Revising my priors

Looking at the desperation with which opponents of climate science, and of sensible policy responses such as Kyoto, are holding on to positions that have clearly become untenable, has prompted me to think about my own views on a range of issues, to see whether I am holding on to beliefs that can’t be sustained in the light of accumulating evidence.

The most obvious problem for me is that of continued macroeconomic stability in the face of trade and current account deficits driven (or so it seems) by speculative asset price booms. I’ve long argued that such deficits can’t be sustained and that neither Australia nor the US is on a path to a smooth adjustment. However, while deficits have continued and, in the case of the US, grown steadily, evidence of anything but smooth adjustment is certainly thin on the ground.

The rapid growth of China, and the apparent willingness of the Chinese government (and maybe also the public) to hold low-return $US assets and to buy large quantities of commodity exports from Australia has rendered previous projections largely irrelevant. While the “Bretton Woods II” story that emerged a couple of years ago seemed implausible to me, it has held up pretty well so far.

While I’m not ready to join the optimists just yet, it’s clearly necessary to rethink the implications of a Chinese economy that is already a substantial part of the global total, and growing rapidly.

A point on which I don’t need to revise my beliefs is my long-running dispute with the Productivity Commission over the productivity ‘miracle’ of the mid-1990s which, I’ve long argued, was a mirage generated by a combination of recovery from the 1989-92 recession and an increase in work intensity. The latter phenomenon, evident to everyone who held a job, but steadfastly denied by a large number of economists, has now been partially reversed.

Given my analysis, I predicted at the time that the productivity surge would turn out to be a blip, and this has indeed been the case. Alan Mitchell has a piece in today’s Fin pointing out that the most recent year of poor data can be explained in part by the mining boom, which has led to growth in marginal mines, but some such explanation (drought, the Olympics, GST-related disruptions and so on) is always possible. After seven years of weak data, the productivity miracle has run out of excuses.

Coming back to work intensity, there’s been an interesting paper recently showing that Australian data on average annual hours of work are biased upwards by the way the ABS treats public holidays (hat tip to regular commenter DD). Claims that we have the longest average hours in the world aren’t sustained once this correction is made. Unfortunately, there’s a bias going the other way, associated with our very high rates of part-time employment. I’d like to see international comparative data on average hours of work for full-time employees (bearing in mind that this is not the well-defined concept it would once have been). Can anyone help on this?

43 thoughts on “Revising my priors

  1. Glad to see you coming around John. Next you’ll be singing the praises of the EMH! 🙂

    The OECD has data on full-time employment by country, but from memory Australia isn’t included (or if it is the data are strange – I did try drag out some numbers a couple of years ago). First sentence, fourth para has a typo ‘Wgile’.

  2. John, on working hours, the Tiffens and Gittins book has an interesting disucussion on page 83, although it does not really answer your question.

    On productivity, the PC figures show that in the last cycle, multifactor productivity, although below the exceptionally high figures of the 94-99 cycle, was still running at pretty respectable levels relative to the 1970’s, 80’s and 90’s. Perhaps the data after 2003-4 may alter the picture. Your point – that there has been no sustained productivity miracle – is valid but nor has there been much decline, as some critics claim.

  3. Fred, I discussed this last year. Although the data is subject to revision, the period since 94, taken as a whole, is indistinguishable from the long-run average. Of course, that’s still better than the first decade or so of micro reform 83-94.

    Sinclair, thanks for the alert. Fixed now, I hope.

  4. The ABS labour force stats have average and total hours for full time and part time, but I’m not sure if they are consistent with the OECD data that SD mentions.

