• Appendix 1. The human development index (HDI)
  
• Appendix 2. Target setting versus a calculus of benefits and costs
  
• Appendix 3. Human development and sustainable development
  

The human development index

Construction of the human development index (HDI)

The HDI has opened new perspectives on measuring and analysing development. But there can be no doubt that the work in this area is still at its beginning. Much more research is needed as is more experience with using the HDI for various practical purposes of assessing, planning or programming development. UNDP would welcome any contributions to this topic, to be taken up in next year's Human Development Report, from interested development research scholars and practitioners.

The HDI includes three key components: longevity, knowledge and income, which are combined in a three-step process to arrive at an average deprivation index (for a full technical description, see Human Development Report 1991, technical note 1, pp. 88-89). Longevity is measured by life expectancy at birth as the sole unadjusted indicator. Knowledge is measured by two educational stock variables: adult literacy and mean years of schooling. The measure of educational achievement is adjusted by assigning a weight of two-thirds to literacy and one-third to mean years of schooling:

For income, the HDI is based on the premise of diminishing returns from income for human development using an explicit formulation for the diminishing return. A well-known and frequently used form is the Atkinson formulation for the utility of income:

Here, W(y) is the utility or well-being derived from income, and the parameter measures the extent of diminishing returns. It is the elasticity of the marginal utility of income with respect to income. If =0 there are no diminishing returns. As approaches 1, the equation becomes:

The value of rises slowly in the HDI as income rises. For this purpose, the full range of income is divided into multiples of the poverty line y*. Thus, most countries are between 0 and y*, some between y* and 2y*, even fewer between 2y* and 3y* and so on. For all countries for which y < y*_that is, the poor countries is set equal to 0. There are no diminishing returns here. For income between y* and 2y*, is set equal to 1/2. For income between 2y* and 3y*, is set at 2/3. In general, if a y* y (a+1) y*, then = a / (a+1). This gives:

So, the higher the income relative to the poverty level, the more sharply the diminishing returns affect the contribution of income to human development. Income above the poverty line thus has a marginal effect, but not a full dollar-for-dollar effect. This marginal effect is enough, however, to differentiate significantly among industrial countries. This method does not take = 1, but allows it to vary between 0 and 1.

For example, Singapore has a real GDP per capita of $15,108. With the poverty line set at $4,829, there are four terms in the equation to determine the well-being of Singapore:

In calculating the HDI of Singapore using the improved variables and applying the methods described here, the following steps are taken:

Making the HDI gender-sensitive

Of the many inequalities in human development, the most striking is that along gender lines. Women typically live longer than men once they have gone beyond the age when differential treatment of boys and girls makes life shorter for girls. They work harder and more hours than men, but they often do work that is unpaid or underpaid. Women cook, take care of children, the elderly and the sick, look after the upkeep of the house and work on the farm or in the shop. Only a small proportion of women find that their work gets paid and recorded as participation in the labour force. Labour force participation as a concept and in its measurement grossly understates women's work even in the productive, commodity-producing sphere. It completely leaves out much work that is useful to the continuing existence of the household.

Any attempt to measure gender inequalities is thus bound to err on the low side. Even allowing for that, the inequalities are striking. An attempt is made here to calculate the ratio of female income to male income. We do not have suitable data on income, but for 33 countries we do have comparable data on the relative wage ratios and the relative ratios for labour force participation.

These data reveal a remarkable pattern of discrimination. The female-male wage ratio for these 33 countries ranges from a low of 50% (Japan) to a high of 89% (Sweden).

In labour force participation, the lowest female-male ratio is 40% (Costa Rica) and the highest is 92% (Sweden). Multiplying these two ratios gives the female-male wage-income ratio.

This wage-income ratio combines two identifiable correlates of gender discrimination. The male wage is greater than the female wage, and the gap in labour force participation rates is even wider. When this is translated into absolute income levels, we see the profound consequences. To do this, a basic assumption has to be made that is clearly gender-biased: that the ratio of non-wage income to wage income is the same for men and women. This ratio therefore understates the inequality.

