Change theme
Change theme

Total factor productivity growth projection for Canada: A sectoral approach

Key messages

We develop a sectoral framework—accounting for the impacts of digitalization, aging and climate change policies—to better understand the sources of total factor productivity (TFP) growth in Canada. We find the following:   

  • TFP is expected to grow by 0.5% on average in the 2020s, slightly below its growth in the previous decade (0.7%) and its historical average (0.6%) (Chart 1).
  • The services sector is expected to account for about two-thirds of TFP growth in the 2020s, replacing the manufacturing sector as the powerhouse of TFP growth (Chart 1, yellow bars).
  • Digitalization of the economy will continue to support TFP growth going forward, accounting for one-quarter of TFP growth in the 2020s (Chart 2, blue bars).
  • Population aging is not expected to put any downward pressure on TFP growth in the 2020s because the shares of young and older workers are projected to decline (Chart 2, yellow bars). Lower shares for these groups boost TFP growth because young and older workers are relatively less productive than prime-age workers.
  • The impacts of implemented and announced climate change policies are positive but small since they are captured only by reallocation across sectors (Chart 2, green bars).

Chart 1: Growth in total factor productivity is projected to slow in the 2020s

Chart 2: Digitalization and demographic shifts will support total factor productivity growth in the 2020s


Labour productivity growth is important for an economy in the long run because it supports living standards and growth in potential output and real wages. But Canada has experienced a long decline in labour productivity growth over the past three decades (Chart 3, panel a). This slowdown accelerated after the 2014 decline in commodity prices as investment decreased significantly in the resources sector and reduced its capital accumulation.1 While business investment is expected to rebound in the coming years (Bank of Canada 2024), this will not be enough to compensate for the “missing” capital accumulation since 2014.2 Going forward, therefore, labour productivity gains need to come from other sources.

This suggests that TFP growth may play a key role in supporting labour productivity growth. While TFP growth has declined steadily since the 1960s, it rebounded in the 2010s (Chart 3, panel b).3 In the 2010s, TFP growth accounted for about one-half of labour productivity growth (compared with slightly less than one-third between 1961 and 1999). The question is how much of this rebound will persist.

To assess the likely path of TFP growth until 2030, we develop a bottom-up sectoral approach that explicitly accounts for three important structural factors:4

  • digitalization
  • aging
  • climate change policies

Overall, our approach involves creating sectoral TFP projections that include impacts from these structural factors. We use G-Cubed, a multi-sector general equilibrium model, to assess the impacts of climate change policies and to project the sectoral shares needed to aggregate the sectoral TFP profiles (for an overview, see Box 1).5

Chart 3: Productivity growth in Canada has been declining

Chart 3: Productivity growth in Canada has been declining

Average annual growth

Sources: Statistics Canada and Bank of Canada calculations
Last observations: panel a, 2023; panel b, 2019

Three structural factors are of particular importance in the current economic context.

First, a growing literature suggests that digitalization (see, for example, Anderton, Botelho and Reimers 2023) and emerging technologies such as artificial intelligence (see, for example, Trammell and Korinek 2024; Comunale and Manera 2024) could boost economic growth. The acceleration of digitalization during the COVID-19 pandemic could offer significant growth opportunities for Canada, especially in areas where Canada lags its peers.6, 7

Second, the recent strong influx of newcomers is affecting the age composition of the Canadian population. Shifts in the composition could affect productivity because workers are more productive in their prime working age (see, for example, Maestas, Mullen and Powell 2023; Vandenbroucke 2021; Tang and MacLeod 2006).

Third, actions taken to mitigate the impacts of climate change will induce significant changes in Canada’s production systems (e.g., reallocation of resources) that could affect their efficiency (Parker 2023).

Box 1: Overview of the framework

Box 1: Overview of the framework

We use a bottom-up approach to estimate overall growth in total factor productivity (TFP) in the business sector. Figure 1-A shows the approach’s four main steps.

Figure 1-A: A visual representation of the sectoral framework

Figure 1-A: A visual representation of the sectoral framework

Step 1 Step 2 Step 3 Step 4 Build a base-case projection of the level of TFP by sector Assess the impacts of digitalization and population aging by sector Use G-Cubed (a multi-sector, multi-country generalequilibrium model with exogenous TFP) to assess the impacts of climate change policies by sector and obtain projections of nominal shares for each sector Obtain total TFP using Domar aggregation

Note: TFP is total factor productivity. G-Cubed is used under license by the Bank of Canada.

Step 1: Construct a base-case projection for the level of TFP for each sector (see Base-case projection).

Step 2: Add the impacts of digitalization and aging to the sectoral base-case projection.

Combine steps 2a and 2b to obtain the impacts of digitalization and aging on the level of TFP by sector. 

