A phone discussion with a reporter this week on trends in Canadian investment and capital formation piqued my curiosity as to what the long-term trends in Canadian gross fixed capital formation have been. Apparently, despite the global financial crisis and associated economic uncertainty, business investment and capital formation is still relatively strong at the moment in Canada (though Statistics Canada did report yesterday that business plant and equipment fell third quarter after six quarterly increases). This recent growth is a function of a number of factors including low interest rates, resource development investment in the oil sands and mining sector, and residential, retail and office construction in the larger urban centres of the country. The question was raised during my discussion as to how this current boom or upsurge compared to past booms. So, naturally, I decided to go to the data and see what insights I could get.
Of course, data availability to put together a consistent series is an important concern but I was able to put together three separate though related data sources. For the period 1870 to 1926, there is the national accounts data put together by Mac Urquhart at Queen’s which provides GNP estimates for Canada as well as an estimate of gross fixed capital formation. For the period 1926 to 1960 there is Historical Statistics of Canada (also reliant on Mac Urquhart), which provides GNP and an estimate of gross fixed capital formation. Finally, for the period 1961 to 2010, I was able to get annual series from Statistics Canada for GDP and gross fixed capital formation. Note that the estimate of gross fixed capital formation is for both business and government.
As the measure of investment activity I have calculated the ratio of gross fixed capital formation to output (GNP/GDP) for Canada for the period 1870 to 2010. The share of GDP going to capital formation is but one indicator of investment activity but it seems like a reasonable one to use. Investment spending is a volatile component of national output and certainly more variable than national output so fluctuations in the ratio should reflect fluctuations in investment rather than GDP changing faster than investment. The mean growth rate of nominal GNP/GDP from 1870 to 2010 was 6.4 percent and the standard deviation 7.8, whereas the mean for nominal gross fixed capital formation was 0.9 percent and the standard deviation 11.0.
The accompanying Figure I plots the investment-output ratio and it shows the most spectacular investment boom in Canadian history was the period from 1896 to 1912 when the investment-output ratio rises from 0.115 to reach a peak of 0.341. This was of course the wheat boom/western settlement period that saw massive capital infrastructure put in place for the western wheat economy. The ratio collapses during the First World War and recovers during the 1920s but modestly before collapsing during the Great Depression to a low of 0.096 in 1934. There is a slight recovery but then the ratio drops during the Second World War to the lowest it has ever been at 0.081. The period 1944 to 1957 sees a steadily rising investment-output ratio –the post-war boom era - and since the 1950s the ratio, while fluctuating, has never since dropped to the depths of the 1930s or the late nineteenth century. No doubt, interventionist government economic policy did much to stabilize fluctuations in the investment-output ratio and the economy in general in the period since World War II.
Figure I
However, the investment-output ratio has never since reached the height achieved during the post-war boom – the post-war peak of 0.259 achieved in 1957. Indeed, the long-term trend since 1957 seems to be a declining investment output ratio despite the period of increase since the late-1990s. I’ve highlighted this in a separate graph with a linear trend (Figure II) for the 1960 to 2010 period. Is the long-term slide in economic growth and productivity since the 1970s simply rooted in the decline in gross fixed capital formation’s share of GDP? Have we simply been investing less in Canada’s economy? Maybe a giant national infrastructure program is what is needed.
Figure II
Thanks for doing this Livio. I didn't know what that graph would look like.
I'm a little surprised at how stable the ratio has been since 1950. I thought it would be more volatile. I expect most of the investment volatility has been in housing. (Investment = fixed capital formation + housing + inventory investment, right?)
Posted by: Nick Rowe | December 01, 2011 at 12:52 PM
Nick:
I do not believe Inventory investment is included in these estimate of gross fixed capital formation - simply residential and non-residential construction, machinery and equipment.
Posted by: Livio Di Matteo | December 01, 2011 at 01:06 PM
Livio: so residential investment is included in your data? Then I'm even more surprised it doesn't fluctuate more.
Posted by: Nick Rowe | December 01, 2011 at 01:44 PM
Question #1:
I'm a little confused as to why a declining ratio is a bad thing.
I invest $100 every year and produce a GDP of $1,000 every year. Ratio is constant at 10%
I invest $100, $90, $80, and produce a GDP of $1000 every year. Ratio of investment/GDP is falling. Good, right?
I invest $100, $110, $120 and produce a GDP of $1000 every year. Ratio is rising. Bad, no?
Question #2: Is this private investment or all investment? In the U.S. data, it makes a big difference, because rapidly falling government investment + constant GDP (which doesn't really do a good job of measuring government output or subtracting wear and tear of public assets from GDP) makes for a falling ratio.
The economy only *seems* to be getting more efficient, but really the infrastructure is just aging and that bill will come due at some time.
Posted by: rsj | December 01, 2011 at 08:21 PM
RSJ:
1. Devoting a larger share of current GDP to capital formation should raise the capital labour ratio and lead to higher future economic growth-which is why a declining ratio is a cause for concern. 2. This investment is capital formation by both government and the private sector.
Posted by: Livio Di Matteo | December 01, 2011 at 09:39 PM
One related question I have is why it doesn't appear that there is a long term relationship between the corporate income tax rate and corporate investment in infrastructure: http://www.theglobeandmail.com/report-on-business/economy/economy-lab/the-economists/five-reasons-to-say-no-to-more-corporate-tax-cuts/article1886449/
Posted by: slantendicular | December 02, 2011 at 12:55 PM
Oh god. Not this meme again.
Posted by: Stephen Gordon | December 02, 2011 at 01:16 PM
Thanks for the link to that post Stephen.
Posted by: Livio Di Matteo | December 02, 2011 at 02:07 PM
I’d like to see more of this type of analysis of investment in the econoblogs in general.
