Friday, 20 July 2012

Sector Financial Balances as a Diagnostic Check


                Martin Wolf has a nice post explaining the financial crisis using sector financial balances. He rightly attributes this way of looking at things to Wynn Godley. It goes way back – I remember using them as a cross-check on forecasts in the UK Treasury in the 1970s, but it was probably Godley’s influence that helped that happen too. They are not a substitute for thinking about macroeconomic behaviour, but they can often be a very useful check on whether your thoughts (or forecasts) make sense.
                Take the example of a balanced budget temporary increase in government spending, which I used recently as a challenge to heterodox economists to come up with an alternative analysis that did not use either representative agents or rational expectations. (I’ve had plenty of responses telling me of all the defects of rational expectations, but no one has as yet given me an alternative account of the impact of this particular policy measure. As Godley is respected among heterodox economists, I thought maybe retelling my analysis using sector balances might help.)
The policy itself does not directly change the government sector’s financial balance (by definition). Theory tells us that consumers will smooth the impact of temporarily higher taxes, so their sector will move into deficit. But if we were foolish enough to think the story stopped there, thinking about financial balances tells us that has to be wrong. Consumers are in deficit, and no sector has moved into surplus. Keynesian theory then tells us what happens to put things right: output and incomes increase until the point that the consumer sector is no longer in deficit. If you think about it (and given consumption would always fall by less than post-tax income because of smoothing), this has to be the point at which income has increased by an amount equal to the tax increase i.e. a balanced budget multiplier of one. We could talk about this as a dynamic multiplier process, or we could talk about rational consumers working this out, and so not bothering to reduce their consumption in the first place.
As Martin Wolf and others have pointed out many times, thinking about financial balances also tells us the foolishness of cutting government deficits when the private sector has moved into surplus to restore their asset/liability position. In a global economy, if governments are successful in cutting deficits then the private sector surplus has to diminish. That makes the idea that nothing will happen to output as a result of deficit reduction rather improbable. With interest rates stuck at zero, real interest rates cannot move to persuade the private sector that they no longer need to correct their financial position. So the only possibility left is that output falls until they no longer want to do so. (Because of consumption smoothing higher short term income would imply a rising, not falling, private sector surplus.)
Looking at sector balances are not a substitute for thinking about behaviour, but they can and should demand that we are able to tell stories about them that make sense. Where I think criticism of the mainstream macroeconomic profession is correct is that there were not enough people telling convincing stories about why the household sector balance was evolving the way it did over the two decades before the recession. (I talk more about this here.) What was I doing? The answer is writing papers looking at the impact of fiscal policy in DSGE models, and not looking at this kind of data at all. In that sense I was definitely part of the problem, although it did kind of come in useful later on.

22 comments:

  1. Why is it that "looking at sector balances are not a substitute for thinking about behaviour"? Would anybody really go to see an economist if they had a question about human behavior? Economists have always had a problem explaining even individual's 'economic' behavior. Isn't that why 'utility theory' (dubious at best) came about? Maybe macro-economists should limit themselves to sectoral analysis. False precision can be worse than an accurate but vague description.

    ReplyDelete
  2. "They are not a substitute for thinking about macroeconomic behaviour, but..."

    This is surely a question-begging statement. As ought to be much better known, Godley's own stock flow consistent analysis enabled him, in 2001, to identify precisely those imbalances in the US economy which would cause the crash. (See here: http://www.lrb.co.uk/v22/n13/wynne-godley/what-if-they-start-saving-again)

    By your own admission, DSGE analysis turned out to have no predictive ability at all. Whereas stock flow consistent analysis turns out to be pretty well equipped to explain how an economy will evolve. So why waste time on the former, when the latter is so much more useful?

    ReplyDelete
    Replies
    1. As anyone doing DSGE analysis will tell you, it is always stock/flow consistent. If DSGE has a bias, it is in assuming that consumers know what they are doing. It is all about theories of behaviour, and in particular what can account for (and therefore what will happen to) sector imbalances.

      Delete
    2. Oliver Rivers21 July 2012 06:10

      Isn't that just because they make the assumption that markets clear? SFC models as developed by Godley (and others) don't make any such assumption.