Step 3: Use the multi-sector general equilibrium model G-Cubed (McKibbin and Wilcoxen 2013) to achieve two objectives:

  • Step 3a: Assess the impacts of the climate change policies announced and implemented so far through reallocation of capital and labour across sectors. This reallocation will affect aggregate TFP by changing the nominal share of each sector. As in step 2, add these marginal impacts to base-case projections (see Impacts of climate change policies).
  • Step 3b: Obtain projection of the nominal share for each sector.

Step 4: Combine sectoral TFP profiles, accounting for the impacts of structural factors, and sectoral nominal shares into aggregate TFP using Domar aggregation (Domar 1961).8

Base-case projection

We begin our analysis by projecting the level of TFP by sector (step 1 in Box 1).9 We construct these projections by assuming a continuation of sectoral trends over previous decades (assumptions are sector-specific; see Table 1). This approach establishes base-case profiles without any impacts from structural factors being above their historical averages.

Table 1: Summary of projection assumptions

Table 1: Summary of projection assumptions
Projection component Assumption for projection Details*
Sectoral base-case profiles of total factor productivity Historical average growth over previous decades, period specific to each sector (see next column) Start of projection: 2020

Historical average period
Agriculture: 1980–2019
Mining, oil and gas: 1985–2019
Gas utilities: 1995–2019
Petroleum refining, durable goods manufacturing: 2009–19
Transportation, services: 1990–2019
Non-durable goods manufacturing: 2000–19
Construction: 0% growth
Electric utilities: anchored to population growth
Digital intensity by sector Average annual increase of digital intensity from 2013–22, by sector Start of projection: 2023
Share of young workers (aged 29 and under) by sector Anchored to the projected share of people aged 29 and under in the total population (see the Appendix) Start of projection: 2024

Population projections (from Statistics Canada)
Share of older workers (aged 50 and over) by sector Anchored to the projected share of people aged 50 and over in the total population (see the Appendix) Start of projection: 2024

Population projections (from Statistics Canada)
Nominal sectoral shares Projections of shares from G-Cubed (general equilibrium model) Start of projection: 2020

* The start of the projection period differs across components and depends on the last data point available (see details in Table A-2).

We use 11 sectors in the analysis that match the sectors in G-Cubed (step 3 in Box 1).10 TFP profiles are heterogeneous across sectors, reflecting the different environments in which firms operate (Chart 4, panels a and b). For instance, the agriculture, forestry and fishing sector has grown the most since 1961 (Chart 4, panel a). Several factors played a role in boosting TFP in this sector:

  • economies of scale due to increased farm size
  • digitalization of farming operations
  • increasing food demand due to population growth
  • a reduction of farmland areas

Expected population growth suggests that TFP gains should continue.

In contrast, crude oil and gas extraction has become much less productive since the early 1970s (Chart 4, panel b). This decline has continued despite massive investments in the 2000s and early 2010s. This suggests that in the run-up to the 2014 decline in commodity prices, investments were geared toward increasing production as opposed to making extraction processes more efficient. Moreover, the growing prevalence of oil sands may have contributed to the poor productivity performance of this sector since this type of extraction costs more and requires more capital (see, for example Loertscher and Pujolas 2024). Because of these factors, we assume that the decline in TFP will continue in the 2020s. A similar story may also explain the dynamics of the mining sector.

Chart 4: Base-case profiles of total factor productivity differ by sector

Chart 4: Base-case profiles of total factor productivity differ by sector

Level of total factor productivity, 1961–2030, index: 1961 = 100

Sources: Statistics Canada and Bank of Canada calculations, estimates and projections
Last data plotted: 2030

Overall, the manufacturing sector experienced steady TFP growth over history, although the durable goods subsector was hit hard by the 2008–09 recession (Chart 4, panel a).11 Services have also grown steadily since 1990, although at a slower pace (Chart 4, panel b). The much lower TFP growth in petroleum refining may reflect the aging of the installed capacity (Chart 4, panel a).12 However, in the 2010s, positive TFP growth resumed in this sector and is expected to continue. Other sectors, such as construction (Chart 4, panel b) and transportation (Chart 4, panel a), experienced little TFP gains over the past two decades—a trend that is anticipated to continue into the 2020s.

Impacts of digitalization

We measure digital intensity by sector as real business investments in computers and software divided by real total investment.13 Digital intensity has increased significantly in Canada since the 1980s, although we see heterogeneity in growth across sectors and periods (Chart 5). Overall, goods-producing sectors experienced a faster pace of digitalization relative to the services sector up until 2019, although goods-producing sectors also tend to have a lower level of digitalization. Digitalization accelerated in the services sector during the pandemic. This likely reflected employers’ incentive to use digital technologies in place of workers in high-contact occupations with a high potential for automation.14

Chart 5: Digital intensity increased faster during the COVID-19 pandemic, especially in services

We project digital intensity by sector using its average annual growth rate over the past 10 years (2013–22).15 The average digital intensity is thus expected to increase slightly faster across the services sector than the goods-producing sector (Chart 5, red bars).

