You are on page 1of 6

American Economic Association

The Interaction of Population Growth and Environmental Quality Author(s): Maureen Cropper and Charles Griffiths Reviewed work(s): Source: The American Economic Review, Vol. 84, No. 2, Papers and Proceedings of the Hundred and Sixth Annual Meeting of the American Economic Association (May, 1994), pp. 250-254 Published by: American Economic Association Stable URL: . Accessed: 19/11/2012 07:30
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact

American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic Review.

This content downloaded by the authorized user from on Mon, 19 Nov 2012 07:30:22 AM All use subject to JSTOR Terms and Conditions


The Interactionof PopulationGrowth and Environmental Quality


The study of interactionsbetween population growth and the environmenthas a long history.Accordingto Malthus,a growing population exerts pressure on agricultural land, forcingthe cultivationof land of poorer and poorer quality. This environmental degradation (broadlydefined)lowers the marginalproductof labor and, through its effect on income, reduces the rate of populationgrowth.The result is an equilibrium population that enjoys low levels of both income and environmental quality. The modern statement of this view replaces agricultural land with nonrenewable resources. In this model, natural resources impose a limit to economic growth, with populationpressuresreducingthe marginal productof labor as scarce naturalresources are exploitedmore intensively. A more recent theme in discussions of population growth and the environmentis the importanceof environmental qualityper se, where environmental quality is measured by the stock of forests or by the absence of air and water pollution. In this view the environment is seen not as a factor that limits productivityas population expands, but as a good whose quality is degradedby a growingpopulation.Population pressures,for example,are frequentlycited

tDiscussants: Sudhir Anand, Harvard University; Sherman Robinson, Council of Economic Advisors; Robert Repetto, World Resources Institute. *The World Bank, 1818 H St., N.W., Washington, DC 20433, and Department of Economics, University of Maryland, College Park, MD 20742. Special thanks to Bill Evans for this helpful comments on this work. 250

as a cause of deforestation: population growth, by increasing the need for arable land, encourages the conversion of forest land to other uses. Populationgrowth,because it places increased pressure on the assimilativecapacityof the environment,is also viewed as a majorcause of air, water, and solid-wastepollution.To some, the logical conclusion of these argumentsis that populationcontrolis an importantmeans of improving environmental quality. While there is no question that population growth contributes to environmental degradation,its effects can be modified by economic growth and modern technology. Consider, for example, two countries with rapid populationgrowthand significantforest resourcesbut with differentlevels of per capita income. The countrywith the higher income is likely to be deforesting less rapidly.As income grows,people will switch to energy sources other than firewood and will use modernagricultural techniquesthat reduce the demand for agriculturalland. Similareffects are likelyto be felt regarding pollution. As income grows, sanitation and waste-water treatment will improve, and pollution will be less of a problem at any level of populationdensity. An important question for policy is whether, holding constant per capita income and other relevantfactors,population pressures have a significanteffect on environmentaldegradation.To the best of our knowledgethere is little empiricalevidence on this point. In this paper we take a first step toward providing such evidence. we examinethe effect of popuSpecifically, lation pressureson deforestationin 64 developing countries.

This content downloaded by the authorized user from on Mon, 19 Nov 2012 07:30:22 AM All use subject to JSTOR Terms and Conditions

VOL. 84 NO. 2



Previous research in environmentaleconomics has uncovered a relationship between environmentalquality, measured by ambient concentrationsof SO2 or particulates, and per capita income (Gene M. Grossmanand Alan B. Krueger,1991).This so-called environmental Kuznets curve shows environmentalquality worsening up until about $5,000 of per capita income (using Robert Summers and Alan Heston [1991] purchasing-power-parity income measures)and improving thereafter.Below, we find a similarrelationshipfor the rate of deforestationin Latin America and Africa. We estimate this relationshipusing pooled cross-sectionand time-series data for each continent for the period 1961-1988, including countrydummiesto capturefactorsthat changeslowlyover time, such as the proximity of forests to cities or rivers. To capturethe effects of populationpressures we include rural population density and the rate of population growth in the equation as well. These variablesthus shift the Kuznets curve for deforestation. It is thus possible for a country that is beyond the level of per capita GDP at which environmentalqualitybegins to improveto have a higher rate of deforestationthan a country that has not yet reached this level of GDP but faces lower populationpressures. This is a simple point, but one that deserves emphasis:the vertical intercept of environmental Kuznets curves is just as important as the level of per capita GDP at which the curvepeaks.
I. Causes of Deforestation

