See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/23722017

Agricultural Development: Issues, Evidence, and Consequences

Article · October 1997

DOI: 10.1007/978-1-349-26188-8_8 · Source: RePEc

CITATIONS READS

17 546

3 authors, including:

Donald F Larson
College of William and Mary
105 PUBLICATIONS   2,389 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Donald F Larson on 26 May 2014.

The user has requested enhancement of the downloaded file.

 Wf's j%11
POLICY RESEARCH WORKING PAPER 1811

Agricultural Development lreg~~ o

Odr he
E or. fto

Issues,Evidence,andConsequences ~
coriuters in theformnof
to

6ow1 prices..Thus
histoiv
| -<t3~~~~~~~~~~~~~~~~~jstles~
pubbcspending onT0
Yair Mundlak -- ratui-rat research.
Donald F. Larson
Al Crego

TheWorldBank
DevelopmentResearchGroup
August1997
I
POLICYRESEARCHWORKINGPAPER1811

Summary findings
A comprehensive examination of data from many
countries shows that in 1967-92, 81 percent of the
world's population lived in countries where agricultural
growth exceeded population growth. Moreover, that
growth occurred as agricultural prices declined.
Productivity gains are a dominant characteristic of
agriculture for the period. Average productivity
increased for land and labor. Moreover, agricultural
productivity gains were greater than average productivity
gains for the economy in 80 percent of the countries
studied.
Measuring the effects of technology choice on
productivity is crucial to understanding the determinants
of agricultural growth. After selectively reviewing
applied production studies, Mundlak, Larson, and Crego
conclude that the choice-of-technique method, which has
its roots in Tinter's early production function studies, is
best suited for examining the determinants of agricultural
growth.
Investments in technology have yielded large gains for
agriculture, and the benefits have been passed on to
consumers in the form of lower prices. Thus history
justifies public spending on agricultural research.

This paper -a product of the Development Research Group - is part of a larger effort in the group to examine the
determinants of agricultural growth. The study was funded by the Bank's Research Support Budget under the research
project "The Determinants of Agricultural Growth" (RPO 679-03). Copies of the paper are available free from the World
Bank, 1818 H StreetNW, Washington, DC 20433. Please contact Pauline Kokila, room N5-030, telephone 202-473-3716,
fax 202-522-3564, Internet address pkokila@worldbank.org. August 1997. (37 pages)

The PolicyResearchWorkingPaperSeriesdisseminatesthe findingsof work in progressto encouragethe exchangeof ideasabout

arelessthanfully polished.The
developmentissues.An objectiveof theseriesisto getthe findingsout quickly,evenif thepresentations
paperscarrythe namesof the authorsand shouldbe citedaccordingly.The findings,interpretations, and conclusionsexpressedin this
paperareentirelythoseof the authors.They do not necessarilyrepresentthe view of the WorldBank,its ExecutiveDirectors,or the
countriesthey represent.

Produced by the Policy Research Dissemination Center

 AGRICULTURAL DEVELOPMENT; ISSUES, EVIDENCE, AND

CONSEQUENCES

Yair Mundlak, Donald F. Larsonand Al Crego

Yair Mundlak is a professor in the Economics Department, University of Chicago;

Donald F. Larson is an economist in the Commodity Policy and Analysis Unit at the
World Bank; and Al Crego is a consultant at the World Bank. The authors would like to
thank Rita Butzer and Nanae Yabuki for valuable research assistance. This research was
conducted at the World Bank as part of a larger research effort on the determinants of
agricultural growth (RPO 679-03).
 CONTENTS

1. THE ISSUES . .............................................. I

2. THE EVIDENCE . .............................................. 2

2.1 Supply and Demand ........................................... 2
2.1.1 The Global Food Experience - an Overview ...... .......... 2
2.1.2 Demand and Agricultural Growth ....... ................. 4
2.1.3 Trade .............................................. 7
2.2.1 Land .............................................. 7
2.2.2 Labor .............................................. 8
2.2.3 Summary ........................................... 11

3. AGRICULTURE AND FARMERS . .................................... 11

4. AGRICULTURE AND DEVELOPMENT .......... ..................... 16

5. RESEARCH IMPLICATIONS .......................................... 17

6. THE PRODUCTION STRUCTURE OF AGRICULTURE ...... ............. 20

REFERENCES .............................................. 26

FIGURES .............................................. 27

i
 AGRICULTURALDEVELOPMENT;ISSUES, EVIDENCE,AND

CONSEQUENCES

1. THE ISSUES

Manyroads lead to the interest in agriculturaldevelopment,and it is only natural

that we find differencesamongeconomistsin their analysesand more so in perspectives,

emphasis,and conclusionsaboutthe processand its outcome. To providea perspective

for our discussion,it is useful to listthe centralissuesrelatedto agriculturalgrowth:

* Agricultureas a supplierof food,

* The welfareof farmersand more generally,the ruralpopulation,

* The role of agriculturein economicdevelopment,

* The impactof governmentinterventionon the performanceof the agricultural

sectoras a subjectof political economy.

* Traditionallyagriculturehas been studiedwith the purposeof checkingthe

validityof generaleconomicpropositions,and this also appliesto the study of

variousaspectsof growth.

Keepingthis diversityof interestsin mind, it is instructiveto conductthe

discussionagainstthe backgroundof the evidence. To do so,we presentresults from a

comprehensiveexaminationof a data base of a large numberof countriesfor the period

1960-1990or thereabout.' The presentationhighlightsimportantempiricalpropositions-

1
mostly knownbut oftenignored- and providesa quantitativedimensionof the cross-

countryvariabilityof levels and rates of changeover time. It is then followedby a short

overviewof the consequencefor policy and analysis. Givenspace limitations,no attempt

is made hereto documentall the assertions. This will be done elsewhere.

2. THE EVIDENCE

2.1 Supplyand Demand

2.1.1 The GlobalFoodExperience- an Overview

There is a keen interestin futuremarketdevelopmentsrelatedto food supply.

This interest is more intensethan in othersectors becauseof the uniquerole of food and

the natureof its demand. Much of the concernwith food supplycan be explainedby

the fear supplyshortagesand hunger at some pointin the near or remote future. In the

dialogueon this issue, referenceis sometimesmade to the experiencesof countriesthat

sufferedfood shortagesfor extendedperiods of time or to episodesof hunger in affluent

economies. Althoughtraumatic,such episodesdo not generalizeto givea correct global

picture. It is here wherethe empiricalperspectiveis essential.

