OLG-CGE Models and Demographic Research

Transcription

1 OLG-CGE Models and Demographic Research Robert E. Wright Professor of Economics University of Strathclyde Presentation, CGE Modelling for Policy Analysis-International Cases, University of Helsinki Ruralia Institute, Seinajoki, Finland, March 16-17, 2015

2 Financial support from the Economic and Social Research Council under the grant: Developing an OLG-CGE Model for Scotland is gratefully acknowledged. Additional financial support was provided by the University of Ottawa and the University of Strathclyde.

3 COLLABORATORS OLG-CGE Modelling of Demographic Change Katya Lisenkova, National Institute of Economic and Social Research, London Marcel Merette, Department of Economics, University of Ottawa Jokke Kinnunen, Statistics and Research Åland CGE Modelling of Demographic Change Peter McGregor, University of Strathclyde Kim Swales, University of Strathclyde Nikos Pappas, Government of Greece Karen Turner, University of Strathclyde Patrizio Lecca, University of Strathclyde Kristinn Hermannsson, University of Glasgow

4 Overview 1. Motivation 2. Scottish Background 3. OLG-CGE Models 4. Three Sets of Results 5. Concluding Comments

5 1. Motivation Population ageing is the shift in the age-structure away from the younger to older age groups It is occurring in almost all countries Most rich countries are ageing rapidly Increase in the number/share of population aged 65 and older Decrease in the number/share of population aged 20 and younger Low, no or slow growth in the number/share of population aged 20 to 64 years Population ageing is NOT population decline

6 Figure 1 Age Distribution, Scotland, 1911 Figure 4 Age Distribution, Scotland, Men Women Men Women Age group Age group Percentage of total population Percentage of total population

7 Accommodating population ageing will be (is) expensive in welfare states Increase in the demand for state-supplied pensions and other agerelated benefits targeting at older people Big expected increases in government expenditure Big expected increases in taxes to pay for it Need high economic growth if T > G If not

8 Dominant view is that population ageing suppresses economic growth Why? One main reason is slow/no/low labour force growth Welfare state: Those in work pay for those not in work Difficult political economy in democratic countries with one man, one vote systems Greying of the electorate coupled with a steep upwards sloping relationship between age and political participation Significant standard of living reductions likely

9 Policy options? 1. Increase immigration 2. Introduce/expand family friendly policies 3. Raise the age of retirement 4. Increase spending on education and human capital investment 5. Older age employment subsidies 6. Subsidise/expand research and development activities 7. Focus on capital-intensive industries 8. Improve the cost-effectivness of government 9. Do nothing 10. Do nothing and hope for a technological leap

10 2. Scottish Background

11 Figure 1 Total Fertility Rate, Scotland and UK, ,50 3,00 Scotland UK 2,50 2,00 1,50 1,

12 Figure 2 Life Expectancy at Birth, Scotland and UK, ,0 0,0 Scotland Men Scotland Women UK Men UK Women 5,0 0,0 5,0 0,0 5,0 0,

13 Figure 3 Net-migration, Scotland,

14 100 Figure 4 Net-migration rate (per 10,000 population), Scotland and UK, UK Scotland

15 Figure 5 Projected Population, Principal Projection, Scotland and UK, (2012=100) Scotland UK

16 120 Figure 7 Projected Population Aged 20-64, Principal Projection, Scotland and UK, (2012=100) 115 Scotland UK

17 85,0 Figure 14 Unadjusted and Adjusted Total Dependency Ratio, Principal Projection, Scotland and UK, ,0 75,0 70,0 65,0 60,0 55,0 Total UK Scotland Adjusted UK Adjusted Scotland 50,0 45,0 40,

18 Demographic shock

19 3. OLG-CGE Models OLG = over-lapping generations CGE = computable general equilibrium

20 What are CGE Models? A computer general equilibrium (CGE) model is a description of an economy using a system of simultaneous equations The general equilibrium idea implies that all the markets, sectors and industries are modelled together with corresponding interlinkages This is opposed to partial equilibrium that only takes into account a part of the system (e.g. labour market), which neglects potential feed-backs This mathematical representation, coupled with a solver algorithm, ensures the computable nature of the model i.e. empirical results are generated

21 Models can be used for what if simulations thereby allowing one to obtain numerical results for endogenous variables based on assumptions about exogenous variables, functional forms, and parameter values Most CGE models rest on neo-classical economic assumptions Consumers are assumed to maximise their utility subject to a budget constraint (demand-side) Producers are assumed to maximise profit, given the prices of goods and factor of production costs (supply side) For each good and factor of production, equilibrium price is calculated where that demand equals supply

22 The first CGE model was constructed by Johansen in Have been largely used to analyse the macro-economic effects of a large number of economic policies such as liberalising foreign trade, decreasing government expenditure and increasing taxation. Main strength of CGE models lies with their flexibility, in the sense that their general nature means that they can be adapted to a variety of problems depending on the creativity and ingenuity of the researcher. CGE modelling is an art

23 CGE models are not in a strict sense a forecasting models. However, given a set of precise assumptions, it can be used to generate expected future time paths of key macroeconomic variables: Output, employment, unemployment, labour force participation, investment, wage rates, inflation and competitiveness. By varying these assumptions alternative time paths are produced, which can be compared in order to evaluate the likely consequences of alternative policies.

