Did the Fed cause the crash? And what does ’cause’ mean?

A few months back I argued that “if policy A would have prevented event B, then not doing policy A caused event B.”  This is what happens when you try to talk about concepts like “causation” without having studied philosophy.  I still haven’t studied philosophy, but at least I have thought about the issue a bit more.

I knew I was in trouble when I tried to apply my concept of causation to the Kennedy assassination.  If the driver of Kennedy’s car had decided to swerve a 1/2 second before the fatal shot was fired in Daley Plaza, would Kennedy still be alive?  (And for those starry-eyed folks who worship politicians, would we have avoided the Vietnam War as well?)  So did this driver’s “policy” of driving straight cause the assassination (and perhaps the war?)  If I’m going to argue the Fed caused the crash of 2008 through errors of omission, I need to come up with a much better notion of causality.  Fortunately, I think I can.

One approach to causality is to discuss necessary and sufficient conditions.  But what does the Fed actually do?  After all, there is no generally established definition of the term ‘monetary policy.’  And it isn’t just me taking my usual counter-intuitive view here; even among mainstream economists there is great disagreement over the term.  Some economists view the fed funds rate as “monetary policy” and thus see a change in the fed funds rate is as change in monetary policy.  Others view a 2% inflation target as the Fed’s monetary policy.  In the former case the Fed didn’t tighten policy in late 2008, in the latter case they did.  Yet without agreement on whether the Fed actually “did anything” or not, we can’t address the necessary and sufficient condition criteria.

Then there is the problem of “root causes” and “proximate causes.”  A government action or inaction may be a proximate cause, whereas there may be deeper political forces that cause those government decisions.  I finally ended up going back to the Rortian argument that “that which has no practical implications, has no philosophical implications.”  And I decided that with respect to monetary policy, the most logical “practical implications” would be policy implications.

More specifically, we need to ask whether there is some alternative policy that could have avoided the crash of 2008, and that also would have been a plausible policy choice at the time.  Thus a debate over “causes” is actually a disguised debate over future policy.  When Friedman and Schwartz argued that the Fed caused the Great Depression, they were implicitly arguing for something like a 4% monetary growth rule.  On the other hand no one would seriously argue for having a presidential car swerve every so often.  But they might argue for having presidents travel in non-convertible, bullet proof cars.

If we are going to look at causality from a “policy implications perspective,” then I see two distinct cases that need to be distinguished.  In one case we might decide that a problem was caused by a policy error that was easily avoidable given what people knew at the time.  A slightly different notion of causality would occur when policymakers did not have enough information to prevent a given problem, but we have learned enough from their error to devise future policies that could prevent similar problems in the future.  Unfortunately, many people (including me) have great difficulty distinguishing between these two cases.  I.e. someone might say “it was obviously stupid to drive the president around in an open car in 1963.”  But was it, or is it only obviously stupid in retrospect?  Why didn’t the president and his “best and brightest” staff see this obvious stupidity?

Niall Ferguson also seems to believe people are far too likely to claim that serious policy errors should have been obvious at the time:

Human beings are as good at devising ex post facto explanations for big disasters as they are bad at anticipating those disasters. It is indeed impressive how rapidly the economists who failed to predict this crisis “” or predicted the wrong crisis (a dollar crash) “” have been able to produce such a satisfying story about its origins.

This is a point I have been trying to make for months, I only wish I could have made it so succinctly.  The crisis was not obvious until it was obvious.  That is, until it exploded into the headlines.  That was when economists realized we had a problem.

This does not mean that we can’t learn lessons that would prevent a repeat, lessons that might not have been obvious before the crash.  But it does mean that we should be careful about blaming one group or another.  So I see two useful types of causality.  A causality that implies moral culpability; and one that merely implies there are lessons to be learned.  At times I have suggested that macroeconomists as a whole have a moral culpability for the crisis—the monetary policy literature (indeed even our textbooks) provides all the tools required for a policy regime that could have easily avoided the crash.  But that view seems difficult to maintain when examining the picture from a more detached perspective.  If almost everyone is to blame, doesn’t that mean almost no one is to blame?

[BTW, with any philosophical maxim you always want to do the “Nazi test.”  If it is any good, the Nazi example should seem to violate the maxim—be the exception that proves the rule.  Sure enough, most people do blame the German people as a whole for supporting Hitler.  So there is the one exception to my argument that if everyone believed something they can’t all be morally culpable.  I developed this test when I noticed that anytime someone brought up the Nazi example in an argument, it invariably led to the wrong conclusion.]

