to-apples.

    … However, I think he is right that “credit” money is independent from “fiat” money. In particular, anyone can credit money. When I work and accept a paycheck weeks after my effort, I am creating money. When I pay with a “charge-card” rather than cash I am creating money. None of these transactions appear as such on a bank balance sheet, therefore they are not regulated by reserve ratios or capital ratios, etc, etc.

    Given that very few monetary regimes have relevant reserve-limits, the only real constraint on bank’s credit creation is the willingness of investors to recycle funds back into the bank in the form of capital.

    Nonetheless, the Fed can shock the system up and down.

  18. Gravatar of Alex Alex
    3. April 2009 at 06:36

    Alex G.

    This is what I think Scott means by “it’s the Feds fault”. One night you are watching TV in your favorite couch having a beer. Your wife walks in from work and she had a terrible day, on top of that she has PMS and her mother called her on the phone to nag about her father. Nothing that happened has to her do with you but she is a time bomb ready to go off and you are the only one that can defuse it. So what can you do, you can stay on you couch having your beer thinking it is better to let her cool off. If you do this things will go very bad for your household, and you wont get any action for a month. On the other hand, you can set her a hot bath with scented candles and offer her to cook dinner why she relaxes and then talk about her problems over dinner (who knows you might actually get lucky). Scott would say that if you follow the first strategy of doing nothing it is your fault if your wife goes ballistic on you even though you didn’t cause her bad mood in the first place but because you could have done something to revert it.


  19. Gravatar of ssumner ssumner
    3. April 2009 at 15:47

    Alex, I don’t like that analogy, it let’s the Fed off the hook too easily. Here’s one I like better. The bus moves along a twisting road. At each turn the driver turns the steering wheel, so that the bus remains on the road. Then at one turn the bus driver decides not to turn the wheel. Rather he decides to keep the position of the wheel constant. He does literally nothing. The bus goes into the ditch. What caused the accident? Is it the road’s fault for turning, or the drivers fault for not turning the steering wheel when he came to a bend in the road? Does one have a right to expect the bus driver to turn the wheel when necessary to keep the bus on the road? But how can it be the driver’s fault? He didn’t do anything.

    TGGP, I had trouble even getting through Keen’s article. Regarding money and endogeniety; if the interest rate is kept fixed, the money supply is endogenous, but it is also indeterminate in the long run (as is the price level.) If you want to control the price level you need to adjust the interest rate when inflation rises or falls. How do you adjust the interest rate? By adjusting the money supply. So like Bill and Jon, I am not convinced he has anything useful to say.

  20. Gravatar of Larry Larry
    3. April 2009 at 18:52

    Fascinating hypothesis. How does it fit with Hamilton’s notion here?

  21. Gravatar of ssumner ssumner
    4. April 2009 at 03:53

    Larry, I agree that high oil prices contributed modestly to the recession. In an earlier post, Hamilton argued that money should have been tighter in early 2008, so that the oil shock would have been smaller. I don’t understand that reasoning, as even tighter money would have meant an even more severe recession. I also don’t think that oil tells us much about why the economy has recently done so poorly. The very bad part of the recession only began last fall, and oil prices were much lower by then. To some extent the causation goes the other direction, oil prices reflect the condition of the economy, although in this case it is the world economy.

    Hamilton also needs to distinguish between the domestic and world economies. In the 1970s high oil prices were caused by less supply, this time around it was more demand. That makes it more likely that causation went from booming economies to high oil prices in 2007 and early 2008 (when the rest of the world was still booming.)

  22. Gravatar of Bill Woolsey Bill Woolsey
    4. April 2009 at 06:43

    With nominal income targeting, an increase in the price of oil would have no impact on the growth path of nominal income or total spending in the economy. Prices, however would rise, reducing real income. Rather than a reduction in total employment, however, there is a shift into producing export goods and import competing goods. (More exports go to pay for imported oil and less for other things.)

    To the degree that the production of compements for oil can shrink faster than exports and import competing goods can expand, then production and employment might shrink. But fundamentally, this is about structural unemployment, a shift in resource allocation.

    On the other hand Americans on average really end up with lower real incomes and can consume less.

    Of course, to the degree those earning the extra money from oil want to invest in the U.S., there may well be an expanded net capital inflow as well. This will dampen the impact on the value of the dollar.

    Anyway, that is the way I see it. To the degree monetary policy is aimed at preventing inflation or else maintaining the value of the dollar, the policy is, of course, inconsistent with nominal income targetting. And then, sales must be depressed thoughout the economy so that prices other than oil will fall (or rise slowly enough) to offset the increase in the price of oil.

    I think both of those policies would be mistaken.

