Employment and Investment: the great Canadian disconnect

[This is a guest post by Hashmat Khan of Carleton University and Nyamekye Asare of the University of Ottawa.]

Can policies stimulating private investment deliver higher employment? Maybe, but investment and employment have become disconnected recently in Canada.

John Taylor has noted a strong negative correlation between investment and the unemployment rate, and argued that higher investment is the best way to reduce unemployment. Paul Krugman has questioned the direction of this causality, and the Canadian evidence seems to be somewhat supportive of his position.

The issue of causality aside, a reduction in the unemployment rate does not necessarily lead to an increase in employment, because the unemployed may stop looking for work and leave the labour force. For this reason, economists pay attention to the employment-to-population ratio as a better gauge of labour market conditions (see for example this post by David Andolfatto, and this Maclean’s Econowatch piece by Stephen Gordon). In this post, therefore, we highlight the relationship between investment and the employment-to-population ratio for Canada.

Starting with the financial crisis period: between the first quarter of 2008 and the last quarter of 2010, the employment-to-population ratio in Canada fell by 2 percentage  points (63.7% to 61.7%) and, since then, it has remained approximately flat (See Figure 1). Meanwhile, the business fixed investment relative to GDP ratio fell by 2 percentage points (19.5% to 17.5%) from 2008 to 2009, but by the fourth quarter of 2013 it was back to the pre-financial crisis level of 19.5%.

It is well known that the Canadian economy weathered the financial crisis of 2008-09 better than the US and European countries, and also in comparison to our own recent history. The decreases in the employment ratio during the recessions of early 1980s and early 1990s were twice as large. The investment ratio also fell by 4 percentage points in the early 1980s, and by slightly over 2 percentage points in the early 1990s.

Although the two ratios tend to move together during downturns in Canadian economic activity, the interesting feature to note is their joint movement during the recovery phase. In particular, starting from the mid-1990s onwards, the two ratios appear to be more disconnected relative to the late 1980s period.

Figure 1:  Data  source:  Statistics  Canada  CANSIM Tables  282-0087 and  380-0064. Employment is number of civilians/non-institutionalized people for the reference  week that either worked for pay/profit or had a job but was away from work due to illness, disabilities, or other  reasons. Population for Canada  is 15+.  I /GDP is computed as the ratio of nominal business fixed investment and nominal chain-weighted GDP. All data are seasonally adjusted.

Figure 2 reports the 10-year rolling correlation between the two ratios, starting in 1981. Between 1995 and 2006, the correlation fell from 0.9 to under 0.4, it then rose back up to 0.8 but from 2010 onwards it has steadily declined. At the  moment, the correlation is at its lowest point of about 0.3, suggesting that the disconnect between the employment and the investment ratios is currently the greatest. To what extent is this disconnect a hindrance for economic policies is an open question at the moment.

Figure 2:  10-year rolling correlations.

[The above is by Hashmat Khan of Carleton University and Nyamekye Asare of the University of Ottawa.]

Comments

[Nick Rowe]

Hmmm. My (tentative) interpretation of Figure 2:

When monetary policy is good, and the economy is stable, we tend to see a low correlation between investment and employment (the composition of NGDP changes, but NGDP and employment grow smoothly).

When monetary policy is bad, we tend to see a high correlation between investment and employment. Monetary policy shocks cause shocks to all components of NGDP in the same direction, and to employment.

[Bob Smith]

Holding relative prcies constant, you'd expect there to be a close correlation between investment and employment. Presumably part of what is going on here is changes in the relative price of labour and capital inputs arising from changing values of the C$ over time. All else being equal, a low dollar would tend to make imported capital goods more expensive and domestic labour cheaper. Faced with that phenomenon, no wonder Canadian employers might choose to hire more workers and invest less in capital.

[Majromax]

> Faced with that phenomenon, no wonder Canadian employers might choose to hire more workers and invest less in capital.

Does that actually make sense with the data? The most recent period of weakest correlation has had the strongest Canadian dollar, and the Canadian dollar was much lower during the 80s and early 90s -- when the correlations were strongest.

[Bob Smith]

Majromax.

Fair point, but I wonder if the issue isn't changes in relative price of capital and labour. With a steady relative price, one might expect the capital/labour ratio to stabilize. Any increase in production requries a corresponding increase in both capital and labour. On the other hand,if the relative price changes, firms will have a different equilibrium capital/labour ratio. In the short run, they might choose to increase production by hiring labour and not capital (or vice versa).

Also, the figure 2 is a bit misleading, because it's a 10-year rolling average correlation (so the high correlation in 2010 isn't neccesarily linked to the high dollar). I'm not sure so the high correlation in 2010, actually reflects the correlation over the prior decade. (Is it me? I'm not at all sure how to interpet that. How does a 10 year rolling average fluctuate so dramatically over 5 year spand from 2006-2011? Is a rolling average correlation even meaningful? A steady moderate correlation looks the same as a time series that fluctuates between zero and 100% correlation).

[Hashmat Khan]

Bob,

The 10-year rolling contemporaneous correlation reported in Figure 2 is helpful in seeing the information in Figure 1 in another way: how employment and investment have moved together over the medium run (a decade) in Canada - a stylized fact. Notice the joint pattern in Figure 1 since about 1995. The shape of the 10-year correlation reflects that. Also, the decrease correlation in the last 10 years is partly driven by the relatively flat employment-to-population ratio since 2010, and the rising I/GDP ratio. The increased contemporary disconnect reflected in the low correlation clearly striking and most likely reflects structural changes in the labour market.

[Ralph Musgrave]

The level of employment relies on the amount of investment? I doubt it. Suppose every investment in the country was destroyed: all houses were demolished, all machinery and factories destroyed. We’d then temporarily go back to the stone age. But why would that preclude full employment?

Everyone would be busy killing wild animals for meat, making fur coats from bearskins etc. And gradually with our knowledge of how to farm, make steel, etc etc, we’d get back to an industrialised economy.

[Jacques René Giguère]

Unfortunally not as all the easy to mine and process ores have all been used up...

[Keishaun D. Mark]

The disconnect could be a reflection of over-investment in production capacity for a commodity exporting economy, which then has to fully utilize said capacity over time before new (over) investments need to be made. The downward trend over time in the correlation however, could reflect the impact of technology in reducing the levels of labor needed to maintain capacity. How have wages responded to relationship between investment and employment?
-IJS