In the literature of deforestation, three reasons are highlightedfor the destruction of tropical forests: the desire to convert forest and woodland areas to pasture and cropland, the harvesting of logs, and the gatheringof fuelwood.Populationpressures are emphasized as an underlyingcause of all three sources of deforestation.Population growth,by increasingthe demand for arable land, encourages the conversion of forests to agriculture. Since it is people living in ruralareas who turn to agriculture as

a livelihood,one would expect deforestation to increase with rural population density. Population growth also increases the demand for wood, both for timber and for fuelwood. The links between population pressures and deforestation are thought to be so strongthat, in a recent assessmentof deforestation in tropical countries, the United Nations Food and AgricultureOrganization (FAO) estimateddeforestationrates using a model of population pressures (FAO, 1993a), Specifically,the FAO assumed that the ratio of forest area to total land area is a logistic function of population density. This model, estimatedusing data for a sample of countries at the subnational level, was used to predict national rates of deforestation for countriesoutside the sample.' The FAO model implies (after some manipulation) that the percentage change in forest area depends on the percentage change in population (the rate of population growth),as well as on populationdensity. This relationship, however, will be modified by a country'sstage of economic development. The relationship between population pressures and deforestation to create arable land is clearly affected by the use of modern agricultural technology, which reduces land requirements.It is also affected by the pace of industrialization, which means that labor will be hired in nonagricultural sectors.Loggingis also likely to be linked to income, interpreted as a proxy for the stage of development. It is likely to grow as a countrydevelops, especially as the abilityto process logs develops, but maywane as industrialization takesover. The demand for fuelwood as an energy source is also a function of income. It may initially rise with income, but is eventually likely to fall with income as more modern sources of energy are used. The previousdiscussionmotivatesthe importance of income and population growth

IWe emphasize that this is not the source of the data used in our analysis.

This content downloaded by the authorized user from on Mon, 19 Nov 2012 07:30:22 AM All use subject to JSTOR Terms and Conditions




as factors underlyingthe rate of deforestation. Other variablesthat may influencethe rate of deforestationare the rate of growth in per capita income (which is correlated with the rate of urbanization) and the prices of forest products,especiallylogs. Letting Fit representforest area in country i in year t, the equationbelow specifies a possible formfor the relationshipbetween the rate of deforestation (minus the percentage change in forest area) and the factors discussedabove:
Fi, tF1it

II. The Data


ao,i + al(RPD)it + a2(A POP)it + a3(TP)

+ a4(A PCGDP)it + a5(PCGDP)it + a6(PCGDP)2+ uit

where RPD = rural population density, APOP = percentage change in population,

TP = timber price, APCGDP = percentage

change in per capita GDP, PCGDP is per capita GDP, and uit is an errorterm. The reason for assuminga quadraticrelationship between deforestation and per capita GDP is that logging and fuelwood uses of the forest are likely at first to increase with income. Agriculturaland fuelwood motives for deforestation, however, are eventually likely to decline with per capita GDP, causing an inverted U-shaped relationship. The intercept of the equation is allowed to vary across countries to capture factors affecting the rate of deforestation that change slowly over time. Deforestation,for example, is more likely to take place the closer forests are to cities and rivers. The size and distributionof forests can also affect the rate of deforestation:forest area that is clusteredis likely to be less vulnerable to deforestationthan fragmentedforest -forest that is interspersedwith agricultural and other land uses. The density of trees likewise affects the profitability of logging. It is these factorsthat the fixed effects are intended to capture.

