Winner of the New Statesman SPERI Prize in Political Economy 2016

Wednesday, 18 February 2015

Endogenous supply and depressed demand

I noted in my last post that without fiscal austerity the US, UK and Eurozone could currently be at output levels that are above current estimates of potential or natural output. (For the US a chart is here.) In other words the output gap would be positive rather than negative. One response to that is to say without any fiscal austerity monetary policy would have raised rates. But are these estimates of potential output really independent of the path of actual output?

In a stylised view of macroeconomics the two are independent. Productive potential calculates how much you could produce if both labour and capital was fully employed using technology that itself is independent of current and past levels of output. You can say the same thing in a more kosher way by talking about natural output involving individuals working the amount they wish given real wages that reflect market clearing etc.

We know that stylised view is wrong for a variety of reasons. Labour that has been unemployed may become deskilled. Firms that are forced to cut back on investment in a recession may take time to rebuild their productive capacity. However there may be other ways it is wrong for reasons that are much more difficult to quantity. In particular, if investment falls in a recession, new technology that has to be embodied in new machines may fail to emerge, so the rate of technological progress may appear to decline.

These processes may not matter too much in normal (mild and short lived) booms and busts. However following a large recession they may become more important. As many have noted (e.g. Larry Ball here), estimates of the growth rate of productive potential made by organisations like the OECD and IMF have been revised down substantially since the Great Recession. The bigger the recession, the larger the fall in potential. As I noted here, to rationalise this as a gradual supply side reduction in the rate of technical progress (i.e. to avoid assuming technological regress), these organisations have had to also revise their view of pre-recession output gaps, to give what are frankly ludicrous numbers. It seems much more probable that estimates of productive potential are strongly influenced by actual levels of output.

If true, this is in one sense very optimistic. The process could be reversible. We could expand the economy by much more than most estimates of the output gap suggest, and estimates of productive potential would to some extent rise too. As I have said many times, given this possibility (and the huge costs of underestimating what potential is) we really should explore it by keeping policy as expansionary as possible until movements in inflation clearly tell us we have gone as far as we can. But this raises another puzzle. If we have the capacity to produce much more, why is demand so weak, when interest rates remain at the Zero Lower Bound (ZLB)? [1]

It is possible to construct sophisticated models of multiple equilibria where beliefs (animal spirits if you like) can shift us between equilibria. Roger Farmer is the most notable example of someone who has explored this possibility (see also David Andolfatto recently). Here I just want to make a simple observation about why we should take such possibilities seriously. The largest component of aggregate demand is consumption, and consumption depends on expected income, which can depend itself on actual output, and therefore on aggregate demand. The macroeconomy is therefore set up to allow self-fulfilling multiple equilibria.

That possibility is plausibly bounded on the up side. Goods have to be produced with capital and labour, and at some point workers at least will start demanding higher wages to work longer, generating inflation, which monetary policy reacts to by reducing demand. But the mechanisms that stop self-fulfilling beliefs on the down side are more problematic. Monetary policy finds it difficult to stimulate demand if interest rates hit the ZLB, particularly in a world of inflation targets. At the ZLB continuously falling inflation would become a liability (pushing up real interest rates), but thankfully inflation may become very sticky near zero. The unemployed may drift out of the labour force, or may become self-employed and produce much less, or real wages may fall such that firms start adopting more labour intensive techniques. For all these reasons, deficient demand may become persistent, to some extent disguised and not obviously self-correcting.

Just think about what has happened in the years following the Great Recession. Central banks and governments have steadily revised down their views of what the long run level of output is. It is hardly surprising in these circumstances, and with real wages stagnant or falling, that consumers would also revise down estimates of their long run income, and adjust consumption accordingly. In that way, demand appears to match a pessimistic view about long run supply.

You should not ask how sure I am about such stories, but how certain you are that they are wrong. If you are not certain, then the moral is the same: after a severe recession which appears to result in a loss of capacity, you use policy to explore the boundaries of just how much capacity has really been lost, and run the risk that inflation may rise as you do so. You do not sit back, tell yourself that below target inflation is probably temporary, and do nothing. And, of course, you do not plan for more fiscal austerity.  