Our analysis shows that in most sectors, increased digitalization is largely associated with higher TFP (see regression results in the Appendix). These results show that productivity in the agriculture sector benefited the most from an increase in digital intensity, followed by manufacturing. A few sectors—including the mining, oil and gas sector and some services subsectors (specifically, professional services as well as accommodation and food services)—exhibit a negative although non-significant relationship between digital intensity and TFP. In these sectors, TFP has been trending down since the 1980s, suggesting that digital intensity may be capturing the effects of other factors not accounted for.

Similar results can be found in the literature. Firm-level evidence suggests digitalization is positively associated with TFP and that this relationship is stronger in the manufacturing sector than the services sector (Gal et al. 2019). However, not all firms are benefiting from digitalization: Anderton, Botelho and Reimers (2023), for example, find negative impacts in about 10% of the sectors they studied.16

Combining the sectoral projections with the estimated historical relationship between digital intensity and TFP (step 2 in Box 1), we find that digitalization has had a positive effect on business sector TFP growth, especially in the 1990s (Chart 6). But the impacts of digital intensity on TFP growth varies significantly across sectors.

Chart 6: Digitalization has had a positive impact on total factor productivity growth

Digitalization in the services sector has made consistently positive contributions to TFP growth since the 1970s. This sector is expected to account for most of the TFP gains from digitalization over the 2020s, likely building on the gains initiated during the pandemic. The manufacturing sector also contributed positively between 1970 and 2010, but its contribution will likely be lower going forward. This may reflect that anticipated investment in digital technology is geared mostly to replacing and maintaining previous groundbreaking technologies introduced into the 1990s and 2000s. The dynamics in the other category are driven by the mining, oil and gas sector, which saw either small or negative returns from digitalization.

Our results also suggest that the contribution of digitalization to TFP growth is expected to increase in the 2020s relative to the previous two decades (Chart 2 and Chart 6). This is mainly due to the acceleration of digital intensity during the pandemic and the continuation of steady digital gains in the services sector (amplified by the increasing nominal share of this sector). These results are qualitatively consistent with findings for the United States: the contribution of information technology to TFP growth moderated in the manufacturing sector and accelerated in the services sector from 1996–2004 and 2005–19 (Byrne 2022).

Impacts of an aging population

The impacts of an aging population are captured through the share of young workers (aged 29 and under) and the share of older workers (aged 50 and over) in each sector. Overall, the share of young workers declined rapidly in the 1980s as baby boomers entered their prime working age (Chart 7, panel a). By the 2000s, they started turning 50, which resulted in a rapid increase in the share of older workers. At the same time, the share of young workers continued to decrease, although at a slower pace relative to the late 1990s. The age composition in most sectors followed the aggregate evolution.

In the future, the shares of both groups are expected to decrease (Chart 7, panel b).17 The influx of newcomers to Canada in 2024 is expected to slightly boost the share of young workers but decrease the share of older workers, as newcomers tend to be younger than the average of the overall Canadian population. After 2024, the shares of young workers are projected to resume their slow decline in most sectors while the shares of older workers will remain flat. 

Turning to the historical relationship between TFP and aging, we find an inverted-U relationship, consistent with findings in the literature (see regression results in the Appendix). That is, an increase in the share of young or older workers has a negative impact on TFP.18 This inverted-U relationship could reflect that young and older workers are less productive compared with prime-age workers, as predicted by the theory of human capital accumulation over the life cycle (e.g., Ben-Porath 1967). Young workers have less accumulated experience and knowledge and thus need to learn on the job, while older workers who are close to retirement have fewer incentives to maintain their skills.

Chart 7: The shares of young and older workers are expected to decline

Chart 7: The shares of young and older workers are expected to decline

Share of workers by age group, annual data

Sources: Statistics Canada and Bank of Canada calculations, estimates and projections
Last observation: panel a, 2021
Last data plotted: panel b, 2030

As for digitalization, we assess the impacts of aging on TFP by combining sectoral TFP projections with the estimate of the historical relationship between TFP and aging (step 2 in Box 1). We find that the impacts of aging on TFP growth were significant over history but are expected to be more limited in the 2020s (Chart 8).

The declining share of young workers contributed positively to TFP growth for most decades analyzed, with the peak impact occurring in the 1990s. In the 2000s, the influx of young workers into the oil and gas sector contributed to the decline in TFP growth. The impact of the share of older workers was limited before the 2000s. However, in the two decades that followed, the increasing share of older workers had a strong negative effect on TFP growth. Based on the population projection, we expect that the peak TFP impact for this age group has passed, leading to a rebound in the 2020s.