Why does establishing moral culpability have practical value?  Because people don’t like to be blamed.  So if our leaders know they will be held morally accountable, they will try harder to do the right thing.  Now let’s assume that I was wrong to imply that the Fed should have known better, that I was wrong to find the Fed morally culpable for their errors.  After all, most private economists made the same mistake.  Does that mean that all I am doing is the equivalent of blaming Kennedy’s driver for not swerving?  No, there is still a useful causality argument if we can devise a way of preventing future crises of this type, without requiring the Fed to predict better than the public.  Using the Kennedy analogy I am calling for a bullet-proof car—make that a bomb-proof car.

Even though the Fed did not foresee the sub-prime crisis, a forward-looking monetary policy would have done two things:

1.  It would have prevented the sub-prime crisis from reducing NGDP growth expectations.

2.  By stabilizing NGDP growth expectations, it would have prevented the sub-prime crisis from paralyzing the banking system in late 2008.

You guys can tell me whether this “policy implications” approach to causality is useful.  Let me end with three observations on what I see as flawed views of causality.  Then I will soon follow with another post putting meat on the bones–explaining exactly why all economists should view monetary policy as the cause of the 2008 crash.

1.  The butterfly effect: I am not interested in what would have happened if the Kennedy car had swerved, or if the German government had not released Hitler from prison, or if bankers had not made so many foolish sub-prime loans.  None of those counterfactuals have policy implications.  Regulators are not going to be able to identify and prohibit foolish loans.  (They might be able to produce useful regulations in other areas, such as capital requirements.)  So don’t tell me about some sort of deep root cause.  Indeed in my Great Depression study I was suspicious of any “root cause” that investors overlooked.  If investors didn’t see it, how could we expect policymakers to see it?  I now have to rewrite my manuscript (at the request of the editors) but I think I will stay with my earlier instinct.  Causal factors are factors that move asset prices.

2.  Moral culpability: I already indicated that this is one type of causality.  But some people seem to think that arguing policy caused a disaster is ipso facto a moral indictment.  I think the Fed did cause the crash, but I no longer believe members of the FOMC should be condemned.  In several earlier posts I discussed research findings by Joshua Knobe which suggested that peoples’ moral views get entangled in views about causation and intentionality.  People are more likely to see causation and intentionality where harm is being done, even if the case is logically no different from an act than does not do harm.

3.  Active vs. Passive: I originally entitled my paper on the crash “Errors of Omission:  How the Fed Caused the Crash of 2008.”  (By the way, any obscure journal out there that wants to publish a contrary take on the crisis?)  But that’s a terrible title, as it suggests the Fed was less morally culpable than if there had been errors of commission.  And because people link causality and culpability, they then dismiss my argument that monetary policy was the cause, or was “to blame,” using a term that conflates causality and culpability.  In fact, there is no such thing as an active or passive policy stance—every monetary policy stance is active.  Why, because every monetary policy stance will move at least some policy indicators.  And since people tend to define passivity as stable policy indicators (i, MB, M2, inflation expectations, etc.), there are no passive monetary policies.

Each period the Fed sets a policy instrument at a given setting.  Whether that setting is more or less than in the previous period is irrelevant.  All that matters is whether it is the right or wrong setting.  If it is the wrong setting (based on information publicly available at the time), then any resulting problems were caused by that policy error.



17 Responses to “Did the Fed cause the crash? And what does ’cause’ mean?”

  1. Gravatar of Phil P Phil P
    19. May 2009 at 06:53

    Scott, I’ve been thinking about this question of cause myself, stimulated in part by an earlier exchange I had with you in the comments. For some reason, the concept of cause seems particularly problematic in economics. Let me wax philosophical for a second. Perhaps ultimate reality is nothing more than a system of simultaneous equations, where everything “causes” everything else. Then once you’ve determined the correct equations, the concept of cause is simply redundant. However, for us human beings, the primary mode of thought and communication is language, which is a linear sequential medium. Translating the reality of simultaneous equations into the language of causation is like trying to convey a three dimensional reality on a two dimensional surface (or maybe a one dimensional surface!).

    And yet, in many areas the concept of cause is not at all problematic. When we ask what caused, say, an airplane crash, there’s no real ambiguity. But the reason for that is we have there a concept of baseline normality, one where most pilots are competent, most airplanes are mechanically sound, and random events other than pilot error or mechanical failure are rare. What is that baseline normality in economics? Maybe there isn’t one. For that you would need more agreement among economists about a common model for the economy, and for monetary policy than evidently exists. (This ties in with what you say about disagreement among economists about what constitutes monetary policy).