A tangential, more global question for you or anybody here: a few years ago when the “global savings glut” was more topical, I recall seeing a global number for saving of about 22 per cent of global GDP, which had also been fairly stable for years. The same number would have to apply to global investment. Looking at your chart, it seems like this is close to some kind of magic level in general.
As I understand it, the savings glut argument was supposed to be based on some sort of ex ante analysis of excess desired saving, etc. Given the apparent stability of investment at around that 20 or 22 per cent number, and the ex post equivalence of investment and saving, just wondered how you interpreted that global savings glut thesis? Depending on your interpretation, was there any sense in which Canada has been affected by it in a material way? More simply put, what do you think the relevance of that thesis is for investment in Canada or globally?
Posted by: JKH | December 02, 2011 at 02:50 PM
Am I alone in thinking that there appears to be a relatively strong relationship between the capital ratio and the value of the dollar vis-a-vis the greenback (high in the late 1980's, dropping off in the 1990s and rising again over the course of the 2000's, with a sharp increase after 2005. To the extent that Canada imports machinery from abroad, but pays wages in Canadian dollars, you could tell a story about a falling dollar making capital relatively more expensive and vice versa.
Of course, the relationship might run the other way, demand for oil both pushes up the dollar and creates increased demand for capital.
Posted by: Bob Smith | December 02, 2011 at 06:04 PM
Would a giant national infrastructure program be bigger than the Build Canada Fund?
Posted by: Ian Brodie | December 05, 2011 at 05:07 PM
Yes, by several orders of magnitude.
To put it in perspective, imagine scrapping the entire Royal Canadian Navy right now and rebuilding it from scratch with thirty surface ships and five submarines; all built entirely in Canada without sparing on capability.
For starters.
That's what a giant national infrastructure program looks like like.
Posted by: Determinant | December 05, 2011 at 05:39 PM
This analysis seems very strange to me. Don't we think we're missing (at least) two factors of first-order importance (or is it just me again?)
1) Population growth. 1960-2010 saw important variations in the rate of labour force growth. Last time I taught macro, that had implications for the rate of desired capital investment.
2) Oil. I admit I have a one-track mind, but the dip in investment rates from the late '80s and the pick-up recently coincide with major sustained movements in oil prices. Would we think that changes in world energy prices should change the optimal level of investment in a country with a large energy endowment?
Posted by: Simon van Norden | December 06, 2011 at 09:00 AM
Am I right in thinking that these numbers show investment/output falling for the first half of WWII? Hard to believe that reflects a fall in total investment....just a surge in output perhaps?
Posted by: Simon van Norden | December 06, 2011 at 09:04 AM
Hi Simon:
The analysis is simply a plot of the investment-output ratio over time to see what the trends and patterns are. There does seem to be a declining trend over the last forty years or so. It also seems to be much more stable relative to the pre-war period. If output rises faster than investment -as is the case during World War II - then the ratio can certainly decline though output rising faster than investment seems to me to be at best short term phenomenon. Are there factors other than a time trend that can explain fluctuations in the investment-output ratio such as population growth, oil price shocks, interest rates,etc...I'm sure there are. Have I run a regression to assess the contribution of these factors? No, I have not. If you can send me oil price and interest rate data for 1870 to 2010 (I already have population) I'd be happy to give it a go. Cheers. Livio.
Posted by: Livio Di Matteo | December 06, 2011 at 10:17 AM
"....output rising faster than investment seems to me to be at best short term phenomenon."
Odd....suppose we see a permanent shift in the relative price of capital and labour...in a small open economy like Canada's, wouldn't we expect to see a shift away from capital towards labour....which could imply what for the investment/output ratio?
Sorry, Livio, I guess I'm thinking that it is a bit early in the day to start thinking about a giant national infrastructure program to cure the problem when I'm not sure one exists. OTOH, if you want to sound alarm bells about the need for urgent action to cure an investment crisis that threatens our standard of living, I'm sure you won't be alone. But I'll pass for now.
BTW, your oil price data can be found at http://chartsbin.com/view/oau
Posted by: Simon van Norden | December 06, 2011 at 10:38 AM
Guess I should have said I was assuming a rise in the price of capital relative to labour.
Posted by: Simon van Norden | December 06, 2011 at 10:39 AM
Thanks for the oil price data link Simon. I was thinking. If oil prices go up, there should be an increase in the demand for our oil which should lead to investment in extraction. Unless I'm mistaken, oil production is fairly capital intensive so should that not spark an increase in capital formation and the investment-output ratio over the 1970-2010 period given that oil prices have trended up? This could also raise the demand for capital relative to labour and raise the price of capital relative to labour which would then act as a drag on capital formation?
Posted by: Livio Di Matteo | December 06, 2011 at 11:29 AM
I agree with your analysis of what an oil price shock should do (remembering that we're really talking about an energy shock; the energy sector is much bigger but given the substitution across oil/gas/coal/hydro we see high correlations between oil price movements and aggregate energy prices.)
However, the 1970-2010 was pretty far from an overall upward trend. (Did you look at the chart that I gave you the link for?) We saw real price spikes in 1970s, followed by a collapse by the mid 1980s (Iran-Iraq war did bad things for OPEC cohesion.) In the mid-1990s, the real price of oil was about as cheap as it was in the 1950s. (I've always blamed that for the rise in popularity of SUVs.) Then we got serious increases only after the 2001 recession.
Posted by: Simon van Norden | December 06, 2011 at 05:32 PM
Good points Simon. I 've put together the data with interest rates, population growth and oil price changes and will be running a regression and doing another post.
Posted by: Livio Di Matteo | December 06, 2011 at 05:46 PM