Monday, May 12, 2014

What Caused the Great Recession: Household Deleveraging or the Zero Lower Bound?

Today I got into a discussion with Amir Sufi on Twitter about what really caused the Great Recession. Was it the vast amount of household deleveraging or the economy being constrained by Zero Lower Bound (ZLB)? Amir Sufi and his coauthor Atif Mian have a new book where they make the argument the key catalyst was household deleveraging.

For example, in a new article they argue the 2001 recession was far milder than the 2007-2009 recession because the related the stock market crash affected mostly rich individuals who had very little debt. On the other hand, during the Great Recession it was a housing market that collapsed and this affected middle and lower-class individuals who were highly indebted. This key difference in debt, they argue, is why the economy contracted so much more in 2007-2009. 

While it is true there was far more U.S. household debt leading up to the Great Recession and that cross country evidence shows that countries with more debt were generally hit harder during the crisis, I think they are confusing a symptom with the cause. In my view, the underlying cause was interest-rate targeting central banks running up against the ZLB. (Yes, there are ways around it for a determined central bank but most did not fully explore these options.) The failure by central banks to get around the ZLB caused most of the household deleveraging, not the other way around. Monetary policy, in other words, was too tight during the crisis.

I have embedded an annotated version of our twitter discussion below the fold:


  1. "for every debtor there is creditor who could provide offsetting spending if the interest rates adjusted down to their natural rate level"

    That's just not right, it seems to me.

    In the housing market at least, for every debtor there is a bank.

    When interest rates decline, banks don't spend more. They don't purchase more newly produced goods and services, the stuff of GDP.

    And as we've seen, when there's lack of quality loan demand (backed by strong income), even with low interest rates they don't lend more either (which would presumably cause more spending).

    This is probably the biggest problem I have understanding and accepting the whole (market) monetarist paradigm.

    The incentives and reaction functions of real-economy, non-financial-corp "lenders" (households and nonfin biz depositors, financial investors, etc.) are completely orthogonal to financial-sector lenders. Nothing like symmetrical as in your (and Krugman/Eggerston's patient/impatient) construct.

    1. Steve, if interest rates were able to fall to their negative natural rate level capital projects become profitable (i.e. positive NPVs) for firms. Consequently, firms sitting on cash would start doing more capital spending and hire more workers. Likewise, households holding lots of liquid assets would start to diversify into riskier assets and durable goods as the rates fell. These creditors would spur an improvement in the economy that would catalyze banks to do more lending.

      In other words, lowering interest rates to their natural rate level would spur inside money creation. The actions of creditors would be an important part of this process.

    2. David:

      That explanation assumes that borrowing/lending and resulting spending are *only* sensitive to interest rates. That all the economic effects are resolved perfectly into interest rates. That, for instance, creditworthiness (based largely on projected incomes), and banks' assessments of creditworthiness, are fully reflected in the interest rate.

      But if the last five years has taught us anything, it's that that's not true. That simplistic clockwork notion of how economies work just doesn't play out in many situations. More an article of faith than an empirically demonstrable reality.

      I very much agree on the portfolio-shifting effects of changing interest rates, but that's starting to talk about a much less direct and more questionable vehicle for policy transmission, one that has a complex and uncertain relationship to real spending.

    3. Steve, it is not just interest rates but the expected path of interest rates relative to their natural rate level. In other words, there is an entire term structure of natural interest rates that can be compared to the observed term structure of interest rates.

      If anything, I think the past five years confirm the view that having the interest rate path above the natural rate level will lead to anemic economic growth and all the other associated problem, like the ones you mention above.

      With that said, I do see the expected path of the interest rate - natural interest rate gap as only a summary statistic that leaves out important details. And this interest rate gap is difficult to estimate. So in practice we have to rely on a host of other indicators.

    4. Steve, for a couple years from 2009 onward, I helped a large real estate fund acquire a couple thousand homes. The group I helped now is managing 5K headed toward 20K single family homes.

      We basically provided the price floor for the housing market.

      In our markets, every auction, every new listing, had the same 10 firms bidding to pay cash. Once the properties are stabilized the cost of capital to lever the assets is much cheaper than a traditional mortgage.

      In some markets I believe this play represents 50%+ of existing home sales.

      My point is that whats really been going on is the creation of a new asset class like commercial, multi-family, storage, but this time for "middle class" Single Family Homes.

      If in 10 years, 3M+ homes in the US are under management this way I wouldn't be surprised.

      So there's "loan" demand, just not consumer driven.

    5. So when those consortia of buyers pick up the borrowing (and resulting lending, hence money creation) slack from households, does that result in more spending on newly created goods and services the way household borrowing does? Je ne sais pas.

