Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Chapter Three:Northern Growth and Environmental Constraints

Andrew Glyn


It is easy to understand why the early 1970s brought predictions that growth in the North would be limited by resource constraints. Over the previous decade use of both energy and steel was growing at more than 5 per cent per year in the OECD. If these growth rates had continued, resource use would have doubled by 1987 and quadrupled by 2000. Such patterns of growth in the North, which were so profligate with resources, seemed evidently unsustainable. In the event by 1990 energy consumed was only 17 per cent higher than in 1973, and steel use was down by around one quarter. Section I of this chapter outlines how the pattern and rate of growth has shifted in the North since the early 1970s so that demands on resources are no longer increasing nearly so fast. Most attempts at modelling Northern pressure on the environment, notably in the context of global warming, assume that the rate of growth will slip down somewhat further in the decades up to the middle of the next century. Examination of the likely pattern of structural and demographic change in the North later in the chapter suggests that these projections may be too optimistic and that growth rates could well decline to something close to zero. If current growth in the North is putting rather little extra pressure on resources, and if growth rates seem destined to fall substantially further, then the task of reducing environmental impact in order to ensure sustainability (see previous chapter) is less demanding, supporting estimates that the costs to Northern GDP of overcoming these pressures will be relatively small

Such a perspective leaves aside the question of growth in the South, however. The final part of this chapter notes that attempts to model global pressure on the environment typically assume very little or no 'catch-up' of Southern incomes towards Northern levels; it is taken for granted that

Table 3.1 Resource use in the OECD, 1960-89

annual %
Energy Steel Non-ferrous metals
Total Per
Total Per
Total Per
1960-73 5.1 4.0 5.3 4.2 5.9 5.0
1973-79 1.5 0.7 -1.4 -2.2 0.9 0.1
1979 89 0.7 - 0.1 -2.2 -2.9 0.1 0.6

Sources: Energy OECD Energy Balances, 196079,1980-90; Metals Tilton, 1990, OECD Historical Statistics, 1990 Notes: Non-ferrous metals refers to unweighted average of five metals. Steel and nonferrous metals final period is 1979-87. Energy refers to Total Primary Energy Supplied.the existing gross disparities in income levels persist in the very long term. If, however, the South made really substantial inroads into these disparities, then pressure on resources would be much increased and the costs of overcoming the resulting problems (pollution, higher costs of resource extraction and so forth) would be correspondingly greater. This scenario is then applied to the large body of work dealing with costs of overcoming global warming. It is argued that serious catch-up of the South to the North would inevitably increase the restraint required in Northern carbon emissions. Like many environmental problems this could in principle be solved without great cost by new technology - in this case the development of carbon-free substitute fuels. But if such technology proved expensive, then the generalization to the South of Northern consumption levels would mean much greater costs to the North of measures to contain global warming. Moreover slow growth in the North would be an unfavourable backdrop against which to gain acceptance that these costs should be absorbed. The conclusion suggests the importance of beginning to think now of how the North could adjust to stable or even falling levels of consumption.


This section records how the pressures on the use of resources deriving from economic growth have diminished in the North over the past two decades. For analysing scarcity and global environmental impact total use of resources is the key indicator; when examining the distribution of resources between North and South this must be supplemented with resource use per capita. Table 3.1 presents some basic series for energy and metals use in the North.

After 1973 the growth rate of energy and metals use dropped very sharply. By the 1980s per capita usage of metals was falling, and was stabilized in the case of energy, compared to increases of 4-5 per cent per year during the 1960s. Similar trends are apparent for other materials; for example per capita use of cement began to decline in the early 1970s (Goldemburg et al 1987) after steady increase earlier.

Table 3.2 OECD energy use and energy intensity, 1960-90

Average annual
changes (%)
Energy use GDP Energy
1960-73 5.1 4.8 0.3
1979-90 0.7 2.8 -2.1
changes -4.4 -2.0 -2.4

Source: as Table 3.1

The stabilization, or even fall in input use per capita occurred while per capita incomes continued to increase (by 1.9 per cent per year since 1973). This suggests that the slowdown in growth after 1973 was only partly responsible and that changes in the pattern of that growth should be examined as well.

Resource Intensity

A conventional 'decomposition' of resource use (R) separates out the influences of GDP and resource intensity (R/GDP):

R = R/GDP x GDP.