The source of our deforestation data is the Food and Agriculture Organization's ProductionYearbook(FAO, 1993b), which providesdataon forestsandwoodlandarea.2 This is a very broad definition of forest land. It includes both closed and open forest, plantations,and land fromwhichforests have been cleared but which will be replanted in the foreseeablefuture.This is an acceptabledefinitionof forest from an economic perspective;however,it is too broad a definitionto be useful for studyinghabitat destructionor the loss of biodiversity. Because deforestationis primarily a problem of developing countries, we have limited our study to non-OECD countries in Africa, Asia, and Centraland South America (hereafter referred to collectively as Latin America) roughlyin the tropicalbelt and containingforest area of over 1,000,000 hectares. Separateversions of the equation are estimated for each continent. The parameters of the equation may vary across continents because the nature of forests varies significantlyfrom one continent to another.Moisttropicalforestscomprisehalf of the forest cover in Latin America, for example, but only 17 percent of the forest cover in Africa. Data on populationand per capita GDP come from Summersand Heston (1991). It is the availabilityof these data that limits the size of our sample: the SummersHeston data are availablefor only 64 countries for which deforestationdata exist, and are availableonly through1988. Since some years are missing for some countries, we have an unbalanced panel. Data on the price of tropical logs are from the FAO (1981, 1990), which reports international pricesfor forest products.These priceswere adjusted by the average mid-year market exchange rate for each country, as pub-

2The 1993 AGROSTATtapes providea consistent time series for forests and woodlandsfrom 1961 to 1991, using data published in the FAO Production Yearbook. It is those data that are used here.

This content downloaded by the authorized user from on Mon, 19 Nov 2012 07:30:22 AM All use subject to JSTOR Terms and Conditions

VOL. 84 NO. 2



increase in the rate of deforestationlevels off as income increases. The coefficient on average rural populaIndependent Latin variable tion density is positive and significant in Africa America Asia Africa, givingthe expected result that rural Per capita income 3.90 6.03 -13.33 ($millions) population density increases deforestation. (2.60) (1.93) (-0.71) Per capita -410.05 -556.29 1,386.77 The magnitudeof this coefficient,which is income squared (-1.68) (-1.54) (0.53) Percentage 0-3 -5.95x 10-2 -1.23x 7.28 x 10 - 3 roughlysimilarfor both continents,implies change in (-2.46) (-3.23) (0.20) that an increase in ruralpopulationdensity per capita income Price of tropical - 1.00ox o01.92 x 10 - 4 -8.70 x 10 of 100 persons per 1,000 hectaresraises the logs ($1,000's) (-0.40) (2.41) (-0.20) rate of deforestation by 0.33 percentage Percentage change 8.33 x 10-3 1.96 x 10 - 2 1.35 x 10-i in population (0.13) (0.39) (0.20) points in Africa. Rural population 3.26 x 10 - 2 3.63 x 10 - 2 -2.09 x 10-3 The rate of growthin per capita income density (3.79) (1.08) (-0.21) Time trend 1.84x 10-5 -6.30x 10-6 3.60x 10-4 also has a significant negative impact on (0.34) (-0.05) (0.79) deforestation, although the magnitude of Number of observations: 862 450 364 this effect is small. In Latin America, for R2: 0.63 0.47 0.13 example, increasing the rate of growth in Turning point: $4.760 $5,420 per capita income by 8 percentage points Notes: The depenident variable is the annual rate of deforestation. reduces the rate of deforestation by only Equations for Africa and Latin America were estimated using the Prais-Winsten technique to correct for autocorrelation. Numbers one-tenth of a percentagepoint. The price below coefficients in parentheses are t statistics. of tropical logs is statisticallysignificantin Latin America but not in Africa, a reasonable result given that logging occurs on a much larger scale in Latin America than in lished by the IMF, to obtain the price faced Africa. by domesticwood producers. The clear anomaly in Table 1 is Asia: To compute rural population density we none of the variables in our equation is obtained the percentageof total population statistically significant for this continent. living in rural areas from the World Bank Breakingthe region into South Asia versus and used this numberto adjusttotal popu- East Asia does nothing to improvethe relation figures from the Summers-Heston sults. A possible explanationfor this finding data. concerns the importance of forest plantations in Asia. The FAO has estimatedthat, III. Results in 1990, natural forest area in Asia decreasedby 3.9 million hectares;however,an Table 1 reports the results of estimating additional2.1 millionhectareswere planted, our equationfor Africa,LatinAmerica,and implying that the decrease in forest and Asia. Only results for Africa and Latin woodland area would amount to only 1.8 America are statisticallysignificantat con- million hectares (FAO, 1993). The factors ventional levels. They suggest, first, that a influencing the destructionof naturalforests hump-shaped relationship exists between are, however, likely to differ from the facper capita income and deforestation;and tors influencingthe growth of plantations. second, that, for Africa, rural population An increasein the price of tropicallogs, for densityshifts this relationshipupward. example,is likely to increaseboth, implying There is, however, a disquietingfeature that it might have no measurableeffect on of the quadraticrelationship between defor- the sum of forest area plus plantations.We estation and per capita income. The levels suspect that, were we able to decompose of income at which rates of deforestation the change in forest and woodlandarea into peak ($4,760for Africa and $5,420for Latin these two components,we would find a reAmerica)are such that most of our observa- lationship similar to that for Africa and tions fall to the left of the peak. It would Latin America for deforestationof natural therefore be more accurateto say that the forests in Asia.