  2. The St. Louis Fed's quarterly newsletter The Regional Economist has an article on the liquidity trap and another on inflation & Fed credibility. Both, unfortunately, lack any citation of market monetarist work.

    1. That is fine. As long as we are pushing the discussion on ideas in the right direction--and hopefully making the world a better place--then I am content. And to be fair, not everything we say is original. Maybe we are just helping folks 'remember' good monetary economics.

  3. "While it is true there was far more U.S. household debt leading up to the Great Recession and that cross country evidence shows that countries with more debt were generally hit harder during the crisis, I think they are confusing a symptom with the cause."

    Actually the premise of this statement isn't even true.

    I got into a conversation with Sufi and Mian at their recent post on the correlation between the increase in Chicago mortgage debt by zipcode between 2002 to 2006 and the decline in auto purchases between 2006 and 2009:

    And that premise was in fact the nub of our argument. What follows are the comments I presented.

    What if we repeated the [Chicago] exercise by looking at international data involving countries with different monetary policies? Would household sector leverage growth in 2002-2006 be correlated with a decline in real household sector consumption expenditures in 2006-2009?

    The best source for such data is the OECD. Household sector leverage can be calculated as the ratio of household sector loan liabilities to household sector Gross Adjusted Disposable Income (GADI) which is equivalent to BEA Disposable Personal Income (DPI). Household Sector Final Consumption Expenditures (FCE) is equivalent to BEA Personal Consumption Expenditures (PCE). To convert it to real consumption it can be adjusted by the Household Sector FCE Deflator.

    Unfortunately this data is only available for ten OECD members with independent monetary policies. The OECD also has this data for 12 out of the 18 Euro Area members (all but Cyprus, Estonia, Latvia, Luxembourg, Malta and Slovenia), so I calculated a weighted averaged for those 12 countries for the Euro Area average.

    Here is the household sector leverage data in percent.

    1.Czech Rep-15.8–30.7–14.9
    2.Euro Area-66.1–77.9–11.8

    Here is the change in leverage with the change in real Household FCE from 2006-2009.

    1.Czech Rep-14.9—7.4
    2.Euro Area-11.8—0.9

    When one regresses the change in real Household FCE on the change in leverage ratio the R-squared value is 0.0211 meaning that only 2.1% of the variation Household real FCE change can be explained by the change in the leverage ratio. In other words there is almost zero correlation when one considers geographic regions which have differing monetary policies.

    The only countries which had above average leverage growth in 2002-2006 combined with slower than average real consumption growth in 2006-2009 are in fact the US, the UK and Hungary. Norway and Sweden had above average leverage growth growth in 2002-2006 and above average real consumption growth in 2006-2009, and the Euro Area and Japan had below average leverage growth in 2002-2006 and below average real consumption growth in 2006-2009.

    1. Mark, where can I get the individual Euro household data you used above to calculate the weighted average?

  4. Sufi and Mian responded by citing four studies to support their claims. Perhaps what they didn't realize was that two of those studies had inspired my above comment. Here is what I said in response.

    Levels of private debt can explain the geographical distribution of the effects of debt-deflation recessions if one controls for monetary policy. But Irving Fisher, the author of the debt-deflation theory of depressions, assigned a pivotal role to monetary policy in causing such recessions and in ameliorating, or preventing them.

    First let’s take a look at some of the supposed empirical evidence presented against this claim.

    1) Jorda, Schularick and Taylor (2012)
    JST examine 67 cases of systemic financial recessions. Of these, 50 occurred under a gold standard regime. Ten occurred in the most recent recession with six of the countries affected either members of the Euro Area or pegged to the euro (Denmark). (The only exceptions are Sweden, Switzerland, the UK, the US.) Of the remaining seven systemic financial recessions, all but Australia (1989) and Japan (1997) involved a fixed exchange rate regime (and it’s also now known that Australia’s recession was largely allowed to occur, in order to achieve significant disinflation). Thus only six out of the 67 systemic financial recessions JST look at involved flexible exchange rate regimes, and consequently occurred under circumstances in which monetary policy was totally free to prevent or respond to the recession.

    2) IMF: “Dealing With Household Debt” (2012)
    Figure 3.2 was one of the two things that inspired my first comment. (The other was Figure 4 from Glick and Lansing.) There are 36 countries in the regression and the time period almost perfectly matches what I did above, except that the IMF look at the change in real consumption through 2010. But 14 of these countries were in the Euro Area, and 4 more were pegged to the euro. Thus 18, or half of the observations, involved countries that had the exact same monetary policy.

    One of the whole points of my exercise was to see if the correlation still existed if you replaced the euro members with a single observation. It clearly does not.

    3) Glick and Lansing (2010)
    Figure 4 was the other thing that inspired my exercise. There are 16 countries of which 9 are Euro Area members and one (Denmark) is pegged to the Euro.

