long-run growth scenarios for the world economy

Journal of Policy Modeling 32 (2010) 64–80

Available online at www.sciencedirect.com

Long-run growth scenarios for the world economy

Romain Duval ∗, Christine de la MaisonneuveOECD Economics Department, 2 rue André Pascal, 75016 Paris, France

Received 1 May 2009; received in revised form 1 September 2009; accepted 1 October 2009Available online 17 November 2009

Abstract

This paper develops and applies a simple “conditional growth” framework to make long-term GDP pro-jections for the world economy, taking as a starting point the recent empirical evidence about the drivers ofexisting cross-country income disparities. Human capital is projected by cohorts, and allowance is implicitlymade for the impact of ageing and potential labour market and pension reforms on employment growth.Leaving aside deeper sources of uncertainty such as model and parameter uncertainty, projections are foundto be sensitive to future economic policies in the areas of education, pensions, labour markets and climatechange mitigation, and even more so to total factor productivity and population trends. A baseline scenarioprojects fairly stable world GDP growth of about 3.5% annually on average (in PPP terms) over 2005–2050.© 2009 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved.

JEL classification: O11; O43; O47; O57; Q43; Q54

Keywords: Growth; Long run; Projections; Human capital; Cohorts

1. Introduction

A long-term growth framework for the world economy is key to exploring a number of long-run economic issues, such as ageing, fiscal sustainability, migrations, or natural resources. Worldeconomic scenarios have proliferated in recent years in the context of climate change projec-tions, such as those developed by the Intergovernmental Panel on Climate Change (IPCC), whichtypically assume a convergence process whereby the income levels of less-developed countriesgradually, and at least partially, catch-up to those of more developed economies (Nakicenovic etal., 2000). The vast majority of projections focus on convergence at the macroeconomic level, in

∗ Corresponding author.E-mail address: [email protected] (R. Duval).

0161-8938/$ – see front matter © 2009 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved.doi:10.1016/j.jpolmod.2009.10.001

mailto:[email protected]

dx.doi.org/10.1016/j.jpolmod.2009.10.001

R. Duval, C. de la Maisonneuve / Journal of Policy Modeling 32 (2010) 64–80 65

terms of GDP per capita or GDP per worker (the “top-down” approach) while a few others assumesome gradual catch-up at the sectoral level (the “bottom-up” approach, see e.g. McKibbin, Pearce,& Stegman, 2004). In both cases, climate modellers have typically relied on simple assumptionsregarding the form and the speed of convergence, without explicitly specifying the policy assump-tions underlying their scenarios. This, and the fact that in the IPCC’s Special Report on EmissionScenarios (SRES) a large number of possible outcomes are presented as being equally likely, mayhave contributed to strengthening the impression of uncertainty that is inherent to any long-runworld economic projections (see Fisher et al., 2007; Nakicenovic et al., 2000).

In this paper, special emphasis is put on setting up a theoretical framework that explicitlyintegrates some of the current theoretical and empirical knowledge regarding the long-run driversof economic growth. At the same time, reflecting both data constraints and the wide diversity ofexisting growth theories, a simple and fairly consensual framework must be retained. While thereis no agreement on any single theoretical model of economic growth, a basic empirical consensusseems to support the so-called “conditional convergence” hypothesis, under which each countrywould be expected to converge to its own steady-state level of GDP per capita, which in turncan be determined by a host of factors including investment rates in physical and human capital,policies and institutions more broadly, demographics, geography, etc. (Barro & Sala-i-Martin,2004). Because these factors may vary permanently across countries, in the long-run differencesmay remain in per capita income levels but not in growth rates.

In order to incorporate the conditional convergence hypothesis into the projections, this paperadopts the most widely used theoretical framework in recent empirical analyses of cross-countrydisparities in per capita incomes (Caselli & Coleman, 2006; Caselli, 2005; Easterly & Levine,2001; Hall & Jones, 1999; Jones, 1997; Klenow & Rodriguez-Clare, 1997; OECD, 2004). Cross-country variation in the levels of output per worker is typically decomposed into parts attributedto the variation in physical and human capital per worker and total factor productivity (TFP).Such exercises have found TFP and – to a lesser extent – human capital to be the main drivers ofcurrent disparities in living standards between developed and developing economies. With thisdecomposition at hand, long-run output per worker scenarios for each country may then be builtup by projecting each of the three components (see e.g. Jones, 1997). Such scenarios can thenbe combined with population and employment rate scenarios to yield GDP projections at bothcountry and world level.

The remainder of this paper proceeds as follows. Section 2 presents the theoretical framework.Section 3 builds up a range of long-run GDP growth scenarios, each based on alternative assump-tions concerning the underlying drivers of growth. Section 4 presents and discusses long-runworld growth projections under each of these scenarios. The exercise illustrates the inherentlylarge uncertainty – and how it has been further amplified by the 2008–2009 financial and economiccrisis – and the influence of economic policies (e.g. in the areas of education, labour markets orclimate change mitigation) on long-run projections. Section 5 concludes.

2. Theoretical framework and basic assumptions underlying its empiricalimplementation

2.1. Theoretical framework

We consider a standard aggregate Cobb-Douglas production function with constant returns toscale featuring physical capital, human capital, and labour as production factors and Harrod-neutral technological progress, and assume that production function is invariant both across

66 R. Duval, C. de la Maisonneuve / Journal of Policy Modeling 32 (2010) 64–80

countries and over time (see e.g. Mankiw, Romer, & Weil, 1992):

Yt = Kαt (AtHt)

1−α = Kαt (AthtLt)

1−α (1)

where Kt, Yt, At, ht, and Lt denote capital, output, TFP, human capital per worker and employ-ment, respectively. The constant-returns-to-scale assumption excludes de facto the possibility ofpermanent growth effects of the human capital stock. While some empirical literature hints atendogenous growth effects of human or even physical capital (see e.g. Bassanini & Scarpetta,2004; Bond et al., 2004), this issue remains fairly controversial in practice (e.g. Sianesi & VanReenen, 2003).

After some basic manipulations, GDP per capita can be decomposed as follows:

Yt

Popt

=(

Kt

Yt

)α/(1−α)

Atht

(Lt

Popt

)(2)

where Kt/Yt, and Lt/Popt denote the capital/output ratio and the employment rate (defined here asthe ratio of employment to total population), respectively.