In adjusted real GDP per capita, female incomes as a percentage of male incomes range from a low of 26% (Costa Rica) to a high of 82% (Sweden). But of the 33 countries for which we have comparable data, only nine have a ratio of 60% or above, while 10 are below 40%. So, even in a statistic that understates the inequality, the differences are stark.

The female HDI gains from the near-equal or better ratio in life expectancy but loses somewhat from unequal access to education, particularly in the developing countries. In education, the developed countries show very little gender difference: though the value for female achievement, as a proportion of male, never goes above 102%, in five countries it goes below 98%. In poorer countries the differences become substantial. Women's educational attainment in Kenya shows a low ratio of 53%. In Myanmar it is 74%, and in Hong Kong 75%. Costa Rica shows a figure of well over 100%.

The overall HDI for men and women calculated separately reflects this pattern. Much of the data, 22 of the 33 countries, relate to the industrial countries. So, technical note table 1.1 does not capture the full extent of gender inequality. Even then, the female HDI as a percentage of the male HDI is as low as 59% in Kenya, 66% in the Republic of Korea and 69% in Swaziland. Of the 33 countries, 10 have ratios below 75%, and only five countries Sweden, Finland, Norway, France and Denmark have ratios of over 90% No country attains full gender equality even in this biased measure, though Sweden comes close with a ratio of 96%.

How should this inequality be reflected in the overall HDI for any country? A simple approach is to multiply the overall HDI for any country by the ratio of female-to-male HDI. If a country has full equality, its HDI is unaffected. Although Japan has the second highest overall HDI (0.981), its low female-to-male ratio brings it down to being 18th of the 33 countries with a gender-adjusted HDI of 0.761. Sweden, by contrast, has a very high female-to-male ratio and moves from fifth to first position with a gender-adjusted HDI of 0.938. Whereas Japan has a high overall HDI with a high degree of gender inequality, Sweden has a high overall HDI with a low degree of gender inequality. Among the 11 developing countries, there is usually both a low HDI and a high degree of gender inequality.

Adjusting the HDI for income distribution

The HDI is a national average, just like real income per capita, one of its components. The use of any such overall average hides the considerable differences in the distribution of the basic indicators, whether by gender, race, region, ethnicity or simply among individuals. The HDI therefore needs to be made sensitive to these distributions.

The HDI has the advantage that two of its three basic variables life expectancy and educational attainment are naturally distributed much less unequally than is income, the third variable. Thus, life expectancy in any population is not likely to be distributed more unequally than, say, three to one. A rich person cannot live a thousand times longer than a poor person, though their incomes may be in that ratio. Across countries, the range of life expectancy is 42 to 79, less than 2:1.

The same is true in educational attainment. The range of the percentage of adults who are literate varies from 18% to 99%, a range of under 6:1. Mean years of schooling show a variation from 0.1 to 12.3, more unequal than life expectancy, and hide even greater variations in the within-country distribution.

Apart from per capita income, all the variables used in the HDI have an obvious maximum. Life expectancy will rarely go beyond 100, literacy never beyond 100% and mean years of schooling seldom beyond, say, 15. Income, however, has no upper bound. For GNP per capita the intercountry range is $80 to $29,880, a range of 375:1. As for real GDP per capita, the range is $380 to $20,998, or 55:1. Such inequalities in income are reproduced just as sharply within countries.

So, a high average value for life expectancy or educational attainment can be obtained only by a reasonably equitable spread among individuals, a result of the fixed maximum possible. Although it is of great interest to know the distributions of those variables, an average is a better statistic for these variables than for income, where it can be very misleading.

The ranking of countries by per capita income could be adjusted if per capita income were multiplied by a factor indicating distributional inequality 1 minus the Gini coefficient. This method can be extended to all the countries having statistics on distributional inequality. Some 41 countries have data on the ratio of the income share of the highest 20% to the lowest 20%. Of these 41 countries, 17 have data on the Gini coefficient as well, and there was found to be a very strong association between the two the logarithm of the ratio being a good predictor of the Gini coefficient. This regression result was used to interpolate the Gini coefficient for the remaining 24 countries. Some countries had data only on the Gini coefficient. In all, 53 countries with directly estimated or interpolated Gini coefficients were available.