A summaryof the growthexperienceof countriesis givenin Figure 1 which

presentsa frequencydistributionof total agriculturalproductiongrowthrates for the

2 The curve to the left is drawnwith equal weight for

period 1967-1992for 130countries.

each country. At any pointon the graph, the readingon the vertical axis showsthe

2
proportionof countrieswhosegrowthrate did not exceed the correspondingvalue on the

horizontalaxis. There is a wide spread in countryperformances,but most countrieshad

positivegrowth,and the medianrate is 1.92percentper year. As some of the low

growthcountriesare small in terms of their agriculturalproduction,this curvedoes not

providea goodview of the changesin global supply. A better view is obtainedwhenthe

relative size of countryis taken into account. This is done in the curve to the right,

where the countriesare weightedby their relativeimportancein world agricultural

production. Countrieswith negativegrowthrates carryvery little weightin terms of their

contributionto overall food production. The medianaverage growthrate of the weighted

distributionis 2.25 percent. This meansthat 50 percentof the food productionis

producedin countrieswhosegrowth ratesexceeded2.25 percent,and the remainingis

producedby countrieswith lower but usuallypositive. Put anotherway, growthrates

havebeen somewhathigherin countrieswith largeagriculturalproduction.

Has supplylaggeddemand? If this were the case,agriculturalprices wouldhave

risen, but this has not happened. This is seen in Figure2, which presentsthe distribution

of the rates of change of agriculturalpricesover the sametime period? Domesticfarm

pricesare deflatedby domesticconsumerprice indices. The medianof the unweighted

distributionis -.45 percentper year and when we weighthe countriesby their importance

in world production,the median changesto -.61 percentper year. Note that the vertical

line at zerogrowthrate cuts the graph of the weighteddistributionat .71, indicatingthat

71 percent of worldproductionin the period 1967-1992was producedin countrieswhere

3
real pricesfell.

Becausethe prices in Figure 2 are of aggregateoutput,their rates of changetend

to be biasedupward. As incomeincreases,there is a shift in consumptionand production

to highervalue products,and this changein compositioncausesthe aggregateprice to

increase. Thus,the fact that we observea declinein prices evendespite an upwardbias,

suggeststhat supplygrew fasterthan demand. This result is consistentwith the finding

reportedin Binswangeret al (1987)that over the period 1900-1984,worldprices of

importantagriculturalcommodities,such as the major cereals,deflatedby US wholesale

prices,declinedat averageannual rates of .5 to .7 percent. This overviewindicatesthat

recent experience,which generalizeswell to the entire post-warperiod, is of faster

growthof supplycomparedto demand.

2.1.2 Demandand AgriculturalGrowth

Whatare the sourcesof the spread in growthrates? This is of coursethe main

questionin the studyof growth in generaland of agriculturalgrowthin particular. The

searchfor answersnaturallyfocuseson the supply side. However,beforeexamining

factors relatedto supply,it is instructiveto examinethe role of demand.

On the whole,agriculturalproductsare tradable,and thereforethere is no need

for each countryto increase its productionto meet domesticdemandchanges. Some

countriescan benefit from comparativeadvantageand reducetheir productionin an

environmentof growthand of reductionof trade restrictions. Suchbehaviormay explain

4
some of the observedspreadin growthrates. For instance,Singapore's agricultural

productionmorethan doubledbetween 1967and 1979,but declinedthereafterto such an

extent that 1992productionwas below that of 1967. Other countriesdo not fully utilize

their potentialin agriculturalproductionin order to protectfarm incomeand are engaged

in policiesthat involvea reductionof their domesticproductionand restrictionson

imports. Togetherthese restrictionsindicatethat there havebeen excess supplies.

Further, importbarriersfurtherrestrict the expansionof productionin exportingcountries

as potentiallyimportingcountriesattemptto use their own marketsfor the development

of their farm sector.

An indirectview of the role of demandcan be obtainedby comparingthe

distributionof growthrates of per capitaoutput,presentedin Figure 3, with total

output, in Figure 1. The medianrate of changeof per capitaproductionof the equal

weightdistributionis minus .03 percent althoughin about 51 percent of the countries

output exceededpopulationgrowth. Taking into accountthe productionweights,the

mediangrowthrate is .7 percent,and outputgrowthexceededpopulationgrowthin

countriesthat accountedfor 81 percentof world production. Sucha pronounced

differencebetweenthe two distributionsunderlinesthe importanceof distinguishing

betweencountryproblemsand global problems,and the pitfall of generalizingfrom

countryperformanceto worldperformance.

We also notethat the graphsin Figure 3 are moresteep thanthe corresponding

graphs in Figure 1.4 This impliesa higherconcentrationof countriesaroundthe median

5
growthrate, that is, the varianceof total growthrates is largerthan that ofper capita

rates. This concentrationin the center indicatesthat manycountriesmet a largepart of

growthin demandthrough home production. In fact, on the whole, productiongrowth

exceededpopulationgrowthsomewhatbecauseof the increasein demanddue to the rise

in incomeas well as the declinein prices. Clearly,this developmentis consistentwith

demandderivedfrom productiongrowth.

A more directview of the role of demand is obtainedby comparingthe rate of

growthof agricultural per capita outputwith that of per capitatotal output,taken as a

measureof income. Ignoringtrade for a moment,we notethat for the closedeconomy

under constantpricesthis ratio is roughly the income elasticityfor food. We say roughly,

becauseagricultureis not identicalwith foodand it includesindustrialcrops whose

demandis not the sameas that of food. However,in most countries,food dominates

agriculturalproduction.

Figure 4 presentsthe distributionof such a ratio for growth in real value for 91

countriesduring 1960-1992.5The medianvalues are .82 and .86 for the weightedand

unweighteddistributions,respectively. Since priceswere not constant,and rather

declined,this resultservesas an upper boundfor the income elasticity.

We emphasizethe role of demandbecausemost discussionsof economicgrowth

deal with a singlegoodeconomywhere there are no prices, exceptfor the discountrate.

However,the single-goodeconomyis an aggregateof many sectors,and thereforethe

discussionof growth at the sectoral level adds a dimensionmissing from aggregate

6
models. Indeed,when it comesto agriculture,it is demandwhich has an enormous

impacton the dynamicsof the sector.