  23. Gravatar of Jon Jon
    4. April 2009 at 11:34

    “In an earlier post, Hamilton argued that money should have been tighter in early 2008, so that the oil shock would have been smaller. I don’t understand that reasoning, as even tighter money would have meant an even more severe recession.”

    Critically exchange values depend on perceptions. Policy was perceived as being loose. Therefore the oil shock was worse than necessary. This does not mean that policy should have been tighter in 2008, nor does it mean that policy was not tight; rather, it reflects a diminishing faith in the Fed’s commitment to a stable dollar. Hardly a trivial concern.

    I disagree with the demand-shock theory of oil prices; I believe the reporting of this was politicized. Demand was on trend, and supply had been on trend to meet it; however, five major supply disruptions took place:

    1) Saudi Arabia was supposed to bring several new fields online. These fields missed their production start dates by about one year, but the Saudis trimmed back production from the Ghawar field on-schedule. (Ghawar is aging field that requires substantial amounts of water injection to maintain production, pumping any field too hard harms long-run extraction). When the Saudis were reported as having their highest ever production, they were scarcely producing above their levels from two years earlier. In essence, they had been at a 1 mb/day deficit.

    2) Nigerian production plunged following terrorist attacks there. This amounted to about 0.5 mb/day loss.

    3) Field mismanagement at PDVSA (Venezuela) due to the diversion of capital investment funds to certain social and farming programs.

    4) Field mismanagement at Pemex (Mexica) due to corruption

    5) Field mismanagement at Lukoil due to “corruption” (excessive taxation, royalty payments) diverting investment funds.

    All told, about 3mbpd were unexpectedly lost from the market. Production decisions that could have compensated for a loss of that scale need to be made several years in advance. Oil companies are pretty aggressive about not building too much production too soon.

  24. Gravatar of Alex Alex
    4. April 2009 at 11:55


    Off the hook too easily? You should have met my wife (now ex wife)…


  25. Gravatar of Larry Larry
    4. April 2009 at 13:34

    The oil price rise wasn’t signaled by futures prices – which I think counts as a shock – inducing consequences in the real economy. That shock has surely worked its way through the system by now. Maybe oil was the camel’s back-breaking straw…

    As an aside, during the Depression the world economy normally operated near subsistence levels in most places. Did that operate to put a floor on consumption? This time around, we are way beyond subsistence. That floor if it is one, leaves us a lot of room to fall. I.e., more of today’s average household consumption could be deferred without real hardship [insert obligatory big-screen tv joke here.] We’re a long way from such a point, but that thought offers no comfort.

  26. Gravatar of Jon Jon
    4. April 2009 at 20:01

    Unexpected shortfall in Saudi oil production:


  27. Gravatar of ssumner ssumner
    5. April 2009 at 05:01

    Bill, I agree, and in fact your analysis provides one of the best arguments for NGDP rules. Regarding Hamilton’s argument, NGDP rose about 3.3% rate in late 2007 and the first half of 2008. That’s certainly not too high, and may even be a bit low.

    Jon, Certainly you know more about oil supply than I do, but I have a few reservations;

    1. One minor point first, the Saudi graph in your second post doesn’t quite fit your argument. It does show a supply dip during the 2007 run-up in prices, but supply returns to the original level by the mid-2008 peak of the oil price.

    2. More importantly, the term ‘demand’ can be very misleading. In industry discussion it usually refers to quantity demanded, not the demand function. Normally that may be OK, but not when prices are running up from $50 to $147. Consider the following hypothetical. Demand from China and to a lesser extent other LDCs shifts sharply right in the world economic boom. Oil consumption also rises sharply in these countries. But oil supply is restricted for two reasons. One, the short run supply is pretty inelastic. Two, the supply disruptions that you mention (that part I agree with you.) So quantity supplied only goes up the usual 1% (say a million bpd) despite the much higher prices. But LDC demand rises much more than this. So oil prices rise sharply enough to reduce western oil demand enough to allow for the big increases in Chinese imports, and yet keep total world consumption rising by only about 1%. It is my argument that this hypothetical is pretty close to what happened, although I may be a bit off in some numbers. But this is a big shift in demand to the right, but only a small rise in quantity demanded as you slide up and to the left on the more stable demand curves for rich countries after oil prices soar.

    3. I also want to briefly comment on “perceptions,” and dollar weakness not being a “trivial concern” It is important to distinguish between actual market expectations and what’s being talked about in the business press. Right now a lot of people are talking about inflation risk down the road because of QE, but the markets don’t show this, indeed they show lower than normal inflation expectations. In mid-2008 inflation expectations did rise a bit, but if we had a NGDP futures market, I think it would not have signaled excessive monetary ease. The falling dollar may or may not be a problem. In early 2008 it was a solution, not a problem. It prevent steep recession in early 2008. Only when the dollar soared after July 2008 did we get a steep recession. Nevertheless, you “perceptions argument explains why I have never been too critical of Fed policy in the 3rd quarter of 2008, even though it was probably already too contractionary

    Alex, I’ll take your word for it on the issue of your former wife.