A word of caution. The Gini coefficients are registered for various years between 1975 and 1988, and the ratios of the top 20% to the lowest 20% are for years between 1980 and 1988. This is not a serious problem, since these coefficients are unlikely to change quickly. But the Gini coefficient is not always truly representative of the entire country. It is sometimes only for a subsection, such as the urban population.

Adjusted income was multiplied by (1 - G) with G being the Gini coefficient to modify income even further. Because this was done for the adjusted income, W(y), rather than for the actual income, the diminishing return effect could be incorporated before the distributional adjustment modifies incomes further. This modified income W(y)[1 - G] is then used as the third variable in addition to life expectancy and educational attainment to compute a distribution-adjusted HDI.

For all but two countries, the HDI is reduced by making it sensitive to income distribution, and in a half of them, it is reduced by 4% or more. This is particularly marked in the developing countries, where 24 of the 32 developing countries have a reduction of 4% or more and seven show a reduction in excess of 10%.

Much better data are needed to pursue the sensitivity of income distribution more thoroughly. The analysis shows that caution is needed in interpreting a country's HDI value as a measure of achieved well-being for all its people.

Measuring progress in human development over time

The human development index (HDI) ranks countries relative to each other for a particular period. The maximum and minimum values that define the maximum distance to be travelled for each variable are specific to that year. Over time, the actual achieved values of life expectancy, literacy and income change, as will the maximum and minimum values of these variables across all countries.

For example, Ruritania's life expectancy in year 1 may be 40, halfway between a minimum of 20 and a maximum of 60. By year 10, Ruritania may have improved its life expectancy to 50, but the minimum may now be 30 and the maximum 80. In such a case, the numerical value of the index indicating Ruritania's life expectancy will drop in the HDI calculations from 0.5 [ = (40 - 20) / (60 - 20)] to 0.4 [ = (50 - 30) / (80 - 30)], despite the 25% improvement in life expectancy.

So, improvements in the components of human development in any country over time may be reflected as a decline in its HDI value, if in the meantime its relative position has deteriorated. To combine a measure of progress over time with intercountry comparisons at one point of time, the HDI has to be modified.

The way to tackle this problem, without changing the logic of the HDI, is to say that the minimum and maximum should be defined, not for each point of time, but over a period of time. Thus if we are measuring progress between 1970 and 1990, the minimum would be the minimum of all values of, say, life expectancy for all countries over the 20 years. Similarly for the maximum. The distance to be travelled is thus stretched out as the maximum over the 20-year period.

In the example of Ruritania, the minimum stays at 20 but the maximum is now 80. In year 1, the life expectancy variable is 0.33[(40 - 20)/(80 - 20)], and in year 10 it is 0.5 [(50 - 20)/(80 - 20)].

With this adaptation, the human development index becomes comparable over time as well as across countries. The difference in the value of the human development index over time can be shown to be a weighted sum of the growth rates in the three variables: the weights are given by the ratio of the initial value of a variable to the maximum range.

To express this algebraically with X1 as life expectancy, X2 as literacy and X3 as income the contribution of each variable to the HDI can be written as Zi, where:

In the formula, j denotes country, t the time period. Note now that the denominator will remain unchanged for all time periods and for all countries.

MHDI stands for the modified HDI since we have a new definition of the maximum and minimum. Countries are ranked by the size of the difference between the 1970 and 1990 values for the MHDI. These differences range from 0.301 for Saudi Arabia to -0.076 for Jamaica. Jamaica is, however, a country where the HDI in 1970 was already quite high, 0.797, and the lack of change does not reflect absolute deterioration.

At the bottom, with Jamaica, are 10 countries that register a negative change, and above them is a group of 28 countries that register a change between 0 and 0.049. Twenty-one of these 38 countries are from Sub-Saharan Africa, seven from Latin America and seven from Asia. These low achievers are usually countries with a comparatively low initial HDI value. Only Poland (0.829 in 1970), Romania (0.798) Jamaica (0.797) and Argentina (0.784) would qualify as such. The remainder are poor initially, and 23 of them remained below 0.300 in 1990. Many of these countries experienced low rates of growth of real GDP per capita over this period, or even had a negative growth. So, income growth may not be sufficient for achieving a high HDI, but it cannot be dispensed with.