Chart 8: The impacts of population aging on total factor productivity growth are expected to become less significant

Impacts of climate change policies

Our analysis focuses on assessing the impacts of sectoral reallocation resulting from announced and implemented climate policies. Climate change policies are simulated in G-Cubed as taxes on the use of fossil fuels, encouraging substitution away from emissions-intensive sectors.19

We find small positive impacts (Chart 2, green bars) of climate change policies in the 2020s. The positive impacts on business sector TFP result from diverting capital and labour away from low-productivity growth sectors (such as mining, oil and gas extraction) toward more productive sectors such as services (excluding transportation) and non-durable goods manufacturing (Chart 9, panel a). These positive impacts are offset by declines in the nominal shares of high productivity growth, such as petroleum refining, durable goods manufacturing and agriculture. The large decline in the nominal shares of these sectors does not lead to significant impacts because their shares in the overall business sector are small (Chart 9, panel b). The small impacts also reflect that these policies were implemented only recently, and resource reallocation will take time.

Chart 9: Climate policies are expected to lower the nominal shares of carbon-intensive sectors

Chart 9: Climate policies are expected to lower the nominal shares of carbon-intensive sectors

Note: A sectoral share is calculated as sectoral nominal gross output divided by business sector nominal GDP. The sum of all sectoral shares in panel b is greater than 1.0 because intermediate inputs are included in gross nominal output.
Sources: Statistics Canada and Bank of Canada calculations, estimates and projections
Last data plotted: 2030

Climate-related risks not incorporated

Climate change could also have direct impacts on TFP, but these are highly uncertain (NGFS 2020; Bijnens et al. 2024). We thus do not incorporate the direct effects of climate change in our TFP projections and instead present risks around these projections. We classify these broadly as:

  • physical risks—chronic and acute effects from the accumulation of greenhouse gas emissions
  • transition risks—changes in policies, technology and markets as economies seek to avoid physical risks by reducing emissions

The impacts of physical risks on productivity are likely to vary across sectors in Canada. For example, chronic risks such as changing seasonal patterns, higher temperatures and higher atmospheric carbon dioxide levels can have positive and negative impacts on agricultural yields.20 Moreover, the increased frequency and severity of extreme weather events could lower TFP by reducing infrastructure resilience and by increasing resource scarcity and supply chain disruptions.

The transition to a lower-carbon economy could affect TFP through several channels. Investments in adaptation measures, such as construction of dikes or flood walls, may mitigate risks from gradual warming and acute events. However, such investments are likely to reduce TFP in the short term by diverting capital away from productive assets. Labour market frictions could impede the reallocation of workers away from emissions-intensive sectors, leading to higher structural unemployment and a drag on TFP through human capital depreciation (Ortego-Marti 2020). In contrast, climate policy could increase innovation in green technologies, and knowledge spillovers could potentially boost overall TFP in the long run (Acemoglu et al. 2023).21

Sectoral analysis

We calculate sectoral contributions to business sector TFP growth by combining the impacts from all structural factors.22 Business sector TFP growth is projected to be lower in the 2020s compared with previous decades and the historical average (Chart 1). The mining, oil and gas sector is expected to be a drag on TFP and the contribution of the manufacturing sector will decline, while the services sector should provide an offset.

Before 2000, the manufacturing sector was the main contributor to TFP growth (Chart 1, blue bars). After a difficult decade marked by the recession in the United States in 2001 and the 2008–09 recession, TFP growth resumed but at a slower pace.23 This slowdown reflects both a decline in average TFP growth and a lower nominal share for manufacturing industries. For instance, TFP growth in the durable goods manufacturing sector in the 2020s is expected to be one-third of its 1970–99 average, while the nominal share of non-durable goods manufacturing is expected to have decreased by 40% over the same period.

In contrast, the services sector has become the powerhouse of TFP growth since the 2000s (Chart 1, yellow bars). This sector was less affected by the 2000s recessions and is expected to benefit the most from digitalization in the future (Chart 6). TFP growth is expected to be about 2.4 times higher relative to its 1970–99 average. The services sector is also anticipated to continue its expansion in the 2020s; its nominal share should be about one-third higher than it was between 1970 and 1999.

The mining, oil and gas extraction sector has weighed on TFP growth for most decades and is expected to remain a drag in the 2020s (Chart 1, green bars). This decline occurred despite massive investment in the 2000s and early 2010s. One hypothesis is that given high oil prices, investments were geared toward extracting more resources rather than enhancing productivity. In contrast, the negative oil price shock of 2014 created incentives to innovate, which raised TFP growth in this sector in the late 2010s.

For the other sectors (Chart 1, red bars), agriculture is expected to account for about 80% of TFP growth in the 2020s.


We propose a tool that decomposes TFP growth into sectoral contributions. The analysis incorporates three structural factors—digitalization, aging and climate change policies—and measures their contributions. Overall, we expect aggregate TFP growth to slow in the 2020s below both its historical average and the average from the 2010s.

While our proposed approach is not intended to forecast annual TFP growth, it is useful to inform the likely path of TFP growth over the medium term and for scenario analysis. In addition to the climate risks discussed above, this projection is subject to many other risks.