Such decomposition can be performed for a base year, and long-run projections then be madefor each of the four components in order to project the future path of GDP per capita.

2.2. Base-year decomposition of cross-country GDP per capita gaps

In order to perform the decomposition featured in Eq. (2) for a base year, comparable data areneeded across countries for labour productivity, TFP, the capital/output ratio, human capital perworker and the employment rate. This in turn implies a number of simplifications to facilitateharmonisation, not least regarding physical capital stocks. Data sources and data constructionmethodology are described in detail in the Working Paper version of this paper (Duval & de laMaisonneuve, 2009).

Two base years are considered, 1995 and 2005. Labour productivity levels are computed foreach country as the ratio of GDP in 2005 constant PPP US$ to employment, using the PPP estimatespublished within the context of the International Comparison Program (ICP) coordinated by theWorld Bank (World Bank, 2007).1

Following Hall and Jones (1999), capital stocks (in constant 2005 national currency prices) arebuilt up from investment series through the perpetual inventory method, assuming a 5% annualdepreciation rate. The rationale for following such a simple approach is to allow capital stocksto be constructed for, and compared across a large number of countries, which is essential in thecontext of convergence scenarios. Long investment time series – dating back at least to the early1970s – are used, so that capital stock estimates in 1995 and 2005 are largely insensitive to thechoice of the initial value.2 The capital stock estimate is then divided by GDP (also expressed inconstant 2005 national currency prices) to obtain the capital/output ratio.

1 The 2007 vintage of PPP estimates include noticeable revisions with respect to past data. For instance, GDP percapita levels in China and India for the year 2005 are now estimated to be equal to about 10% and 5% of the US level,respectively, versus 16% and 8% previously.

2 The initial value of the capital stock is computed as I0/(δ + g), where I0 denotes investment for the first available year,δ is the depreciation rate of the capital stock (set here at 5%) and g is average GDP growth rate between periods 0 and10. This is the capital stock that would prevail along a balanced growth path where GDP growth and the investment ratewould be constant.


Human capital stocks are constructed in two steps. First, data on the average number of yearsof schooling across population aged 25–64 are assembled for a wide range of countries, usingprimarily the dataset by 5-year age groups constructed by Cohen and Soto (2007), which in turnis consistent with, but improves on the Barro-Lee dataset (Barro & Lee, 1993, 2001). Second,the average number of years of schooling across the population is converted into a human capitalstock based on an assumption regarding returns to education. Following Hall and Jones (1999),and relying on microeconomic evidence on returns to schooling for many countries surveyed inPsacharopoulos (1994) and Psacharopoulos and Patrinos (2004), the marginal return to schoolingis set equal to 13.4% for the first four years of education, 10.1% for the next four and 6.8% beyondthe eighth year. While the magnitude of social returns to education – and of possible externalitiesto education in particular – remains subject to uncertainty, the general pattern of falling returns bylevel of education is well established, and a 6.5–7% average return to upper secondary and tertiaryeducation is broadly consistent with both microeconomic and macroeconomic evidence (seeBassanini & Scarpetta, 2001; Gemmel, 1996; Harmon, Oosterbeek, & Walker, 2003). Formally, itis assumed that human capital per worker can be written as ht = eφ(S), where S denotes the numberof years of schooling, and φ(S) is chosen to be a piecewise linear function in order to reproducethe three different marginal returns to education used here for three different levels of education.Bils and Klenow (2000) argue that such specification is the appropriate way to incorporate yearsof schooling into an aggregate production function.

TFP is then derived from GDP per capita, physical capital stock, human capital stock andemployment rate data, re-arranging equation (2) as follows:

At = (Yt/Popt)

[(Kt/Yt)α/(1−α)ht(Lt/Popt)],

where the capital share α is set equal to 1/3.Results from this decomposition are presented in Table 1 for the year 2005. In line with

findings from recent literature – which Easterly and Levine (2001) labeled the “new stylised factsof growth” –, and in contradiction with basic neo-classical growth theory (see e.g. Mankiw etal., 1992), TFP appears to be the main driver of existing cross-country differences in GDP percapita. Cross-country variance in the logarithm of TFP is found to account for over half of thecross-country variance in the logarithm of GDP per capita. Human capital is also found to playan important role. This suggests that both these factors are likely to be the major drivers of anylong-run convergence scenario.

3. Economic scenarios for the world economy up to 2050

With the decomposition presented in Table 1 at hand, long-run scenarios can now be drawnfor each of the four drivers of GDP per capita, namely employment, human capital accumulation,capital intensity and TFP growth. Up to 2009, GDP, investment and employment projections aretaken from the OECD Economic Outlook for OECD countries as well as for Brazil, China, Indiaand Russia, and from the IMF World Economic Outlook for all other countries. This allows someextension of the GDP decomposition up to 2009. Starting from 2010, alternative scenarios arethen built up for each of the four GDP components for each country up to 2050, which can thenbe combined at will. These scenarios are described below, and summed up in Table 2. This isfollowed by a description of the approach taken for those countries – representing about 10% ofworld GDP – where the projection framework cannot be applied for lack of accurate data, and bya discussion of sectoral issues of relevance for long-run economic projections.


Table 1Decomposition of cross-country differences in GDP per capita into their broad determinants, 2005a,b (United States = 100).

GDP PPP per capita TFP Human capital Physical capital EmploymentY/Pop A h (K/Y)α/(1−α) L/Pop

United States 100.0 100.0 100.0 100.0 100.0Canada 83.5 72.0 103.3 105.8 106.0Japan 72.6 52.6 100.4 130.7 105.1China 9.8 13.6 57.3 105.2 119.5India 5.2 12.7 47.7 98.3 87.1Brazil 20.5 29.3 70.1 103.1 96.8Russian Federation 28.6 31.5 84.9 97.4 99.3

Australia-New Zealandc 78.3 64.1 101.5 114.8 104.5EU27 + EFTAc 64.7 67.8 91.2 114.1 91.3Rest of the Worldc 12.3 20.9 59.7 103.6 81.7Total Worldc 22.8 27.9 64.2 104.2 95.8

a While equal in principle, Y/Pop and the product of A, h, (K/Y)α/(1−α) and L/Pop can differ in practice for two reasons:first, for countries where fossil fuels extraction makes a sizeable share of overall output (Russian Federation and a numberof countries in the Rest of the World aggregate), TFP levels were estimated for total output excluding the mining andquarrying sector, for reasons explained in the text. Second, geographical area aggregates are computed as arithmeticaverages, while geometric means would have to be used for the equality Y/Pop = Ah(K/Y)α/(1−α)L/Pop to hold.

b The long-term growth framework is applied at the individual country level. The geographical disaggregation of theworld economy presented here matches that of the OECD ENV-Linkages model, as used in Burniaux et al. (2008).

c Population-weighted arithmetic averages.