    The story is of course the same. Remove the countries in the Euro Area, or pegged to the euro, and replace them with a single Euro Area observation, and the correlation totally vanishes.

    4) Mervyn King (1994)
    Figure 7 has ten countries of which six, or over half, were part of the European Exchange Rate Mechanism (ERM). Remove those countries and the correlation totally vanishes (and the number of observations becomes ridiculously small).

    An interesting thing about this particular example is that the decline in consumption is calculated through 1992, which happens to be the very year that three of the countries (Norway, Sweden and the UK) abandoned the ERM and significantly devalued their currencies. This was in fact what turned those economies around.

    In Figure 8 on the following page King repeats the exercise with regional UK data, in a fashion very similar to the Chicago diagram. Pointedly, King states that the reason why he used data from the same country was because it “helps to control for differences in national fiscal and monetary policy shocks.”


    1. (conclusion)

      In the conclusion, while paying Irving Fisher homage as the author of the debt-deflation theory of depressions, King notes:

      “I have argued that debt-deflation should be seen as a real business cycle [RBC] rather than a monetary phenomenon.”

      This is certainly not what Irving Fisher himself thought, who states on page 346 of “The Debt-Deflation Theory of Great Depressions”:

      “On the other hand, if the foregoing analysis is correct, it is always economically possible to stop or prevent such a depression simply by reflating the price level up to the average level at which outstanding debts were contracted by existing debtors and assumed by existing creditors, and then maintaining that level unchanged.

      That the price level is controllable is not only claimed by monetary theorists but has recently been evidenced by two great events: (1) Sweden has now for nearly two years maintained a stable price level, practically always within 2 per cent of the chosen par and usually within 1 per cent. Note Chart IV. (2) The fact that immediate reversal of deflation is easily achieved by the use, or even the prospect of use, of appropriate instrumentalities has just been demonstrated by President Roosevelt. Note Charts VII and VIII.”

      It’s a tragic shame that we’re still debating Irving Fisher’s original point 81 years later.

    2. Mark, those are great comments. You need to write those up as a note.

      I looked up one cross-country study and found Australia's incredible performance discussed nowhere. Did any of the ones you covered address this it?

    3. David,
      Glick and Lansing (2010) and IMF: “Dealing With Household Debt” (2012) include Australia in their respective data sets but its effect is massively outweighed by the Euro Area members, who of course all have the same monetary policy.

      I could not find GADI and/or Household FCE for Australia, Canada, Israel, New Zealand or Taiwan at the OECD otherwise I would have included those countries. Data for Croatia, Macedonia and Romania is probably available at Eurostat but I have not got around to it yet. In my opinion including any or all of these countries would only further challenge Sufi and Mian's claim about the cross country evidence.

  5. Amir Sufi:
    "but be careful to mistake cause for effect. we hit ZLB because indebted households massively cut back on spending."

    And why did indebted households cut back spending?

    Sufi obviously believes that the initial shock came from debt. I emphatically believe this is false. The initial shock came from monetary policy.

    The US yield curve became inverted in August 2006 and stayed that way through May 2007:

    Every US recession since WW II has been preceded by an inverted yield curve in the previous 6-18 months. An inverted yield curve is strictly a matter of monetary policy choice. In this case the fed funds rate had been raised steadily in quarter point increments over the course of two years until short term rates were above long term rates.

    Year on year nominal GDP growth fell from 6.5% in 2006Q1 to 5.3% in 2006Q3 to 4.3% in 2007Q1 to 3.1% in 2008Q1 to 2.7% in 2008Q2:

    Falling NGDP growth literally means falling nominal income growth. It's not surprising that more indebted households cut back spending more than less indebted households, but that doesn't explain why nominal incomes fell in the first place. The cause is monetary policy.

    Debt-deflation is a channel of monetary policy. The Unanticipated Price Level Channel of the Monetary Transmission Mechanism (MTM) is discussed in nearly every intermediate textbook on monetary economics. Debt-deflation is one explanation of how monetary policy shocks get propagated. But it does not claim that debt is the source of the shock.

    The debt-deflation theory of depressions was born in the Great Depression. And it is now conventional wisdom, thanks to Fisher, Friedman, Romer, Eichengreen and many others, that the initial prick that popped that bubble was monetary policy.

    Bubbles don't pop themselves.

    1. Bubbles do pop themselves once customers run out and there is nobody else to service. The cash flow halts, the debt built up becomes exposed and financial companies can't revolve the debt anymore. This leads to a crash in all leverage, as it becomes suspect to crash, even outside where the bubble built up. Monetary policy really can't stop this nor does it matter in its creation. Central banks could have in 2003-5, talked about bubble doom, jacked up the fed rate sky high quickly and knocked out the last years of the bubble, but all it would have done is expose the leverage, create a recession quicker and still bare the underlying structural problems. Monetary policy is not the answer.