One positive risk relates to our definition of digitalization, which does not directly include emerging technologies such as artificial intelligence or information technology cloud services. Both of these could improve productivity more than we expect.24 In contrast, the rapid digitalization that occurred during the COVID-19 pandemic may limit future gains if firms have already completed much of their planned digital transformations.

Demographic forces driven by newcomers to Canada (including non-permanent residents) may impact TFP growth if they further shift the age composition of the Canadian population. This risk could be either positive or negative, as it depends on changes in immigration policies. Policies that further reduce inflows of immigration—for example, lowering immigration targets—could dampen the expected decline in the share of older workers and drag on TFP growth.25 Reversal of such policies would have the opposite effect.

Finally, our base-case outlook assumes that sectoral TFP projections are expected to grow in line with their historical averages. Moreover, we also assume that the historical relationship between TFP, digitalization and aging remains unchanged going forward. Any departure from these assumptions would lead to a different path for the TFP projection. For instance, the emergence of artificial intelligence could increase TFP gains per unit of investment relative to the pre-pandemic average.


Estimating the historical relationship between total factor productivity, digitalization and aging

This relationship is estimated using a fixed-effect regression with annual sector-level data:

\(\displaystyle\, TFP_{j,t} \) \(\displaystyle=\, \beta_{j}DI_{j,t-1} \) \(\displaystyle+\, γ_{1,k}S_{j,t}^k \) \(\displaystyle+\, γ_{2,k}S_{j,t}^k D_{2000} \) \(\displaystyle+\, \delta X_{j,t-1} \) \(\displaystyle+\, \alpha_{j} \) \(\displaystyle+\, \theta_{t} \) \(\displaystyle+\, \epsilon_{j,t} \) \(\displaystyle,\) \(\displaystyle (1)\)

where \(\displaystyle\, j\) denotes sector and \(\displaystyle\, t\) time, \(\displaystyle\, \alpha_{j}\) and \(\displaystyle\, \theta_{t}\) are sector and time-fixed effects, and \(\displaystyle\, k\ \epsilon\ (young, old)\). \(\displaystyle\, D_{2000}\) is a binary variable equal to 1 for the year 2000 and after, and X controls for past total factor productivity (TFP), research and development (R&D), intangible investment and education.26 All variables are in level.

Following Anderton, Botelho and Reimers (2023), we define digital intensity (DI) as real investment in computers and software in a given sector divided by real total investment in this sector. The digital intensity parameter is sector-specific.  

Aging is captured by the share \(\displaystyle\, S_{j,t}^{young} \) of young workers (aged 29 and under) in each sector and the share of older workers \(\displaystyle\, S_{j,t}^{old} \) (aged 50 and over). The estimated relationship of aging is linear by segment as indicated by the interaction between the shares and \(\displaystyle\, D_{2000}\).

Estimated parameters for DI turn out to be mostly positive, although not all of them are significant (Table A-1). Taking the estimated parameters for the manufacturing sector, an increase of 1 unit in the digital index of this sector generates a 1.49 unit increase of the TFP index. A joint test of the significance of all DI parameters rejects the null hypothesis that they are all equal to zero. R&D and intangible investment do not play a sizable role, although the estimated coefficients are significant.27

Turning to aging, we find that younger workers are relatively less productive than prime-age workers. As a result, an increase in the share of young workers reduces the level of TFP. The share of older workers also has a large negative impact, but only after the year 2000. Older workers were found to be as productive as prime-age workers in the previous decades.

Table A-1: Estimated coefficients of equation (1)

Table A-1: Estimated coefficients of equation (1)
Variable Estimated coefficient Standard errors
Digital intensity (DI)
DI – agriculture 16.17*** 1.827
DI – mining, oil and gas -2.17 2.193
DI – utilities 0.01 0.287
DI – construction -0.44 0.328
DI – manufacturing 1.49*** 0.293
DI – wholesale 0.64*** 0.117
DI – retail 0.64*** 0.136
DI – transportation 0.24 0.245
DI – finance and insurance 0.07 0.127
DI – remaining services -0.08 0.057
DI – information and recreation 0.02 0.102
Share of young workers -32.22*** 9.138
Share of young workers post-2000 1.65 7.599
Share of older workers 4.03 14.180
Share of older workers post-2000 -27.28* 15.360
Other variables    
Lagged TFP 0.82*** 0.021
Lagged R&D intensity 0.03* 0.016
Lagged R&D intensity post-2000 -0.13** 0.065
Lagged intangible intensity 0.00* 0.002
Lagged intangible intensity post-2000 0.02** 0.007
Share of university graduates -25.40 16.731
Share of university graduates post-2000 11.31 8.170
Binary variable 1970s -3.03 5.957
Binary variable 1980s -2.09 5.849
Binary variable 1990s -6.27 5.663
Binary variable 2000s -2.13** 1.032
Constant 30.29*** 6.320

* significant at 0.1 level; ** significant at 0.05 level; *** significant at 0.01 level
Note: TFP is total factor productivity; R&D is research and development.