3.1. Employment, human capital accumulation, capital intensity and TFP growth scenarios

3.1.1. Employment scenariosFuture employment is projected by decomposing employment into population, the participation

rate and the unemployment rate:

• Population scenarios are taken from the United Nations (UN), which publishes baseline, highand low case scenarios that differ according to their fertility assumptions. In the baseline,fertility in all countries converges eventually – beyond the 2050 projection horizon – towarda level of 1.85 children per woman, while the low (high) case scenario assume that fertilitywould constantly remain 0.5 children lower (higher) than in the baseline.

• Two unemployment scenarios are considered, each relying implicitly on different assumptionsregarding future labour and product market policies: a baseline scenario assumes that futurereforms will gradually bring unemployment down to 5% by 2050, while a low case scenariostabilises unemployment in all countries at their average 1995–2005 rates.

• Two labour force participation scenarios are constructed. A baseline scenario assumes boththat pension and labour market reforms will be undertaken in order to remove early retirementincentives where these remain high, and that effective retirement ages will be indexed to lifeexpectancy. Concretely, in the top quintile of OECD countries where participation is currentlyhighest (Canada, Denmark, Iceland, New Zealand, Norway and Sweden), future effective retire-ment ages are partially indexed to life expectancy, so as to maintain a constant share of lifespent in retirement. In all other countries where data by cohorts are available, participation foreach age group is then assumed to converge gradually to the average of the top quintile, implic-

R.D

uval,C.de

laM

aisonneuve/JournalofPolicy

Modeling

32(2010)

64–8069

Table 2Alternative long-run scenarios for each of the four drivers of GDP per capita.

Employment Education TFP growth Physical capitalintensity (Worldsaving rates)

Pension and labourmarket reforms toraise labour forceparticipation

Labour marketreforms to lowerunemployment

Population Human capitalaccumulation

TFP catch up Frontier TFPgrowth

Long-runcapital/outputratios

A B C D E F G

Baselineassump-tions

1 Partial indexationto longevity inhigh-participationcountries; othersgradually convergeto levels inhigh-participationcountries

Unemploymentrates graduallyconverge to 5% by2050

Baseline UNpopulationscenario (Baselinefertility scenario)

Pace ofaccumulation byyounger cohortstends to averageworld pace over1960-00

Convergencespeed tends to 2%

TFP growth at thefrontier is 1.5%

Capital/outputratios converge tocurrent US level

Low caseassump-tions

2 No indexation tolongevity inhigh-participationcountries; othersgradually convergeto levels inhigh-participationcountries

Unemploymentstabilised ataverage 95–05rates

Low fertility Pace ofaccumulation byyounger cohortstends to zero(cohort effectsonly starting from2025)


TFP growth at thefrontier is 1%

Capital/outputratios converge to40% below currentUS level

High caseassump-tions

3 Full indexation tolongevity inhigh-participationcountries; othersconverge to levelsinhigh-participationcountries fasterthan in baselineand low case

High fertility Pace ofaccumulation byyounger cohortstends to averageOECD pace over1960-00


TFP growth at thefrontier is 2%

Capital/outputratios converge to40% above currentUS level


itly embedding the effect of future reforms.3 Assuming convergence in participation by agegroup rather than on aggregate allows taking into account permanent cross-country differencesin participation that may still be associated with country-specific demographics. Aggregateparticipation rates are then projected using a cohort approach, as used for instance in Burniaux,Duval, and Jaumotte (2003). By contrast, a low case, “business-as-usual” scenario assumes nopension and labour market reforms, as well as no indexation of effective retirement ages onlife expectancy.

3.1.2. Human capital accumulation scenariosTwo main stylised facts stand out regarding world accumulation of human capital, which

can be useful for the projection. First, based on the historical dataset by cohorts used in thispaper, educational attainment of the 25–29 age group increased on average at a fairly regularpace of about 1.1 years per decade between 1960 and 2000 in less-educated countries, definedas all countries except those 15 countries where educational attainment was highest back in1960. Second, educational attainment has tended to level off in those countries where it washighest, e.g. Australia, Switzerland or the United States. A hypothesis that may be drawn fromthis observation is that the speed of human capital accumulation typically slows as educationalattainment increases.

Against this background, the baseline human capital scenario assumes that educational attain-ment of the 25–29 age group remains constant in the future in the country where it is currentlyhighest (South Korea, with 14.4 years of education in 2000). Up to 2015, recent trends are extrap-olated by assuming that all other countries converge to the above “frontier” at their average1990–2000 speed (zero if educational attainment actually declined over this period). Between2015 and 2025, the speed of convergence is assumed to converge gradually to the average speedobserved in the average “less educated” world country over the period 1960–2000. Based on thisscenario for the 25–29 age group, future educational attainment across the population aged 25–64is projected through cohort effects, and then converted into a human capital stock based on thereturns to education assumptions described above.

Two alternative scenarios are also considered, each corresponding to different assumptionsregarding future education policies. A low case scenario assumes no further progress in education,so that human capital increases only through cohorts effects beyond 2025. By contrast, in the highcase scenario, the pace of human capital accumulation by new cohorts would tend towards theaverage pace observed across the OECD over 1960–2000, as opposed to the lower pace observedacross the world in the baseline. This would result in a total increase of about 4.2 years between2005 and 2050 in the average educational attainment of the 25–29 age group in less-educatedcountries, as opposed to about 3 years in the baseline.

3.1.3. Physical capital intensity scenariosIn a world where international capital is at least partly mobile, future physical capital accumu-

lation at the country level should be driven at least partly by real interest rates at the world level,which in turn should reflect the world saving-investment equilibrium. In the absence of a trulyglobal, integrated world growth framework, world saving trends cannot be factored in the analysis

3 Countries for which participation data by cohorts are available include the OECD countries, Brazil, Russia, India andChina. For those countries where data by cohorts are not available, the convergence assumption is similar but is applieddirectly to aggregate participation rather than to the participation of each individual age group.


undertaken here, so that their impact on future world investment via real interest rates cannot beexplored. Nevertheless, it is still possible to incorporate the fact that capital/output ratios shouldnot diverge permanently across countries in a world of integrated capital markets, as this wouldimply permanent cross-country differences in marginal returns to capital αY/K.