[1] Some central banks, most recently the Bank of England, appear to be revising what they think the actual lower bound is. According to Britmouse, that means I should no longer talk about either the ZLB or fiscal policy. Or as the skipper of the Titanic might have said, that iceberg really shouldn’t have been there.


25 comments:

  1. Britmouse's analysis of Carney's comments is so bizarre, that I assume he must be being sarcastic.

    ReplyDelete
    Replies
    1. Indeed, I did wonder. But remember the cornerstone of market monetarism - whatever you do, try and avoid talking about fiscal policy.

      Delete
    2. Given that we cannot go back in time, it doesn't much matter what the BOE believed at the time. The question is what would you suggest the authorities do the next time we are at 0.5% rates and there is insufficient demand. I see 3 alternatives:
      1) cut rates all the way down to -0.5%
      2) do QE
      3) do fiscal stimulus

      I would argue that they should definitely do the first given that we have seen other central banks do that successfully, Do you argue differently? If so why?

      I cannot see what is the difference between cutting rates from 1.5% to 0.5% and cutting them from 0.5% to -0.5%. Unless you argue we should do fiscal policy even when rates are at 1.5% or higher...

      Delete
    3. Whatever Britmouse may mean, he does not say SWL should no longer talk about either the ZLB or fiscal policy. He simply says it is no longer necessary. That i no more than a matter of opinion (and he has a lot going for it).

      Delete
    4. This is the thing about market monetarists - they always pretend that people like me never talk about monetary policy. Except we do, all the time. But even if the Bank now thinks the ZLB is 0.25% or even zero, it seems an unnecessary risk to undertake further fiscal austerity. Simply changing the position of the ZLB does not make it go away.

      Delete
    5. On the contrary Anonymous! Britmouse says, "No more talk of building schools to create jobs.... no need to talk about fiscal policy" Are you Britmouse being naughty?!

      Delete
  2. These are core UK CPI prints:

    2008 1.1%
    2009 2.8%
    2010 2.9%
    2011 3.0%
    2012 2.4%
    2013 1.7%
    2014 1.3%

    The average core CPI between 1997-2007 (I haven't got data before that) was around 1.2%. So we did have above target inflation for a few years by 1-1.5%, we did try to stimulate demand until inflation started showing up. Maybe we should have been willing to pay a higher inflation cost. But those are the numbers. UK is not the EU.

    I agree many of the fiscal choices were bad (like cutting investment and transfers to the poorest). But I think you are overselling your argument. The policy mix could have been a bit better, but I don't think it's reasonable to assume it would have been as successful...

    If you are arguing about the EU, then I agree fully, but that was a more clear monetary policy mistake (they hiked rates, so it's hard to argue they wouldn't have done more)...

    You really need to argue that even higher inflation would have been better, but I really think that needs to argue for a different monetary framework... I think the BOE did as much as was reasonable given the target.

    The government has the ability to change the target so it's hardly with no responsibility anyway. They still choose how much inflation to experience in the end. They clearly choose to raise inflation too little in your mind, but I think that's a clearer way to state their mistake...

    ReplyDelete
    Replies
    1. This post was general, but lets talk about the UK. I basically agree with what you say. Given the inflation target (set by government), the MPC did a lot with QE (even though inflation was above target part of that time), and schemes like Funding for Lending were also helpful. I would have been very critical if they had raised rates in 2011, but to their credit they did not. I think the monetary framework could have been changed (dual mandate, nominal GDP target), but it was not in the MPC's remit to do this.

      I am less happy about policy over the last year or so, but that is not really what your comment is about.

      Delete
  3. Simon, to what extent do you think countries like China and India are "picking up the slack" when actual GDP in the West falls below potential GDP? Naturally China and India cannot pick up demand for localized services (a British teacher or lawyer won't move to China or India that easily), but they can import natural resources and they can increase demand for technologies and global services when that demand falls in the West.