  5. Pr Q, I think you do need to revise your view on micro reform. While the benefits of micro reform may have been temporary, it is hard to argue that there aren’t any benefits at all.
    The data shows a very strong increase in Multifactor productivity (MFP) in the 1990s (see below). It is very hard to dismiss this as being entirely due to a recovery from the recession or work intensity.
    Re work intensity – Pr Q raises mentions some survey data showing that work intensity has increased in his 2001 Agenda paper. However, this data doesn’t help unless it can also be shown that the pace of work didn’t change as much in earlier periods. Otherwise, this explanation merely states that all MFP calculations are overestimates, it does nothing to remove the strong relative MFP performance in the 1990s. In other words, the data not only needs to show that work intensity increased, but it increased much more between 1993 and 1998 than the earlier periods.
    Another source of data errors could be the information on hours. The ABS data shows hours worked declining reasonably consistently over time (see link in Pr Q’s original post). The data on hours uses worker surveys and it is hard to see how workers would substantially underestimate their hours worked between 1993 and 1998 while this underestimate did not occur prior to 1993.
    Re part time work, again the level of part-time work would not affect the results, it is the change in part time that would. I’m not sure that it has increased at a faster rate after 1993 than it did before.
    Re recovery from recession – It is not clear why this would be an explanation for the recovery from the 1990-91 recession, when multifactor productivity (MFP) growth was below average in the recovery from the 1974 recession and below average in the recovery from the 1982-83 recession. It needs to be shown why the 1990-91 recession was different, with more MFP growth in the aftermath than after previous recessions.
    Re the data: The latest data (updated since JQ’s blog of 2005 and 2001 Agenda article) shows that MFP growth was very strong in the 1990s – MFP growth from 1993 to 1998 was 2.3 percent, well above the average (from 1964 to 2003) of 1.3 percent. The latest cycle, from 1998 to 2003, had MFP growth of 1.1 percent, only a bit below the average.
    The average from 1993 to 2003 is 1.7 percent, which is still above the average.
    Even then, these MFP numbers significantly understate the ‘real’ productivity numbers, with unemployment falling over the 1990s and 2000s. Productivity data does not adequately deal with the effects of bringing unemployed resources into production. If an unemployed worker enters the workforce and produces below average output (which is likely), then this will reduce the productivity figures. Pr Q acknowledges this in his 2001 Agenda article.
    This effect occurs despite the fact that a person who was producing nothing is now producing something. The productivity numbers misleadingly indicate that bringing unemployed resources into production is a bad thing.
    Note that it is very difficult, if not impossible, to correct for this.
    JQ raises a valid question, why didn’t we see the benefits of micro reform earlier than 1993? Firstly, there can be a considerable lag from policy changes to productivity changes. Secondly, it could be argued that the very heavily regulated labour market prior to 1993 meant that it was very hard to achieve productivity gains, even with reforms outside the labour market.

  6. A bunch of points here. I’ll just reply to one for now. As regards work intensity, I think it’s pretty clear that we saw a sign change in the 1990s. That is, up to that point, work intensity had generally declined for a century or more – this was an unmeasured productivity gain. By contrast, work intensity rose in the 1990s, an unmeasured productivity loss.

  7. “The average from 1993 to 2003 is 1.7 percent, which is still above the average.”

    I’ve never believed this “productivity cycle” stuff, so I prefer not to throw out data, while waiting for a supposed end of cycle.

    Taking the entire period since 1993, the average rate is almost exactly equal to the long term average.

  8. Having read the ABS paper on the changed method of estimation, I’m now not sure whether the new method is the same as other OECD countries use. Can we now say we really have productivity of 83% of the USA etc.?

  9. “I’ve never believed this “productivity cycleâ€? stuff…”

    Pr Q – you seem to (implicitly) accept it in your Agenda paper of 2001.

  10. #10 I didn’t think it was worth fighting over this in Agenda, but I’ve expressed my views many times.

    #11 This is well known to everyone except economists. Which part do you want to deny?

  11. On hours of work, it’s important to distinguish between part-time and full-time work so you can look at the trend in average hours for full-time workers. This fell consistently until around 1990, rose in the 1990s and has fallen back since, following exactly the same path as subjectively reported work intensity. There are obvious reasons for supposing that full-time hours of work will be positively correlated with work intensity.

    The graph here, headed Longer Hours for FT workers tells the story.

  12. #12: It is well known to everyone except economists that Australians work longer than the rest of the developed world. Yet this is wrong. Since when does blind faith trump facts?