This is implemented here by running three alternative capital intensity scenarios. In a baselinescenario, the US investment rate path gradually stabilises the capital/output ratio at its currentlevel, implicitly assuming that the United States is on a balanced growth path. Investment ratesin all other countries are then projected to vary in such a way that capital/output ratios convergegradually to the US level, with full convergence being achieved only by 2100, i.e. beyond the 2050horizon of this paper.4 Low and high case scenarios implicitly make alternative world saving rateassumptions. All capital output ratios – including the United States’ – converge to 40% below andabove current US levels, respectively. In the production function framework adopted here, thiscorresponds to steady-state world saving and investment rates being about 5 percentage pointslower and higher than in the baseline, respectively.5

3.1.4. TFP growth scenariosFuture TFP growth rates are modeled as being driven by two parameters, namely the speed of

technological progress in innovating countries that are close to the productivity “frontier” and thespeed of catch-up of countries that are lagging behind. Because technology circulates freely acrossthe world over long horizons, it might seem reasonable to expect TFP levels in lagging countries togradually catch up to those of technological leaders. At the same time, a host of factors – includingpersistent cross-country differences in policies and institutions, e.g. regulatory, trade and FDIrestrictions, property rights, institutions, etc. – may slow down technology diffusion and preventfull TFP convergence (see e.g. Acemoglu, 2009). The scenarios implicitly assume that such factorswill continue to play a role in the future, albeit to different degrees across different scenarios.

Against this background, the “frontier” here is assumed to be not just the United States but ratheran (lower) average of TFP levels in “high-TFP” OECD countries.6 Three alternative TFP growthscenarios are considered for these most productive economies, which are meant to illustrate in astylised way the consequences of uncertainties associated with the long-term pace of innovation.In a baseline scenario, TFP is assumed to grow at a 1.5% annual rate between 2010 and 2050.This is slightly above the 1.3% average annual growth rate over 1990–2006 observed for theUnited States in the dataset, and yields medium-term potential GDP growth estimates for thiscountry which are consistent with OECD and IMF medium-term projections as published in theirEconomic Outlook and World Economic Outlook, respectively.7 Low and high case scenarios thenassume 1% and 2% average annual TFP growth over the projection period, respectively.

4 Formally, the investment rate path chosen for each country is the smoothest possible path that meets the constraintthat the capital/output ratio is equal to the US level in 2100. Convergence is assumed by 2100 rather than by 2050because convergence by 2050 would imply large upward shifts – over and above those driven by TFP convergence andimprovements in human capital – in investment rates over the coming decades in those (developing) countries wherecapital/output ratios were particularly low in 2005.

5 This can be inferred from the fact that at the steady-state, the world saving rate is k*(g + δ)/(1 + g), where k* is thesteady-state capital/output ratio, g is the world GDP growth rate – which in turns equals the sum of world TFP growthand population growth – and δ is the depreciation rate of the capital stock.

6 Based on estimated TFP levels in 2005, these countries are Austria, Belgium, Canada, Denmark, Finland, France,Greece, Ireland, Italy, the Netherlands, Norway, Spain, Sweden, the United Kingdom and the United States.

7 Note that in our framework, the 1.5% TFP growth assumption implies that in the long run, assuming the educa-tional attainment of younger generations ultimately levels off and the capital/output ratio reaches its steady-state, labourproductivity will also grow at 1.5% a year.


The resulting TFP path in most productive countries under each scenario defines the frontiertowards which all lagging countries converge. Such convergence is assumed to be driven by thefollowing dynamic equation:

d ln(TFPt) = d ln(TFP∗t ) − β[(ln(TFPt−1) − ln(TFP∗

t−1)], (3)

where TFP* is the “frontier” described above and β is the speed of convergence.Up to 2015, β is set equal to its actual average value observed over the period 1995–2005

(zero in case of divergence). Between 2015 and 2025, β is assumed to converge gradually tothree alternative values, each embedding implicitly different assumptions regarding policy andinstitutional drivers of technology diffusion. A baseline scenario sets the eventual value of β equalto 2% – meaning that the TFP gap with respect to frontier closes by 2% each year –, and to 1% and3% respectively in low and high case scenarios. The baseline value corresponds to the averageestimated speed of convergence in GDP per capita across a wide range of datasets and econometricmethods reported by Barro and Sala-i-Martin (2004).8 Applying such speed of convergence toTFP seems reasonable given that TFP is the main driver of cross-country differences in GDPper capita levels (Table 1; Hall & Jones, 1999) and growth rates (Easterly & Levine, 2001). Thisfigure is also not inconsistent with the 3% TFP convergence speed estimated in recent analysisof TFP convergence across OECD countries (Nicoletti & Scarpetta, 2003). It is also close to thespeed of convergence observed in the data for China over 1995–2005.

3.1.5. Baseline scenario for those countries where the framework cannot be appliedThe above approach is applied to 76 individual countries covering over 90% of world GDP

and world population in 2005. For all other countries, human and/or physical capital data aretoo scarce or too unreliable to apply the framework with a reasonable degree of confidence. In anumber of cases, for instance, investment time series are too short to estimate capital stocks witha reasonable degree of confidence. The approach followed for these countries is to apply to labourproductivity the methodology applied to TFP for those countries where the above frameworkcould be applied.

3.2. Sectoral issues

One missing element from most aggregate projections, especially those made in order toexplore issues related to natural resources or climate change, is that GDP is seldom endogenisedfor those countries where fossil fuels extraction makes a sizeable share of overall output. Forexample, it makes little sense to project future GDP in OPEC countries regardless of the futurepaths of oil supply. In order to overcome this issue, for those countries where fossil fuels matter,the convergence scenarios described above are in fact applied not to GDP but rather to GDPexcluding the mining and quarrying sector.9 GDP excluding mining and quarrying is projectedusing the approach described in Section 3.1, and the value added in mining and quarrying – andtherefore overall GDP – is then determined by running a global, multi-sector, general equilibriummodel, the OECD’s ENV-Linkages (Burniaux & Chateau, 2008). With its nested-CES structurefeaturing a detailed representation of energy inputs at the sectoral level, the model is particularlysuitable to project energy supply, demand and prices.