2.1.3 Trade

Agricultureproducesa tradableproduct,but the foregoingreviewindicatesa

tendencytoward self sufficiency. This is reflectedin the fact that trade accountsfor only

10-12percentof total world production. Thistendencytoward self-sufficiencyis in part

due to policies as indicatedabove,but is also due to objectivemarket conditions. The

marginbetweenimportand exportpricesmay be wide, distributioncosts withinthe

country,both for the exporterand the importer,canbe high, especially in countrieswith

inadequateinfrastructure.

2.2 Inputsand ProductivityChanges

Turningto the supplyside, the joint event of output growthand decliningprices

indicatesthat productivitygrowthmore than offsetsthe effect of the decliningpriceson

profitability. To evaluatethe sourcesof growthwe examinechangesin land and labor,

the two inputsthat togetheraccountfor the large proportionof output.

2.2.1 Land

It is oftenthoughtthat agriculturalland is a fixed quantity. This view is true for

countries' total physicalareaswhereascultivatedland,hereon"land," is an economic

7
quantityand as seen in Figure 5, was subjectto changes. In 65 percent of the countries

the cultivatedarea increasedduring 1967-1992,with a median averageannual growth

rate of .4 percent. For the world as a whole, the averageannualgrowthrate was .58

percentwhich amountsto about 15.7percentgrowth for the period as a whole. Land

growthwas fasterin tree crops, 1.48,than in annual crops, .52, percentper year. Thus, in

spiteof the lowerprices for agriculture,it was still profitableto expandthe area of

agriculturalland.

The growthrate of land is muchsmallerthan that of outputand this reflects an

increasein land productivity.Figure 6 presentsthe distributionof growthrates in

averageland productivityfor 87 countries. In 11countries(13 percentof the countries)

the average productivitydeclined,but thesecountriescarrysmall productionweights

(less than 2 percent of production)and the weighteddistributionshowsa concentrationin

the positivegrowthrates. Thus,the bulk of world productionwas obtainedwith an

increasein land productivity.The mediangrowthrates are 1.8and 1.92percent for the

unweightedand weighteddistributionsrespective!y.

2.2.2 Labor

Changesin the agriculturallaborforcehave taken a somewhatdifferentpattern.

In 40 percent of the countries, total agriculturallabor declined. Figure 7 presents the

distributionof changesin the laborforce for 148countriesduring 1950to 1990. The

mediangrowthrate was .56 percentwhile similarto that of harvestedland,there is a

8
substantialspread acrosscountries.

Severaltechnologicalfactorsaffected labordemand. In part, the labor

requirement,particularlyharvestlabor,is increasingwith the level of outputand as such

increaseswith yields. Similarly,when the compositionof outputchangesfrom cerealsto

fruits and vegetables,labor requirementsincreaseas well. Also,an increasein livestock

productionfrequentlyincreaseslaborrequirementsfor agriculturallabor. Finally,the

increasein cultivatedland addsto the labordemand. On the otherhand, mechanization

of agriculturereducesthe laborrequirements.All of this regardsdemandfor labor input

but the data report a differentvariable:the laborforce in agriculture. This measure

includesworkersnot fully employedin agricultureand manyof them eventuallyleave

agriculture,but the timing and the degreeof such migrationdependson the off-farm

work and incomeopportunities.

With all thesequalifications,the net effect of mechanizationon agriculturallabor

input is detectedin the labor-to-landratio which are presentedin Figure 8. In abouthalf

of the countriesthis ratio declined,and the rate of declinewas strongerthan the rate of

increasein the other half of countries. This uneveneffect is attributedto the dependence

of the degree of mechanizationon the economicenvironment,which determinesthe

profitabilityof mechanization.It is the poorer countries,with lowerwage-rentalratios,

that rely moreheavily on laborthan on machines. In those countries,the off-farm

migrationwas not sufficientfor the data to show a declinein the actual labor inputin

agriculture.

9
 When outputper unit of land increaseswhilethe laborper unit of landis

constant,there is an increasein productivity.The distributionof the growthrates of

average laborproductivityin 87 countriesis drawn in Figure9. The growthratesof

average laborproductivityare somewhathigherthan that of average landproductivity.

The median ratesare 2.0 and 2.6 percentfor the unweightedand weighteddistributions,

respectively. A comparisonof the weighteddistributionsof average laborand land

productivityfor 87 countriesis presentedin Figure 10.6

Changesof these partial-productivitymeasuresreflectchangesin the agricultural

labor force,cultivatedland,other inputs and technology. The majorinput that is missing

from the discussionso far is capital. Data on this variableis limitedbut preliminary

estimatesindicatethat it grew fasterthan output. However,its factor share in total output

can be quite high and hence it can have a strongdirect effect on productivity.In any

case, it is importantto emphasizethat perceivedproductivitychangesare outcomesof

decisionson inputs and implementedtechnology. Thesedecisionsare endogenousand

have to be studiedin conjunctionwith, and as a part of, the productivityanalysis. We

returnto this subjectbelow.

How does the productivityperformancein agriculturecomparewith that in non-

agriculture? Figure 11 presentsthe distributionof the differencesin the growthrates of

averagelaborproductivityin nonagricultureand agriculture. The median valuesare -

1.58and -1.1 for the unweightedand weighteddistributions,respectively,which implies

a highergrowthrate in averagelaborproductivityin agriculture.However,what is

10
strikingis that laborproductivitygrew faster in agriculturethan in the rest of the

economyin nearly 80%of the countries.

2.2.3 Summary

What comesout clearlyfrom this descriptionis the existenceof a large spread in

inputand outputchangesacross countries. The importantquestionhere is what accounts

for sucha spread and more specifically,to what extentdoes the economicenvironment

contributeto this spread. The term 'economicenvironment'is used here to representthe

set of all the variablesthat affect the productiondecisionsmade by firms. These include

prices,interestrates, price uncertainty,availabilityof inputs,the competitivenessof the

inputmarkets,availabilityand quality of infrastructure,the accessto new information,

the legal and organizationalinfrastructureneededto conductbusinessactivity,the

stabilityof the economicvariables,tax structureand alike. Not all of these variables

changeto a degree that affects productionin a givencountryin any givenperiod,and the

strategyof researchis to concentrateon isolatingthe effectsof the importantvariables

that did change. The list of pertinentvariablesmay vary from countryto countryand the

purposeof a countryanalysisis to identifythem and therebysupplementthe results

obtainedin a cross-countryanalysis.

