Intertemporal and Inter-Industry Effects of Population Ageing: A General Equilibrium Assessment for Canada Nabil Annabi, Maxime Fougère and Simon Harvey November, 2008 Policy Research Directorate, Labour Market Research and Forecasting
The views expressed in this research are solely those of the authors and do not necessarily reflect the views of HRSDC, nor those of the Government of Canada. 2
Projected old-age dependency ratio (OADR) and population growth rate in Canada Note: OADR = Population aged 65+ relative to the working-age population 3
Outline Objective Methodology Model features Data & Calibration Simulation Results Concluding Remarks 4
Objective Simulate the inter-industry and labour market effects of population ageing using a multi-sectoral life-cycle overlapping-generations (OLG) model. Account for both the supply and demand effects of population ageing. Supply effect: decline in the labour force growth. Demand effect: structural changes in final demand due to different consumption preferences of older generations. Examine the impact on living standards by taking into account not only the change in the OADR but also the lower population growth rate. 5
Model Features Extended version of Fougère, Mercenier and Mérette (2007) model. Calibration along a balanced-growth path: account for (Harrodneutral) labour-augmenting technical progress. Assume the presence of unemployment on the labour market. Increased heterogeneity among households: incorporate earning profiles for 25 occupations (National Occupational Matrix, NOC). Increase the number of generations from 7 to 15: 4-years instead of 10-years model. Taking into account the difference in the effective age of retirement among occupations. 6
Model Features cont d. Households (generations) maximize an intertemporal utility function subject to a lifetime budget constraint. The representative household values consumption and bequests. A representative competitive firm in each of the 14 industrial sectors using intermediate goods, hired labour and rented physical capital in the production process. The aggregate intermediate input is a CES function of goods produced across industries. Labour factor is composed of 25 occupations, combining 10 occupational groups and 5 qualification levels. Aggregate labour is also represented by a CES function of 3 types of labour with high, medium and low elasticity of substitution. 7
Figure 1. Nesting in Sectoral Production Production Cobb-Douglas Intermediate Consumption Value Added Cobb-Douglas CES : constant elasticity of substitution L : low elasticity of substitution M : medium elasticity of substitution H : high elasticity of substitution B,C,D: qualification levels in the NOC matrix Prod. 1 CES Capital Labour CES-L Type 3 Prod.14 CES-M Prof. 7 Prof. 8 Prof. 9 CES-H CES-H CES-H B C D B C D B C D 8
Model Features cont d. Following Blanchflower and Oswald (1994), a Wage curve depicts a negative relation between the real wage rate and the occupation-specific unemployment rate. Local unemployment rate is considered as one of the determinants of the wage rate. Wages tend to be higher on a labour markets with lower unemployment rate. Wage-unemployment elasticities based on the estimates in Decaluwé, Lemelin, Bahan and Annabi (2005). A consolidated government levies taxes to finance expenditures. Public education and health expenditures fixed per head in each age group budget shift in favour of health as population ages. An intermediary represents the pay-as-you-go pension system. 9
Data & Calibration Table 1. Base run statistics Share of sectoral GDP at factor cost Share of sectoral GDP at factor cost in gross output Share in sectoral GDP at factor cost of Capital Labour Primary 3.3 34.4 34.4 65.3 Man. & utilities 20.8 32.0 53.2 46.6 Construction 14.8 66.8 77.3 22.6 Transp. & stor. 2.2 24.4 65.0 34.8 Communication 1.8 42.5 48.6 51.2 Wholesaling & ret. 9.2 59.0 74.0 25.9 Fin., ins., & RS serv. 10.8 55.7 29.6 70.1 Serv. to firms 4.8 75.7 70.2 29.6 Comput. & serv. 3.2 77.4 70.2 29.6 Public. adm. 11.0 69.4 85.6 14.4 Education 3.4 63.4 84.1 15.9 Health 6.6 64.8 70.3 29.6 Accom. & leis. 1.5 16.5 75.0 24.9 Other serv. 6.7 73.4 79.4 20.6 Source: Input-Output tables, Statistics Canada 10
Table 2. Labour demand by sector and occupation (share, %) Source: Labour force survey, Statistics Canada 11
Table 2. cont d. Source: Labour force survey, Statistics Canada 12
Figure 2. Distribution of public expenditures on education and health per age group (share, %) Source: Annabi, Harvey and Lan (2007), HRSDC. 13
Figure 3: Spending shares by sector and age group (%) Source: Survey of Household Spending, Statistics Canada 14
Figure 4. Occupation-specific earnings profiles Source: Census 2001, Statistics Canada 15
Simulation Results: transmission channels Population ageing implies a slower labour force growth which causes a change in factors remunerations: rise in wage rates; capital becomes relatively more abundant and its rental rate declines. Combined change in wages and rental rate would affect production costs and market prices. Change in factors reallocation across sectors (subtit. of capital to labour) Substitution in consumption. On the other hand, changes in remunerations will affect household s income: change in aggregate private demand and savings; and ultimately investment. 16