Table A-2: Data sources

Table A-2: Data sources
Variable Source Period
Total factor productivity Statistics Canada, Table 36-10-0208-01: Multifactor productivity, value-added, capital input and labour input in the aggregate business sector and major sub-sectors, by industry 1961–2019
Gross nominal output and nominal GDP Statistics Canada, Table 36-10-0208-01: Multifactor productivity, value-added, capital input and labour input in the aggregate business sector and major sub-sectors, by industry 1961–2019
Tangible investments Statistics Canada, Table 36-10-0097-01: Flows and stocks of fixed non-residential capital, by sector of industry and type of asset, Canada (x 1,000,000) 1961–2022
Intangible investments Statistics Canada, custom tabulations 1976-2016
Share of workers by age and sector Statistics Canada, Labour Force Survey: Public Use Microdata 1976–2023
Share of university graduates by sector Statistics Canada, Labour Force Survey: Public Use Microdata 1976–2023
Population projection Statistics Canada, custom tabulations 1976–2030

Projecting the share of workers by age and by sector

While Statistics Canada does not project the share of workers by age group and sector, it does produce projections of total population by age. These two variables turn out to be strongly correlated over history (over 0.9 on average across sectors).

We thus exploit this strong correlation between the sectoral share of young (older) workers and the share of young (older) people in the total population. For each sector, we thus regress the share of young (older) workers\(\displaystyle\, \left(Share_{j,t}^{W,k}\right) \)on the share of young (older) people in the population
\(\displaystyle\, \left(Share_{t}^{Pop,k}\right) \):

\(\displaystyle\, Share_{j,t}^{W,k} \) \(\displaystyle=\, \alpha_{0} \) \(\displaystyle+\, \alpha_{1}Share_{t}^{Pop,k} \) \(\displaystyle+\, \alpha_{2}\alpha_{3}\left(Share_{t}^{Pop,k}\right)^2 \) \(\displaystyle+\, \alpha_{3}\left(Share_{t}^{Pop,k}\right)^3 \) \(\displaystyle+\, \epsilon_{j,t} \) \(\displaystyle,\) \(\displaystyle (2)\)

where \(\displaystyle\, j\) denotes sector, \(\displaystyle\, t\) time and \(\displaystyle\, k\ \epsilon\ (young, old)\). We then project \(\displaystyle\, Share_{j,t}^{W,k} \) over 2024–30 by using the estimated parameters and Statistics Canada projections for \(\displaystyle\, Share_{t}^{Pop,k} \).


Acemoglu, D., P. Aghion, L. Barrage and D. Hémous. 2023. ”Green Innovation and the Transition Toward a Clean Economy.” Peterson Institute for International Economics Working Papers 23-14.

Anderton, R., V. Botelho and P. Reimers. 2023. “Digitalisation and Productivity: Gamechanger or Sideshow?” European Central Bank Working Paper Series No. 2794.

Bank of Canada. 2024. Monetary Policy Report (April).

Barr, J., P. Foltin and J. Tang. 2023. “Recent Productivity Trends in Canada: Navigating the Twin Transitions of Green and Digitalization.” International Productivity Monitor 45: 93–119.

Bellatin, A. and S. Houle. 2021. “Overlooking the Online World: Does Mismeasurement of the Digital Economy Explain the Productivity Slowdown?” Bank of Canada Staff Analytical Note No. 2021-10.

Benatti, N., M. Groiss, P. Kelly and P. Lopez-Garcia. 2023. “Environment Regulation and Productivity in the Euro Area: Testing the Porter Hypothesis.” European Central Bank Working Paper Series No. 2820.

Ben-Porath, Y. 1967. “The Production of Human Capital and the Life Cycle of Earnings.” Journal of Political Economy 74 (4): 352–365.

Bijnens, G., S. Anyfantaki, A. Colciago, J. De Mulder, E. Falck, V. Labhard, P. Lopez-Garcia, N. Lourenço, J. Meriküll, M. Parker, O. Röhe, J. Schroth, P. Schulte and J. Strobel. 2024. “The Impact of Climate Change and Policies on Productivity.” European Central Bank Occasional Paper Series No. 340.

Byrne, D. M. 2022. “The Digital Economy and Productivity.” Federal Reserve Board Finance and Economic Discussion Series 2022-038.

Chernoff, A. and C. Warman. 2023. “COVID-19 and Implications for Automation.” Applied Economics 55 (17): 1939–1957.

Comunale, M. and A. Manera. 2024. “The Economic Impacts and the Regulation of AI: A Review of the Academic Literature and Policy Actions.” International Monetary Fund Working Paper No. 2024/065.

Domar, E. D. 1961. “On the Measurement of Technological Change.” The Economic Journal 71 (284): 709–729.

Feyrer, J. 2007. “Demographics and Productivity.” The Review of Economics and Statistics 89 (1): 100–109.