      I'd be interested to know what you think about my other point: an approach to analysis which simply analyses the data through the lens of quadruple-entry turned out to be a much better guide to the actual behaviour of the economy than DSGE. So which is the better approach to economics?

      Delete
    3. Oliver,

      Godley's book is languishing unread in my to-read pile, so I might have got this completely wrong, but from what I've seen of it, I think the point it is not a substitute for thinking about behaviour is valid. From what I've seen, the "thinking about macroeconomic behaviour" boiled down to Wynne looking at the stocks and flows, and thinking "this can't go on - this is where trouble is going to come from", that is to say, Wynne used his judgement and did not formally model behaviour. Is that fair?

      If I've understood it correctly, his approach is more like an informative accounting exercise that provides a useful picture of the economy to guide and constrain your thinking, but if you were to write down any given description of any given economy, there would be any number of ways it could evolve in the future, depending on how agents behaviour from there on. That is to say, all the problems macro economists grapple with, trying to model behaviour, are side-stepped by Wynne. Different people with the same data (stock flow consistent picture of the economy) might come to different conclusions about the future path of the economy and, crucially, the impact of various policy options, without there really being any way to distinguish between competing interpretations/predictions.

      I wouldn't be surprised if I've got this wrong: look forward to being corrected if so.

      [fwiw, my impression is that Wynne's approach is powerful and deserves to be taken seriously]

      Delete
    4. Luis,

      Ch.3 would give you a brief overview of the kind of model that Godley used. There is also an EViews programme file for the model (model SIM) knocking about the internet somewhere.

      There are plenty of behavioural equations, but no microfoundations. You will be familiar with the sort of thing from undergrad macro. Consumption as a fixed share of disposable income, etc, etc. There is also the use of what the authors call "stock-flow norms" to characterise the steady-states of the models.

      Delete
    5. thanks Vimothy, I should have guessed there would be some behavioural equations. Personally I think you can get a long way with simplistic assumptions like that, but they are not immune from criticism.

      Delete
    6. Very much agree.

      Delete
  3. "As anyone doing DSGE analysis will tell you, it is always stock/flow consistent. "

    I'd like to know how this is true. Can you please refer to a paper where this can be seen. I'd like to see it - especially on how money is treated if there is stock-flow consistency.

    ReplyDelete
    Replies
    1. Listen buddy, why don't you send us a paper showing a DSGE model that is not stock-flow consistent. You are making the claim, and the burden of proof is on you.

      Delete
    2. i would guess that most dsge modelers do not not know what stock flow consistency means

      Delete
    3. You didn't have to take it the wrong way Pontus. Just looking for some big DSGE model which really has a counterpart for every asset or liability.

      Delete
    4. Anonymous - You're right. In DSGE we simply call it "consistency". There are no leakages. A dollar spent is a dollar income for someone else. A dollar lent is a dollar borrowed, and so on. I'm amazed that post-keynesians just discovered that the equations must add up! What were you guys doing before? Stock-flow inconsistent models?

      Ramanan - Sorry, didn't intend to. Feel free to take a look at any DSGE model published in a respectable journal. They will all have the stock-flow consistency. But beware, they won't put up nice Bill Mitchell excel sheets. But the equations will hold, rest assured.

      Delete
    5. Pontus, you got the burden of proof backwards. The original claim is that DSGE models are stock-flow consistent. Ramanan is asking for the proof of that. The burden of proof is on the claimant, not the person who asks for proof.

      Delete
    6. "A dollar lent is a dollar borrowed, and so on."

      I'm guessing most DSGE papers get the above process wrong, as most mainstream economists don't understand modern banking. I don't know if this violates "consistency," but it does violate reality. This is maybe what Ramanan is referring to.

      Delete
    7. Anonymous - Wow, so when I lend you/a bank/the government a dollar, you/the bank/the government has not borrowed a dollar?

      Min - Allright. Here's an example http://www.econ.nyu.edu/user/gertlerm/gertlerkiyotaki102409paper.pdf

      Please let me know if you find any inconsistencies.

      Delete
    8. "Wow, so when I lend you/a bank/the government a dollar, you/the bank/the government has not borrowed a dollar?"