  13. I hope that’s not the case JQ. It was your paper, The Unsustainability of US Trade Deficits, that landed me here in the first place. As that paper lays out, it’s pretty difficult to get around the arithmetic. IMO, you were right then even if the Wile E. Coyote market, (Jeremy Grantham’s terminology), still doesn’t know it.

    That euro yen is at an all time low is one indication that the monetary disorder that has fueled and been fed back into by global imbalances, (to say nothing of its role in anomalous productivity growth, profit margins and the labor share), is still in full swing. Credit spreads outside US subprime mortgage sector are another. I would recommend waiting out a break in that momentum before you throw in the towel…

  14. I too have been wondering about the unsustainablility of Australia’s and America’s CAD. I admit to having been a pessimist for a while on this issue and being wrong for a long time now. But I can’t honestly say that I am admitting my views are wrong as I still believe there has to be an adjustment, just that it is taking longer than I thought it would.

  15. Pr Q, if you are arguing that increase in work intensity occurred over the 1990s at a faster rate than beforehand, it is you who have to provide the facts to support that argument.

  16. So productivity hasn’t risen a great deal recently. It’s not hard to think of reasons why productivity may decline in the future, and if it did, after a while, succeeding generations would be getting poorer…

  17. John,

    I’ve never really understood your complaint about work intensity. So what if we work harder now, couldn’t this be a positive of micro reform?

    For example, back in the day when effort was not correlated with reward (due to a raft of trade, IR and other protections) why work harder when you weren’t rewarded for individual effort. The micro reform process which removed some of these restrictions gave more fredom for workers to decide whether working harder was worth the increased reward (now permitted to them). Consequently, pareto improving work-leisure trades were made and as theory would predict the greater rewards led to greater effort. Isn’t this a positive productivity benefit?

  18. Matt, there’s a Potential-Pareto improvement, but that doesn’t change the fact that working harder is costly. There are gains from trade if we, for example, sell coal and buy TV sets, but you still have to count the cost of the coal. The PC numbers don’t do this.

    MP, I’ve given plenty of evidence in my papers that work intensity increased in the 1990s, and that this increase was correlated with micro reform. Are you seriously claiming that work intensity was higher in 1990 than in, say, 1900? If so, I’m happy to provide evidence, but I want to check that someone would make such an obviously untenable claim before bothering to refute it.

  19. Not at all. The key period is after 1964, when the TFP data starts.
    Work intensity may have reduced from 1900 to 1964, and then increased steadily since then. If it did, then none of the above average TFP performance in the 1990s can be explained by greater work intensity.

  20. Pr Q, to be absolutely clear, you need to show that work intensity increased at a (significantly) faster rate during 1993-98 than over 1964-1993.

  21. Again, I think it’s pretty clear that there was no general increase in work intensity over this period, and as far as I know no-one has ever asserted that there was. If you want to make the claim, I’ll refute it, but, as I’ve already said, I’m not going to bother refuting an untenable claim that no-one has made.

  22. We seem to be going round in circles.
    I thought it was pretty standard for the proponent of an argument being required to present the evidence for it, rather than asserting the argument and requiring opponents to present arguments against it.
    But for the sake of argument, let me assert the contrary point and you argue against it.

  23. MP, there were numerous improvements in working conditions over the period from 1964 to the late 1980s that had the effect of reducing work intensity. The removal of these improvements, called “restrictive work practices” was a major goal of microeconomic reform, and of employers in wage bargaining from the late 1980s onwards. Hence, work intensity almost certainly declined over this period.

    So, the productivity surge of the 1990s is doubly overstated – productivity growth in earlier periods is understated because of the failure to take reduced work intensity into account, while productivity in the 1990s is overstated because of increased work intensity.

  24. Pr Q says:

    the productivity surge of the 1990s is doubly overstated – productivity growth in earlier periods is understated because of the failure to take reduced work intensity into account, while productivity in the 1990s is overstated because of increased work intensity.