3. AGRICULTUREAND FARMERS

In evaluatingthe impactof technicalchangeon agricultureit is importantto

11
differentiatebetweenagricultureas an economicactivity(or a sector)and farmers. The

welfareof the sector is measuredby the returns to land,which is the factor specificto

agriculture. The main welfare issuerelatedto farmersis their incomelevel and, to a

lesser extent, its stability. Farm operatorsbenefitfrom increasesin landrent, residual

income,and laborincome. In this discussion,we ignoreincomefrom other formsof

capital withoutaffectingthe main conclusionsof the discussion.

As economictheory and the empiricalliteraturetell us, the returns to land are

reflected in landprices. Thus, if the agriculturalsectorbenefitsfrom the technical

change,land priceswould be expectedto increase,and it is therefore of interestto

observethe changesin land pricesover time. Unfortunately,there is no readily available

comprehensivedata set with time series of land pricesfor variouscountries. In view of

the importanceof this variable,togetherwith Rita Butzer,we are assemblingdata by an

intensivelibrary search. At this pointwe reviewsome of the data assembledthus far.

A cross-countrycomparisonof landprices is a difficulttask becauseof quality

differences,notwithstandingthe difficultyof quantifyingthe importanceof the economic

environmentthat affects such prices. Nevertheless,the time path of landprices in the

various countriesis telling. Thereare two ways,pertinentto our discussion,to look at

land prices. The first is to deflateland pricesby the price of the agriculturalproduct,and

the secondis to deflatethem by the price of the consumptiongood as measuredby the

"CPI." The first, referredto as the outputmeasure,providesthe price of land reflecting

its productivity,whereasthe second,referredto as the consumptionmeasure,indicates

12
the value of land in terms of the purchasing power it has over an aggregate consumption

good, whose price is measured by the "CPI."

Figure 12 shows the pattern of real agricultural land prices for the United States,

Canada, South Africa and Japan countries for which extended time series are currently

available. In Figure 12, land prices are reported as indices with 1986 = 1. The upper

panel shows the output measure of land prices whereas the lower panel shows its

consumption measure. By both measures, land prices declined in the pre-war period.

They were gradually increasing in the immediate postwar period. This rise gained

impetus during the inflationary period of the 1970s and came to an end in the early

1980s, when the economies started to deflate.

Comparing land prices at the beginning and end of each series, we see that by the

output measure prices are considerably higher, but according to the consumption measure

there is little change. In the US, the consumption-land-price index was .84 in 1992,

compared to .73 in 1910. For Canada the values for l990 and 1914 are .87 and .76,

respectively. The series for South Africa begins in 1940 at a level of .88, whereas the

1993 value is .65. Data for other countries show the same trend. On the whole, prices in

the 1990s are historically relatively low, and it is remarkable that land prices today are

not much different from those at the beginning of the century. Moreover, land prices

reflect subsidies to agriculture. If account were made for subsidies (which are relatively

new), then land prices would exhibit even less growth over their historic levels.

Another striking observation is the high correlation between the price movements

13
for the four countriesplotted in Figure 12,as well as for most countries. The correlation

is strongerfor the output measurethan for the consumptionmeasure. This correlation

indicatesthat muchof the impactof the dramaticchangesin agriculturaltechnologywas

transmittedto the various countries(or at least to those observedhere). However,the

benefitswere not capturedby agricultureand eventuallywere distributedto the

consumers. This was done through a declinein agriculturalprices relativeto prices of

the otherconsumptiongoods.

The changesin the welfareof people in agricultureis a different,and far more

complex,story. The incomeof farmersis the returnto their laborand factorsin their

possession. Aside from the returnsto land discussedabove,whichapplies mainlyto land

owners,laborincomeis the main income sourcefor farm operatorsand primarilythe

only sourcefor landlesslabor.7 Dataon returns to labor are deficient. Therefore,we will

examine changesin the structuralcompositionof the labor force and inferthe impactof

the changesin technologyon labor.

Eventhough in most countriesthere was an increasein the agriculturallabor

force,this increaselaggedbehindthe increase in the total laborforce due to off-farm

migration. This is seen by comparingthe growthrates of agriculturallaborwith that of

the population. Under the assumptionof equal fertilityrates in agricultureand

nonagriculture,the growthrate of agriculturallaborless that of the populationgivesthe

rate of labormigrationfrom agricultureto nonagriculture,expressedas a percentageof

the agriculturallaborforce.8 The distributionof this differencein growthrates for 148

14
countriesis plotted in Figure 13,which showsthat such migrationtook place in

practicallyall countries. The median averageannualmigrationrate for the period 1950

to 1990 is 1.97percent. That is, at the medianthe agriculturallabor forcedeclinedat an

annualaveragerate of about2 percent.

In interpretingthis result,it shouldbe notedthat off-farmmigrationis

determinednot only by changesin agriculturalprofitabilitybut also by changesin labor

demandin nonagricultureas well as sectoraldifferencesin the level of standard-of-

living. A detailed studyof off-farmmigrationbased on the cross-countrydata for the

period 1950-1990appearsin Larson and Mundlak(1995). Naturally,when labor income

in agricultureis low,peoplewill migrateto othersectors. Thus, off-farmmigrationis

probablythe most importantsingleprocessthat contributesto the alleviationof rural

poverty. This resultshouldbe kept in mind in judging not only economicprocessesbut

also institutionaland legal reforms. There is nothingmoreeffectivein improving

people'swelfarethan allowingthem to choose freelybetweenopportunitiesand thus

escape areasof hardship. Thetraditionalexploitationof agriculturallabormay have

taken place becausemigrationopportunitiesfor such labor were restrictedby institutional

arrangement,by lack of infrastructure,or by lack of opportunitiesoutsideof agriculture.

For migrationto take place, opportunitiesin nonagricultureare required. This is where

economicgrowthcomesin as a factor in the alleviationof agricultural-basedpoverty.

