Why do we care about the labour share of income?

And by 'we', I mean 'Canadians'.

A lot has been said and written about the decline in the labour share of income, usually calculated as total employee compensation divided by nominal GDP. This decline is generally regarded as a negative development: the reduction in the share of income going to workers is interpreted as a symptom of suppressed wage growth and of increased income inequality.

I don't doubt that this is a useful narrative for understanding what has been happening in many countries, the US in particular. But I can't see how it fits the Canadian experience. Movements in the Canadian ratio of wages to national income  appear to be a story of the denominator, not the numerator.

Here is the labour share from 1947 to 2018; the data are taken from the most recent update of Project Link:

 

If we were going to interpret the labour share as an indicator of workers' collective bargaining power, the story would go something like this. Workers' bargaining power increased steadily during the thirty years after the Second World War, attained its peak during 1975-1995 (with substantial fluctuations), and then has declined.

The problem with this story is that it would have us believe that the period 1975-1995 was some sort of Golden Age for Labour, when in fact the exact opposite was true. Those peaks in the mid-1970s, the early 1980s and again in the early 1990s weren't periods where workers' power to extract wage gains was its strongest; they are all associated with recessions (you can see another spike in 2009). Capital income is far more procycical than labour income, so the percentage fall in GDP is greater than the percentage fall in labour income during recessions. Since the numerator is falling less quickly than the denominator, the labour share becomes countercyclical. Those spikes are an artifact of the cyclical properties of the components of total income, not a sign of workers' bargaining power. I will have difficulty believing that workers' bargaining power increases when unemployment rates go from 7% to 13%.

Now let's set aside the recessions and look at the generally high labour shares that prevailed over this period. Were higher labour shares associated with stronger wage growth over 1975-1995?

No, they weren't:

Canada did have an extended period of stagnant wages, but it occurred between 1975 and 1995, when the labour share fluctuated around its post-war highs. The resumption of real wage growth in the past 20 years coincided with the decline in the labour share.

What about income inequality? Once again, a story in which a high labour share acted as a force for keeping income inequality in check doesn't fit the data:

The Gini coefficient jumps in the first part of that sample, and only start to level off when the labour share started to decline in the mid-1990s. Statistics Canada's Gini coefficients only start in 1976, but there's good reason to believe that income inequality had been stable before then.

Finally, what about the steady increase in the labour share during the post-war boom? You could tell a story here of how labour's increasing bargaining power resulted in sustained real wage growth, but then you'd have to explain why this stopped being the case over 1975-1995.

But there's a much simpler explanation for why total employee compensation grew more quickly than GDP during the post-war expansion:

The increase in female participation rates meant that the number of people earning wages grew much faster than the population. If Canada's production function were literally of the Cobb-Douglas form, the expansion of the labour force would have pushed down wages in order to keep the labour share constant. But if labour demand is more elastic than what the Cobb-Douglas model predicts, then total employee compensation would increase: the downward pressure on wages would be more than offset by the increase in the number of workers.

The years in which the Canada's labour share was its highest were years in which unemployment was high, wage growth was nonexistent and income inequality was increasing. The decline in the labour share since 1995 coincided with a resumption in real wage growth, a stabilisation of income inequality, and generally lower unemployment rates.

So what's so great about high labour shares? 

 

Comments

[Jussi]

Really interesting, thanks. How is it in other countries, are their data similar?

[Nick Rowe]

Good post.
Can't help answering: "Because economists used to insist on writing about the Functional Distribution of Income rather than the Personal Distribution of Income, thinking the first told you what you needed to know about the second".

[Frances Woolley]

Steve,

Population aging is steadily increasing the number of people supported by income from capital, as opposed to income from labour (think of all of those people relying on their pensions). If the labour share doesn't fall, what are you and I going to live on if/when we retire?

[Stephen Gordon]

Dang. I meant to add that point and forgot.

[KIEN CHOONG]

Hi, a falling share of income going to labour ought to prompt public discussion on the reasons for this trend. But like so many things in life, a high share of labour income is not something that we value per se.

An issue I would raise is that it is sometimes not easy to separate income into labour vs capital. Warren Buffet notionally earns a lot of income from capital but he also says that his job is to read all the time, and occasionally take action.