This content downloaded by the authorized user from on Mon, 19 Nov 2012 07:30:22 AM All use subject to JSTOR Terms and Conditions



MA4Y 1994

IV. Implications

Macroeconomicrelationshipsof the type reportedin Table 1 are sometimesmisinterpreted as indicatingthat income growth,if fast enough, will solve environmental problems. This is clearlynot the case for deforestation in Latin Americaor Africa. For a countryin Latin America with an interceptthat is zero, the rate of deforestation at the peak of the curve is 1.63 percent per annum.Even at a per capita income of $8,000,the rate of deforestationis 1.26 percent-surely not an indication that economic growth will solve the problems of deforestation! In Africa, the situation is yet more grim. Rural populationdensityshifts the relationship between income and deforestationupward,so that a countrywith a ruralpopulation density equal to that of Kenya (0.3 persons per hectare) has a peak deforestation rate of 1.91 percent per year, while a countrywith the ruralpopulationdensityof Malawi(0.7 personsper hectare)has a peak deforestationrate of 3.21 per year. The implied trade-offbetween per capita income and ruralpopulationdensityis large: at a per capita income of $4,760, a country with a rural population density of 0.1 persons per hectare(the averagefor the African countries in our sample) has a peak deforestation rate of 1.26 percent per year. A country with a population density of 0.7 persons per hectare requires an income of $11,650per year to achievethe same rate of deforestation! In spite of these grimpredictions,it would be inappropriateto conclude that reducing the rate of populationgrowthis necessarily the best method of reducing the rate of

deforestation. Deforestation in developing countriesis very much a problemof market failure. Because property rights are often not definedor not enforced,the privatecost of deforestation is effectively zero. Put somewhat differently,because people have no rightof ownershipin the land, they have no incentiveto make efficientland-usedecisions. It is this problem that must be addressed, as well as the problemsof poverty and populationgrowth. REFERENCES
Food and Agriculture Organization. Forest

productprices, 1961-1982. Rome: Food and Agriculture Organization of the United Nations, 1981.

Forest product prices, 1969-1988. "Forest Resources Assessment

Rome: Food and AgricultureOrganization of the United Nations, 1990. 1990: Tropical Countries."Forestry Paper No. 112, Food and AgricultureOrganization of the United Nations, Rome, 1993a.
. FAO production yearbook. Rome:

Food and Agriculture Organization of the United Nations, 1993b.

Grossman,Gene M. and Krueger,Alan B. "En-

vironmentalImpacts of a North American Free Trade Agreement." Working Paper No. 3914, National Bureauof Economic Research (Cambridge, MA), November1991.
Robertand Heston,Alan. "The Penn Summers,

World Table (Mark 5): An Expanded Set of International Comparisons, 1950-1988." QuarterlyJournal of Economics,May 1991, 106(2), pp. 327-68.

This content downloaded by the authorized user from on Mon, 19 Nov 2012 07:30:22 AM All use subject to JSTOR Terms and Conditions