Contents - Previous - Next

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

Chapter Four:Can The North Stop Consumption Growth? Escaping The Cycle of Work and Spend

Juliet B Schor


In the last half century consumption has increased dramatically in the industrialized North. The most rapid increases were during the 1950s and 1960s, the so-called 'Golden Age of Capitalism.' Beginning in the 1970s, the growth of GDP and consumption per capita slowed. But even so, since 1973 the average consumer in these countries found his or her consumption rising by an average of more than 2 per cent per year. In the United States, the country with which this paper is mainly concerned, real per capita consumption expenditures rose 2 per cent per year between 1973 and 1989 (see Table 4.1), and despite the current global slowdown, the future is likely to bring further increases in income. Productivity should increase rapidly on account of further innovations in and diffusion of new technologies. And, in much of the South, income is rising even faster than in the North. In countries such as Brazil, India and China, growing middle classes will continue to emulate western standards of living. As the 21st century approaches it is difficult to avoid the conclusion that increasing numbers of people will be consuming a bundle of goods and services resembling that of northern populations.

Table 4.1 Northern consumption per capita, 1960-89

  Annual rates of growth (%)
1960-1989 1973-89
Europe 2.9 2.0
Japan 4.7 2.4
US 2.5 2.0

Environmentalists have made the argument that current patterns of consumption are not environmentally sustainable - in fact the world faces significant threats to survival (see Ekins and Jacobs, Chapter 2). While it is impossible to construct an index of the impact of an entire lifestyle on the environment, it is clear that northern lifestyles, and particularly that of the United States, are highly environmentally damaging. Affluence is causing air and water pollution, soil degradation, ozone depletion, species extinction, greenhouse gas emissions, toxic chemcial wastes, deforestation and other environmental disasters.

One approach is to change the way we produce and what we consume. Firms can switch to environmentally more benign production technologies. Consumers can alter the pattern of goods and services they purchase and use, for example by switching to smaller automobiles (or even to public transportation or bicycles). Consumers can also avoid buying toxic products, recycle, and insulate their homes. By shifting to less damaging methods of production and patterns of consumption, the overall (negative) environmental effect of any given level of consumption can be reduced. In the terms depicted in Figure 4.1, such changes will result in an inward shift in the growth/environment tradeoff.

Figure 4.1 Growth versus environmental quality

In the short to medium term, it is not likely that these kinds of changes will produce enough improvement in the earth's ecology. In part this is due to linkages among components of consumption 'bundles', in particular linkages between housing and transport which representsubstantial expenditures and are two major sources of environmental degradation. Consider American suburbia. Houses are large and therefore costly to build. They are energy inefficient and cannot be kept cool without air conditioning. They have large lawns, often with ecologically destructive landscaping. Because of the distances and the land-use patterns, automobile travel (extremely damaging) is almost imperative. To dramatically improve environmental impact, Americans would have to shift to smaller, differently designed houses, with new transportation systems and new conceptions of land use around households. These extraordinarily costly changes would take decades to complete. Linkages among consumption decisions as well as the fixed costs embodied in past decisions are therefore formidable barriers to rapid reduction in the 'environmental intensity' of each dollar of consumption spending.

This suggests that the road to sustainability will also include a change in the current trajectory of consumption levels. By lowering the rate of growth of consumption per capita, perhaps even to zero, the amount of 'environmental degradation' generated by each consumer will be less than it would be if consumption grew at current or historical rates. That is because it is not only what we consume, but how much we consume that matters for the ecology. That is, we need to move rightward on the growth-environment frontier. A recent poll suggests considerable awareness of these issues among the American people. Thirty-seven per cent strongly agree and 40 per cent agree somewhat that 'American overuse of resources is a major global environmental problem that needs to be changed.' (Pew 1994)

But altering the trajectory of private consumption is not an easy task. Economic growth plays a central political, economic and psycho-cultural role in western society. Since the Second World War, growth has been the foundation of political consensus and stability in the North (Marglin and Schor 1990). The postwar regime was premissed on steady increases in income and consumption, in that the Keynesian alliance between business, labour and government was essentially an agreement to avoid conflict over shares by ensuring higher absolute levels for all. Similarly, redistribution among income classes was small in comparison to the impact of growth on the standard of living of lower income groups. Prosperity became key to the virtual elimination of political challenges to the system. Increasing consumption also solved the 'economic problem', that is, fear of stagnation. Productivity increases were channelled into growth in real wages and hence private consumption. Consumer credit enhanced consumer demand. The gap between what the economy was capable of producing and what people were willing to buy - the nightmare of prewar theorists - was kept to a minimum.