I suspect that going forward in the future, with AI and machine learning, it is likely more of our income will come from investments in capital, informed by continuous learning and independent thinking.

[Bernard Leikind]

Prof. Gordon, (Am I correct that this post is not by Prof. Rowe?)

I can well imagine that Canadians are not feeling well disposed toward Americans these days, what with our president declaring you a national security threat and planning to force PM Trudeau into concessions while offering none in return.

Nevertheless, would you be willing to tell us what a similar analysis of the US labor market and labor's income shows?
I thought your explanation and reasoning was greatly aided by the graphs.

Bernard Leikind
Tampa, FL

[Pierre Fortin]

Steve. I always wonder why people divide by GDP instead of net domestic income (GDP less Cap consumption less indirect taxes) and avoid adding the wage component of "Mixed income". Further, since the overall compensation of employees includes the highly-paid 10%, not sure variations in labour share (in addition to your own remarks) is a good indicator of changes in income inequality. I think Andrew Sharpe's et alii in their 2016 report of 2016 is a good way to approach the inequality question. Pierre

[mulp]

Canada is basically in a balance trade position.

So, how is 50% of GDP paid for when workers get paid only 50% of the price of GDP?

Do capitalists, say 10% of the population, personally consume 50% of GDP? Ie, about 10 times as much as worker. Do capitalists eat 5-10 times as many burgers and fries as workers?

Or is 50% of GDP going to government to pay for consumption by workers? Ie, government takes 10-12% of GDP in kind, as health care production to provide health care for free to workers and their families, which is mostly timeshifting, ie, workers work in midlife to provide health care for birth and childhood and then later in life when no longer working, to themselves. Ditto for steel production in bridges and school buildings, etc.

50% of work is not building capital assets for consumption or of growth, unless assets like bridges are extremely low quality, falling down after only a few years.

So, if 50% of work is done for government, does Canada's governments own all the capital, or take in taxes all profits/rents plus returns to invested capital (distinct things to Keynes)?

I'm a throw back who believes zero sum is not only a good thing, but the law. Costs must equal price, wages paid must equal consumption, thus higher standard of living cost means higher income to pay the costs of consumption. Tanstaafl

Post WWII, I believe Canada like the US was repaying the cost off high national consumption (war) that consumed future production in debt, plus consuming capital, PLUS building new capital to grow productive capital for the population boom and higher worker consumption. To increase worker consumption, in contrast to national consumption, workers must be paid more as a share of gdp. As working to benefit other discourages work, working simply to increase national consumption, eg more wars, is not pursued with enthusiasm to increase gdp rapidly. Thus, increasing worker income with policy like higher minimum wages to increase consumer demand driving up gdp.

Maybe the shift was to have capitalists keep more of GDP, then lend money to workers exponentially increasing debt to fund consumer demand as if workers will keep producing after death to repay debts after they stop consuming. Or we can view bankruptcy debt write down as deferred wages/compensation. Except, consumers going deepest in debt are paid for working hard to consume GDP far beyond their means.

I'm just trying to get to the zero sum. The total price of GDP must equal the incomes of the buyers of GDP. Selling $100 of goods produced for $50 adds only $50 to GDP. If not, gdp can be doubled every year by pricing the same production at twicee the price and then exponentially discounting the price to sell it to consumers with static incomes.

[robots2005 AI32080]

The Unions got higher wages at the wrong times in the 80's and 90's. The Steel Unions got their wages and quickly saw losses posted. The car industry always had high wages but assembly work moved to the USA south and Mexico, etc. That is part of the employment falls.
I am interested for two reasons. It might be the easiest way to an ethical economy. A past French leader was looking for the next big thing while chasing women, and missed the France NDPer Hamon figure a robot tax. This is the easiest strategy to begin to prevent factories from being hacked by AI. Staffed by people, you can turn off the machines and if necessary break the back-up power source.
Money seems to be part of the reason people fund WMDs. The insurance industry isn't to worried about hacking or AI now. But if there are fewer rich CEO's and Bankers, I'd expect this to change. Worker's share represents marshmallows getting less syrup.

[RebelEconomist]

A good, thought-provoking post, thanks. I dropped by to re-read an old Nick Rowe post on functional finance and saw this.