Canada-U.S. Productivity Comparisons

StatsCan released a new analytical study today on the decline of Canadian labour productivity relative to the U.S., up to 2003.

http://www.statcan.ca/Daily/English/080721/d080721a.htm

Main findings are not surprising: Canadian business sector productivity has slipped relative to U.S. productivity (to 87% by 2003).  (We know it’s fallen significantly further than that since — Canadian labour productivity has hardly grown at all since 2003, and by nothing since 2006, while U.S. productivity is growing at something close to 2% per year).  The StatsCan numbers are less dramatic than information published by Andrw Sharpe’s Centre for the Study of Living Standards, according to whom Canadian productivityin the business sector had fallen to 74% of U.S. levels by 2006.

http://www.csls.ca/data/iptjune2007.pdf

The StatsCan report argues that most of the decline in Canada’s relative productivity gap is located in the goods side of the economy, which is somewhat surprising.  (There’s been a lot of attention paid to service sector productivity growth in the U.S., both the alleged “Wal-Mart” effect and the indirect impact of the spread of information technology through service industries.)  In this case, the resource-led structural shift in Canada’s good sector (booming petroleum exports, shrinking manufacturing) is very bad news indeed on two fronts: productivity in resources production is falling rapidly (as record prices lead producers to develop increasingly marginal deposits), and hundreds of thousands of high-productivity manufacturing jobs are being wiped off the map.  In the StatsCan report, mining (including petroleum) is included in what they call an “engineering” sector; declining relative productivity in that sector as well as the traditional goods sector explains most of the decline in overall relative productivity since 1999.  Our relative services productivity is even worse — but it hasn’t been deteriorating since 1999 as our goods-sector comparion has been.

 

The StatsCan report also highlights the importance of Canada’s lower investment in machinery and equipment. We have more capital in use, proportionately, than the U.S., but that’s all due to our reliance on mega-structures (mines, pipelines, etc.).  We use less machinery and equipment than the Americans, and it’s M&E capital that economists have found is especially closely linked to productivity.

One other interesting tidbit: in the non-business sector (government, public services, non-profits) Canadian productivity is 6% HIGHER than U.S. productivity.  That offsets a portion of our inferior business sector performance, so that the overall productivity gap is slightly smaller than the business sector productivity gap.  This productivity advantage was achieved despite a dramatically lower use of M&E by the Canadian non-business sector (which StatsCan pegs at only a third of U.S. equipment-intensity — a finding that I find surprising and questonable).

The StatsCan report pins the main blame for Canada’s poor productivity performance on what it calls “multifactor productivity.”  They thus commit a strange but increasingly common error in the productivity literature.

 

 

 

 

 

 

 

The report explicitly says Canadian productivity is lower “because of lower multifactor productivity,” which makes it sound like MFP is an independent, measurable causal factor.  In fact, however, MFP is an econometric artifact of a particular method of attempting to disaggregate productivity into its traditional factor-driven “causes”.  More specifically, it is the residual left over in a regression of productivity on the stocks of capital and labour (sometimes, not always, adjusting the labour input for skill and education levels).

The fewer explanatory variables you consider, or the less accurate is the specification of the relationship among them, the “more important” will this MFP residual seem to be.  The StatsCan paper only considers a gross measure of capital intensity (not specific types of capital or labour quality) – so their MFP residual seems especially large, thus their blanket conclusion that it’s Canada’s MFP performance (whatever the heck that is) that’s at fault.  This is silly.  Saying that “Canadian productivity is lower mostly because of lower MFP” is utterly equivalent to saying “Canadian productivity is lower and we don’t really know why”!

In my view the MFP approach misses broad complementarities and structural changes that are associated with M&E investment, which I would view as being even more important to productivity growth than the traditional Addison-style growth-accounting regressions make it seem.  This view was enunciated by my old professor Edward J. Nell in his work on transformational growth (in the Polanyi tradition): new capital investment is essential to incorporate changes in product, process, and work organization that are not at all captured in a simple quantitative measurement of the amount of capital in use at any point in time.  (The best reference:  The general theory of transformational growth : Keynes after Sraffa By: Nell, Edward J. Cambridge ; New York : Cambridge University Press, 1998, 784 pp.)