The centrality of growth in our political and economic culture means that moral or pragmatic environmentally-motivated appeals to reducing consumption may not be successful. But, with widespread perceptions of cultural and economic decline, the promise of a higher 'qualify of life' may be. In the United States, where working hours have been rising for twenty five years, and in Japan, where working hours are extremely high, there
may be considerable receptivity to 'trading consumption for time'. I believe we now have the possibility to create a 'politics of time' which has sustainability at its core. Doing this requires extracting ourselves from what I have called the 'cycle of work and spend'. Difficult as it may be to tame northern appetites for consumer goods, there may be no other choice if the planet is to escape environmental crisis and at the same time experience significant growth in consumption per head in the South. This is the view that appears to have emerged from the Rio summit. In the pages which follow, I explore the obstacles and opportunities to a trajectory in which productivity growth in the North is channelled into shorter working hours, rather than higher income.


Table 4.2 shows the fraction of growth in labour productivity since 1870 which has gone toward reducing hours. (These estimates are somewhat imprecise because the hours calculations, from Maddison, are rough. However, if anything, they overstate the decline in hours.) As the table clearly shows, the 'extent' of hours reduction has fallen substantially in all countries except Germany. In an absolute sense, the extent of hours reduction can be described as virtually non-existent to modest, eg a low of 4 per cent in Japan to a high of 28 per cent in the UK.

Table 4.2 Changes in hours and productivity, 1870-1984

  France y German Japan Netherlands UK US
1. Annual hours (% reduction)
1870-1938 45.8 23.8 20.8 27.6 27.3 35.9
1950-84 17.1 32.1 6.3 29.5 25.3 13.4
2. Labour productivity (% change)
1870-1938 119.0 109.0 131.0 83.0 79.0 114.0
1950-84 128.0 136.0 154.0 108.0 91.0 64.0
Hours reduction as a fraction of productivity change: row (1) divided by row (2)
1870-1938 0.385 0.218 0.159 0.332 0.346 0.315
1950-84 0.134 0.236 0.041 0.273 0.277 0.20

Source: Angus Maddison, 'Growth and Slowdown in Advanced Capitalist Economies' journal of Economic Literature, vol XXV, no 2, June 1987. Tables A-5, A-9 (pp 683-686). Notes: Annual hours are hours worked per person per year. Labour productivity is GDP per hour worked

My own estimates for the United States suggest that Maddison may have overstated the hours decline. Although there was some decline between 1950-1969, I have found that between 1969 and 1989 average hours for those who were not involuntarily working part-time or part-year have actually risen by an average of 138 hours, thereby cancelling the pre-1959 gain (Schor 1992). Similarly, in Japan, hours have most probably risen since

1973. The postwar experience therefore raises the question of why, if leisure is a 'normal' good, hours have not fallen more (or at all in some cases).

The failure of the 'market in hours'

Neoclassical economics has a straightforward explanation for situations where hours fail to fall with productivity increase: hours merely reflect the expression of consumers' preferences for income rather than leisure. The simple neoclassical model begins with individuals' preference functions. These individuals choose a level of working hours which (with a given technology) determines the level of income and free time. Employers offer job schedules which conform to individuals' preferences. This so-called 'worker sovereignty' in the labour market is ensured by competition. Should the market fail to provide the schedules people want, new employers would come along and outcompete existing firms.

This simple neoclassical story is not well supported by the evidence, at least for the United States and Japan. Most important for our purposes is the question of whether or not firms allow workers to choose hours. Studies of US workers show significant constraints on choice of hours. Altonji and Paxson's work (1988) provides evidence that workers have difficulty changing hours within jobs. Among a sample of low-income married men, Moffitt (1982) found a large discrepancy between mean desired hours (21 per week) and mean actual hours (39 per week). A second type of evidence comes from surveys. When questioned, workers report that they have little freedom to vary hours. In one study only 15 per cent of the sample were free to adjust hours up and down (Kahn and Lang 1987).

One explanation for hours constraints is the structure of firms' costs. If costs do not vary by the hour, but are incurred on a per employee basis, then firms will not be indifferent to workers' schedules (this point was first made by Lewis 1969 and subsequently by Schor 1992). In the postwar period, the 'hours-invariant' component of labour costs has risen significantly in most countries of the North. The rise of hours-invariant costs has meant that employers have had a growing incentive for longer working weeks and a disincentive for reducing hours. In the US, fringe benefits (comprising mainly medical and pension payments, as well as paid time off) as a per centage of total wages and salaries rose from 17 per cent in 1955 to 39 per cent in 1992 (US Chamber of Commerce, 1993 Table 17). Employer tax contributions which are capped (such as unemployment insurance, workers' compensation etc will have a similar effect. These hours-invariant costs are probably the major reason why the current US recovery had so little job growth for so long, and why levels of factory overtime continue to be sustained at record levels.

Variation in these hours-invariant costs may be useful in explaining cross-national trends in hours in the OECD. Those countries with a higher share of hours-invariant welfare spending borne by employers tend to have considerably higher hours (eg United States, Japan). Countries with large public welfare states, by contrast, have lower hours (Scandinavia, West Germany). Canada and the United Kingdom are intermediate cases on both counts.

