Charles St-Arnaud: The K-shaped economy is real and labour is on the wrong end of it

The U.S. economy has been growing strongly since the pandemic, outpacing most developed economies. However, this outperformance masks conflicting trends across economic groups, a phenomenon dubbed the K-shape economy, in which some have benefited significantly from the economic expansion, while others have struggled and even fallen behind.

This trend has been in the making for decades. Economic growth is increasingly attributed to capital rather than

labour , a shift that has been underway for decades and is now being reinforced by

artificial intelligence . While GDP measures the value of what is produced in an economy, how that value is distributed matters, as expressed by GDP by income. In the U.S., labour’s share of economic output has declined by roughly seven percentage points since 1980, while the share belonging to corporate profits has nearly doubled.

Several structural forces explain this trend. The erosion of

unions and the expansion of outsourcing have weakened workers’ bargaining power, while automation has reduced the need for human labour, first in manufacturing and increasingly in services. At the same time, the composition of capital has changed. Investment has shifted away from labour-intensive factories toward software, intellectual property and platforms that scale rapidly without large workforces. As a result, today’s dominant firms are highly profitable but employ relatively few workers, even as revenues expand.

While labour shortages during the pandemic briefly reversed these dynamics, driving up wages, higher inflation absorbed much of the gains, leaving the purchasing power of many unchanged. Profits, on the other hand, quickly rebounded to record levels. Rising equity valuations have further amplified the role of wealth in driving consumption, especially among higher‑income households that own most financial assets. As a result, consumer spending has become more sensitive to market movements than to job or income growth, and has been driven mainly by higher earners. Artificial intelligence could accelerate these trends, as task automation and AI agents increasingly substitute for workers, tilting the distribution of income further toward capital and away from labour.

After looking at the situation south of the border, many are wondering whether Canada is also on the cusp of seeing a shift in income away from labour towards capital. The data in Canada does not indicate a shift similar to that seen in the U.S. in recent years. Yes, it indicates a reduction in the share of income going to labour and an increase in the share going to corporate profits, but most of the changes occurred in the 1990s. As such, between 1980 and 1993, about 53 per cent of national income went to labour; it is now about 50 per cent. Similarly, the share of income going to corporate profits was about 23 per cent in the 1980s, and it is now closer to 28 per cent.

In short, the Canadian economy is not seeing a significant shift of income away from labour to corporate profits in recent years, like the U.S. is experiencing. This likely reflects that the Canadian economy has a much smaller tech and AI industry than south of the border, and, as a result, these sectors have a much smaller influence on the broader economy. However, this does not mean that the Canadian economy has been spared inequality of outcomes.

One can look at the aggregate data on household disposable income for some hints of divergence. I have been tracking real disposable income per person, adjusted for inflation, to assess changes in household purchasing power over recent years. An interesting pattern has emerged in the post-pandemic recovery: while disposable income per person, adjusted for inflation, has increased, albeit only modestly, compensation of employees per person, adjusted for inflation (i.e., wages and salaries), has stagnated. This suggests that households dependent on labour income as their main source of income have seen their purchasing power stagnate in recent years.

Looking more closely at the data, the evidence suggests that lower-income households have seen their nominal disposable income stall in recent years, meaning a loss of purchasing power once inflation is considered, mainly because compensation has stalled. On the flip side, wealthier cohorts have seen their disposable income increase over the same period. But interestingly, most of the increase in disposable income comes from robust increases in compensation, supplemented by an increase in net property income or income from assets (i.e. investment income) minus payments on liabilities (i.e. debt repayment).

This result suggests that it is not just the fact that higher-income cohorts usually hold more assets with high returns and also have a lower level of debt that explains their higher income, but also that they are employed in sectors where compensation has increased at a much faster pace than for lower-income cohorts. Property income for the lower-income bracket increased over the period, but not as much as for the higher-income. Asset composition likely matters here, as higher-income earners are more likely to own stocks and higher-yielding assets than lower-income earners, who rely more on fixed-income assets.

This has an implication for the wealth of these various cohorts. In recent years, the net worth of lower-income sectors has been declining slightly, as the loss of purchasing power due to stagnating disposable income is offset by increased borrowing. Conversely, higher-income earners are seeing an increase in net worth as the value of their assets rises with rising equity valuations.

However, unlike in the U.S., there doesn’t seem to be any significant impact of income and wealth divergence on the consumption patterns of these cohorts. If anything, despite stagnating disposable income, the lower-income group has seen the biggest increase in consumption across all cohorts, while higher earners have seen a more modest increase in expenditure. Nevertheless, this is a clear sign that the squeeze on purchasing power over recent years likely affected lower-earning households more acutely and provides some evidence of the affordability crisis many households are facing.

Looking ahead, the pressures that AI adoption will generate on the economy are unlikely to change the situation. With new technology displacing workers in favour of capital, wages are likely to continue to stagnate. However, the displacement caused by AI could disproportionately affect white-collar workers, rather than blue-collar workers. Ultimately, we could see the share of income flowing to corporations and the holders of capital (i.e. shareholders), increase, benefitting those holding the assets.