    The claim that labour intensity has increased since the late eighties is clearly consistent with the evidence from a variety of sources:

    – the decline of unions and “featherbedding” replaced by Human Resources,

    – just-in-time and other forms of time-and-motion labour speed-ups;

    – down-sizing, out-sourcing and sub-contracting;

    – increase in illegal immigrant “sweated” cheap labour,

    – the 24/7 digital work cycle means that you can never get away from the job;

    – reduction and constriction of leave entitlements (sickies, week-ends, public holidays)

    – slash-and-burn management practices to “clean out the dead wood” of workers who were not pulling their weight. These semi-disabled men have now been tossed onto the scrap heap.

    Of course pay rates have improved, although not noticeably faster than the Golden Age of 1943-1973. People seem to be retiring earlier. It seems that people are willing to work like stink for a decade or so and then pull the pin for a bit of sea-change down-shift.

  25. Pr Q, that makes it clearer what you are arguing.
    So the argument is that productivity did grow in the 1990s, but this didn’t mean an improvement in welfare because of the increase in work intensity.

    However, if we switch from measuring productivity to measuring welfare, then it would actually be easy to argue that the the 1990s onwards saw a very large increase in welfare because of the large falls in unemployment. So I’m not sure that switching from productivity to welfare helps your argument that there wasn’t a benefit from micro reform.

  26. I think its pretty clear from the post that I don’t deny that good macroeconomic performance improves economic welfare. But there’s no obvious basis for the claim that micro reform is the cause of good macro outcomes. NZ had even more micro reform from 83-95 or so, along with disastrous macro outcomes, and has had good macro performance since the Labor government partially reversed the reforms.

    And, as I’ve pointed out previously, belief in the “micro reform = macro flexibility” idea was one of the factors that led to policy misjudgements in 1989, and the deepest recession since the 1930s.

  27. jquiggin Says: February 24th, 2007 at 12:20 pm

    as I’ve pointed out previously, belief in the “micro reform = macro flexibility� idea was one of the factors that led to policy misjudgements in 1989, and the deepest recession since the 1930s.

    The “recession we had to have” was, in part, caused by the mistaken Keynsian belief that macro monetary policy had to ensure micro balances in the sum of external accounts. As Pr Pitchford forcibly argued, and Pr Q would himself now admit, this macro argument rested on some kind of economic fallacy of composition.

    The external account, whatever its financial valency, is the sum of countless consenting capitalist acts and is presumptively optimal. Monetary policy should concentrate on keeping inflation low and stable, such that real economic growth is maximised.

    Surely the “reformist” equation was “micro-flexibility = macro-stability”? The validity of this proposition is substantated by the fact that the 1983 liberalised currency markets (micro flexibility) smoothed out the adjustment of the Australian economy to the 1998 Asia crisis (macro stability). The steep decline in the AUD in the wake of this crisis acted to cheapen the international value of our exports, demand for which would otherwise have crashed leading to the domestic recession that we did not have.

  28. “NZ had even more micro reform from 83-95 or so, along with disastrous macro outcomes”.

    This claim, or more specifically the claim that NZ macro performance worsened after labour market reform, is debunked by Perry (2006) “New Zealand lessons for workplace reform – a critique of 151 academics” Economic Papers 25(4)and Perry (2006) “Do Workplace Contracts Harm Labour Productivity Growth – a reconsideration of the macro evidence from New Zealand” Australian Economic Review 39(4).

    Perry uses new data to show that NZ’s labour productivity increased at a faster rate after LM reform than before.

    NB JQ was a signatory to the 151 academics paper – perhaps another revision of priors is called for…

  29. MP, although this paper has “Macro” in the title, it doesn’t sound as if it actually talks about macro outcomes in the sense in which we were discussing the terms.

    So, are there any other points you want to raise on the relationship between micro reform and macro outcomes in the ordinary sense, or do you want to move on to NZ labour productivity?

  30. I don’t really mind the order in which you respond to comments. But let’s move on to NZ if you want.

  31. MP, if you have any unanswered comments on micro reform and productivity in Australia, I’m not aware of it. I’d rather finish with that issue before moving on to NZ. Do you have anything more to say on this, or any points you think I haven’t replied to?