    Personally I don’t think the concept of culpability is useful, as that is a moral and legal rather than a scientific concept. Let me suggest one way of conceptualizing cause in relation to business cycle contractions. I believe economists sometimes think of contractions as being caused by “shocks” (unless the impulse-propagation view is out of date). The magnitude of a contraction could then be the product of the magnitude of the “shocks” and the adequacy of the policy response. I have no opinion about Fed monetary policy in 2007-2008, which is beyond my competence. It does seem that in traditional terms, the Fed acted aggressively in response to the credit crisis, and certainly they thought they were acting aggressively. But if you are correct in your analysis, then their response was inadequate in relation to the magnitude of the shocks, because they were not using the correct model.

    I want to emphasize that I’m not trying to debate economic theory but simply suggest a language for discussing questions of causation in relation to business cycles. I think my suggestion is in the spirit of Friedman and Schwarz. What they were doing in the MH was developing a model of the relation between money and output in which individual business cycles, or individual contractions were observations in the statistical sense. They believed there was a relationship between money supply decline and the the severity of contractions. They blamed the Fed for allowing the money supply to decline in the GD, and implicitly, for following an incorrect model of monetary policy. You are saying something like this with respect to the current recession, only employing a different monetary model.

  2. Gravatar of Alex Golubev Alex Golubev
    19. May 2009 at 07:50

    I do like this version of your theory a lot more. I absolutely agree that there wasn’t a foreseeable “deep root” cause, but I think the role of leverage needs to be discussed. Shadow banking is not inherently evil and anyone that suggests such is misapplying the benefit of hindsight. Leverage isn’t evil either. Neither are guns. Guns don’t kill people, bankers do (or so I hear nowadays although i promise i was following direct orders as i testified in my Killinger trials). Hitler was dismissed as a joke by the French (in real time), so I’m the last person to suggest that WWII or most messes are ever foreseeable). However, we DO have a democratic system precisely because of the fallibility of any potential regulator no matter how brilliant. So I really don’t understand why ANY entity would be allowed to have leverage to the degree where systemic risk becomes an issue. We’re not allowed to drive commercial vehicles without a special license. We have speed limits for all cars for the same reason. So putting incredible leverage in the hands of any single entity IS an equivalent of APOINTING an entity to collect pennies at the risk of bringing down the whole system. It’s an equivalent of a financial meth lab. How much leverage is right? I don’t know. Are out of the money options and futures used just as irresponsibly – yes. Can we grow as fast without leverage – absolutely not. I’m here to identify the CORRECT issue. A little time and a few fresh brains will hopefully think up a slightly better system.

  3. Gravatar of David Ortmeyer David Ortmeyer
    19. May 2009 at 11:28

    There is another concept of “causality” that at least deserves a mention and that is time-ordering. If movements in “A” always predictably precede movements in “B” and if movements in “B” do not predictably precede movements in “A”, you would say that “A” causes “B”. You would not say that “B” causes “A”. Now this isn’t necessarily true causality because of the omitted variable problem, but it is suggestive of more than just association.

  4. Gravatar of Leigh Caldwell Leigh Caldwell
    19. May 2009 at 14:43

    Interesting point (and also Phil P’s comment in particular). Although I just gave Niall Ferguson’s book a slightly nasty review, I do agree with him (and Scott) on this point: we make up stories to suit our moral outlook.

    I’m not sure if I’m overstating the case – some economic narratives are probably defensible – but here’s my article on the matter: http://www.knowingandmaking.com/2009/05/seven-basic-plots-and-narrative-of.html

  5. Gravatar of Devin Finbarr Devin Finbarr
    19. May 2009 at 16:21

    I still think that using the word “cause” to describe a sin of omission is an abuse of words. You can call the Fed’s failure a “shocking dereliction of duty”. You can say the Fed allowed the crisis. But it’s simply a misuse of language to say the Fed caused the crisis by not acting. No common definition of the word cause involves inaction: http://www.google.com/search?rlz=1C1GGLS_enUS326US326&sourceid=chrome&ie=UTF-8&q=define:cause Using the word “cause” continues to confuse me greatly, because I’m never sure if you are talking about something the Fed did or failed to do. General rule of English, if a word is so confusing that you have to write a blog post defending its usage, pick another word and move on.