8 Summing up the evidence, the authors argue p. 497 that “one surprising result is the similarity of the speed of β

convergence across data sets. The estimates of β are around 2–3% per year in the various contexts”.9 These countries are OPEC countries as well as Norway and Russia.


ENV-Linkages being a multi-sector model, the variables projected in Section 3.1 for the econ-omy as a whole are not sufficient to run it, and sectoral assumptions are required. Such assumptionsare also useful when long-run economic projections aim at exploring issues that have a majorsectoral dimension, such as the economic cost of possible future international policies to reducegreenhouse gas emissions, which will be discussed in Section 4. The starting point for projectingsectoral output and value-added growth is that history consistently points to different productivitytrends across sectors, as illustrated for instance by the long sector-level productivity growth timeseries assembled by the Groningen Growth and Development Centre for a wide range of countries(Groningen Growth and Development Centre, 2006; Timmer & de Vries, 2007; Van Ark, 1996).Productivity growth has typically been found to be faster in agriculture and manufacturing than inconstruction, transport and – to an even greater extent – other services. In order to ensure the con-tinuation of these historical patterns over the projection horizon, sectoral productivity growth inthe ENV-Linkages model is calibrated in such a way that particular relative sectoral productivitygrowth patterns and the aggregate productivity growth scenario (excluding mining and quarrying)both hold ex post (for details, see Duval & de la Maisonneuve, 2009).

The approach followed in this paper also addresses the criticisms made recently in the cli-mate change economics literature towards using market exchange rate (MER)-based economicprojections (Castles & Henderson, 2003a, 2003b; Henderson, 2005; Nordhaus, 2007). In most ofthe scenarios published over the past two decades, including in the IPCC SRES, current cross-country differences in income per capita levels have been over-estimated due to the use of MERs toconvert national GDPs into a common currency, which ignores the well-known “Baumol-Balassa-Samuelson” effect. Within the context of any convergence scenario, such over-estimation is likelyto translate into an over-estimation of future GDP and greenhouse gas emissions growth, ceterisparibus. Here, no such problem arises because PPPs, not MERs, are used to compare initialincome per capita levels and compute the economic convergence scenario.

4. Results

Table 3 presents the main features of a baseline world economic scenario that combines thebaseline employment, human capital, TFP and capital intensity scenarios described above andsummarised in Table 2. In this scenario, world GDP growth would remain fairly stable at about3.5% annually (in PPP terms) on average over the coming decades. When expressed in constant2005 MER (market exchange rate) US$, baseline world GDP per capita growth up to 2030 fallsroughly in the middle of the 1–3.1% range provided in the IPCC SRES, which also relies onconstant MER US$. This baseline scenario also has a number of interesting country-specificimplications, such as the fact that growth could be lower in China than in India over the comingdecades. This is because compared with India, China is already fairly capital intensive, has virtuallyno room for further raising labour force participation, and is bound to face a significant slowdownin population growth.

Interestingly, looking at alternative scenarios points to some influence of future policies inthe areas of education, pensions and labour markets on projected world GDP growth up to mid-century, despite the fact that the growth impact of such policies ultimately fades out and cannot bepermanent in our theoretical framework (Table 4). Compared with the baseline scenario (Column4), a scenario under which unemployment rates would remain at their current levels rather thandecline to 5% would have virtually no impact on projected annual world GDP growth (Column3). However, a scenario where no labour market and pension reforms are undertaken to raiselabour force participation and index effective retirement ages to life expectancy would shave 0.2

74R

.Duval,C

.dela

Maisonneuve

/JournalofPolicyM

odeling32

(2010)64–80

Table 3Baseline world economic scenario: main features (average annual growth rates).

GDP perworker

GDP percapita

GDP

2000–2005

2005–2015

2015–2025

2025–2050

2005–2050

2000–2005

2005–2015

2015–2025

2025–2050

2005–2050

2000–2005

2005–2015

2015–2025

2025–2050

2005–2050

United States 1.8 1.2 1.7 1.5 1.5 1.5 1.0 1.6 1.6 1.5 2.5 1.9 2.4 2.2 2.2Canada 0.7 1.0 1.7 1.6 1.5 1.5 1.1 1.5 1.5 1.4 2.5 2.0 2.2 2.0 2.1Japan 1.7 0.7 2.0 2.1 1.8 1.3 0.6 1.9 1.7 1.5 1.4 0.5 1.5 1.0 1.0China 8.4 7.6 5.2 3.7 4.9 8.8 7.7 4.6 3.3 4.6 9.5 8.3 5.0 3.2 4.7India 4.6 4.8 4.4 4.6 4.6 5.2 5.9 5.5 5.0 5.3 7.0 7.4 6.7 5.5 6.2Brazil −0.4 1.9 2.6 3.2 2.8 1.3 2.2 3.1 3.4 3.0 2.7 3.5 4.0 3.8 3.8Russian

Federationa5.3 2.7 3.3 3.0 3.0 6.6 3.7 3.3 2.9 3.1 6.1 3.1 2.6 2.1 2.5

Australia-NewZealand

1.0 1.2 1.7 1.6 1.5 2.0 1.5 1.5 1.6 1.5 3.3 2.4 2.3 2.1 2.2

EU27 + EFTA 1.1 1.2 1.9 1.8 1.7 1.5 1.5 2.1 1.9 1.8 1.8 1.6 2.1 1.7 1.8OPEC + other

oilproducersa

1.9 2.0 2.9 4.0 3.3 2.7 2.9 3.4 4.4 3.9 4.5 4.6 4.7 5.3 5.0

Rest of theWorld

1.8 1.9 2.8 3.4 2.9 2.5 2.3 2.9 3.6 3.2 4.2 3.9 4.4 4.7 4.4

Total World 1.8 1.9 2.5 2.8 2.5 2.2 2.3 2.7 2.9 2.7 3.5 3.4 3.6 3.4 3.5

a For countries where fossil fuels extraction makes a sizeable share of overall output, the GDP excludes the mining and quarrying sector.

R.D

uval,C.de

laM


Modeling

32(2010)

64–8075

Table 4The impact of education, labour market and pension policies (GDP growth, 2005–2050).