24 Some Useful Features: Substitution in production or consumption Non-passive supply side i.e. both demand and supply considered so supply matters (no the case in I-o) Relative prices become endogenous (e.g. the real wage and regional competitiveness are determined within the model) Producers and consumers allowed to respond to relative price changes through substitution in production or demand More flexible technology and demands structures (non-linearities)

25 More flexible technology and demands structures (nonlinearities) Over lapping generations (OLG) structure makes them more demographically friendly Can be used to price shocks, changes and policies e.g. In s or per cent output loss Consistency with microeconomic circular flow underpinnings.

26 Figure 1. Micro-economic Circular Flow Firms Factor Mrkts Goods Mrkts Households

27 Computable Calibration Accuracy is evaluated by back forecasting (backcasting)

28 Not surprising that more recently such model have been applied to a series demographic problems largely (but not exclusively) around the impacts of population ageing The type of CGE model that is best suited to demographic research builds in the notion of over lapping generations (OLG-CGE) This approach more explicitly allows the interaction of age effects, which in most demographic problems is of central importance.

29 Firm-side One sector. One good. Cobb-Douglas production function Factor prices equal to their marginal product Labour demand disaggregated by types of labour Three types of assets Household assets Government assets Foreign assets One region (RUK only to close the model)

30 Household-side 21 generations from 0-4 to 100+ First 4 generations only affect public consumption Starting from generations optimise their consumption/savings Four types of households depending on level of qualification Three working qualifications (high, medium, low) One non-working Two types of nationalities British born Foreign born Age/qualification-specific productivity (age-earnings profiles) Age-specific labour force participation rates

31 Time Overlapping generations structure Age A1 A2 A3 A4 A5 T1 G1 T2 G2 G1 T3 G3 G2 G1 T4 G4 G3 G2 G1 T5 G5 G4 G3 G2 G1 T6 G6 G5 T7 G7 G5 T8 G8 G5 T9 G9 G5 T10 G10 G9 G8 G7 G6 T11 G11 G10 T12 G12 G10 T13 G13 G10 T14 G14 G10

32 Demography: Standard approach Generations are born when they enter labour force They live for certain number of years At specified age they die with certainty Fixed (certain) life expectancy One variable parameter fertility rate

33 Problems with standard approach Population structure is a result of three demographic processes Fertility (people appear at the beginning) Migration (people appear/disappear in the middle) Mortality (people disappear in the middle and at the end)

34 Why use standard approach? Certainty is required to use standard household optimisation problem Household Utility Function U Household Budget Constraint Euler Equation a 1 (1 ) a 1 1 cat 1 A a+1,t+1 = A a,t (1+r t+1 )+W a,t -C a,t æ = 1+ r ö t+1 ç C a,t, è 1+ r ø C a+1,t+1 1/q

35 Demography: Our Approach Variable mortality rate (includes both mortality and migration) Life expectancy uncertainty on individual level but not on generation level, as mortality schedule is known today and at every point in time in the future Perfect annuity market (accidental bequests distributed implicitly) Alternative bequests distributed between children (e.g., Hans Fehr, Sabine Jokisch, Laurence J. Kotlikoff, 2003, 2004, 2005)

36 New household problem Household Utility Function U 1 t a, s a 1 a, t (1 ) 1 a 1 Household Budget Constraint A a+1,t+1 = 1 s a,t a c ( A a,t (1+ r t+1 )+W a,t -C a,t ) s a,t -- probability of survival for age a at time t s a,t =pop a+1,t+1 /pop a,t

37 New household problem Household Utility Function U 1 t a, s a 1 a, t (1 ) 1 a 1 Household Budget Constraint A a+1,t+1 = 1 s a,t a c ( A a,t (1+ r t+1 )+W a,t -C a,t ) s a,t -- probability of survival for age a at time t s a,t =pop a+1,t+1 /pop a,t Euler Equation æ = 1+ r ö t+1 ç C a,t, è 1+ r ø C a+1,t+1 1/q

38 Government revenues Two taxes Wage tax Consumption tax Separate pension contributions Return on government assets UK fiscal transfer

39 Government expenditures Three types of public consumption 1. Age-independent (fixed level per capita) 2. Health expenditures (mostly for old age) 3. Education expenditures (mostly for young age) Transfers (qualification-specific) Pensions (for 65+ year old)

