Productivity and Pay

This is a very nice piece from Dean Baker of the CEPR on the delinking of real worker pay from labour productivity growth in the US.

I’ve argued for years, with much of the left, that average worker pay has lagged productivity growth mainly because of the increased bargaining power of capital vis a vis labour due to “globalization”, attacks on unions etc etc. That story is true, and capital’s share has risen sharply in Canada as across all OECD countries, but things are also a good deal more complicated.

I crunched the Canadian numbers a few years ago and found more or less what Dean documents for the US:

For a start, productivity growth (real output per hour) is exaggerated, in that capital depreciation rates have increased in the computer/infotech age. Growth of net output per hour or “usable” productivity growth thus lags growth of gross output per hour. (When I last checked this was broadly true of Canada as well.)
Second, real wages (adjusted for consumer price inflation) lag the growth of real output per hour, since consumer prices have been rising faster than economy-wide prices in recent years. Again computers play a major role since their real price, and thus the real price of investment goods, have been declining. (I think in the case of Canada there is also an important potential discrepancy between consumer prices and the GDP deflator since resources loom high in GDP but are largely absent from personal consumption.)

Third, hourly wage growth has lagged total labour compensation growth, since compensation has become more tilted to benefits (especially health benefits.) (This I’ve checked recently, and is also true of Canada though probably to a lesser extent than in the US. EI and CPP premium increases a few years back also squeezed the money wage portion of labour compensation.)

Dean’s overall conclusion is well-taken – yes, labour’s share has been squeezed by returns to capital, but this does not explain a large chunk of the gap between pay and productivity growth.

Finally, as Dean reminds us, the widely-noted trend in average real pay relative to productivity does not capture a key trend of concern – the shift of labour compensation to the very high end of the wage distribution.

I encourage anyone with the time and inclination to reproduce Dean’s study for Canada


  • For an account of these trends in Canadian manufacturing See:”Globalization and Wage Inequality in the Canadian Manufacturing Sector: A Time Series Analysis.” Gilles Grenier Akbar Tavakoli†.

    I have a series of slides on these issues for Canada (excluding the decompression of hourly pay).

  • Dean’s overall conclusion is well-taken – yes, labour’s share has been squeezed by returns to capital, but this does not explain a large chunk of the gap between pay and productivity growth.

    I blogged about this awhile ago, and a plausible explanation might be the reduction in the US labour terms of trade: the ratio or producer to consumer prices.

    Real wages and the terms of trade: Productivity growth is little help if you’re making stuff no-one wants to buy

  • I’ve just done the same exercise for Canada

    If you use the output deflator, you get a real wage series that tracks output per hour worked not too badly. But if you use the CPI, you see no increase in buying power between 1981 and 1996.

  • Stephen, there are a couple of issues here. First if you want to track workers consumption you have to use the CPI. Workers buy wage goods not producer goods to reproduce themselves. Capitalist of course buy both because they buy capital and labour power. So a better metric would be to take a weight of capital investment and consumption along with the a weight of the labour input and then derive a sectors price index from there. That could then be compared to the purchasing power of workers wages.

    Second, an even better way to get at wage inequality is to use value added series such as I have done and look at movements of employment and in income shares of value added over time for any given sector of the economy. To my mind this provides much more information about the dynamics on the ground so to speak.

  • The point I’m trying to make is that firms and workers are looking at different prices when they calculate the measure of real wage that interests them. When you calculate real wages as firms would see them – that is, using output prices – you get a real wage series that more or less tracks productivity. This means that the standard story we tell about how firms set wages (real wage = marginal product of labour) is actually holding up pretty well: there isn’t any significant lag or gap between real wages and productivity.

    But it doesn’t mean that workers will see the same thing, of course. What happens to their buying power will depend on the labour terms of trade.

  • Not to throw the baby out with the bathwater, Stephen, Baker still sees a significant shift of productivity-generated income from labour to capital. But, yes, the CPI is the right measure to deflate wages and, yes, we should not deflate real output by the CPI. The difference in the deflators does explain part of the apparent shift.

  • Stephen did you really just write this: “real wage = marginal product of labour” ?? Play nice or we going to have a holy war.

Leave a Reply

Your email address will not be published. Required fields are marked *