  6. Gravatar of ssumner ssumner
    19. May 2009 at 17:59

    Phil P, I’m afraid “ultimate reality” is above my pay grade. I agree that humans don’t really know what the term ‘causality’ means, and we are sort of groping in the dark about the basic ontological questions. That’s why I want a pragmatic, useful definition. And it sounds like you do as well.

    “In traditional terms” the Fed was also very aggressive in 1929-33, but completely ineffective. F&S tried to create a new “tradition” a tradition that said don’t just look at interest rates and the base, look at M1, M2. I am saying don’t just look at money and interest rates, look at NGDP expectations.

    Alex, We may want to regulate leverage; but once again I don’t think our current crisis was caused by bad banking, it was caused by bad central banking.

    David, Yes, I forgot to mention Granger causality. My sense is that the variables that predictably precede other variables usually don’t cause them, rather they predict them. Thus stocks predict GDP growth. Maybe there are some cases where Granger causality is useful, but I can’t think of any in economics. But this is not my area of expertise.

    devin, My argument is that every Fed policy stance is equally “active.” The Fed can’t sit on the sidelines and do nothing. It doesn’t have that option. You may disagree with that view, but you can’t argue I am blaming the Fed for doing nothing. I am blaming them for adopting the wrong monetary policy stance. If they don’t change the money supply at all, then interest rates will change. That is a policy decision. That is an action.

    Lots of economists blame the Fed for the Great Depression. Almost none blame the Fed for the crash of 2008. Why the difference? The Fed was no more “active” in 1929-33.

    BTW, did you notice that I criticized the title of my earlier paper (“Errors of Omission”?

  7. Gravatar of ssumner ssumner
    19. May 2009 at 18:28

    Leigh, That’s a very interesting post. I do think that people use narratives to understand what is going on with the economy, and I see that as the heart of the problem. When I hear others talk about the crisis I think “yes that sounds like common sense; we got too greedy, what goes up must come down,you can’t push on a string, etc.” But then I recall that monetary theory is like nothing in our everyday life, it can’t be understood by these narratives.

    Early on I did a post on how Puritan attitudes got in our way (“It can’t be as simple as printing more money”)

    I have to run now, but I will give your post more thought. Ultimately we need a psychological theory of this recession. Macro needs a shrink. Or a literary theorist.

  8. Gravatar of David Ortmeyer David Ortmeyer
    20. May 2009 at 02:06

    A plausible scenario is that the sub-prime crisis and financial meltdown, collapsing housing and stock markets caused velocity to fall and the Fed failed to counteract. If an exogenous monetary policy shock had knocked us off our NGDP growth path, I would say that the Fed “caused” the recession. But Fed policy reacted endogenously (and inadequately) to the sub-prime mess. They were culpable and they worsened it, but didn’t “cause” it. (Here’s where a time-ordering notion of causality might be useful).

  9. Gravatar of ssumner ssumner
    20. May 2009 at 04:44

    David, As soon as you mention velocity you are already implicitly assuming that “monetary policy shocks” are changes in the money supply. But most economists don’t agree. Most would say if the Fed raised its fed funds target from 2% to 8% and the money supply was stable, that that would still be a very tight money policy. A monetary shock. My point is that the term ‘monetary shock’ is almost impossible to define. The concept of shock is completely arbitrary. I say a change in expected NGDP growth is the most USEFUL definition of a monetary shock. Others say it is a change in short term interest rates. In both cases we are talking about variables that the Fed does not directly control, but rather influences through changes in its various “tools.” My view may seem weird, but it is because people are looking at it the wrong way. Keynesian economists are doing the exact same thing when they define monetary shocks as sudden changes in the FF rate.

  10. Gravatar of Alex Golubev Alex Golubev
    20. May 2009 at 06:05

    In a 0 leverage world, the central bank would be out of a job. Disagree? Was the velocity of money collapse not a consequence of a reluctancy to lend that occured seemingly overnight in September-Nov (TED and Commercial paper spreads skyrocketed in September). That’s the liquidity that disappeared practically overnight and spread through leverage (OPM). You suggest more of a blanket approach with NGDP targetting, which might very well work even though it would punish savers and help borrowers, which isn’t necessarily fair, but I think we SHOULD be a little more surgical. Of course my approach would probably mean reducing leverage even further, which would kill the velocity of money and prolong the current recession, but i think it would be more FAIR. I guess i’m more of a 1931 type of guy. I’ll take 10 years of pain, if it teaches folks a lesson for 70 years.