Transmission channels - cont d. Population ageing would shift private demand in favour of the service goods. At the sectoral level, production would expand more in sectors with lower labour share in value-added (GDP at factor cost). Demand effect is expected to mitigate the long-run negative impact on the (labour intensive) service industries compared to the rest of the economy. Lastly, the profession used intensively in the more expanding industries will benefit the most in terms of remuneration rate. The decline in population growth rate will be accompanied by a decrease in the unemployment rate which will further increase pressures on wages. 17
Macro effects 18
Figure 5: Macro effects (Annual average % growth rate) Simulation results. 19
Figure 6: Impact on factors remunerations (% change from 2006) Simulation results. 20
Figure 7: Impact on the unemployment rate (Percentage point change from 2006) Simulation results. 21
Figure 8: Impact on GDP and GDP per capita (Annual average % growth rate) Note: The exogenous labour productivity growth is set equal to the 1996-2006 historical average of 1.9% per year. Simulation results. 22
Inter-industry effects 23
Figure 9: Base run sectoral labour share and long-run impact on market price Simulation results. 24
Figure 10: Long-run impact on sectoral production and market price Simulation results. 25
70 60 50 40 30 20 10 0-10 -20 Figure 11: Long-run impact on factors reallocation (% change from 2006) Production (detrended) Labour demand Capital demand (detrended) Note: Data sorted by increasing change in production. Simulation results. 26
Labour market effects 27
Figure 12: Impact on unemployment by occupation and skill level (level, %) Management & skill level A Skill level B Simulation results. 28
Figure 12: Impact on unemployment by occupation and skill level cont d. Skill level C Skill level D Simulation results. 29
Table 3: Impact on unemployment and wages by occupation Skill level Base run unemployment rate (2006) Unemployment rate (% change from 2006) Wage rate Management - 1.9-84 17 Bus., fin. & adm. A 1.7-88 16 B 3.0-87 16 C 4.2-86 15 Nat. & app. sciences A 1.7-65 13 B 3.7-62 12 Health occupations A 0.2-50 12 B 1.2-75 15 C 1.8-67 14 So. scien., edu., gov. A 2.6-65 13 B 2.8-75 16 Art, culture, rec. & sports A 2.9-72 16 B 6.0-55 10 Simulation results. 30
Table 3: Impact on unemployment and wages by occupation cont d. Skill level Base run unemployment rate (2006) Unemployment rate (% change from 2006) Wage rate Sales and services B 3.6-69 12 C 5.5-67 11 D 6.3-57 9 Trades, tra.& equip.oper. B 5.2-44 10 C 5.5-55 10 D 9.9-27 7 Occup. in prim. ind. B 5.8-71 15 C 5.8-74 13 D 22.0-36 6 Processing, man. and ut. B 3.0-67 11 C 7.5-49 8 D 11.1-43 7 Simulation results. 31
Sensitivity tests Lower labour productivity growth Average growth is much lower, however, the long-run drop in growth rates remain similar to that under the baseline scenario; The decline in unemployment rate is larger leading to higher pressure on wages. Higher intertemporal elasticity of substitution Average growth rates are lower, but the difference with respect to the baseline is small; Slower growth in investment which mitigates the relative scarcity of labour, resulting in lower pressure on wages. Higher elasticities of substitution in the CES (labour) nesting The macro and sectoral effects are similar to the baseline scenario; Small differences in real wages at the occupational level. 32
Concluding remarks At the aggregate level, population ageing lowers effective labour supply and decreases saving and investment in physical capital. The unemployment rate decreases by 2.7 percentage points in the long run, leading to an increased pressure on wages (+12% on average). In the long run, GDP and GDP per capital growth rates would fall by nearly 1.5 and 1 percentage point, respectively. At the sectoral level, production costs would increase more in the labour intensive sectors Production would expand more in sectors with lower labour share in value-added (GDP at factor cost) Primary Industries, and Finance, Insurance and Real Estate. Demand effect would mitigate some of the long-run negative effects on labourintensive service industries Health, Accomodation and Leisure, Other services and Transport and Storage industries. 33
Concluding remarks cont d. Real wage pressures will rise across all occupational groups. Wage pressures in Management occupations, Business, Finance and administration, health, social science, and education as well as in Occupations in Primary industries would be well above average. Wage pressures in Processing, Manufacturing, Sales and Trades could be below average. Under various assumptions about the model parameters, the sensitivity tests show that the conclusions remain similar to those under the baseline scenario. 34
Annex 1: Household s optimization program TC, Rbeq g 15 1 1 1 θ θ 1 θ t ( g, t+ g 1 βg g, t+ g 1) βg 15 βg= 15 1 θ g = 1 1+ ρ Max U = TC + RBeq, = 0, > 0 t t s.t Lend Lend = (1 + + τ ) R int Lend k g 1, t 1 gt, t t gt, + (1 τ cr ) Linc + (1 τ ) Pens + Inh Beq (1 + τ ) Pc w w t t gjt, t gt, gt, gt, TC c t g, t g, t ( 1 ) Linc = w u L EP AAR qual sup gj, t itype, iprof, iqual, t itype, iprof, iqual, t itype, iprof, iqual, gj gj itype, iprof, iqual, gj, t itype iprof iqual 35
Annex 2: Spending shares by sector and age group Calibration data. 36