What I’m reading

I’ve been looking at an interesting report from the National Office for the Information Economy on Productivity Growth in Australian manufacturing (PDF file). Released with no fanfare (not even a press release) recently, it presents results which accord with everyday observation, but not with the ideological assumptions that dominate the policy debate in Australia. The approach used was to examine labour productivity growth in a wide range of manufacturing industries and use regression analysis to explain differences in rates of productivity growth. By far the most important factor was the use of advanced information and communications technology, with additional explanatory power being gained when the education level of the workforce was taken into account. Claims that reductions in tariffs would expose inefficient industries to the “cold wind of competion” (or sometimes the “hot blast”), forcing them to increase productivity and therefore induce dynamic efficiency gains got no support. The coefficient on the associated variable was both economically and statistically insignificant (negative in most cases) [1].

A notable technical feature is the point, which has been emphasised by Brad de Long that it’s not appropriate to focus on multifactor productivity when most technological progress is embodied in capital items with rapidly declining prices, such as computers.

fn1. A mildly surprising result in view of the Thatcher effect, by which average productivity is automatically increased when inefficient factories close.

15 thoughts on “What I’m reading

  1. Perhaps whilst studying the workforce for productivity measures the time-servers at NOIE examined their own level of productivity and found it wanting. Or maybe they started to think a Labour Government was on the cards and they’d better get out while the going was good.

    Whatever the reason they’re already folding up the card table and disappearing before anyone starts asking hard questions like: Did they do anything useful?

  2. A notable grammatical feature is that authors agree that data is a plural, as in this example on page x of the Executive Summary

    “Some relevant data are also available on multi-factor productivity (MFP) growth and were used as
    dependent variables in some regressions.”

    Well done NOIE.

  3. Multifactor productivity on Sunday night? So what are weekends for then? Is this penance for the excess of birthday frivolity on Monday?

    By the way, if we are to have a rematch of ‘data is’ versus ‘data are’, may I request fresh arguments on both sides rather than a rehash of this.

  4. “Claims that reductions in tariffs would expose inefficient industries to the “cold wind of competion” (or sometimes the “hot blast”), forcing them to increase productivity and therefore induce dynamic efficiency gains got no support.”

    So what?

    PrQ is indulging in a little selective reporting here. For example the report also states that “It should be noted that tariff reductions had a beneficial effect on economic performance not only by increasing competition and productivity in the industries affected but also by improving the allocation of resources, by channelling them toward activities where
    Australia has a comparative advantage in international trade. In the National Accounts
    this is reflected in productivity improvements in all sectors and not only in those directly affected.” (Page 56).

  5. Ron, if you’d been following this debate a bit more closely you’d be aware that everyone agrees that the ‘static’ gains in allocative efficiency from tariff reform are real but modest. The question of whether there are, in addition, dynamic efficiency gains is central. You can read my argument on this here.

  6. Dynamic gains involve the improved innovation and immitation of production and business practices – generally attributed to the increased pressure on business caused by competition.

    The NOIE study points out that much of Australia’s growth can be explained by the improved innovation and immitation of a production and business practice – specifically related to the use of ICT. This story is perfectly consistent with the idea that tariff reductions result in dynamic gains.

    Before any pro-competitive measure, it is near impossible to determine exactly what innovation will result in improved production and/or business practices. After we’ve seen the improvement, we can identify the innovation. It is not true that identifying the innovation therefore destroys the argument for the vaguely named ‘dynamic benefits’.

  7. John H, you don’t seem to have responded to the observation that productivity gains from ICT are uncorrelated with changes in tariff protection.

    It’s a semantic point whether the adoption of exogenous technological improvements is what is meant by “dynamic”, but however you label it the evidence suggests it is unrelated to microeconomic reform, or at least to changes in tariffs which were the reforms most directly relevant to manufacturing.

  8. The results cited in this report are a bit surprising. There were large productivity gains in car assembly following the Button plan and the large reduction in tariff protection that forced producers to produce fewer models and to aim for scale economies. These are pro-competitive effects that involve dynamic efficiencies.

    But as the report points out there were no great productivity gains in TCF even though tariffs fell almost as dramatically. Maybe the sector just got smaller because the more capital intensive technological options called for by reduced tariffs were limited.