Fifty-six countries show a moderate improvement, between 0.05 and 0.15, 16 show a greater improvement and 43 show a total increase in excess of 0.10. Since the maximum distance to traverse is 1.00, some 40% of the countries, starting at different levels, covered 10% or more of the maximum distance. Almost all the countries with high HDIs in 1990 are in this group. They have continued to improve despite their already high levels for 1970. But, a few countries in this fast-moving group began in 1970 with low levels of HDI, notably Yemen (0.093 in 1970), Kenya (0.253), Morocco (0.268) and Indonesia (0.316).

As should by now be apparent, the HDI is a flexible tool of analysis which can be useful in monitoring the progress of human development. It can readily be disaggregated, assuming the data are available in the required form. In large countries with considerable geographical diversity, it might be useful to produce regional HDIs, using the same method applied to country HDIs. In pluralistic societies with ethnic diversity, it might be useful to produce an HDI for each ethnic group, using the same method used to produce female and male HDIs. In other words, the HDI can be disaggregated in a number of ways and each country can adapt it to suit its own policy purposes.
Source: Human Development Report 1992, published for the United Nations Development Programme (UNDP), by Oxford University Press, New York

Target setting versus a calculus of benefits and costs

We have advocated in the text that expenditures on human development should be based in part on a calculus of benefits and costs. We are of course aware that it often is difficult to measure accurately the benefits of investing in people--the costs are relatively easy--but none the less we believe that all investment projects, be they additions to the stock of human, physical or natural capital, should be subjected to similar scrutiny and should be required to pass the same test. The intellectual discipline required by a calculus of benefits and costs forces analysts to make hidden assumptions explicit, enables policy makers to compare on an equal basis the alternative uses of public funds and can help prevent a waste of scarce investment resources. Whether the decision rule is a benefit-cost ratio greater than one, or a positive net present value or an internal rate of return greater than a predetermined value is a secondary matter. The important thing is that a genuine effort be made to calculate benefits and costs rather than rely on intuition, common sense or political expediency.

The alternative approach, common in international circles as well as in individual countries, is to set specific targets and then in effect attempt to minimize the cost of attaining the target. The United Nations, for example, has elaborated an "international development strategy" for each of the last four decades, beginning with the 1960s. The international development strategies specified global or universal targets that set identical standards for all countries regardless of circumstances, e.g., "free education at all levels". Moreover, the targets became increasingly numerous, rising from four in the first Development Decade (the 1960s) to more than 23 in the third Development Decade (the 1980s). Finally, the targets became increasingly ambitious and even unrealistic. For example, the growth target for the 1960s was 5 per cent; in the 1970s it was raised to 6 per cent; and in the 1980s it was raised again to 7 per cent. Actual rates of growth diverged increasingly from the targets.

Targets tend to proliferate because they specify levels of specific benefits desired--rates of growth, levels of literacy, extent of health coverage, school enrollment rates, degree of industrialisation, share of world trade, etc.--and there is no limit to the number of good things that can be desired. Similarly, targets tend to be excessively ambitious because there is nothing in the process of target setting that forces planners to estimate the cost of meeting the target. Target setting, in other words, does not require planners to compare alternative goals and to weigh the benefits against the costs.

It is often implicitly assumed when setting a target that the unit cost is constant. In the case of a mass literacy campaign, for instance, it is assumed that the additional or marginal cost (MC) of providing literacy is independent of the size of the programme. The unit cost (or cost per person included in the programme) is the same whether the programme aims for universal adult literacy or, say, aims at a more modest target of only 80 per cent of the eligible population. This assumption is depicted in Figure 1(a).