15
4. AGRICULTURE AND DEVELOPMENT

The technological change in agriculture had two important effects on the

economy as a whole. First, it improved the overall food supply while prices declined and

thus improved consumer welfare. Second, it made it possible to produce more output

with relatively less labor and thus facilitated the development of nonagriculture. As a

result of the migration, the share of agriculture in the total labor force has declined. This

is seen in Figure 14 in which the share is plotted for 148 countries for the years 1950 and

1990. The share varies considerably among countries. It is high in less-developed

economies and declines with the level of development. Even so, it has declined in all

countries. The median share was 70 percent in 1950, but only 33 percent in 1990. The

proportion of countries with a share below 10 percent increased from 1.3 percent in 1950

to 15 percent in 1990. Few indicators are better able to tell the story of the role of

agriculture in economic development. The freeing of labor that is a necessary condition

for development. Therefore, understanding of the process of off-farm migration, and the

determinants of its rate, contributes to our understanding of the process of development.

The net outcome of the processes reviewed above is a decline in the relative

importance of agriculture in total output. The change in the distribution of this share in

the 40 year period between 1950 and 1990, as seen from Figure 15, is indeed noteworthy;

the median for this group of 67 countries declined from 30 to 8 percent over this period.9

In summary, technological change in agriculture serves the rest of the economy

by increasing supplies at lower prices and contributing labor to the development of

16
nonagriculture.At the sametime, in the longerrun, none of the benefitsare capturedin

agriculture. Who wouldstand to benefitfrom furtherincreasesin agricultural

productivityin the future? There is no reason to think that the futurewill be any different

from the past.

This welfareconclusionhas an importantpolicy implicationfor the financeof

agriculturalresearch. Historically,muchagriculturalresearchwas publiclyfinanced

becausethe fruits of the researchcan not be internalizedby the researcher. The

foregoingdiscussionindicatesthat it is indeedin the interestof the publicat large to

promoteresearchthat will producetechnologicalchangein agriculture. Therefore,the

burden of this researchshouldbe carriedby the public at large and not by agriculture,

which does not seemto captureand maintainthe benefit from it. Furthermore,because

the fruits of the researchare eventuallyspreadto all countries, a strong case can be made

for financingresearchinternationally.But countriescan take lead in advancing

technologyand benefit from it until the new technologyspreadsto other countries. Also,

for countriesto take advantageof global developments,they must have an ongoingactive

researchprogram. Thus, there is a strongcase for countryfinanceof its own researchas

well.

5. RESEARCHIMPLICATlONS

An importantconclusionto be drawnfrom the foregoingreviewis that the

underlyingprocessof agriculturalgrowth is universalbut its pace varies over time and

17
acrosscountries. The basic questionis how to account for this variability. There is no

simpleanswerto this questionbecausethere is no singlefactorthat can providethe

answer. This is also true when it comesto the evaluationof the consequencesof policies.

Much researchon agriculturalpolicy concentrateson the implicationsof sector-specific

policies,which are pertinentbut by themselvesinsufficientto unfoldthe whole story.

The process of developmentis a processof continuousadjustmentsmade by

householdsand firms. These adjustmentsand their consequencesare spreadover time

and as such are dynamic in their nature. They are carriedout in responseto changesin

the economicenvironmentsubjectto the prevailingconstraints. Thusthe key to

understandingof the economicperformanceof agriculture,as well as othersectors, is in

the studyof the impactof the economicenvironmenton such decisions. The task is to

formulatea frameworkthat will channelthe net effect of the variousvariablesforming

the economicenvironmentinto measuresconduciveto empiricalanalysis.

The approachis based on the simplepremisethat an increasein output is

obtainedby a change in resources,in the efficiencyof their use or in the implemented

technology. For most countries,the limitingfactorto growthis not the available

technologyas determinedby the frontierof knowledge,becausethey are far from the

frontier. The pressingresearch issueis to identifythe constraintsthat preventthe country

from taking full advantageof the availabletechnology,and the determinantsthat control

the pace of convergenceto the frontier. As these constraintsare affected by the

economicenvironment,a successfulmappingof the economicenvironmentto variables

18
considered by producers in making their production decisions can form the framework

for the analysis.

There are several important blocks in this formulation. First is the

mapping of the economic environment to sectoral incentives. Second is the

impact that sectoral incentives have on factor demand and factor supply.

Third is the dependence of the implemented technology on incentives and on

resource and other constraints. In this, an explicit account should be

taken of the fact that the decisions have consequences in more than one

period. The dependence of factor demand on incentives is obvious. A good

illustration of the dependence of factor supply on incentives is that of

labor where the off-farm migration is affected by the intersectoral income

differentials. A good illustration of the dependence on constraints is that

of the choice of techniques. Different techniques have different resource

requirements, where on the whole, the more advanced techniques are more

capital, human and physical, intensive. Hence, capital scarcity determines

the pace of implementation of new techniques. As decisions on the

implemented technology are made jointly with the decisions on input demand,

that demand also depends on the same constraints that affect the implemented

technology. The research and policy implications of this point can not be

overemphasized.

Underlying the process of implementation of new techniques is the flow of new

19
techniques, or changes in the available technology. Such a flow is correlated with

investment in research and development but, more profoundly, it depends on the

advancement in basic knowledge which, by its very nature, follows a random walk. The

voluminous literature that aims to explain this flow, for which agriculture is a good

example, provides plenty of anecdotal evidence but no structural model, and therefore

little predictive power.

The empirical knowledge of the various blocks makes it possible to simulate the

economy and to evaluate the consequences of various policies. This is the approach

taken in Mundlak, Cavallo and Domenech for Argentina and Coeymans and Mundlak for

Chile. These studies construct simulation models along the lines indicated above which

are fitted to the historical data. Thereafter, the model is simulated to evaluate the

consequences of alternative policies of interest within a historical context in order to

understand the dynamic, quantitative effect of various policies and to infer from them

desirable courses of action for future development.

6. THE PRODUCTION STRUCTURE OF AGRICULTURE

Technology and competitive conditions plays a role in the determination of

supply. Almost any of the foregoing discussion involve attributes of technology.

Technical change is the key factor for growth and affects income distribution. Factor

shares determine the relationships between factor prices and product prices and the

competitive position of a sector. Returns-to-scale affect the size distribution of finns.

20
Elasticities of substitution affects the functional distribution of income due to capital

accumulation. Not independent of all of the above is the question of supply response

and factor demand. Hence, the empirical analysis of technology, and its changes, is of

cardinal importance.

The literature on production functions and related topics deals with various

aspects of the subject but comes short of providing an integrated framework which is

readily applicable in the evaluation of the dynamics of agriculture. To provide an insight

into this literature we summarize some key findings on the central issues pertinent to the

construction of such a framework based on a recent literature review (Mundlak 1997).