    Larry, as I said to John, the key is to look at NGDP. I don’t think the oil shock would raise or lower nominal GDP, for reasons outlined by Bill. But in that case I don’t see the mechanism for what you describe, this looks like a demand side recession, not a supply side recession (circa 1973).
    Subsistence is always relative, what is subsistence for the U.S. might be normal with a country of half our income (S. Korea), and what is subsistence for Korea might be normal in Mexico (one quarter of our real income.) And what is subsistence for Mexico might be normal for China, and what is subsistence for China might be normal for India, and what is subsistence for India might be normal for Bangladesh, and what is subsistence today in Bangladesh might have been normal 30 years ago in that country. There is no meaningful floor on consumption, but fortunately the Fed won’t let nominal GDP fall anywhere near Bengali levels.

  28. Gravatar of Jon Jon
    5. April 2009 at 07:04

    #1) Yes, I admit that my memory (original description) and the data did not line-up perfectly; however, I do not think it damning that oil prices peaked once Saudi oil returned to its normal trend. Indeed, that’s precisely what one expects in an inelastic situation. Nonetheless, your comments have merit. The initial rise up to $70/barrel is better explained by the output loss than the subsequent rise past that point

    #2) My concern with the notion of a demand shock is with the idea that the shift was unexpected. To be a shock, you need to have an unexpected event. But the unexpected event was the supply loss, not the surging demand.

    2b) The elasticity of supply is not a constant. In particular, there is usually a buffer of 1-2mbp of easy expanded production, and the supply curve is flat in that region–just like any other industry.

    3) That’s an interesting remark about perceptions. You are dead-on, but would you agree that during 2008, the markets were signally a perception of dollar weakness, not just the press?

  29. Gravatar of Larry Larry
    5. April 2009 at 09:17


    On oil, the original shock (per the other commenters) seems to have been on the supply side, exacerbated by continuing demand increases. We certainly have demand-side problems now…

    Agree that subsistence is relative, but my thought was that my hypothesized floor would simply be different for each country – as it was then. In each case, the distance between full employment and that floor is much greater now than then.

    I agree that the Fed will act. I’m conjecturing that what might have been a significant stabilizing force then doesn’t have as much effect now.

  30. Gravatar of ssumner ssumner
    5. April 2009 at 15:50

    Jon, The concept of a “shock” is something I overlooked. You are partly right. All I’d add is that places like China and India are somewhat opaque. Add in the huge implicit subsidies (during rising market prices) from administered prices in many LDCs like China, and I’d still attribute some of the surprise to the demand side.

    On perceptions, I agree that the dollar was perceived as being weak, it’s just that I think that is only a problem when it pushes inflation expectations too high. And that was only a modest problem in 2008 (and not at all after August.) I assume your reference to the dollar was to the spot rate. The market clearly never expected the dollar to depreciate much more than it already had against the euro, as the interest parity condition prevents that expectation from developing. Instead, dollar pessimism shows up one for one in the spot market. The $1.60/euro exchange rate was already ridiculously low, I don’t think it was expected to go lower (and it didn’t.)

    Larry, I did understand that you were making a pragmatic argument on living standards, but I just wanted to point out that one has to be careful. No pragmatic person would have foreseen even the drop in living standards of 1929-33 , but it happened. So I’m not putting much weight on that factor in today’s economy.

    I see the point you and Jon are making about oil, and in the very short run it has some validity. But let’s also be clear that world oil consumption rose steadily until recently, despite the huge price increases, so there were some significant demand shifts to the right, hidden by movements up and left along the demand curve. There is a sort of optical illusion issue here, which I discussed in an earlier post arguing that S&D is very tricky, and often not what it seems. So I encourage people to keep an open mind about demand shifts, even if it is not as visible as supply.

  31. Gravatar of Alex Golubev Alex Golubev
    6. April 2009 at 08:22

    Mild or severe? What if a minority lost everything (leveraged in housing) while another minority made millions (mortgage brokers, execs). While this is a mild slowdown, it’s hugely unfair. While systemic slowdown that hits unemployment in a major way, in my view is much less unfair considering the tradeoffs in a normal employment contract. I’m suggesting another parameter to judge what ought to be done about misallocation of assets. On top of that, my point is that distrubances that might seem minor can create extraordinary systemic risks. And my argument is that allowing short term incentives for agents in highly leveraged and thus systemically important institutions will lead to highly unfair payoffs and some of the time major crisis. On top of that, adressing those issues will prevent us from ever getting to major crisis. But i guess our arguments do come into play at different points of the crisis on the severity scale. (just not the fairness scale. which some might say IS pretty important for the survival of the system that is capitalism)

  32. Gravatar of ssumner ssumner
    6. April 2009 at 17:23

    Alex, I need more specifics. Where are the “short term incentives?” If they are there, let’s address them by getting to the root of the problem. Some people wrongly argue that stock investors have a short term focus (which is nonsensical if you look at biotech stocks) so I want to make sure your problem is real before I address it.