  11. Gravatar of 123 123
    20. May 2009 at 13:55

    Scott, Fed’s actions and statements during the run on Bear Stearns say that if a too big to fail financial institution fails, monetary policy becomes less effective or more risky. This means that Fed made two errors – it let Lehman fail, and it did not respond appropriately after that. You are recognizing only the second error. Are you saying that NGDP targeting means that no institution is too big to fail?

  12. Gravatar of ssumner ssumner
    21. May 2009 at 03:39

    Alex, A Fed could exist in a world without leverage, or without banks, or without any financial system. Just cash and goods.

    No, the drop in lending should not have caused a collapse in velocity. That was caused when interest payments by the Fed led banks to hoard reserves. With a slight negative interest rate on reserves, banks would not have hoarded them, they would have bought T-securities.

    1931 was a very bad year; no long term benefit (which doesn’t exist in my view) was worth the cost. 1931 led directly to the rise of the Nazi’s in Germany, as just one example. And what lesson was learned by the Depression? During and after the war we went right back into inflation. The Depression also led to bad statist policies. Any legacy of the Depression was negative, not positive. It would have been much better if 1920s-style policies continued indefinitely, as they could have.

    Because of the Fisher effect, NGDP targeting doesn’t help borrowers and hurt savers. The 5% expected NGDP growth rate is factored into nominal rates, leaving real rates unchanged.

    123, Yes, ending too big too fail would be a huge advantage of NGDP targeting. If we could go back and do it all over again, it would have been better to have let LTCM fail, to reduce the reckless behavior of other large financial institutions. Ditto for Bear Stearns. Plus if we had to have a financial crash, it would have been better to have it in 1998, or early 2008, rather than late 2008 when the economy was much weaker. Oddly, however, I think that given what had come before, they probably made a mistake in letting Lehman go under. I know they really didn’t have any good options, but given that we’d already created moral hazard through earlier bailouts, letting Lehman fail simply confused investors even more.

    But with NGDP futures targeting, financial crises would have relative little effect on the real economy (at most shaving 1% off RGDP growth, and adding 1% to inflation.)

    Good macro policies lead to good micro policies, and vice versa.

  13. Gravatar of 123 123
    22. May 2009 at 13:18

    So if you have enough helicopters with cash you can solve the moral hazard problem. Actually this was the opinion of Fed and treasury when they let Lehman fail.

  14. Gravatar of ssumner ssumner
    22. May 2009 at 16:35

    123, You don’t need a helicopter. The public really doesn’t want to hold that much cash, and the banks only hold it because you pay them to. Charge a penalty rate on ERs, and promise credible inflation or NGDP targeting, and I doubt you’d need to increase the base by more than 10-20%. To repeat, however, I said given what had come before, Lehman was a mistake. (I was wrong at the time too.) I still think I was right about the first two cases. It would have been way better to have this crisis when the economy was not teetering on the edge. BTW, many people don’t realize that 9/11 happened at a really bad time as well. If it had happened in 1995 or 2005, it is very possible that it wouldn’t have caused a recession. The economy actually started growing soon after 9/11. Like Katrina, it was just a blip in the data. I still think Lehman itself wasn’t so bad, but rather the way the Fed handled it. The market didn’t crash on Lehman news, it crashed on news of some pretty inept Fed policy responses. And I’m not just referring to the issues John Taylor complains about, but also the excess reserve policy. But given the response, it would have been better to save Lehman, I agree.

  15. Gravatar of 123 123
    23. May 2009 at 01:25

    small note – credit markets crashed right after Lehman. Taylor uses wrong market indicators (Libor is a lagging indicator)

  16. Gravatar of ssumner ssumner
    23. May 2009 at 05:12

    123, OK, but that makes me feel better about my view, which differs from Taylor. Stocks crashed in early October, so that’s probably when the public lost confidence that the Fed could keep NGDP rising. I think it’s very revealing that stocks didn’t crash during the post-Lehman credit crunch that you mention.

  17. Gravatar of TheMoneyIllusion » In praise of backseat drivers TheMoneyIllusion » In praise of backseat drivers
    24. May 2009 at 18:36

    […] but what about the question of causality, did the captain cause the accident?  In my earlier post (here) I argued that it was hard to assign moral culpability if most other experts would have done the […]

Leave a Reply