      Well here's one answer that stands your rhetorical question on its head:-

      http://neweconomicperspectives.org/2013/03/the-i-o-u-in-the-u-s-dollar.html#more-5001


      Delete
  4. Lars P Syll writes: "Tom, I agree with what you say in this post. And as you and other readers on this blog knows I have been pretty tough about his microfoundationalism. But I think we should also give Simon due credit for doing something that almost no one else of his colleagues ever do: admitting he was wrong. So when he writes "What was I doing? The answer is writing papers looking at the impact of fiscal policy in DSGE models, and not looking at this kind of data at all. In that sense I was definitely part of the problem" it shows a kind of intellectual honesty that is a rare animal in mainstream neoclassical establishment today!"
    http://mikenormaneconomics.blogspot.se/2012/07/sector-financial-balances-as-diagnostic.html

    ReplyDelete
  5. Martin Wolf sent me the following comment, which I am sure others will also find interesting:

    "I used sectoral financial balances before the crisis, following Wynne. I argued that what was going on in the US external and household sectors were evidently unsustainable. This allowed me to argue that when the latter's deficits were eliminated, there would be a recession and a huge fiscal deficit. What I had not expected was that the turnaround in the household sector would trigger a meltdown of the financial system.

    "This makes it clear that one has to link the flow sectoral balances to the balance sheets in the economy. In this case, my mistake was not looking closely enough at the balance sheet of the financial sector. Good macroeconomic analysis has to examine the flows and stock meticulously and seek to assess whether the behaviour we see is sustainable. The assumption that private agents cannot make huge mistakes about the sustainability of what they are doing is, in my view, the biggest mistake in macroeconomics."

    ReplyDelete
    Replies
    1. There is indeed a strong bias in today's macro against modelling situations in which agents make systematic mistakes. There is some work, but even analysis of bubbles tends to involve a good deal of rationality.

      But I also think the methodology of DSGE modelling, with its focus on microfoundations rather than matching data, meant that too few macroeconomists had to confront the data on balances or savings rates, and therefore come up with explanations. This is the point I tried to make in this post: http://mainlymacro.blogspot.co.uk/2012/04/microfoundations-and-evidence-1-street.html

      I think Paul Krugman and others have written in a similar way to you: they argued at the time that household deficits (or associated house price movements) were not sustainable, but failed to see the impact this would have on the financial sector. However if you look at data on trends in leverage, the vulnerability of finance is (in retrospect) pretty clear. I think an interesting question is why (most of) those who were paid to monitor such things did not raise the alarm. Was it, as I speculated at the end of this post http://mainlymacro.blogspot.co.uk/2012/07/crisis-what-crisis-arrogance-and-self.html, because of 'political' pressures, or was this too a belief that this sector could not systematically undervalue risk? Were the regulators subject to capture, or to the famous Keynes dictum about academic scribblers, where the scribbler in this case was the rational expectations revolution?

      Delete
    2. "But I also think the methodology of DSGE modelling, with its focus on microfoundations rather than matching data, meant that too few macroeconomists had to confront the data on balances or savings rates, and therefore come up with explanations."

      It is a big deal that you're conceding this (and apologies for being a tad rebarbative in my first response). Have you read Godley/Lavoie's Monetary Economics? From a micro point of view some of the behavioural assumptions in the SFC models they develop in that book are crude. In particular, their consumption function ignores diminishing marginal utility, and there isn't much attempt to think about consumption smoothing. Both of those are remediable faults.

      The only logical constraint imposed in the G/L type SFC models is that of the accounting identities; all funds flow from somewhere to somewhere. There is no further assumption made about equilibrium (on the contrary, the assumption is always that there will be buffers in an economy, of cash, of inventories, and hence that markets don't clear).

      The models are very well suited to answering questions about the macro impact of behavioural changes (e.g. changes in households' portfolio allocation preferences, or firms' investment rates, or banks' capital adequacy ratios). Isn't that what macro is supposed to do?

      Delete
  6. "This is just the information I am finding everywhere. Thanks for your blog, I just subscribe your blog. This is a nice blog. " market timing

    ReplyDelete