    I thought about this in the context of global oil consumption. This fell in 2009 but has since been rising with ~1% per year (after an initial strong growth in 2010 that made up for the fall in 2009).

    As a second question. Part of this "take over" at the expense of the West must have been inevitable (the West undoubtedly would have lost some of its competitive edge even without the great recession, with pre-recession southern Europe perhaps only having had the appearance of competitiveness), do you think this has been properly accounted for in the models for potential Western GDP and actual-Western-GDP-if-there-had-been-no-austerity? If so, what are the major factors/assumptions of these models that convinced you of that?

    ReplyDelete
  4. I remember attending a presentation on the so-called "productivity puzzle" by some BoE person in 2012/13, talking about the huge observed decline in labour productivity during the crisis. De-skilling and cuts to investment, as you mention, was definitely a part of that, but there was still a large drop left unexplained ... haven't heard much about it lately, or paid much attention to how productivity's come on since then. What are your thoughts on it?

    ReplyDelete
  5. I have to admit that you calling it the "ZLB" does seem a bit odd now that the Swedish Riksbank has set a negative rate of interest. I'm not denying that that there is a lower bound, it just doesn't appear to be zero. Perhaps you could follow the example of Lars E.O Svensson
    in this paper and call it the Effective Lower Bound (ELB)?

    ReplyDelete
  6. "You should not ask how sure I am about such stories, but how certain you are that they are wrong. If you are not certain, then the moral is the same: after a severe recession which appears to result in a loss of capacity, you use policy to explore the boundaries of just how much capacity has really been lost, and run the risk that inflation may rise as you do so. You do not sit back, tell yourself that below target inflation is probably temporary, and do nothing. And, of course, you do not plan for more fiscal austerity. "

    A new tack, just as well.

    Before, SWL believed there existed tools to help every household to have 10,000 pounds more than under the present government.

    Let me refer him to

    http://andolfatto.blogspot.de/2015/01/on-want-of-bold-persistent.html

    ReplyDelete
    Replies
    1. No new tack - entirely consistent. One is about the past, the other the future.

      The reference is about experimentation. That is the wrong way to think about it. Its simply maximising expected welfare.

      Delete
  7. Are there ways that estimates of potential output can be lowered when actual output is below natural output for a long period of time, even if natural output has not been lowered?

    Anecdotally, worries about secular stagnation seem to come up in the history of economics whenever there are multiple and/or persistent negative AD shocks. Can estimates of potential output avoid this sort of confusion?

    ReplyDelete
  8. NGDP targeting plug: for NGDP targeting, the output gap doesn't matter for monetary policy, so misestimating it is not an issue.

    ReplyDelete
    Replies
    1. Unfortunately not the case - with levels targeting you need to start with the right level, which means making an assumption about the output gap. Would you start with a NGDP target based on today's level, or extrapolating pre-recession levels? I'm in favour of NGDP targets, but they do not make the basic problems go away.

      Delete
    2. Ok, fair point, they do matter initially.

      I don't like the way the discussion in many countries has been framed in terms of unemployment levels rather than hours worked. There are certain policies for which unemployment levels are what we should be looking at, but hours worked seems to be the more important statistic for output gaps.

      Delete
  9. At the LRB, apart from your 'The Austerity Con', Stefan Collini was also penning an essay. On another occasion Collini has said 'it's all in the language'.

    So, despite the strong mathematical data economics has produced over the last seven years, the gimcrack response by politicians, most journalists and the overwhelming mass of the electorate still seems trapped in simplisms, ensnared in a language game they don't seem able to extricate themselves from.

    ReplyDelete
  10. Simon,

    I think you're right that the MM'ers have been too dismissive of fiscal policy as a practical tool, and that Keynesians have been relatively more embracing of "unconventional" monetary policy. However, Keynesians have not been very specific about how monetary policy is supposed to operate at the ZLB.