40 Age-specific public expenditures At the moment calibrated on the National Transfer Accounts (NTA) for Canada NTA measure economic flows across age groups in a manner consistent with National Accounts. This is a very new approach to measuring intergenerational transfers UK NTAs are in the process of preparation David McCarthy and James Sefton at Imperial college represent the UK in the NTA project First estimates of UK National Transfer Accounts (April 2011)

41

42 Age-related components of the model Age-specific labour force participation rates Age-qualification-specific productivity profile Age-specific government spending on health and education Age-specific consumption profile (from calibration stage)

43 ASIDE: Ageing and Consumption A key part of the ageing impacts puzzle is missing: (1) Consumption of almost any good or service differs by age (2) Age-consumption profile is inverted U- or J-shaped With population ageing there will be increases in demand for some goods/services and decreases in demand for other goods and services Average consumption increases and then decreases with age Clear supply-side effects. Requires multi-sector modelling

44 5. Results 1: Size versus Age-specific Effects Hypothesis: Relative importance of size versus age-specific effects" Demographic shock: ONS 2010-based principal population projection Run model with age-specific effects turned off Run model with age-specific effects turned on Macro-economic variable of interest: Output per person

45

46

47 5. Results 2: Immigration and Economic Growth Grow the labour force Hypothesis: Relative importance of different levels of net migration Demographic shock: 2010-based variant population projection with different level of net-migration Run model with different levels of net-migration Macro-economic variable of interest: Output per-person

48 Additional Assumptions Levels of net-migration: 1. Zero 2. 10, , , , ,000 Sex ratio: 50% female, 50% male Age structure of immigration: < age 45 distributed equally

49

50 5. Results 3: Do Nothing and Hope for Technological Change How big of a leap is needed? What rate of technological change is needed

51 Solow Residual estimate of TFP growth Rise in output with constant capital and labour Rate of growth in total factor productivity Not an explanation Growth accounting exercise or growth decomposition Question: What causes the residual?

52 Estimate of TFP growth Solow (1957): USA, , 1-2% per year Maddison (1987): , France 4.0% Germany 4.3% Japan, 5.8% UK, 3.1% USA, 1.9% Kohli and Werner (1997): Korea, , 3.5% 1971, 4.2% 1991, 2.4%

53 Technological shocks Positive shocks steepens the production function Examples of big shocks: 1. Information technology/computing revolution Krueger (1993): USA, : Wages 10-15% higher for jobs that involve using PCs 2. Reduction in shipping costs: Wind-to coal-powered freight ships 3. Large reductions in mortality: Labour productivity increase caused by improved health. Healthy workers are more productive workers, increased participation, increased labour supply

54 Baseline scenario: circa 15% welfare loss How much technological change is needed to counteract this welfare loss? Arithmetic suggests total factor productivity growth of 0.14% year is needed to keep output per-person constant Over the past two decades in the UK total productivity growth over the past two decades has averaged around 2% per year So about 7% of the recent past historical growth is needed.

55 Technological change needed to counteract population ageing is considerable. What is going to be the source of this change? Complications: Age-bias in technological change Skill-bias in technological change

56 6. Concluding Comments?

57 Special issue of Journal of Economic Modelling Volume 35, 2013

58 Demographic Applications of Over-lapping Generations Computable General Equilibrium (OLG-CGE) Models Giorgio Garau, Patrizio Lecca and Giovanni Mandras, The Impact of Population Ageing on Energy Use: Evidence from Italy Katerina Lisenkova, Marcel Mérette and Robert Wright, Population Ageing and the Labour Market: Modelling Size and Age-specific Effects Patrick Georges, Katerina Lisenkova and Marcel Mérette, Can the Ageing North Benefit from Expanding Trade with the South? Axel Börsch-Supan and Alexander Ludwig, Modelling the Effects of Structural Reforms and Reform Backlashes: The Cases of Pension and Labour Market Reforms

59 CGE-OLG Demographic Modelling Teams/Research Groups Department of Economics, University of Ottawa: Marcel Mérette National Institute for Economic and Societl Research: Katya Lisenkova Max Planck Institute for Social Law and Social Policy: Axel Börsch-Supan Vienna Institute of Demography: Alexia Fürnkranz-Prskawetz INGENUE Team, France: CEPII, CEPREMAP, MINI-University of Paris X and OFCE: Vincent Touze Department of Human and Social Sciences, University of Calgliari: Giorgio Garau Ministry of Economy, Trade and Industry: Kazuhiko Oyamada Regional Economic Applications Laboratory, University of Illinois at Urbana-Champaign: Geoff Hewings Department of Economics, Loyola University, Seville, Alejandro Cardenete