However, it is unlikely that this explanation is exhaustive, especially for adult men, whose scheduling options are most severely circumscribed. With hours-invariant costs average hours would be longer but worker sovereignty should still rule. Employers would offer short schedules and workers would pay a wage or benefit penalty to get them. One would expect far more diversity in scheduling than exists. By contrast, scheduling options have tended to be discrete: full-time positions offering full benefits and part-time positions with severe wage and benefit penalties. For example, in the US in 1985, 54 per cent of all non-farm employees had a 40 hour work schedule. By contrast, only 2.3 per cent worked 25-29 hours, 4.1 per cent worked 30-34 hours and 7.4 per cent worked 35-39 hours. Twelve per cent of workers were located in the 1-24 hours category (Smith 1986: 8). If hours-invariant costs were the only factor at work we would not expect to see clumping at 40 hours. Furthermore, it is implausible that only 2.3 per cent of workers independently prefer 25-29 hours, for example.

An alternative explanation breaks with the neoclassical model and argues that the 'market in hours' is largely absent. For the most part, employers set hours and give employees limited choices, most often 'all or nothing' (that is, work the expected hours or quit). Most workers do not have the right to refuse overtime.

Schor (1992), Seo (1993) and Schor and Seo (1994) argue that for salaried workers there is an hours externality: additional hours are free to the employer because the pay itself is hours-invariant. We find that salaried workers have significantly longer hours than hourly workers, after accounting for other determinants of hours (income, occupation, age, etc). In the United States, among male heads of households, the additional hours due to payment by salary were 140 (Seo 1993). A reasonable interpretation of these results is that employers require long hours from these employees. In addition to the fact that pay is fixed, employers may use hours as a signalling device to identify loyal or hard-working employees. As Bailyn (1993) has argued, this does not mean they are actually more productive, only that the corporate culture requires them to work long hours.

Finally, in cases where firms do not have flexibility in terms of shift work, or there is a limited supply of labour, and where capital investment is substantial, management will oppose hours reductions in order to maintain high capital utilization. This may explain why capital intensive industries such as auto and steel have tended to have relatively high hours. The combination of hours-invariant costs, the hours externality of salaried employment and capital investment create powerful incentives for employers to schedule long hours and to oppose hours reductions.


The failure of hours to fall in response to productivity gains is in contrast to the late 19th and early 20th centuries, when a larger fraction of productivity growth was channelled into shorter hours (see Table 4.2). In the United States and Japan, the postwar era has brought a cycle of work and-spend, as shown in Figure 4.2. Employers set schedules and workers conform to them. When productivity rises, employers pass the gains along in the form of higher wages, rather than reduced worktime. Workers take the extra income and spend it. They become accustomed or habituated to the new level of spending and develop aversion to their previous, lower level of spending. Preferences adapt such that workers are unwilling to reduce current income in order to get more leisure. They respond to surveys saying they are satisfied with their current levels of hours and income. With respect to these surveys work and spend is observationally equivalent to neoclassical theory: in both, workers are satisfied with current hours and income. That is, the surveys cannot distinguish between workers 'getting what they want' or 'wanting what they get'.

Figure 4.2 Models of work and consumption

The existence of a work and spend cycle has a number of implications. First, it suggests an addictive aspect to consumption. Over time, people become habituated or 'addicted' to the level of consumption which they are attaining. Goods which are originally experienced as luxuries come to be seen as necessities. Habit formation can imply the endogeneity of preferences, that is, the ability of preferences to change as consumption occurs. This seems to have occurred over time in individuals' preferred tradeoffs between hours and income (see below).

Work and spend also provides a dynamic context for status emulation, or 'keeping up with the Joneses', as described by Veblen, Duesenberry and others. In these approaches satisfaction is derived not from one's absolute level of consumption, but from one's level relative to others. In Duesenberry (1949), consumption is driven by 'demonstration' cum 'emulation' - Smiths seeing and then trying to keep up with Joneses. With rising consumption, participants in status processes mistakenly perceive their own increases in consumption as relative improvement, when they are only absolute gain. No one is made relatively better off by higher consumption, yet that is only apparent after the game has been played. If free time is less of a status good than commodities, and yields greater intrinsic satisfaction, then people would be better off with slower consumption growth and more free time than in the work and spend cycle (see Frank 1985 on this point).

To understand the power of work and spend, we must also recognize the extent to which consumption has come to assume a central 'psychocultural' role in western societies and Japan. In the United States, where this process has arguably gone farthest, society and culture are increasingly commercialized. Advertising and marketing encourage consumers to be dissatisfied with what they have and to covet more. Culturally, consumerism is remaking the landscape as the shopping mall takes over as the dominant public space, equivalent to the cathedrals of the medieval era. Consumption has also become a crucial determinant of identity: 'I shop, therefore I am.' Cutting edge advertisers, especially those whose products are symbolically intense, have begun marketing an identity concept, rather than the qualities of products. Soft drinks represent the 'new generation'. Sneaker ads have evolved into sophisticated visions of desirable lifestyles. Perhaps more than any other activity, consumerism has become the lynchpin of modern economic, political, social and personal life.

Contents - Previous - Next