 

At any rate, the more evidence comes out, the more it is clear that Canada’s status as a leading developed economy is very much at risk as our level of productivity and innovation fades, and as we become ever-more-dependent on the extraction and export of the stuff that happens to be buried beneath our feet.  Most damning of all, the erosion of Canada’s relative productivity performance (as demonstrated most convincingly by Chart 3A in the CSLS package mentioned above) coincides precisely with the arrival of hard-edged neoliberalism within our fair borders.  Until the early 1980s Canada was rapidly catching-up to the rest of the industrialized world — thanks (in my view) to deliberate, interventionist policies.  Since then, in a context of hands-off free trade, privatization, and deregulation, we’ve been fading fast.

4 comments

  • MFP. don’t even get me started on that acronym! as you would have to censor this.

    As you mention productivity, when aggregating to this level can get quite lost in what is meaningful and what is not.

    I have found over the years that many attempts at measuring such constructs simplify to a point that the underlying heterogeneity is restructured into some nice fitting packages that do nothing but mislead the results.
    For example working on the workplace and employee survey data that I did for several years, attempting to cluster many of the different shop floor dimensions of machinery and equipment was difficult within the same NAICS code. At this micro level, the qualitative can barely be resuscitated. At higher levels, I find that one is easily tempted to start agreggating everything into one nice fitting cookie cutter. It is a fundemental flaw in the data quantification process. And then to try and attempt this across nations. I must conclude that one must stand back and have a good long hard look at a project and the data before even thinking about such exercises.

    I know John B. has done some great work in the innovation circles. However I feel this paper is not his best work and it is precisely the idea that one can achieve all the niceties that he tries to with the data. I find it quite a stretch to push the data that far.

    Any data on productivity research is best performed at a closer to the shop floor, more homogeneous, micro level. Otherwise I find one is basically just putting up numbers in the political debate about how productive we are at the macro level and how meaningless it all seems at the end of the day. The analysis at this level is pretty much lost in translation. I am surprised that John tackled such an issue at this height.

    paul

    pat

    although his group has done some good work on innovation and such, however when it comes to concrete measurement

  • I’ll have to take a closer look but the non-business sector productivity numbers are curious. Generally, productivity in government is just wages and salaries paid, so there is no “real” sense of the contribution of the public sector with regard to productivity (an important shortcoming of Ottawa’s obsession with productivity. So it could just be the Canadian average compensation is 6% higher than in the US.

    The other thing that bears repeating is that most detailed studies of Canada-US business productivity find that the gap is due to a couple sectors in which Canada has a much smaller share of the population and economy than the US. For most sectors, there is not much difference compared straight up. That reinforces Jim’s point that it is industrial structure that matters most for the bottom line productivity numbers.

  • How does productivity in the Health Sector compare between Canada and the US? It should be much better in Canada because we get better health outcomes and spend far less then the US. Is this reflected in the productivity numbers?

  • I wouldn’t be too worried about this since Republicans took office in 2001. Likewise if they win again. Republicans rack up about $500 billion in federal debt per year (more if you subtract B.Clinton’s surpluses). The annual interest service charges on their debt is just under 3% of their GDP. In contrast Canada, both Conservatives (at least until their latest GST cut) and Liberals, have paid down about $100 billion(?) over the last 8 years. That’s just under 1% of GDP annually saved.
    Productivity isn’t directly equatible with GDP, but going bankrupt with productivity gains as a side effect isn’t a good Canadian measuring stick. We should use Northern Europe as our benchmark.
    The real productivity gain would be investment in capital; new machines and stuff. S.Harper’s accelerated capital depreciation tax write-off is good, but I’d think P.Martin’s free 1st and last year of university would’ve been just as good.

Leave a Reply

Your email address will not be published.