1 2 3 4 5 6No educationreforms

No pension andlabour marketreforms to raiselabour forceparticipation

No labourmarket reformsto lowerunemployment

Baseline(%)

Fastereducationreforms

More radicalpensionreforms

(Baseline but D2instead of D1 inTable 2)

(Baseline but A2instead of A1 inTable 2)

(Baseline but B2instead of B1 inTable 2)

(Baseline but D3instead of D1 inTable 2)

(Baseline but A3instead of A1 inTable 2)

(Difference with respect to baseline, in percentage points) (Difference with respect to baseline, in percentage points)

United States 0.0 −0.3 0.0 2.2 0.0 0.0Canada 0.0 −0.1 −0.1 2.1 0.0 0.0Japan 0.0 −0.1 0.0 1.0 0.0 0.0China −0.1 −0.1 0.1 4.7 0.2 0.1India −0.2 −0.4 0.0 6.2 0.2 0.1Brazil 0.0 −0.2 −0.1 3.8 0.1 0.1Russian Federationa −0.1 −0.3 −0.1 2.5 0.1 0.1Australia-New Zealand 0.0 −0.1 0.0 2.2 0.0 0.0

EU27 + EFTA 0.0 −0.3 −0.1 1.8 0.0 0.1OPEC + other oil producersa 0.0 −0.1 −0.1 5.0 0.1 0.1Rest of the World −0.1 −0.1 −0.1 4.4 0.2 0.1Total World −0.1 −0.2 0.0 3.5 0.1 0.1



percentage points off annual world growth (Column 2), while failure to raise the human capitalof future generations from current levels would reduce growth by 0.1 percentage points annually,with larger effects in some developing regions (Column 1). Overall, a low-case policy reformscenario featuring more pessimistic education, pension and labour market reform assumptionsthan implicitly embedded in the baseline would lead to future world GDP growth being about 0.3percentage points lower than otherwise. By contrast, more ambitious efforts to raise the educationalattainment of new generations (Column 5) or more radical pension reforms that would fully indexeffective retirement ages to life expectancy (Column 6) could each boost world growth by 0.1percentage points relative to baseline.

Capital intensity – and implicitly world saving – also has only transitory growth effects inour framework, but its impact is still significant over the projection horizon, with alternativeassumptions leading projected annual world GDP rates to vary by 0.1–0.2 percentage pointsrelative to baseline (Table 5, Columns 4 and 6). Even more influential on the projections areassumptions made regarding TFP and population growth, unsurprisingly so given that they affectsteady-state growth. A low-case assumption regarding TFP growth at the frontier brings downannual world growth by 0.4 percentage points (Column 3), while slower catch-up to frontier thanembedded in the baseline could cut future growth by 0.2 percentage points (Column 2). Underthe UN low-fertility scenario, world growth would be reduced by 0.2 percentage point (Column1). High-case TFP and fertility scenarios have the opposite effect (Columns 7, 8 and 9). Overall,annual world GDP growth could easily vary by 3/4 percentage points over the projection horizonif low-case or high case scenarios materialise for both TFP and population simultaneously.

Natural resource and environment issues are another source of uncertainty surrounding futureworld growth. Most importantly, the above scenarios incorporate neither the possible damagesfrom climate change nor the costs of addressing it. Damages are likely to remain limited over therelatively short time horizon considered here, however. Plugging the baseline projection above inthe ENV-Linkages model yields a projected temperature increase of about 2 ◦C (relative to pre-industrial levels) by 2050. According to the range of estimates presented in Smith, Schellnhuber,and Monirul Qader Mirza (2001), such a temperature rise might reduce the level of world GDP byless than 1% by the time it occurs, implying a fairly negligible impact on world growth over thenext few decades. By contrast, the cost of addressing climate change could be more significant ata 2050 horizon. For instance, plugging the baseline projection in ENV-Linkages, and simulatinga world carbon tax policy that would cut world emissions by 50% in 2050 relative to 1990 levels –an often debated target in the international climate policy debate10 –, is estimated to reduce annualworld GDP growth by about 0.2% over the projection period (Table 5, Column 10). This estimatefalls roughly in the middle of the range available in the literature (for details, see Burniaux,Chateau, Duval, & Jamet, 2008), and is comparable in magnitude to the growth impact of mostof the alternative assumptions discussed above – e.g. regarding future population growth.

Tables 4 and 5 illustrate some of the uncertainties involved in long-run projections. The mag-nitude of these has arguably been increased by the consequences of the 2008–2009 financialcrisis, which in most industrialised countries resulted in the deepest recession since the GrandDepression. In particular, one open question is whether the financial crisis will induce permanentoutput losses compared with a no-crisis situation and, if so, whether those permanent effects willbe on the level or the growth of output. While it seems too early to make a definitive judgment,

10 This emission reduction target would also be consistent with meeting another often debated target, namely stabilisationof long-run CO2 concentration in the atmosphere at 450 ppm (for a recent example of focus on such a target in this journal,see Bosetti et al., 2009).

R.D

uval,C.de

laM


Modeling

32(2010)

64–8077

Table 5The impact of TFP, capital intensity, population and climate policy assumptions (GDP growth, 2005–2050).

1 2 3 4 5 6 7 8 9 10Lowfertility

Slower TFPcatch-up

Low frontierTFP growth

Low capitalintensity (lowsteady-stateworld savingrates)

Baseline(%)

Highcapitalintensity(highsteady-stateworldsavingrates)

High frontierTFP growth

Faster TFPcatch-up

Highfertility

Climatepolicy

(Baseline butC2 instead ofC1 in Table 2)

(Baseline butE2 instead ofE1 in Table 2)

(Baseline butF2 instead ofF1 in Table 2)

(Baseline butG2 instead ofG1 in Table 2)

(Baseline butG3 instead ofG1 in Table 2)

(Baseline butF3 instead ofF1 in Table 2)

(Baseline butE3 instead ofE1 in Table 2)

(Baseline butC3 instead ofC1 in Table 2)

(Difference with respect to baseline, in percentage points) (Difference with respect to baseline, in percentage points)

United States −0.2 0.0 −0.4 −0.1 2.2 0.1 0.4 0.0 0.2 −0.1Canada −0.2 0.0 −0.4 −0.1 2.1 0.1 0.4 0.1 0.2 −0.1Japan −0.2 −0.2 −0.4 −0.1 1.0 0.1 0.4 0.2 0.2 −0.1China −0.2 −0.3 −0.4 −0.1 4.7 0.2 0.4 0.4 0.2 −0.4India −0.2 −0.5 −0.4 −0.1 6.2 0.2 0.4 0.5 0.3 −0.2Brazil −0.2 −0.2 −0.4 −0.1 3.8 0.4 0.4 0.3 0.3 −0.1Russian