  9. The results cited in this report are a bit surprising. There were large productivity gains in car assembly following the Button plan and the large reduction in tariff protection that forced producers to produce fewer models and to aim for scale economies. These are pro-competitive effects that involve dynamic efficiencies.

    But as the report points out there were no great productivity gains in TCF even though tariffs fell almost as dramatically. Maybe the sector just got smaller because the more capital intensive technological options called for by reduced tariffs were limited.

  10. As far as I can tell (and I haven’t read the whole report), the study showed a relatively small correlation (0.15 using the weighted numbers) between labour productivity growth and tariff reductions.

    My point was that what is often referred to as dynamic benefits would now be reported in other areas, as we know what technology has been used to improve productivity (the advantage of hindsight).

    The study does not attempt to correlate tariff reductions with increased use of ICT – though it’s hard to believe they wouldn’t be correlated. It would also be interesting to see the correlation between tariff reduction and R&D spending etc.

    The report also notes that the relationship between productivity and tariff reform exists for the motor vehicle manufacturing industry, and several other industries already had low protection (as so the benefit of reducing protection would obviously be low).

  11. “The study does not attempt to correlate tariff reductions with increased use of ICT – though it’s hard to believe they wouldn’t be correlated. It would also be interesting to see the correlation between tariff reduction and R&D spending etc.”

    The whole point of multiple regression is to take correlations between explanatory variables into account. The regression analysis confirmed the finding from the partial correlations that tariff reductions were unrelated to productivity growth.

  12. Ah – I see. Chart F1 (showing no correlation between ICT and tariff reductions – unfortunately it is unweighted). Still, I would prefer to see ICT regressed against a number of explainatory variables. I am surprised that there is no correlation shown between ICT and tariff reduction – but there are several potential reasons for this (ICT is likely influenced by many, sometimes offsetting, variables), and we won’t know the real relationship until ICT is explained in more detail.

    Same holds true for the other explainatory variables, such as R&D. I note that the Melbourne Institute has previously shown a relationships between competitiveness and R&D investments using multiple regression.

  13. Just on a tangential point related to MFP: has any work been done on the fact that the same factors which have been driving down the prices of ICT hardware have also presumably assured that their effective depreciation life has shortened? It struck me a while ago that there would be room for all sorts of double-counting fun and games if you used hedonic pricing for computer gear while not changing your estimates of the average life of the capital stock, but I kind of stuck it in the loft to think about later.

  14. John,

    I’m a bit late coming to this debate. I suspect that the contrast between the MFP impacts of tariff reform and ICT adoption is unwarranted.

    Regressions notwithstanding, my own experience of industry policy changes viewed from a Canberra perspective (the once highly protectionist Trade department) over 20 years from the mid-70s inclines me to think that there was a direct relationship between the increasing trade exposure of key services and manufacturing industries (transport, finance, wholesaling, base metals, metals fabrication, electrical machinery) in the 1980s and early 1990s and the rapid adoption of ICT technology in Australia in the 1990’s by those same industries.

    I think that the evidence does, indeed, show that MFP improvements are strongly linked to the ICT uptake. But I think that the tariff reforms layed the groundwork both historically and in the impact that they had on industry margins, structures, investor profiles and executive composition.

    I’m afraid this is speculative, because I can’t cite objective evidence. But Gary Banks (Chairman of the Productivity Commission) in a very good speech at ANU in August last year on ‘Australia’s Economic Miracle’ (available on the PC site somewhere) argued along similar lines, I think, with more scholarship.

    Of course, a cynic might say that both NOIE and the PC are burnishing their own ‘scutcheons by these arguments. But it may be that both views deserve some creedence.

    Best wishes,

    Peter

  15. Sorry, I’m late here too but the ‘both views desrve some credence’ seems right.

    Correlation and causation are often confused, regression techniques notwithstanding. The implementation of technological improvements depends upon the independent existence of technological improvement. People started using computers because they had a capital advantage over cheaper competing sources which could be realized as a technological advantage.

    The difference between the TCF and Automotive industry examples of tarrif reduction is about 10 years which correlates with the development of IT and robot techonology which the auto makers could make use of, but the TCF producers had already gone broke.

Comments are closed.