Applying this to changes over time, the growth in energy use can be split into the growth of output and the growth in energy intensity. Table 3.2 shows that just over half the change in the trend in energy use in the 1980s as compared to the 'Golden Age of Capitalism' (1950s and 1960s), represented the change in the trend in energy intensity. Energy intensity fell sharply after 1973; declines of around 2 per cent per year after 1973 were very common in OECD countries, including those with high (eg Belgium), middling (eg the US, UK and Germany), and low (eg Japan) energy intensity at the end of the golden age; rises in energy intensity after 1973 were exceptional (Portugal, New Zealand, Greece and Spain).

It might be expected that the improvement in the trend in energy/materials intensity was closely connected with 'deindustrialization' - a switch away from resource using industry towards services. Industry (together with its associated transport) is about eight times as energy intensive as services and a switch of 1 per cent of GDP from industry to services would reduce energy demand by about 1 per cent of total energy used

Sectoral output growth, OECD, 1960-1990

While industrial output grew very slowly between 1973 and 1979, as investment spending stagnated, since 1979 its shortfall behind GDP growth has been small and accounted for by the decreasing importance of net exports of manufactures to the rest of the world (Figure 3.1 shows sectoral growth rates for the OECD since 1973). So 'deindustrialization' is a term properly applied to the structure of employment, not output. There has indeed been a sharp fall in the share of industrial employment (from 36.4 per cent of the total in 1973 to 29.6 per cent in 1990), but this overwhelmingly reflected faster productivity growth in industry than for the economy as a whole rather than slower output growth (Baumol et al 1989).

The shift in production from industry to services is relatively slight, therefore, and could account for only around a 0.1 per cent annual fall in overall energy intensity. So if structural change was important in explaining the break in the trend of input intensities after 1973 it must have operated largely at a more disaggregated level than captured by the notion of the 'service economy' - that is a switch away from heavy (materials and energy intensive) manufacturing towards 'lighter' sectors. The World Resources Institute believes that this has been important in relation to materials:

the demand for a wide range of materials, both traditional and modern, is no longer increasing in physical terms [per capital. Both traditional and modern materials are being used more efficiently -for example through the development of higher strength or more durable products....Although materials substitution and increased efficiency of materials use are clearly contributing to the shift away from materials, today these factors are probably not as important as the saturation of markets for bulk materials and heavy consumer goods and the shifting of consumer preferences to products characterized by a higher ratio of value added to materials content.'

Goldemberg et al 1987:36-7

Energy used per unit of industrial output fell by about 2.6 per cent per year after 1973; here the OECD (1992b) suggested that 'efficiency gains [within individual industries] influenced energy intensity trends more than structural changes.' The pattern of decline over time is interesting; falls of 2 per cent per year between the Oil Shocks, 3.9 per cent per year 1979-86, but back to 1.3 per cent per year (the pre-1973 rate) after the oil price fall in 1986. This illustrates the build up of price-induced substitution, superimposed on quite marked underlying technological trends tending to reduce energy intensity in industry. The latter are reflected in estimates in the global warming models that 'autonomous' energy intensity (without changes in relative prices) will decline by 0.5 per cent to 1 per cent per year for the economy as a whole over the very long term (though the weak basis for quantitative prediction is generally admitted).

While use of materials takes place overwhelmingly within industry, this is less true of energy. Residential uses and private transport (about two thirds of total transport) each consume about 20 per cent of total energy directly (see Table 3.3). Per capita energy use for residential purposes has declined by 0.8 per cent per year (and twice as fast per unit of floorspace). Energy use for transport has grown by 1.8 per cent per year since 1973, a per capita increase of 1 per cent per year; transport accounts for all the increase in OECD energy use since 1973.

Output and Productivity growth

The previous sub-section showed that the slowdown in output growth accounted for nearly half of the decline in the growth of energy use in the 1980s. Perhaps surprisingly, given the upward trend in unemployment, none of this reflected slower growth of total labour input. Total hours worked in the North actually rose faster in the 1980s than in the 1960s as employment growth was maintained and hours worked per worker fell more slowly (0.4 per cent per year after 1979 as compared to nearly 1 per cent per year previously). Thus all the slowdown in output growth (and consequent effects on resource use) reflected the productivity slowdown (nearly 2 per cent annual growth of hourly productivity for the period 1979-90 as compared to nearly 5 per cent per year for the period 1960-73).