There are two fairly distinct periods in the study of agricultural production functions,

before and after duality. The changing of the guards was in the early 1970s. Duality is a

powerful theory to analyze supply issues using primal or dual functions. However, its

empirical use is more limited. As a result, the appearance of duality changed not only the

method of estimation but also the questions asked to the extent that there is little

continuity in the subjects of interest. This is illustrated below.

The empirical work on agricultural production functions originated in a

methodological paper by Tintner (1944) and an application by Tintner and Brownlee

(1944). This work was influenced by the work of Paul Douglas and it thus took fifteen

years to adopt the work of Douglas in agricultural economics application. The primal

estimation of the Cobb-Douglas function was the centerpiece in the pre-duality work.

The approach does not impose competitive conditions on the estimation but instead

21
submitsthem to empiricaltesting. Such testingoften showsa differencebetween

observedfactorshares and the estimatedproductionelasticities. This result is not an

absoluterejectionof the prevalenceof the competitiveconditionsbut rather a conditional

result,based on the modeland the statisticalprocedureused. Still, it is indicativeof the

existenceof deviationsfrom such conditions.

Empiricalresults maintainthat the elasticityof laborneverexceeds0.5, and in

most cases it varies in the rangeof 0.25to 0.45. This value is less than the elasticity of

labor in nonagriculture.If all non-laborincomeis consideredto be capitalincome,then

the result supportsthe positionthat agricultureis cost-capital-intensiveand therefore less

susceptibleto increasesin the wage rate than nonagriculture.Also, the factor share of

labor in agriculturehas declinedwith time, indicatingthat technical changewas labor-

saving.

The elasticityof landvaries betweenzero, in somecases, and aboutone third.

We interpretthis elasticityto be a measureof the competitivepositionof agriculture.

From the perspectiveof farm incomeit is meaningfulto look at the sum of laborand land

elasticities,and this sum oftenfluctuatesaround0.5.

One motivationfor estimatingproductionfunctionsis to provideweights for the

computationof technicalchange. However,this approachhas not providedany

substantiveadvantagecomparedto the use of factor shares,even thoughthey may not be

the same as the productionelasticities. The reason is specificationerrors and

interpretationof statisticalstudies are often greaterthan the discrepanciesbetweenthe

22
factor shares and the true production elasticities. An example is the error involved in

finding increasing returns to scale and its subsequent incorporation in the computation of

total factor productivity. This leads to the elimination of residuals in comparisons of

growth over time or productivity differences across countries. Such a procedure was

motivated by the belief that all growth can be accounted for and therefore there should be

no residual (Griliches, 1963). However, this belief has no foundation and it is

inconsistent with theory.

An important feature common to many studies is a lack of robustness in the

estimates and their dependence on the selected variables and sample coverage. This

result is consistent with the interpretation that the implemented technology is sensitive to

variations in the economic environment.

What distinguishes the dual approach from the primal is the appearance of prices

in the empirical equation. Hence, in evaluating the performance of this approach we

address the following questions:

* What has been the contribution of prices to the empirical equation?

* What additional information is obtained from the dual equations and how can

they be interpreted?

* Are the underlying assumptions of duality met?

* What are the statistical benefits of this approach?

The dual estimates are obtained by regressing factor shares on prices, a time

trend and sometimes output. When the change in use of inputs is decomposed to price,

23
technology(morecorrectly,time) and outputeffects,trend and outputaccountfor most

of the changeswhereasthe role of prices is the least important. Thus the apparentdirect

contributionof pricesto the explanationof inputsor outputvariationsis rather limited.

There is an indirectcontributionthat worksthrough the impacton the quasi-fixedinputs,

not capturedin the usual dual estimates,but this is not discussedhere.

On the whole, the own-priceelasticitiesreportedin the literatureare less than

one. There is no uniformityin the signsof the cross-elasticitiesbut in general,most

inputs appearto be substitutes. The price elasticitiesof factor demandand product

supplyare usuallyobtainedunder the assumptionthat factor supplysare perfectlyelastic.

In practice,this is not the case. Consequently,the magnitudeof the estimatedelasticities

are affectedand the resultsneed not representdemand-drivensubstitution. This is

particularlyrelevantto elasticitiesrelatedto labor,land and capital.

Interestingly,on the whole,the studiesbased on duality do not show increasing

returnsto scale. Moreover,technicalchange,obtainedby includinga time trend in

regressionsof factor shares,is largely labor-saving,capitalusing,and fertilizerusing.

The results on land are somewhatambiguous.However,this is reflectiveof the data.

Therefore, the effect of includingpriceswas not sufficientto changeconclusionsthat

could be drawnfrom the raw data. This does not give a strongmarkto the analysisin

that the results are obviouswithoutit.

Dualitybetweentechnologyand prices holdsunder well definedconditionsthat

can be tested empirically. In most studies,these underlyingconditionsare not fully met,

24
particularlyconcavityof the cost functionand convexityof the profit function.

Therefore,the estimatedtechnologyis inconsistentwith the basic premisesof the model.

In a way, this is the most disappointingresultbecause duality is a very powerfultheory.

An importantquestionthen is whythe theoreticalresults are inconsistentwith the

empiricalanalysis. There is morethan one reason, but probablythe most importantone

is the impropertreatment of technology.

One of the expectedvirtuesof duality has been its apparent solutionof the

simultaneousequationbias realized in some primal estimators. However,a closer

examinationof this issue indicatesthat in general dual estimatorsare inferiorto primal

estimators(Mundlak,1996).

In summary,a close examinationof the literaturesuggeststhat variousanomalies

and apparentinconsistenciesin empiricalresults can be interpretedwithinthe framework

of the choice of techniqueswhich viewsthe implementedtechnologyto be sensitiveto

the economicenvironment. This providesan appropriateframeworkfor empirical

analysisand providesthe neededinsightfor the evaluationof changesin policies.