      Then you have the foundational economy, or the guts of it. The United States has seen a structural slowdown there as well with the end of Peak Boomer spending, the end of the computerization/information innovation wave of the latter 20th century and the rise of globalism which gave markets a huge one time wealth transfer in the 90's.

    2. Anonymous, did you not read the part about creditors/savers providing an offset?

  6. You could argue there wasn't even a recession in the early 2000's. A couple "micro recessions", one in 2001 and another in late 02/early 03.

    The demand post-2000 was completely consumption driven, which means that party will end at some point. That point ended. The "0 bound" has nothing to do with it.

  7. David,
    Mian and Sufi have erased the entire conversation with me that took place on their blog save for my initial comment.

    Disappointing to say the least.

  8. Curvehospitality provides extensive range of Casegoods,Seating,Lighting,Artwork,Bathroom fixtures,Electronics,Drapery and Flooring.It saving valuable space.
    Best Western electronics

  9. Thanks.I love reading through your blog, I wanted to leave a little comment to support you and wish you a good continuation about financial & non financial advice.

    Independent Financial Adviser Weston-Super-Mare & Pension Adviser Bristol

  10. All recessions since the Great-Depression have the same cause (a short-fall in the money stock). I.e., all these recessions were huge mistakes made by the Fed. It is so simple it wouldn't tax a 6th grader.

  11. The cause & effect was extraordinarily simple. Money was exceptionally tight. Bankrupt U Bernanke drained legal reserves (based on transaction type deposit accounts 30 days prior), for 29 consecutive months (using the 24 month rate-of-change in the proxy for inflation). This turned previously “safe assets” into “impaired assets” on financial institutions balance sheets (principally within the investment banks). This policy change coincided with the turn in the Case-Shiller home price indices in Feb 2006.

    But monetary policy blunder (the short-fall in the money stock), was exacerbated by a reversal in financial innovation (velocity). Definitions of the money supply are not timeless. The extension of the scope and practices of the GSEs along with their Federal Government guarantees - assured a secondary market for housing - the "store of purchasing power" attribute of money (e.g., Fannie Mae & Freddie Mac ensured a secondary mortgage (resale) market for residential mortgages).

    I.e., there are various types of "tertiary money". These assets possess general liquidity. They do not bear a direct, unit for unit, unvarying relationship, to the primary money supply. Legislation which established a new classification of tertiary money made these Government guarantees inflationary (by underwriting & insuring with the credit of the U.S. Government).

    Thus financial engineering had the initial impact of transforming non-negotiable illiquid assets to negotiable (tradable), higher yielding, lower risk rated, i.e., highly liquid asset-backed securities. In effect, this vastly increased liquidity (vastly increased the transactions velocity of money via refinancing & reselling). I.e., our means of payment money began to approximate M3 instead of M1. Then as liquidity dried up, this moneyness property reverted back.

    But the coup de grĂ¢ce for the Great-Recession was the trajectory for short-term monetary flows in Dec 2007 (our means-of-payment money times its transactions rate-of-turnover). The 10 month rate-of-change in the proxy for real-output (using MVt or aggregate monetary purchasing power), pointed to an outright economic contraction beginning in July 2008 & culminating in an economic free fall during the 4th qtr of 2008 thru the 1st qtr of 2009.

    But in the 4th qtr of 2008, instead of easing monetary policy, Bankrupt U Bernanke tightened again. In Oct 2008, the Fed introduced the payment of interest on excess reserve balances which destroyed non-bank lending/investing, etc., etc.

  12. The Fed reacted to the economic downswing in the 4th qtr of 2008 very timidly. The FRB-NY's "trading desk" didn't purchase any significant volume of SOMA securities until April of 2009 (when stocks & gDp bottomed):

    Securities held outright - H.4.1 (Factors affecting Reserve Balances)

    07/2/2008 -- $478,838.00
    08/6/2008 -- $479,291.00
    09/3/2008 -- $479,701.00
    10/1/2008 -- $488,541.00
    11/5/2008 -- $490,027.00
    12/3/2008 -- $488,445.00
    01/7/2009 -- $495,383.00
    02/4/2009 -- $511,440.00
    03/4/2009 -- $581,721.00
    04/1/2009 -- $773,497.00

    I.e., Bankrupt U Bernanke executed a qualitative or "credit easing" program (sterilized balance sheet expansion via emergency credit facilities involving the substitution of risk quality & change in the gov’t vs. private-sector composition of the Central Bank’s assets), instead of quantitative easing (an outright expansion of the money stock).

  13. Unless the money stock increases at least at the rate prices are being pushed up, output can't be sold, & thus jobs will be lost. Thus the Fed's inflation mandate is a positive 2 plus percent.