In practice, however, unit costs are likely to rise as the size of the programme expands. And as one approaches complete coverage of the eligible population, the marginal cost may rise sharply. Returning to the case of a mass literacy campaign, some individuals may be more strongly motivated than others and hence may put more effort into learning to read and write, some groups (e.g., the young as compared to the elderly) may be able to learn more quickly and easily than others, and some people (e.g., those living in remote locations) may be more difficult to reach than others. If so, the marginal cost curve of a mass literacy campaign will begin after a point to rise, as depicted in Figure 1(b).

In Figure 1(b) the marginal cost more than doubles when the literacy campaign is enlarged from 80 per cent coverage to universal coverage. As a result, the total cost of the campaign rises more than proportionately. Policy makers in circumstances such as these should ask themselves whether it is worthwhile aiming for 100 per cent coverage, or whether instead the extra funds needed to achieve universal adult literacy would be better spent on some other programme, such as a programme to reduce the infant mortality rate.

The answer depends on the expected benefits of the additional expenditure. Once again, it is often implicitly assumed when setting a target that the marginal benefits (MB) are high and constant, easily exceeding the marginal costs. On this assumption it clearly is worthwhile to aim for universal coverage. This is the situation depicted in Figure 2(a).

It is not difficult to imagine however that just as marginal costs tend to rise as coverage increases, so marginal benefits tend to fall. Some people need literacy skills more than others to earn a living; some people obtain more pleasure from leisure reading than others; some people in the early stages of life will benefit longer from the possession of literacy skills than those who are well past the prime of life. For all these reasons the benefits on the margin of expanding the coverage of a mass literacy campaign are likely to decline. This is the situation depicted in Figure 2(b).

Note that in Figure 2(b), if the target is set at universal adult literacy, the marginal costs will exceed the marginal benefits. Total benefits from the programme will be greater than total costs, but the last dollars spent will have been wasted: they yielded fewer benefits than costs. The optimal size of the programme occurs where marginal benefits and costs are equal. This is point T in the figure. At this point net total benefits, i.e., total benefits minus total costs, are at a maximum and any additional resources that might become available would be better spent on some other programme. A target of T, rather than universal adult literacy, would therefore be ideal.

Target setting, in other words, is not incompatible with a calculus of benefits and costs. Targets have their virtues--they are clear, easy to understand and can be used to mobilise public support--and these virtues can be preserved, but the targets should be set after the benefits and costs have been calculated and not plucked arbitrarily out of thin air. If the targets are set where net total benefits are maximized, policy makers can have the best of both worlds.

Human development and sustainable development

Historical patterns of economic development have produced rapid economic growth in a number of countries for an extended period of time, but these patterns of development also have caused the stock of natural capital in the world to contract considerably and often to become degraded. In addition, there has been a rapid loss of biodiversity. Lastly, there has recently emerged a serious threat of global warming. In the last two decades people have become increasingly aware of the social costs associated with production processes and consumption patterns that harm the environment and this has given rise to demands that henceforth growth should be sustainable.

At the same time historical patterns of development have been criticized for their failure to put people first and this has led to demands for a change to a human development strategy. There is a danger that human development and sustainable development will be seen as competing strategies, as alternatives between which we must choose. This would be a mistake, for if development is seen as a process that widens people's choices and increases their capabilities, then it must do so not only for the current generation but also for future generations. In this sense, development must be sustainable. Human development thus embraces sustainable development, the latter being particularly concerned with the environmental and inter-generational aspects of human development.

There are indeed environmental dangers inherent in current rates and patterns of growth in global production and consumption. Yet there is no inherent conflict between development and the environment, between the reduction of poverty and the maintenance of the stocks of natural, physical and human capital at optimal levels, between sustaining biodiversity and increasing the capabilities of people. Insofar as the stock of natural capital is damaged or consumed at an excessive rate, the prospects for future development are reduced. In that sense good environmental policies are part of good development policies. This is true whether one is concerned with degradation of the environment at the local level (such as pollution of fresh water supplies), at the international level (such as damage to forests in neighbouring countries downwind of factories emitting sulfur into the air), or at the global level (such as destruction of the ozone layer in the upper atmosphere caused by high emissions of CFCs). Sustained development requires that the basis of life on this planet be preserved and that any depletion of the natural stock of capital be transformed into an equivalent value of human or physical capital. The central point is that if the value of the aggregate stock of capital falls--natural, physical and human--the prospects for development are reduced, but if the value of natural capital falls, development prospects can be maintained provided physical and human capital are increased in compensation. 104