    I wish to propose a New Monetary and Fiscal Policy Consensus (NMFPC) with the following elements:

    1. NGDPLT. We all recognize that inflation targeting via interest-rate targeting (Taylor rule) hits a major roadblock at the ZLB. NGDP level targeting has no such discontinuity, and avoids the muddle of QE (where the markets can't understand what the Fed is trying to achieve, so they don't know how much of the monetary base expansion is permanent.)

    2. Automatic fiscal stabilizers. If the CB falls X short of its NGDP level target for Y amount of time, and actual output is at least Z below potential output, then the fiscal authority automatically pays T worth of tax credits distributed to taxpayers according to policy P.

    I think all reasonable economists (meaning not Austrians or RBC types) should be able to agree that the above would be a dramatic improvement over current policy, with any half-sensible values of X, Y, Z, T, and P. So we should team up here, not argue about monetary offset. If the economics community could speak with one voice, we could be more effective at influencing policy.

    I think it's incredibly important to the world to make something like the above happen, and I am prepared to fund any credible effort to do so.

    What do you think?

    -Ken

    Kenneth Duda
    Menlo Park, CA
    kjd@duda.org

    ReplyDelete
  11. It is valuable to link Professor Farmer to this subject of missed output. Trends don't really mean anything if the current generation of people, responsible for most of consumption, change the way they and the previous generations did things because of some shock factor. I suppose though it can be countered that this change should show up in some stats somehow, e.g., saved more of disposable income than before should show an increase in savings as a reason for drop in demand. But there can be many factors all adding up to a drop in output and that is beyond most pocket calculators. However the econ super calc machines should be able to identify the various elements of change somehow if the data is available.

    Professor Farmer touches on self-fulfilling problems following shocks. I don't know if he says anything about what mum says to the kids making up national characteristics. Different household saving rates are found throughout the nations. People, of roughly the same means, even eat differently, in general. Meat and two veg is not for everyone. It must mean more money for other things.

    ReplyDelete
  12. Simon, what about the methods the BoE use at the moment for estimating output gaps "from first priciples"? This seems to mean trying to measure slack in the labour market (from unemployment, participation, underemployment/part time/low hours) and slack within firms (mostly survey data, agents reports, possibly industrial output data). This is how they have got to the 0.5% estimate given most recently. I completely agree with you on the deep flaws in using regressions on actual output to estimate potential, but the "first priciples" arguments are harder to counter. Thanks

    ReplyDelete
  13. This post is extremely important. The assumption that potential output is exogenous makes a huge difference. I guess DeLong and Summers on fiscal stimulus that pays for itself is a recent illustration of this point.



    I'd go a bit further (as usual). You suggest that technological progress might not be measured because it has to be embodied. I would guess that actual technology progresses less in depressed economies. Technology does not develop itself -- it is affected by dedicate research (as well as on the job tinkering and finding shortcuts). The only reason it is exogenous to standard models is that it is hard to model. Exogenous to my model doesn't mean really exogenous. The fact that I choose to treat something as given doesn't mean it is given. In particular it doesn't mean it isn't effected by policy or that I can treat it as given and correctly evaluation policy.


    I think that another part of the problem is attachment to accelerationist Phillips curves. Steady inflation convinces people that the output gap is small so assumptions about technological progress are adjusted to give a small output gap. Even if the final product consists of estimates of available factors of production and total factor productivity Solow's residual has been chosen based on the assumption that only the change of inflation is informative.

    This implies downward biased estimates of potential output if there is downward nominal rigidity.

    I want to advertize my posts (if this isn't OK use of comments section please delete and tell me not to do it again)

    On the unreliable shared assumption about the long run see
    http://angrybearblog.com/2014/01/john-quiggin-is-as-usual-brilliant.html

    It is an assumption and one which is no one asserts to be plausible if taken literally.

    I have a model with endless depressions where the key assumption is downward nominal rigidity
    http://angrybearblog.com/2013/12/macroeconomic-dynamics-with-downward-nominal-rigidity.html
    and
    http://rjwaldmann.blogspot.it/2013/12/dreadful-plumbing-with-downward-nominal.html

    * OK now the usual rant.
    Those bored and irritated by my usual rant might benefit from skipping on to the next comment.