Table 3.3 OECD energy uses

  Share of total 1990 % Annual growth
rate per capita, 1973-90
Industry 33.6 -1.2
Residential 17.9 -0.8
Commercial/public 10.6 0.4
Transport 31.3 1.0
Total   -0.3

Source: as Table 3.1

There has been an enormous literature on, and little agreement on the causes of, the productivity slowdown (see Englander and Mittelstadt 1988). But Nordhaus' conclusion (1992) that resource depletion, including pollution reduction, played a subsidiary role (he estimates the effect at about 0.25 per cent per year) is widely accepted. To this direct effect should be added the impact of the oil and materials price increases in the early 1970s and 1980s in exacerbating profit squeezes and inflation. These in turn inhibited capital accumulation both directly and via the deflationary policies which resulted. But resource price increases were only one contributory factor to the turbulence of this period, arguably less important than the enhanced economic power of labour as a result of the long-boom (Marglin and Schor 1990). Thus the conclusion that resource costs and prices were not the dominant cause of the productivity slowdown is still plausible when indirect effects are included.

Evidence of Resource Scarcities

Even if resource scarcities have not played the major role in the growth slowdown it is important to have some perspective on their magnitude. The prices of materials and energy, relative to the implicit deflator for GDP say, provide evidence of cost trends. There is general agreement in a range of studies (Barrett and Morse 1963, Barnett 1979, Slade 1982, Baumol et al 1989, Nordhaus 1992), that relative prices of resources generally trended downwards until the 1970s (timber being an exception); the early 1970s saw sharp upward movements (relative to the prices of manufactures), succeeded by large falls in the 1980s (one half for petrol, food, one third for coal and agricultural commodities and around one quarter for metals and timber). By 1991 petroleum and timber prices in real terms were still well above (double and 50 per cent higher respectively) the levels of the 1960s, whereas agricultural commodities (both food and other) were only half the level of the 1960s and metals and minerals about two thirds of the earlier level.

Such prices give information on scarcity at a world level. But what of Northern production itself? Here total factor productivity (TFP) is a simple indicator of the real costs (in terms of capital and labour inputs) of resource extraction. Data for the US show rising TFP (and thus declining real input costs) in the extractive industries up to the 1960s (Barrett and Morse 1963, Barnett 1979). But the increases slowed down in the 1960s and after 1970 mining TFP declined (Kendrick and Grossman 1980), by around 5 per cent pa from 1973-83 (Englander and Mittelstadt 1988). Imports may limit productivity declines and thus the extent of increasing real costs of domestic production, as in Ricardo's discussion of the Corn Laws; but imports appear not to have been a major factor in the US up to the 1950s (Barrett and Morse 1963). Thereafter the huge increase in oil imports played an important role; labour productivity in natural gas and petroleum mining rose by about 4 per cent pa up to 1973, fell at 9 per cent pa from 1973-82 as the oil price increases stimulated domestic production and then increased by 3 per cent pa for 1982-89 (Nordhaus 1992).

The US is an interesting case because it comprises such a substantial share of Northern production and because it is thought of as a resource rich country. Data for other OECD countries (Englander and Mittelstadt) suggest that TFP declines in mining were general over the period 1973-83, with the exception of countries benefiting from North Sea Oil.

Despite this evidence of scarcity within the North, the slowdown of growth in the world economy and opening up of reserves in other parts of the world has held at bay or reversed the symptoms of global scarcities which characterized the early 1970s. Nordhaus' conclusion that the evidence does not indicate that the 'major appropriable resources have taken a major turn towards scarcity during the last century' (1992:28) seems justified. But the importance of imports of fuel and materials into the North must be underlined - they still represented some 1.5 per cent of OECD GDP in 1990 despite the generally low prices described above (in 1974 the figure was 3.6 per cent). If the South achieved its aspiration of industrializing to Northern levels, where would be the hinterland which would supply cheap energy and materials?


The pattern of Northern growth since 1973 appears much less threatening to the environment than that of the golden age. Further reassurance comes from findings (Shafik and Bandyopadhyay 1992, Holtz-Eakin and Selden 1992, and Grosmann 1993) that the income elasticity of pollution generally declines with per capita GDP and even becomes negative in a number of cases where higher incomes generate financeable demands for effective pollution control policies. Nevertheless if growth could be expected to accelerate sharply back to golden age rates there would be recurring worries about resource availability, the extent of emission cutbacks necessary to contain global warming and so forth.