25
REFERENCES

Binswanger,H. P, Y. Mundlak,C. Young and A. Bowers(1987) 'On the determinants

of cross countryaggregateagriculturalsupply', Journal of Econometrics,vol. 36,
pp. 111-131.
Coeymans,J. E. and Y. Mundlak(1993) 'Sectoral Growthin Chile: 1962-82',
InternationalFood PolicyResearchInstitute,ResearchReport95.
Griliches,Z. (1963) 'Estimatesof the AggregateAgriculturalProductionFunctionfrom
Cross-Sectional Data', Journal of Farm Economics, vol. 45, pp. 419-28.
Hayami,Y. and V. Ruttan(1985)AgriculturalDevelopment- an International
Perspective(Washington,DC): JohnHopkinsUniversityPress.
Larson,D. and Y. Mundlak(1995) 'On the IntersectoralMigrationof Agricultural
Labor', WorkingPaper 1425(The WorldBank, InternationalEconomics
Department,Washington,DC).
Larson,D. and Y. Mundlak(1997) 'On the IntersectoralMigrationof Agricultural
EconomicDevelopmentand CulturalChange,vol. 45, number2, pp. 295-319.
Mundlak,Y., D. Cavalloand Domenech,R. (1989) 'Agricultureand Economic
Growth,Argentina1913- 1984, ResearchReport76 (InternationalFood Policy
ResearchInstitute, Washington,DC).
Mundlak, Y. (1997) 'AgriculturalProductionFunctions- a CriticalReview', Working
Paper 9702 (TheCenter for AgriculturalEconomicResearch,Rehovot).
Mundlak, Y. (Januaryor March 1996) 'ProductionFunctionEstimation:a Revival of
the Primal', Econometrica,vol. 64, no. 2, pp. 431-438.
Mundlak, Y. (Forthcoming)Agricultureand EconomicGrowth(HarvardUniversity
Press).
Tintner,G. (1944) 'A note on the derivationof productionfunctionsfrom farm
records', Econometrica,vol. 12, pp. 26-34.
Tintner,G. and O.H. Brownlee(1944) 'ProductionFunctionsDerivedfrom Farm
Records', Journal of Farm Economics(August, 1944),vol. 26, pp. 566-571(a
correctionin JFE Feb. 1953,p.l23).

26
FIGURES

Figure 1 Total Agricultural Production

1967-1992 130 countries

0. - ------ ----------
--- - - -- - - - I - - --- - -r----- --------

0.7
9 -- -
0.6 -- - - - - - - - -
l 0.5
* 6r
0.4
_ > Prod'nShares

E 0.2- _Eqt !WeiWght -

0 .1 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

0
-2.6 -1.3 0 1.3 2.6 3.9 5.2 6.5
Annual Growth Rate (percent)

Figure 2 Agricultural Prices

112 Countries

, 0.8t- Prod'n Shar

. 1 J Equi it Weights

E 0.2; ; '; ;

-10 -5 0 5 10
Annual Growth Rate (percent)

27
 Figure 3 Per Capita Agricultural Production
1967-1992 130Countries

07
3 0.5
0.6 ..... ...
,,,,,,;............................
0.1
028
Annual
Growth
Eqnual Rt(Pedent)
Weigh
Rat hre
d' Shamsn

S 03 .......... - .... 2..

 6Z

I S @ 1 B I}MO19 JO 0fl3B
X~~~~~63

9L tl ZWaL 1 80 90 t0 Z0
C)

3
.... ... ... ... ... ... ................................ .................. . / -9

...................................
......................
...... ..................
..........I.......... - 8-

s.1'*unoO16 Z66 096

jlol ol o}anXlnoijB
41MOJ9
01181S da9 vl]dLo Jadt asnBI
 Figure 5 Harvested Land
1967- 1992 131Countries