  32. OK, I don’t think you’ve dealt with the following issues:
    * The fact that TFP fell after other recessions, but increased after the 1990s recession, suggesting the recovery from recession argument actually increases the amount of productivity in the 1990s unexplained.
    * Why you don’t think that productivity should be measured peak-to-peak or trough-to-trough
    * Why micro reform wasn’t responsible for good Aust unemployment performance (unless your only evidence for that is NZ).

  33. On (1), I think there were specific features of the 1990s recession, namely overinvestment in physical capital that led to a spurious recovery in measured capital productivity. However, this point is a minor one.

    I do think productivity should be measured peak-to-peak or trough-to-trough. However, I think the relevant cycles are business/macro cycles. I don’t see any reason to believe in a separate productivity cycle.

    On micro reform and macro outcomes, I’ve already mentioned, as well as NZ, the 1989-92 recession (I argue this point in more detail in my Agenda paper). Then there’s the strong macro performance of the postwar boom, when micro policy was the exact opposite of what is now called reform. You could also look at World Bank studies imputing strong Asian growth to micro reform (in particular financial liberalisation). These came out early in 1997, IIRC.

  34. I will address your points in order.
    1. I’m not sure why you use recovery from the recession as a reson to dismiss productivity growth in the 1990s if you think the point is minor. NB I thought we were talking about MFP, not capital productivity.

    2. If this is the case, it is not clear why you measure productivity from 1993 to now if we aren’t at the same point in our economic cycle as we were in 1993.

    3. I would argue that the very bad performance in the 1990 recession was due to a lack of labour market reform. The postwar boom is explained by the Blanchard & Wolfers (2000) thesis that it was interaction of institutions and shocks that caused poor performance in economies. Increase in regulation didn’t cause poor performance, until those institutions interacted with economic shocks (particularly the oil shock).
    The were obvious problems with the WB imputation of Asian performance to micro reform, but that doesn’t mean that all imputations are wrong.

  35. Capital productivity is a component of MFP. The recession point is a minor one, since my main concern is with mismeasurement of labour productivity, which is the larger component of MFP.

    Again, I’m bending over backwards to be fair to the PC in starting at 1993. I would favour measuring productivity from 1989 to now (peak to presumptive peak). What I don’t agree with is ignoring recent bad productivity data until it can be shown ex post that a cycle has been completed, by which time the bad news is old enough that it can be ignored. Note that the PC didn’t follow this rule when the news was good.

    How does the BW story work for 1989-92? There was no external shock and the recession was driven by policy misjudgements. And of course, there’s still the NZ recessions to explain.

  36. We can indeed measure from 1989 to today, but then some reforms had only recently occurred, and some very important ones were yet to occur (NCP, privatisation, tarrif reductions and LM reform). So measurement of the effects of these reforms should occur from later. The question is when.

    The recession was indeed driven by policy misjudgements, but this is almost entirely a macro problem (tho I argue that it was made worse by the lack of LM reform). Exactly the same argument could be made for NZ.

  37. MP, on your first point note that micro reform continued well after the end of the “productivity surge” – notable examples are the GST, considered the jewel in the crown by reformers like Hewson, and the privatisation of Telstra. Similarly, NCP didn’t really start coming into effect until after the surge.

    On the recessions, I agree that macro outcomes are driven primarily by macro policy. The NZ recession, after radical LM reform shows this, as do recessions in the US where the labour market has always been lightly regulated.

  38. I don’t consider the GST to be a useful micro reform, and it is unlikely that the privatisation of Telstra would have had a large effect either way. I don’t know about NCP, but the agreement did occur in 1994.

    Re NZ, I thought their recession was in 1991, before LM reform.

  39. I agree with you about the GST and privatisation, but I could have used some support on these points at the time. Most micro reformers were strongly in favour of privatisation, and a lot were annoyed when I said the GST (which I supported, BTW) was a second-order issue. I was involved in implementing NCP in Queensland and it didn’t start, there or in the other states, until the late 1990s. The establishment of an access regime for QR wasn’t done until about 2003 and this was a high priority item.

    On NZ, they had the 91 recession, then LM reform and “the two good years” when success was claimed in very strong terms, then another recession induced by bad monetary policy at the time of the Asian crisis.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s