Source: see text

Figure 3.2 OECD output projections, 1973-2050

Far from assuming faster Northern growth some further slowdown is usually built into long-term assessments of environmental impacts such as global warming. Thus the OECD Model Comparisons project assumed that OECD growth would decline from 2.7 per cent per year 1979-90 to 1.5 per cent per year 2025-2050 (Dean and Hoeller 1992 Table A2), whereas Cline (1992 Table 7.2) assumes the decline will be to 1.2 per cent per year. These projections are typically presented without much discussion. Here we examine likely trends in the population and in the industrial structure which confirm the likelihood of a distinct slowdown in Northern growth. Firstly, population growth is expected to decline and the proportion of population of working age to fall considerably. Second, the continued shift in the employment structure towards services, where productivity grows slowly, will tend to reduce average productivity growth. In combination they actually suggest a greater slowdown than is commonly assumed.

The projections for OECD output growth shown in Figure 3.2 are based on the following assumptions about labour productivity growth within each sector, the pattern of output growth and the growth of labour input:

  1. Labour productivity growth in each sector (agriculture, industry and services) continues at the rates observed over the period 1973-90 (3.8 per cent, 2.2 per cent and 0.8 per cent pa respectively);
  2. Agricultural output grows in line with population while industrial output grows slightly slower than GDP and services output slightly faster;
  3. Employment growth is built up from OECD projections of total population, and of the ratio of population of working age to total population, together with assumptions that the post-1979 trends continue for the ratio of employment to population of working age and average hours worked.

These assumptions generate a slowdown in GDP growth from 2.7 per cent per year during 1973-90 to 0.5 per cent per year for 2030-2050. The first two assumptions imply that the shift of employment from industry (and less importantly agriculture) continues. By 2050 the share of industry has shrunk by more than one half and provides jobs for only 14 per cent of those in work as compared to 29.7 per cent in 1990 (services' share rises from 65 per cent to 85 per cent). As in the past, the main force for the swing out of industry is its much faster labour productivity growth, with slower output growth playing a subsidiary role. The increasing share of services employment, where productivity is assumed to continue growing more slowly, pulls average productivity growth down, from 1.6 per cent per year for 1973-90 to 1.0 per cent per year for 2030-2050. Thus of the 2.2 per cent points decline in the projected growth rate of GDP, 0.6 per cent points, or around one quarter, flows from the impact of structural change on average productivity growth.

The remainder of the slowdown derives from declining growth of labour input. Indeed employment growth declines from 1.1 per cent per annum for 1973-90 to -0.6 per cent per year during 2030-50. Figure 3.3 shows the components of this decline; population growth slips from 0.6 per cent per year during 1973-90 to a decline of 0.1 per cent per year for 2030-2050. The further implication of OECD's demographic assumptions is that the growth of population of working age declines much faster from 0.9 per cent per year in 1973-1990 to -0.8 per cent per year for 2030-2050. Employment is assumed to fall a little less fast than population of working age (bringing the average employment rate to 76 per cent in 2050, still below the levels achieved in Sweden and Denmark in 1990).

Thus it is the decline in population of working age which drives the decline in labour input, which in turn explains more than two thirds of the projected decline in the growth rate. Since the decline in population of working age flows mainly from the increasing 'dependency ratio' (those under 16 and especially those over 65) rather than from slower total population growth, then the fall in per capita GDP growth is nearly as steep as for GDP as a whole (from 2.1 per cent per year in 1973-90 to 0.5 per cent per year during 2030-2050). This is important since funding expenditure on the environment may well be more difficult the slower the growth of per capita income.