a
0.8

-15 -10 -s o 6 10

, ..
S

0. Figure 6 Land Productivity

i - 199287 Countries
~~~~~1960

-15 -10-5 0 5 10
Annual Growth Rate(Pecn)

O~~~~~~~~
~ ~~~~~- - -
Ij 0.2 -I -, - - - - - - - - - - -

-6---- -- -4-- -- --- -- -- -024

E Growt Ra
eq~~~~~nual
tPtcn
0~~~~~~3
 (UlJgd) *;el %m0oja lenuuV
tZ O Z- ~ ~ ~~~~~~tI
9-

.,, , ...
I , prT~<.4 ZOo~
S14OIOmibt 4seJu,powd
|80

l~ ~~ ____ ___I___ 1

se!SJunooL8 z66I - 0961

ol;elE puel o; °oqe,l 8 eJnB_

(4uooJed) OPH q;mojg IDlnuuY

Z O Z- t-09-

...........
.........................
...X...I.................
............ .............
...............
..............................
.......
Z._90

..........................
.. ................ ................
...,............. ................................
.......
'-

seujunoo 8Vo O66M-OM

.ioqel ejvn;jn*pBV L ein6B:

 Figure 9 Average Labor Productivity
1960 - 1992 87 Countries

1-II

o 02 08 2--.;

-4 -2 0 2 4 6 8
Growth Rat* (percent)

Figure 10 Labor and Land Productivity

1960 - 1992 87 Countries

0.8 ------------ --- ---- ----------

Is 0.6 -------- - ----------l ... .. .. ... .. .. ..

0.4 - ------- -------------- - - --- - --------. ..-.. .. -- -- -- -- -- -------

I-
s ~~~~~~~~~Land
f Labor

,O 0.2 - - - -- - - -. -- -------------------

-6 4 -2 0 2 4 6 8 10
Annual(r Growth Rate

32
 Figure 11 Differential Growth of Labor Productivity
Non-Ag Minus Ag 1960 -1992 88 Countries

0.8+ ....... .... .. .

C + equal weights prod'n shares
0 .6 - --..---. .- .-- .....
- ....
.......
--- --. ......
j
-----
.
U._
l _ _ _ _ _ _ _ _ _ _ _

> 0.2- ---- - - ....

E l
0.2- ...... -- ;... ........ ..........................-----

0 t l i i I

-15 -10 -5 0 5 10 15 20
Annual Growth Rat (V)

33
 ................... ............................ ,e .......

ea^4 wnoS ~~~~~......................... .............. I.......... . . . ...... ... ... EX.

w-e; P.flUf \lL

*wqmv q no ...........
. .:. . .. . . . ... .. .. ..

epeuoO .on
............ 0 *a
'..

*o1#1f
4in°S .... .... .. ..... .. a e

se 0*1J d isJLun:> uo o, eAlawleM

.... .... ....... - ...... ............. ....

-a.J.s Pealun

N*W1V Wn0lS

P'.

SOOIJd tjvlnZlniJSV o:j OAli8iO

u! puu-1JOOO!Jd
OJflhflO!JOV O41

ZlonBi:
 Figure 13 Off-Farm Migration Rates
1950-1990 148 Countries
I

p06

4- 8 62
0

Annual Migration Rates (Percent)

Figure 14 Share of Agriculture in Labor Force

1950,1990 148 Countries

351990 1950
E
IL 0.8Lf/
* 0.4 .....

6 0.2-

0 0.2 0.4 0.6 0.8

Labor Ratio (Agriculture to Total)

35
 Figure 16 Share of Agriculture in GDP
1950, 1990 67 Countries

O.8{ ~1990
U ~ ~~~ ~ ~ ~~~j ~~~~
1950
0.6 ,

IL 0.4-.

E 022 |

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Output ratio (Agriculture to total)

36
 Notes

1.This researchhas been conductedat the WorldBank. The datawere assembledfrom the
known internationalsourcessuch as FAO,ILO, the WorldBank, IMF as well as from
countrypublications.The results are presentedin terms of empiricaldistributions. The
numberof countriesand the periodcoveredin the figuresvary accordingto data availability.

2. Unless indicatedotherwise,all growthrates are computedfrom a semi-logarithmic

regressionof the variablein questionon time.

3. The numberof countriesvaries across figuresbecauseof incompletedata sets.

4. In making sucha statement,we note that the verticalaxis in the two figuresis the same
and that the units of the horizontalaxis are also the same
5. Based on nationalaccountsdata, convertedto dollars and deflatedby the US GDP deflator.
6. The varianceof the averageland productivityis lower than that of the averagelabor
productivity. This is due to biggervariancein the rates of changeof the labor forcethan of
land. The pattern of changesin land and labor productivityis discussedby Hayamiand
Ruttan (1985).
7. Part of the rent (residualincome)may also be sharedby tenants,dependingon the nature
of the contract.

8. The assumptionof equal fertilityrates is made here for purposeof illustration. The
migrationrates couldbe estimatedunder the assumptionof differentfertilityrates.
9. The medianin 1990for a larger set of 110 countriesis 9 percent. Figure 15 containsfewer
countrieson whichwe have data also for 1950.

37
I
 Policy Research Working Paper Series

Contact
Title Author Date for paper

WPS1787 Trading Arrangements and Diego Puga June 1997 J. Ngaine

Industrial Development Anthony J. Venables 37947

WPS1788 An Economic Analysis of Woodfuel Kenneth M. Chomite June 1997 A. Maranon

Management in the Sahel: The Case Charles Griffiths 39074
of Chad

WPS1789 Competition Law in Bulgaria After Bemard Hoekman June 1997 J. Ngaine
Central Planning Simeon Djankov 37947

WPS1790 Interpreting the Coefficient of Barry R. Chiswick June 1997 P. Singh

Schoolingin the HumanCapital 85631
Earnings Function

WPS1791 Toward Better Regulation of Private Hemant Shah June 1997 N. Johl
Pension Funds 38613

WPS1792 Tradeoffs from Hedging: Oil Price Sudhakar Satyanarayan June 1997 E. Somensatto
Risk in Ecuador 30128

WPS1793 Wage and Pension Pressure on the Alain de Crombrugghe June 1997 M. Jandu
Polish Budget 33103

WPS1794 Ownership Structure, Corporate Xiaonian Xu June 1997 J. Chinsen

Governance, and Corporate Yan Wang 34022
Performance: The Case of Chinese
Stock Companies

WPS1795 What Educational Production Lant Pritchett July 1997 S. Fallon

Functions Really Show: A Positive Deon Filmer 38009
Theory of Education Spending

WPS1796 Cents and Sociability: Household Deepa Narayan July 1997 S. Fallon
Income and Social Capital in Rural Lant Pritchett 38009
Tanzania

WPS1797 Formal and Informal Regulation Sheoli Pargal July 1997 E. de Castro
of Industrial Pollution: Hemamala Hettige 89121
Comparative Evidence from Manjula Singh
Indonesia and the United States David Wheeler

WPS1798 Poor Areas, Or Only Poor People? Martin Ravallion July 1997 P. Sader
Quentin Wodon 33902

WPS1 799 More for the Poor Is Less for the Jonath B. Gelbach July 1997 S. Fallon
Poor: The Politics of Targeting Lant H. Pritchett 38009

WPS1800 Single-Equation Estimation of the John Baffes August 1997 P. Kokila

Equilibrium Real Exchange Rate Ibrahim A. Elbadawi 33716
Stephen A. O'Connell
 Policy ResearchWorking Paper Series

Contact
Title Author Date for paper

WPS1801 RegionalIntegrationas Diplomacy MauriceSchiff August 1997 J. Ngaine

L. Alan Winters 37947

WPS1802Are There SynergiesBetween Harry Huizinga August1997 P. Sintim-Aboagye

World Bank Partial Credit 38526
Guaranteesand PrivateLending?

WPS1803 Fiscal Adjustmentsin Transition BarbaraFakin August 1997 M. Jandu

Economies:Social Transfersand the Alain de Crombrugghe 33103
Efficiencyof Public Spending:A
Comparisonwith OECDCountries

WPS1804 FinancialSectorAdjustmentLending: RobertJ. Cull August 1997 P. Sintim-Aboagye

A Mid-CourseAnalysis 37644

WPS1805 RegionalEconomicIntegrationand Junichi Goto August 1997 G. llogon

AgriculturalTrade 33732

WPS1806 An IntemationalStatisticalSurvey SalvatoreSchiavo-Campo August 1997 M. Gueverra

of GovernmentEmploymentand Giulio de Tommaso 32959
Wages AmitabahMukherjee

WPS1807 The Ghosts of FinancingGap: William Easterly August 1997 K. Labrie

Howthe Harrod-DomarGrowth 31001
ModelStill Haunts Development
Economics

WPS1808 EconomicTransitionand the FranciscoH. G. Ferreira August 1997 M. Geller

Distributionsof Incomeand Wealth 31393

WPS1809 Institutionsin Transition:Reliability Aymo Brunetti August 1997 M. Geller

of Rules and EconomicPerformance GregoryKisunko 31393
in FormerSocialistCountries BeatriceWeder

WPS1810 Inspectionsand Emissionsin India: SheoliPargal August 1997 E. Krapf

PuzzlingSurveyEvidenceabout MuthukumaraMani 80513
MainulHuq

View publication stats