Federationa−0.2 −0.1 −0.4 −0.1 2.5 0.4 0.4 0.1 0.3 −0.6

Australia-NewZealand

−0.2 0.0 −0.4 −0.1 2.2 0.1 0.4 0.1 0.2 −0.2

EU27 + EFTA −0.2 0.0 −0.4 −0.1 1.8 0.1 0.4 0.0 0.2 −0.1OPEC + other

oilproducersa

−0.2 −0.2 −0.4 −0.1 5.0 0.2 0.4 0.2 0.2 −0.8

Rest of theWorld

−0.2 −0.2 −0.4 −0.1 4.4 0.3 0.4 0.3 0.2 −0.1

Total World −0.2 −0.2 −0.4 −0.1 3.5 0.2 0.4 0.3 0.2 −0.2



recent empirical evidence on the long-term output effects of crises point to level – but not growth –effects of about 4 to 10%, depending on the magnitude of the crisis (Serra & Saxena, 2008; Furceri& Mourougane, 2009). These figures are consistent with the 6% estimate implicitly embeddedin the baseline projection of the present paper, as inferred by comparing the baseline level ofworld GDP in 2050 with its value under a similar scenario where the projection framework isapplied starting from 2007 rather than 2010, i.e. under a scenario that comes close to the baselineprojection that would have prevailed in the absence of the 2008–2009 crisis.11

Finally, it should be acknowledged that long-run economic growth projections are inherentlyspeculative, with uncertainties compounding at several levels. First, there is model uncertainty,i.e. the most appropriate growth model is unknown and could feature a very wide range of deter-minants, as the empirical growth literature illustrates. Most recently, this has led to an empiricalquest for the “right” model. For instance, taking a Bayesian Averaging of Classical Estimates(BACE) approach, Sala-i-Martin, Doppelhofer, and Miller (2004) conclude to the existence of 18significant growth determinants among a pool of 67 potential explanatory variables. Second, thereis parameter uncertainty, i.e. the magnitude – and in some cases even the nature – of the growthimpact of any given determinant is uncertain. Third, even under the very optimistic assumptionthat the theoretical framework developed in Section 2 builds on the right model and parameters,all of the growth drivers it features are hard to predict. Because the prior probability distributionof these parameters is unknown, a full probability distribution of projected growth cannot be com-puted, and in this regard the alternative scenarios featured in Tables 4 and 5 are only illustrativeof the uncertainties involved.

5. Conclusion

This paper has presented a simple, “conditional convergence” framework for projecting long-run GDP growth at the world level, taking stock of some of the current theoretical and empiricalknowledge regarding long-run growth drivers, including the role of human capital accumulation.Special emphasis has also been put on the role of ageing populations and potential labour marketand pension system reforms for future employment growth. This framework has been applied toproject world GDP growth up to 2050, although extension to longer horizons is fairly straight-forward. According to the baseline scenario constructed here, world GDP would grow by about3.5% per year on average over the period 2005–2050. Long-run growth projections are inherentlyspeculative due to model and parameter uncertainty, but also because the various determinantsof growth would remain very hard to predict even in a hypothetical case where the “true” modelwere known. In this regard, the range of plausible alternative scenarios explored in this paperconfirms the impact on future world GDP growth of the “usual suspects”, namely total factor pro-ductivity and population trends. However, it also shows that over an horizon of several decades,policy-related factors in the areas of education, labour markets and pensions systems also mattera good deal for world growth, even though they do not – at least in the framework of this paper –have steady-state growth effects. Policies to mitigate climate change would also have a sizeable(negative) impact on world growth over the coming decades – even though that impact would beexpected to be offset by gains beyond the 2050 projection horizon considered here, both in terms

11 This alternative baseline scenario is not reported here but is available upon request. This figure basically correspondsto the output loss between 2007 and 2010 that is not recovered later on. The theoretical framework of this paper allowssuch level effects, but precludes permanent growth effects of crises.


of conventional GDP, and even more in terms of broader income measures taking account of theavoidance of the non-market impacts of climate change on health, biodiversity, etc.

A limitation of the simple long-term growth framework presented in this paper is that tradelinkages across countries are ignored, and financial linkages are only roughly and indirectlyincorporated. The single most important extension of this framework would be to model andproject national saving rates – taking at least into account the influence of future demographictrends – and, via the world saving-investment equilibrium, world real interest rates and physicalcapital accumulation. A more comprehensive but costlier approach could be to plug the projectionsof growth drivers produced here – not least human capital and labour supply projections by cohorts– exogenously into a global macroeconomic model with overlapping generations, so as to projectinvestment, saving and interest rates within a fully consistent framework.

Acknowledgments

The authors would like to express gratitude to Jean Chateau for running the ENV-Linkagesmodel simulations. They also want to thank Jorgen Elmeskov, Giuseppe Nicoletti and Jean-LucSchneider for helpful comments. Thanks are due to Marcelo Soto for providing the human capitalstock data. The authors retain full responsibility for errors and omissions.

References

Acemoglu, D. (2009). Introduction to modern economic growth. Princeton: Princeton University Press.Barro, R., & Lee, J.-W. (2001). International data on educational attainment: Updates and implications. Oxford Economic

Papers, 53(3), 541–563.Barro, R., & Lee, J.-W. (1993). International comparisons of educational attainment. Journal of Monetary Economics,

32(3), 363–394.Barro, R., & Sala-i-Martin, X. (2004). Economic growth (second ed.). Cambridge, MA: The MIT Press.Bassanini, A., & Scarpetta, S. (2001). The driving forces of economic growth: Panel Data Evidence for the OECD

Countries. OECD Economic Studies, (33), 9–56.Bassanini, A., & Scarpetta, S. (2004). Does human capital matter for growth in OECD countries? A pooled mean group

approach. Economics Letters, 74(3), 399–405.Bils, M., & Klenow, P. (2000). Does schooling cause growth? American Economic Review, 90(5), 1160–1183.Bond, S., Leblebicioglu, A., & Schiantarelli, F. (2004). Capital accumulation and growth: A new look at the empirical

evidence. Boston College Working Papers No. 591.Bosetti, V., Carraro, C., & Massetti, E. (2009). Banking permits: Economic efficiency and distributional effects. Journal

of Policy Modeling, 31, 382–403.Burniaux, J-M., & Chateau, J. (2008). An overview of the OECD ENV-Linkages Model. OECD Economics Department