    Macroeconomists had the broadest consensus regarding and the most confidence in their views on the long run. This is an area where the data tell us little, and also where macroeconomists who studied effects of aggregate demand filtered out the little evidence there was. This is the pattern one would expect of a research program based on fundamental errors. There were strong claims. Those that could be tested were rejected by the data. There are a variety of responses to that failure and disagreement about which is bets. Those that can't be tested were not tested and are still held with confidence.

    I think this can be summed up simply. The more you know about a question, the worse mainstream macroeconomic's answers look. This is absolutely not true of human inquiry in general. It is not at all true of many research programs in the natural sciences.

    ReplyDelete
  14. This post is extremely important. The assumption that potential output is exogenous makes a huge difference. I guess DeLong and Summers on fiscal stimulus that pays for itself is a recent illustration of this point.



    I'd go a bit further (as usual). You suggest that technological progress might not be measured because it has to be embodied. I would guess that actual technology progresses less in depressed economies. Technology does not develop itself -- it is affected by dedicate research (as well as on the job tinkering and finding shortcuts). The only reason it is exogenous to standard models is that it is hard to model. Exogenous to my model doesn't mean really exogenous. The fact that I choose to treat something as given doesn't mean it is given. In particular it doesn't mean it isn't effected by policy or that I can treat it as given and correctly evaluation policy.


    I think that another part of the problem is attachment to accelerationist Phillips curves. Steady inflation convinces people that the output gap is small so assumptions about technological progress are adjusted to give a small output gap. Even if the final product consists of estimates of available factors of production and total factor productivity Solow's residual has been chosen based on the assumption that only the change of inflation is informative.

    This implies downward biased estimates of potential output if there is downward nominal rigidity.

    I want to advertize my posts (if this isn't OK use of comments section please delete and tell me not to do it again)

    On the unreliable shared assumption about the long run see
    http://angrybearblog.com/2014/01/john-quiggin-is-as-usual-brilliant.html

    It is an assumption and one which is no one asserts to be plausible if taken literally.

    I have a model with endless depressions where the key assumption is downward nominal rigidity
    http://angrybearblog.com/2013/12/macroeconomic-dynamics-with-downward-nominal-rigidity.html
    and
    http://rjwaldmann.blogspot.it/2013/12/dreadful-plumbing-with-downward-nominal.html

    * OK now the usual rant.
    Those bored and irritated by my usual rant might benefit from skipping on to the next comment.


    Macroeconomists had the broadest consensus regarding and the most confidence in their views on the long run. This is an area where the data tell us little, and also where macroeconomists who studied effects of aggregate demand filtered out the little evidence there was. This is the pattern one would expect of a research program based on fundamental errors. There were strong claims. Those that could be tested were rejected by the data. There are a variety of responses to that failure and disagreement about which is bets. Those that can't be tested were not tested and are still held with confidence.

    I think this can be summed up simply. The more you know about a question, the worse mainstream macroeconomic's answers look. This is absolutely not true of human inquiry in general. It is not at all true of many research programs in the natural sciences.

    ReplyDelete
  15. Are you familiar with the work of Michael Pettis? http://press.princeton.edu/titles/9936.html
    Now that I have read this, I see these questions entirely differently. He would answer it by saying that capital flows and trade imbalance make it impossible for the US and peripheral Europe to raise employment. Under the Euro, Greece will never be able to generate a trade surplus while Germany retains its surplus. I find his models very convincing.

    ReplyDelete
  16. Are you familiar with the work of Michael Pettis? http://press.princeton.edu/titles/9936.html
    Now that I have read this, I see these questions entirely differently. He would answer it by saying that capital flows and trade imbalance make it impossible for the US and peripheral Europe to raise employment. Under the Euro, Greece will never be able to generate a trade surplus while Germany retains its surplus. I find his models very convincing.

    ReplyDelete

Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time.