Damage to the environment is not caused by human activity as such, i.e., consumption, production and reproduction--any more than it is caused by the activities of other species--but by a set of incentives which induces humans to neglect or overlook some of the costs of economic activity, be it production or consumption. All environmental problems--all apparent conflicts between development and the protection of the environment--can be seen as arising ultimately from either (i) government failure, (ii) market failure, (iii) missing markets arising from undefined property rights, or (iv) high discount rates of the poor because of their inability to sustain life without depleting the stock of natural capital.

Government failure and class bias

Government failure in the context of sustainable development refers to actions taken by local, provincial or national governments which create incentives to use the natural stock of capital wastefully and inefficiently. These defective incentives may reflect ignorance on the part of policy makers, but they may also reflect a class bias in policy making. That is, government intervention may be designed to serve particular economic interests and policies conventionally described as "government failure" may be quite successful in their own terms. In what follows we shall use the phrase government failure in the broadest sense to cover errors committed out of ignorance as well as waste and inefficiency knowingly created while favouring the interests of particular classes or groups in society. Examples of government failure in this broad sense include policies which subsidize the cost of water to users, policies which subsidize the cost of land to ranchers and agriculturalists and often encourage otherwise unprofitable economic activities in areas where the resource base is fragile or marginal, and policies which subsidize logging, and provide little incentive for replanting, in the tropical forests of developing countries.

Correction of government failure would automatically reduce the incentive to over-exploit the environment and make investment in physical and human capital relatively more attractive. That is, adoption of a more sustainable path to development would tend to accelerate human development.

A special case of government failure concerns the failure of the standard national income and production accounts to record the depreciation of the value of natural capital. The standard accounts consequently overstate both the rate of growth of the economy and the level of capital formation. In Costa Rica, for example, it is estimated that unrecorded depreciation in the value of the country's forests, soils and fisheries was on average five per cent of GNP per year between 1970 and 1990. 105 In Indonesia, the unrecorded depreciation of its forests, soils and petroleum resources was about nine per cent of its GDP between 1971 and 1984. 106 These failures to record the consumption of natural capital mean that the official national accounts can be highly misleading, giving a false picture of the level of development.

Market failure

The second source of defective incentives arises not from government subsidies and other forms of government failure but from market failure. This was discussed at some length in the main body of the text in the context of market signals failing to reflect accurately the costs of land use and the benefits of increasing the stock of human capital. In the context of sustainable development, market failure occurs whenever prices generated by the market mechanism fail to reflect fully the costs to the environment and the natural stock of capital of production and consumption activities. In such cases negative externalities ("bads") are produced as well as goods. Where environmental costs are understated by market prices, producers and consumers have an incentive to use resources wastefully and to over-produce those goods that use intensively natural capital.

Unfortunately, many economic processes generate negative externalities and this is particularly true in technologically advanced economies. Examples include the pollution of the air by trucks and automobiles, pollution of rivers and lakes by factories dumping noxious waste in inland waterways, salination of the land in areas where irrigated agriculture unaccompanied by proper drainage is practised, and pollution of coastal waterways by the discharge into the sea of untreated sewage from large cities. Market failure is widespread in the industrialised countries and this accounts for the fact that most global environmental damage is the result of past and current production and consumption patterns in the advanced economies. The industrialised countries, for instance, account for about 75 per cent of the carbon and 90 per cent of the CFC emissions, the two gases believed to cause global warming. In addition, the industrialised countries consume about ten times as much energy, water, minerals and biomass as do the developing countries. Thus the developed countries bear primary responsibility for the present global environmental problems and it is they who have the major obligation to repair the damage. This is a straightforward application of the polluter pays principle.