Source: see text

Figure 3.3 OECD employment projections, 1973-2050

The history of economics is littered with incorrect predictions that capitalist economies will converge towards zero growth. Any such projections are no more convincing than their assumptions. The growth of services productivity is a key issue because, if it is both low and below that in industry, then it persistently drags down average productivity growth provided demand for services grows in line with incomes. A case can be made for there being systematic underestimation of services productivity growth in the national accounts; in the US, for example, the output of significant parts of the finance sector and the general government is estimated by labour input implying zero labour productivity growth, while direct estimates suggest productivity growth in these areas of 1-1.5 per cent per annum. Baily and Gordon (1988) give plausible examples for retailing of output growth being underestimated by ignoring improved quality in the form of expanded convenience and choice; but there are contrary dangers that measured increases in the productivity of activities, where personal contact forms an essential part of the service, may hide a real deterioration in quality. There seems no reason to believe that underestimation of output changes apply more to services than to industry (the evidence on US capital goods suggests persistent underestimation of quality changes and thus output growth of about 3 per cent per year - Gordon, 1990). So slowdown of average productivity growth as a result of the shift to services has not been put in question by available evidence on mismeasurement. Nor does there seem any hard evidence that the problem of underestimating quality improvements has been increasing, to judge by evidence on US capital goods (Gordon 1990). So measurement problems do not seem able to explain the productivity slowdown since 1973, nor do they undermine the idea that structural change will further reduce average productivity growth.

The possibility that technological breakthroughs will transform the prospects for services productivity growth cannot be discounted. The increasing importance of the sector will tend to generate more research effort directed to that end. Much has been expected from computerization, in particular, the results of which, however, have been notably disappointing thus far (see Englander and Mittelstadt, 1988).

While the possibility of revolutionary technologies in the services sector cannot be ruled out, there are several reasons for believing that projecting on recent productivity performance may in fact exaggerate growth prospects. Firstly the tendency for employment to gravitate towards sectors with low productivity growth applies within services, and not just between industry and services (see Baumol et al, 1989). The share of community, social and personal services in total employment in the OECD rose from 21.8 per cent to 29.3 per cent between 1970 and 1990. Although the increase slowed in the 1980s, reflecting the widespread assault on the welfare services, it was not halted. These are precisely the sectors where productivity is inherently most difficult to increase (and not just to measure). The trend towards a larger proportion of elderly people will stoke up the pressure for more spending on these services and this will tend to drag down average productivity growth within services as a whole.

Secondly the demographic changes making for slower output growth will reduce the scope for further exploiting economies of scale. This will tend to restrain productivity growth both within industry and also in those sectors of services (eg retail trade, goods transportation) closely dependent on throughput from industry (currently comprising around 20 per cent of total employment - Blades 1987). It is notable that the decline in TFP growth after 1973 in retail and wholesale trade was as great as in manufacturing (Englander and Mittelstadt 1988).

Finally there is the international dimension. If projections are made separately (according to the same assumptions) for the US, Japan and Europe, then pushing the existing growth differential on for 60 years inevitably leads to a vast relative expansion of Japan (whose GDP would quadruple, as compared to a less than a doubling in the US). Now there can be little doubt that there is still an element of catching up in Japan's growth performance; while Japan leads the world in much of the machinery sector, in other parts of the economy, particularly services and agriculture, productivity levels are quite low by international standards. Part of the higher growth of productivity in much of the Japanese economy involves a process of convergence. As this 'growth bonus' exhausts, productivity growth in Japan would tend to decline, pulling down that for the North as a whole. Acting in the contrary direction, when there are more countries at the technological frontier in a particular sector they may push it along faster.

The demographic projections that the population of working age, which was growing by 0.9 per cent per year during 1973-90 will be declining by 0.8 per cent per year during 2030-50, are very striking. They are based (Hagemann and Nicoletti, 1989) on assuming that fertility rates, most of which are currently below replacement in the North, converge to replacement levels by 2050; that average life expectancy rises by a further two years and that international migration continues at current rates. There seem no obvious reasons to regard these projections as too low. Indeed if fertility did not converge back to replacement the fall off in population growth would be sharper still. The projection of current trends in employment rates and working hours again seems reasonable (though the slow rate of growth of real incomes would certainly generate pressures for more work).

In respect of productivity particularly, it seems just as likely that the projected growth rates may be over-optimistic than the reverse. Thus there seems a real possibility that the internal logic of Northern growth will pull the growth rate of conventionally measured GDP close to zero in the middle of the next century. The reasons for this demographic and structural change have played no substantial role in the growth slowdown since 1973. The social and political problems of adjusting to such trends will be very severe, as discussed in the next chapter. But the rather reassuring trends in the environmental impact of Northern growth, suggested by the examination earlier in the chapter of recent patterns of resource use, is, if anything, reinforced by analysis of this section which suggests a further slowing of the growth rate


Contents - Previous - Next