Gal, P., G. Nicoletti, C. von Rüden, S. Sorbe and T. Renault. 2019. “Digitalization and Productivity: In Search of the Holy Grail—Firm-level Empirical Evidence from European Countries.” International Productivity Monitor 37: 39–71.

Government of Canada. 2020. “Climate Change Impacts on Agriculture.” Agriculture and Agri-Food Canada.

Government of Canada. 2022. “Canada’s 8th National Communication and 5th Biennial Report.” Environment and Climate Change Canada.

Gu, W. and M. Willox. 2023. “The Post-2001 Productivity Growth Divergence Between Canada and the United States.” International Productivity Monitor 45: 27–60.

Gugler, K., F. Szücs, and T. Wiedenhofer. 2024. “Environmental Policies and Directed Technological Change.” Journal of Environmental Economics and Management 124: 102916.

Haun, C. and T. Sargent. 2023. “Decomposing Canada's Post-2000 Productivity Performance and Pandemic-Era Productivity Slowdown.” International Productivity Monitor 45: 5–27.

Jaumotte, F., L. Li, A. Medici, M. Oikonomou, C. Pizzinelli, I. Shibata, J. Soh and M. Mendes Tavares. 2023. “Digitalization During the COVID-19 Crisis: Implications for Productivity and Labor Markets in Advanced Economies.” International Monetary Fund Staff Discussion Notes No. 2023/003.

Lanoie, P., J. Laurent-Lucchetti, N. Johnstone and S. Ambec. 2011. “Environmental Policy, Innovation and Performance: New Insights on the Porter Hypothesis.” Journal of Economics and Management Strategy 20 (3): 803–842.

Loertscher, O. and P. Pujolas. 2024. “Canadian Productivity Growth: Stuck in the Oil Sands.” Canadian Journal of Economics (April 11, Early View): 12707.

Maestas, N., K. J. Mullen and D. Powell. 2023. “The Effect of Population Aging on Economic Growth, the Labor Force and Productivity.” American Economic Journal: Macroeconomics 15 (2): 306–332.

McKibbin, W. J. and P. J. Wilcoxen. 2013. “A Global Approach to Energy and the Environment: The G-Cubed Model.” In Handbook of Computable General Equilibrium Modeling, Volume 1, edited by P. B. Dixon and D. W. Jorgenson, 995–1068. Elsevier, 2013.

Mollins, J. and T. Taskin. 2023. “Digitalization: Productivity.” Bank of Canada Staff Discussion Paper No. 2023-17.

Network for Greening the Financial System (NGFS). 2020. Guide to Climate Scenario Analysis for Central Banks and Supervisors. NGFS technical document.

Organisation for Economic Co-operation and Development (OECD). 2024. Going Digital Toolkit: Canada.

Ortego-Marti, V. 2020. “Endogenous TFP, Labor Market Policies and Loss of Skills.” University of California at Riverside, Department of Economics, Working Papers 202007.

Parker, M. 2023. “How Climate Change Affects Potential Output.” European Central Bank Economic Bulletin 6/2023.

Rosell, C., K. Dowsett and N. Paterson. 2023. “A Critical Juncture: Assessing Canada’s Productivity Performance and Future Prospects.” International Productivity Monitor 45: 61–92.

Syverson, C. 2017. “Challenges to Mismeasurement Explanations for the US Productivity Slowdown.” Journal of Economic Perspectives 31 (2): 165–186.

Tang, J. and C. MacLeod. 2006. “Labour Force Ageing and Productivity Performance in Canada.” Canadian Journal of Economics 39 (2): 582–603.

Trammell, P. and A. Korinek. 2024. “Economic Growth Under Transformative AI.” Centre for Economic Policy Research Discussion Paper No. 18780.

Vandenbroucke, G. 2021. “The Baby Boomers and the Productivity Slowdown.” European Economic Review 132: 103609.