  33. Gravatar of Alex Golubev Alex Golubev
    7. April 2009 at 07:38

    Scott, Cash bonuses for decision makers based on volume of sales and/or accounting(aka imaginary) earnings over a period <= 1 year. A ton of long-run risky assets very created because they generated a high “gain on sale” for accounting/bonus puproses. (It just may be that stock investors have a short term horizon, BECAUSE incentives within companies are also so short term. But that isn’t my concern.)

  34. Gravatar of ssumner ssumner
    7. April 2009 at 17:33

    Alex, Stock investors don’t have a short time horizon, or else nobody would want biotech stocks. Some of these companies go more than 20 years without any profits at all.

    I doubt companies would offer those bonuses unless they were in the long run interest of the company, but obviously each case would need a lot of study.

  35. Gravatar of Alex Golubev Alex Golubev
    8. April 2009 at 06:35

    The second argument sounds like – I doubt there is $20 on the ground cause someone would have picked it up by now?! Companies don’t offer short term bonuses. The board of directors negotiates packages with executives. Once short time incentives are created for the top, they design bonus packages for their subordinates to align with their’s. And son on. The board of directors and executives consistently show evidence of (rightly) pursuing their own interests/incentives vs company. Poison pills would be one example. How can it make sense from an investors point of view to create an antidillution clause if anyone buys more than x% of the company. I know the management can spin a great story about protecting the investors from the evils of a takeover. My point is that I don’t doubt it. First argument – biotech stocks have intrinsic value that’s tied to the long run, but that doesnt’ mean that investors aren’t playing hot potato with the stocks. The float turnover for quite a lot of companies would suggest an average holding days in days and months. There’s plenty of turnover data fund turnover ratios out there (http://thefloat.typepad.com/files/kasten-on-portfolio-trading-costs-in-401k-plans-1.pdf). 90% in a year?! I don’t know of too many studies showing anything predictable about stock returns over a course of the year. My point is that maybe if the board of directors actually created long run incentives for the company, things would become mroe stable and holding periods longer. As it stands currently, the incentives just might actually be aligned with short time horizon investors and management. Unfortunately people holding the hot potato are now seen as victims and people that threw them the potato are all of a sudden evildoers. Don’t hate the player, hate the game. It’s all about the asymetrical incentives. I think the argument against mine is that short time incentives actually help in GROWTH and not STABILITY and i cannot argue that STABILITY is in any way more important than GROWTH. I think growth for the country is more important than stability for any unfortunate individual. So Denmark just may be a mroe balance place to live, but it’s obviously not as darwinian as the US.

  36. Gravatar of ssumner ssumner
    10. April 2009 at 15:28

    Alex, You make a good argument about poison pills, doesn’t this suggest there is a flaw in the market for corporate control. There are laws that make it harder to take over companies, such as one that requires disclosure once you buy more than 5% of stock in a corporation. This is not my area of expertise. But it seems to me that if the owners were given control of the companies they own, it would at least reduce the agency problem.

    But back to your original point. The villains, who were the bankers and others in related fields, lost huge amounts of money. So I don’t see this as primarily misaligned incentives, but rather bad decisions.

  37. Gravatar of TheMoneyIllusion » The League of Monetary Cranks TheMoneyIllusion » The League of Monetary Cranks
    25. April 2009 at 05:31

    […] information, only asset prices.  Does this sound familiar?  In two separate posts here and here, I collected graphs and tables showing the movement in six key asset prices in late […]

  38. Gravatar of Das Lehman-Märchen « Kantoos Economics Das Lehman-Märchen « Kantoos Economics
    28. November 2010 at 05:13

    […] Scott Sumner (Bentley), Große Depression (2 Links) […]

  39. Gravatar of TheMoneyIllusion » From TheMoneyIllusion to conventional wisdom TheMoneyIllusion » From TheMoneyIllusion to conventional wisdom
    15. September 2011 at 05:50

    […] 2009 I made a number of claims that seemed almost preposterous.  One of those was the claim that people had reversed causation; the housing crisis didn’t cause the recession, it was mostly a product of the recession.  […]

Leave a Reply