Note however that the existence of negative externalities and market failure do not imply that either developing or developed countries should reduce their rate of growth in the name of sustainable development. If an economic activity produces a negative externality, the activity in question (say, the output of a specific product) should be reduced. Efficiency in resource allocation requires a changed composition of output, not a reduction in aggregate output. The implication of market failure is that governments should intervene to correct market prices so that producers and consumers respond to a set of incentives that accurately reflects all costs, including of course environmental costs. If this were done, human development activities would become even more attractive than they are now. The general point, however, is that market failure is a cause of defective incentives not a symptom of excessive growth. The solution is to correct the incentive system, not to restrain the level or rate of growth of net output and income. Indeed, well conceived environmental policies--sustainable development--would enhance the overall rate of growth, provided the national accounts are improved and growth is measured correctly by taking into account all benefits and all costs of production.

Missing markets

The third source of defective incentives occurs whenever a market for an environmental service is missing. As explained in the main body of the text, these so-called missing markets arise when property rights are undefined or unenforced. Examples include unclaimed forest land, common pastures under the control of no particular group or individual, fresh water aquifers from which anyone may extract water by digging a well, and at the global level, the high seas, the upper atmosphere and the polar regions. In cases where property rights are undefined no one has an incentive to manage the natural asset to ensure a sustained flow of income and everyone with access to the asset has an incentive to extract maximum benefit as quickly as possible. The results are over-cutting of forests, over-grazing of pastures, withdrawal of water in excess of the recharge rate, over-fishing the oceans, destruction of the upper atmosphere, etc. Here again, however, the solution to the problem of missing markets is to correct the incentive system, by creating enforceable property rights--either individual, collective, state or supra-national--not to retard the process of human development or diminish efforts to reduce poverty.

High discount rates

A major cause of environmental problems in developing countries is restricted access by the poor to income earning opportunities, including opportunities arising from ownership or control of sufficient natural capital to sustain life above a subsistence level. It is becoming widely recognized that in many cases the alleviation of poverty by providing greater access by the poor to resources and employment, and by promoting human development, is probably the most effective policy of sustainable development. A guarantee to the poor of employment, a redistribution of land through a land reform, and a substitution of human capital for natural capital are illustrations of how human development is intimately connected to sustainable development.

Land degradation and desertification are concentrated in areas where poverty leaves the population with little alternative to the excessive exploitation of resources. Deforestation, too, is often the last resort of people who cannot find other ways to generate a minimum level of income. Over-fishing of coastal waters and inland lakes and rivers often occurs because poor fisherfolk have no alternative means of support. That is, because of their very low levels of income the poor often have high discount rates. Their concern is immediate survival, not maintaining the stock of natural capital so that it can generate a sustained flow of income for the indefinite future. As long as acute poverty persists, the interest of future generations will be ignored and the natural stock of capital will be consumed and damaged. Rural poverty and environmental damage thus often go together, and the solution to the latter is to eliminate the former by adopting a human development strategy.

In conclusion, provided sustainable development is interpreted to mean the preservation of the natural stock of capital at a level which is capable of sustaining indefinitely the human population at current standards of living, there is no conflict between sustainable development and human development. Of course one aspires to more, namely, a substantial reduction in poverty achieved either through a redistribution of income or through growth of average incomes per head. Sustaining the present population at current standards of living is too modest a goal. This does not detract from the fact however that sustainable development is necessary to permit long run human development. Equally, human development is essential if the damage currently being inflicted on the environment is to be controlled and sustainable development is to become a realistic possibility.

NOTES

104. This leaves unanswered the difficult question of how changes in the stock of natural capital are to be valued. Various techniques have been developed, e.g., estimates of how much people would be prepared to pay for an environmental service or how much they would be prepared to accept for the loss of an environmental service, none of which are fully satisfactory. The crucial point however is a conceptual one: changes in the stock of natural capital should be subject in principle to what we have called in the text the calculus of benefits and costs.

105. BioScience, Vol. 42, No. 4, April 1992, p. 325.

106. Jan Pronk and Mahbub ul Haq, Sustainable Development: From Concept to Action, New York: UNDP, March 1992, p. 9.