Working Papers No. 653.Burniaux, J-M., Chateau, J., Duval, R., & Jamet, S. (2008). The economics of climate change mitigation: Policies and

options for the Future. OECD Economics Department Working Papers No. 658.Burniaux, J-M., Duval, R., & Jaumotte, F. (2003). Coping with ageing: A dynamic approach to quantify the impact of

alternative policy options on future labour supply. OECD Economics Department Working Papers No. 371.Caselli, F. (2005). Accounting for cross-country income differences. In P. Aghion, & S. Durlauf (Eds.), Handbook of

economic growth (pp. 679–741). Amsterdam: Elsevier.Caselli, F., & Coleman, W. (2006). The world technology frontier. American Economic Review, 96(3), 499–522.Castles, I., & Henderson, D. (2003a). Economics, emission scenarios and the work of the IPCC. Energy and Environment,

14(4), 415–435.Castles, I., & Henderson, D. (2003b). The IPCC emission scenarios: An economic-statistical critique. Energy and

Environment, 14(2–3), 159–185.Cohen, D., & Soto, M. (2007). Growth and human capital: Good data, good results. Journal of Economic Growth, 12(1),

51–76.


Duval, R., & de la Maisonneuve, C. (2009). A long-run GDP growth framework and scenarios for the world economy.OECD Economics Department Working Papers No. 663.

Easterly, W., & Levine, R. (2001). What have we learned from a decade of empirical research on growth ? It’s not factoraccumulation: Stylized facts and growth models. World Bank Economic Review, 15(2), 177–219.

Fisher, B., Nakicenovic, N., Alfsen, K., Corfee-Morlot, J., de la Chesnaye, F., Hourcade, J.-C., et al. (2007). Issues relatedto mitigation in the long-term context. In B. Metz, O. Davidson, P. Bosch, R. Dave, & L. Meyer (Eds.), Climate Change2007: Mitigation. Contribution of Working Group III to the Fourth Assessment Report of the Inter-governmental Panelon Climate Change. Cambridge: Cambridge University Press.

Furceri, D., & Mourougane, A. (2009). The effect of financial crises on potential output: New empirical evidence fromOECD countries. OECD Economics Department Working Papers No. 699.

Gemmel, R. (1996). Evaluating the impacts of human capital stocks and accumulation on economic growth: Some newevidence. Oxford Bulletin of Economics and Statistics, 58(1), 9–28.

Groningen Growth and Development Centre. (2006). Data Sources and Methodology of the 60-Industry Database of theGroningen Growth and Development Centre, Groningen Growth and Development Centre, mimeo.

Hall, R., & Jones, C. (1999). Why do some countries produce so much more output per worker than others? QuarterlyJournal of Economics, 114(1), 83–116.

Harmon, C., Oosterbeek, H., & Walker, I. (2003). The returns to education: microeconomics. Journal of Economic Surveys,17(2), 115–156.

Henderson, D. (2005). The treatment of economic issues by the intergovernmental panel on climate change. Energy andEnvironment, 16(2), 321–326.

Jones, C. (1997). Convergence revisited. Journal of Economic Growth, 2(2), 131–153.Klenow, P., & Rodriguez-Clare, A. (1997). The neoclassical revival in growth economics: Has it gone too far? In B.

Bernanke, & J. Rotemberg (Eds.), NBER Macroeconomics Annual 1997 (pp. 73–103). Cambridge: The MIT Press.Mankiw, G., Romer, D., & Weil, D. (1992). A contribution to the empirics of economic growth. Quarterly Journal of

Economics, 107(2), 407–437.McKibbin, W., Pearce, D., & Stegman, A. (2004). Long-run projections for climate change scenarios. Lowy Institute for

International Policy, Working Papers in International Economics No. 1.04.Nakicenovic, N., Alcamo, J., Davis, G., de Vries, B., Fenham, J., Gaffin, S., et al. (2000). Special Report on Emission Sce-

narios. Working Group III Inter-governemental Panel on Climate Change (IPCC). Cambridge: Cambridge UniversityPress.

Nicoletti, G., & Scarpetta, S. (2003). Regulation, productivity and growth: OECD evidence. OECD Economics DepartmentWorking Papers No. 347.

Nordhaus, W. (2007). Alternative measures of output in global economic-environmental models: Purchasing power parityor market exchange rates? Energy Economics, 29(3), 349–372.

OECD. (2004). Sources of Economic Growth, Paris.Psacharopoulos, G. (1994). Returns to investment in education: A global update. World Development, 22(9), 1325–1343.Psacharopoulos, G., & Patrinos, H. (2004). Returns to investment in education: A further update. Education Economics,

12(2), 111–134.Sala-i-Martin, X., Doppelhofer, G., & Miller, R. (2004). Determinants of long-term growth: A Bayesian Averaging of

Classical Estimates (BACE) Approach. American Economic Review, 94(4), 813–835.Serra, V., & Saxena, S. (2008). Growth dynamics: The myth of economic recovery. American Economic review, 98(1),

439–457.Sianesi, B., & Van Reenen, J. (2003). The returns to education: Macroeconomics. Journal of Economic Surveys, 17(2),

157–200.Smith, J., Schellnhuber, H.-J., & Monirul Qader Mirza, M. (2001). Vulnerability to climate change and reasons for

concern: A synthesis. In J. McCarthy, O. Canziani, N. Leary, D. Dokken, & K. White (Eds.), Climate Change 2001:Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Third Assessment Report of theInter-governmental Panel on Climate Change. Cambridge: Cambridge University Press.

Timmer, M., & de Vries, G. (2007). A Cross-Country Database for Sectoral Employment and Productivity in Asia andLatin America, 1950–2005, Groningen Growth and Development Centre, mimeo.

Van Ark, B. (1996). Sectoral growth accounting and structural change in post-war Europe. In B. Van Ark, & N. Crafts(Eds.), Quantitative aspects of post-war European economic growth. Cambridge: CEPR/Cambridge University Press.

World Bank. (2007). 2005 International Comparison Program: Preliminary results, 17 December 2007. Washington,D.C.: The World Bank.

long-run growth scenarios for the world economy

Documents