  1. 1. Labour productivity growth can be seen as the sum of two components: the contribution from capital accumulation (measured as the capital-to-GDP ratio) and TFP. The latter measures the change in output not attributed to changes in capital or labour inputs (in terms of either quantity or quality). TFP growth is often associated with technological change, organizational change or economies of scale.[]
  2. 2. If investment had continued growing at its historical average (quarterly growth of 0.9%) after 2014, the level of business investment in the fourth quarter of 2023 would have been about 65% higher than its actual level.[]
  3. 3. The data used in this analysis do not include 2020 data nor the historical revisions released by Statistics Canada on April 16, 2024. However, the results presented in this paper are not affected by the new data vintage.[]
  4. 4. See Rosell, Dowsett and Paterson (2023) and Barr, Foltin and Tang (2023) for a policy discussion about the role of structural factors for the future of productivity growth.[]
  5. 5. The G-Cubed model has been extended to assess the impacts of climate policy (McKibbin and Wilcoxen 2013). A version of G-Cubed has been developed by McKibbin Software Group to be used under license by the Bank of Canada.[]
  6. 6. Areas where Canada falls behind include machine-to-machine communication, the share of firms buying cloud services, and investment intensity in information and communication technologies. See OECD (2024).[]
  7. 7. Some argue that digitalization of the economy could explain the decline in productivity because digital activities are mismeasured. Evidence so far suggests that the mismeasurement hypothesis cannot account for most of the long-term decline in productivity in Canada (Bellatin and Houle 2021) and the United States (Syverson 2017).[]
  8. 8. Sectoral nominal shares (also called Domar weights) are calculated as sectoral nominal gross output divided by nominal GDP in the business sector.[]
  9. 9. We chose the level of TFP because annual TFP growth is too volatile to forecast.[]
  10. 10. G-Cubed has 12 sectors, but TFP for one sector—coal mining—was not available because of data limitations. This sector is included in the analysis as part of the mining, excluding crude oil and gas extraction sector.[]
  11. 11. While all durable goods manufacturing industries experienced negative TFP growth in 2009, the largest declines were in transportation equipment and primary metal manufacturing. This most likely reflects the difficult conditions of auto manufacturing around 2009 and the general decline in export values for Canadian primary metal (e.g., steel and aluminium).[]
  12. 12. The Sturgeon Refinery, which opened in late 2017 (in Redwater, Alberta), was the first refinery to open in Canada in 30 years.[]
  13. 13. We use this definition because it provides sector-level data over the longest period. Other possible measures of digital intensity include spending on information and communications technology, size of the digital workforce and spending on digital services.[]
  14. 14. Occupations in the services sector (e.g., retail salespersons, cashiers, secretaries and administrative assistants) were most vulnerable to pandemic-induced automation because of increased risk of transmission of the COVID-19 virus and the potential for automation in these jobs (Chernoff and Warman 2023). A growing body of literature is studying the impacts of the pandemic on digitalization (e.g., Jaumotte et al. 2023; Mollins and Taskin 2023).[]
  15. 15. This period notably includes the pandemic, during which Canada saw a strong increase in digitalization. The 10-year assumption is a middle ground between the view that the pandemic should be excluded because of its extraordinary nature and the view that the pandemic has become the new normal.[]
  16. 16. For instance, waste collection, travel agency and tour operator activities, and electricity and gas sectors.[]
  17. 17. See the Appendix for details on the projection for shares of workers by age and sector.[]
  18. 18. See, for instance, Feyrer (2007) for OECD countries and Tang and MacLeod (2006) for Canada.[]
  19. 19. We first simulate currently announced carbon taxes (including the investment tax credits announced in the 2023 federal budget), then capture the remaining climate policy measures needed to reach our projected emissions targets through an additional (shadow) tax on emissions. The taxes are adjusted such that projected emissions align with Environment and Climate Change Canada’s modelling of currently announced and implemented climate policies (Government of Canada 2022).[]
  20. 20. Agriculture and Agri-Food Canada provides a thorough overview of possible impacts of climate change on Canadian agriculture (Government of Canada 2020).[]
  21. 21. There is support in the literature for the theory that climate policy encourages green innovation, but it provides little evidence that this raises overall TFP growth in the long run (see Lanoie et al. 2011; Benatti et al. 2023; and Gugler, Szücs and Wiedenhofer 2024).[]
  22. 22. Gu and Willox (2023) and Haun and Sargent (2023) offer alternative sectoral decompositions of labour productivity by source of growth. They also compare productivity growth in Canada and the United States.[]
  23. 23. Notably, TFP growth in the durable goods manufacturing sector rebounded strongly in the early 2010s following the 2009 recession, but its average growth settled down below its long-term average in the following decade.[]
  24. 24. Some of these activities are presumably captured in our analysis through investment in computers and software.[]
  25. 25. This is because newcomers tend to be younger than the Canadian population.[]
  26. 26. Mineral exploration investments are combined with R&D expenditures since only the mining, oil and gas sector conducted such investment.[]
  27. 27. We do not incorporate R&D and intangible investments (e.g., architectural and engineering design, organization capital, firm-specific human capital and advertising) into our measure of digital intensity because either they are not contributing directly to digitalization or they could potentially do so with a significant lag.[]


We thank Alexander Ueberfeldt and Craig Johnston for their helpful feedback and support during the drafting process. We would also like to thank Rachit Lumb and Mallory Long for their excellent research assistance. Finally, we thank Colette Stoeber and Meredith Fraser-Ohman for their great editing and Judith Lefebvre, Sylvie Vancappel and Eric Bannem for the excellent translation.


Bank of Canada staff analytical notes are short articles that focus on topical issues relevant to the current economic and financial context, produced independently from the Bank’s Governing Council. This work may support or challenge prevailing policy orthodoxy. Therefore, the views expressed in this note are solely those of the authors and may differ from official Bank of Canada views. No responsibility for them should be attributed to the Bank.


On this page
Table of contents