the annals of the american academy of political and social science-2001-rao-85-104.pdf

8/14/2019 The ANNALS of the American Academy of Political and Social Science-2001-Rao-85-104.pdf

1/21

http://ann.sagepub.com/of Political and Social Science

The ANNALS of the American Academy

http://ann.sagepub.com/content/573/1/85Theonline version of this article can be foundat:

DOI: 10.1177/000271620157300105

2001 573: 85The ANNALS of the American Academy of Political and Social ScienceVijayendra Rao

Poverty and Public Celebrations in Rural India

Published by:

http://www.sagepublications.com

On behalf of:

American Academy of Political and Social Science

can be found at:ScienceThe ANNALS of the American Academy of Political and SocialAdditional services and information for

http://ann.sagepub.com/cgi/alertsEmail Alerts:

http://ann.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://ann.sagepub.com/content/573/1/85.refs.htmlCitations:

What is This?

- Jan 1, 2001Version of Record>>
http://ann.sagepub.com/http://ann.sagepub.com/http://ann.sagepub.com/http://ann.sagepub.com/content/573/1/85http://ann.sagepub.com/content/573/1/85http://www.sagepublications.com/http://www.aapss.org/http://ann.sagepub.com/cgi/alertshttp://ann.sagepub.com/cgi/alertshttp://ann.sagepub.com/subscriptionshttp://ann.sagepub.com/subscriptionshttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsReprints.navhttp://www.sagepub.com/journalsPermissions.navhttp://ann.sagepub.com/content/573/1/85.refs.htmlhttp://ann.sagepub.com/content/573/1/85.refs.htmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://ann.sagepub.com/content/573/1/85.full.pdfhttp://online.sagepub.com/site/sphelp/vorhelp.xhtmlhttp://ann.sagepub.com/content/573/1/85.full.pdfhttp://ann.sagepub.com/content/573/1/85.refs.htmlhttp://www.sagepub.com/journalsPermissions.navhttp://www.sagepub.com/journalsReprints.navhttp://ann.sagepub.com/subscriptionshttp://ann.sagepub.com/cgi/alertshttp://www.aapss.org/http://www.sagepublications.com/http://ann.sagepub.com/content/573/1/85http://ann.sagepub.com/


2/21

85

Poverty and PublicCelebrations in Rural India

By VIJAYENDRA RAO

Vijayendra Rao is an economist with an interest in anthropology. He received hisPh.D. in economics at the University of Pennsyluania, heldpostdoctoralpositions at theUniversities of Chicago and Michigan, and taught at Williams College before moving tothe Development Research Group of the World Bank. He mixes the field-oriented collec-tion of qualitative and quantitative data with economic methods to gain a morecontextualized and grounded understanding of household behavior inpoorcommunities.

ABSTRACT: This article examines the

paradoxthat very poor house-

holds spend large sums of money on celebrations. Using qualitativeand quantitative data from South India, it demonstrates that expen-ditures on weddings and festivals can be explained by integrating anunderstanding ofhow identity is shaped in the Indian context withan economic analysis of decision making under conditions ofextremepoverty and risk. It argues that publicly observable celebrations havetwo functions: they provide a space for maintaining social reputa-tions and webs of obligation, and they serve as arenas for status-

enhancing competitions. The first role is central to maintaining thenetworks essential for social relationships and coping with poverty,while the second is a correlate of mobility that may become more

prevalent as incomes rise. Development policies that privilege indi-vidual over collective action reduce the incentives for the former

while increasing them for the latter, thus reducing social cohesionwhile increasing conspicuous consumption.

NOTE: I am indebted to Sita Reddy for invaluable discussions and to Francis Bloch andSonalde Desai for ourjoint work that informs part of the analysis in this article. I am grateful toRama Ranee,A. C. Komala, and the National Council forApplied Economic Research for helpwith conducting the fieldwork and to Christopher Clague, Karla Hoff, Michael Woolcock, and ananonymous referee for constructive comments. The findings, interpretations, and conclusion ofthis article are those of the author and should not be attributed to the World Bank.


3/21

86

P ARADOXICALLY,families who

-C earn barely enough to survive

spend vast amounts of moneyon

cele-brations. In the rural South Indian

data I analyze in this article, a typi-cal household spends approximatelyseven times its annual income on a

daughters marriage-on celebratingthe wedding and on dowry transfersto the grooms family. Moreover, 15

percent of its expenditures, on aver-

age,are

spenton

celebrating villagefestivals. Poor households tend tofind such large sums ofmoney by go-ing severely into debt at interestrates that range from 100 to 300 per-cent per year, which leads to chronic

indebtedness and, sometimes,bonded labor. Despite the importanceof these expenditures, the economics

literature has largely ignored them,perhaps because they fall into therealm of nonrational behavior. How-

ever, given their pervasiveness, andthe dominant role accorded to eco-

nomics in the formulation of policy,this omission can have serious conse-

quences for the design of policy inter-ventions. Money spent on celebra-

tions is, after all, moneynot

spenton

food, education, health, and other

productive inputs.What explains this paradox?Are

celebrations wasteful? In this article,I will examine the factors underlyingexpenditures on festivals and wed-dings with qualitative and quantita-tive data from South India. In this

introductory section, I sketch a gen-eral conceptual model of publiclyobservable events and the role theyplay in Indian village life. I will thenillustrate this conceptual frameworkby examining aspects of wedding-and festival-related behaviors. While

the article relies on fieldwork from

the South Indian state of Karnataka,

I believe that the analysis is applica-ble to other parts of SouthAsia. With

changes in structure and context, thebasic points are perhaps also rele-vant to human behavior in other non-

Western societies. Working within acontext with which I am relativelyfamiliar, however, allows me to pro-vide a more nuanced perspective onan

important aspectof

the culturaleconomics of behavior in poor house-holds.2

2

In rural Indian life, celebrationstend to mark important transitions.

A wedding marks the transition of awoman from daughter to wife and isthe central event in her life cycle(Fruzzetti 1990). Festivals mark

importantevents in the life

cycleof a

village-a religious event or a har-vest (Fuller 1992). They are rituals,marking events that shape identityand meaning. They are also, neces-sarily, public. If they were private,the change in status would not be rec-ognized, because others would notobserve the ritual. The public natureof celebrations thus

providesan

opportunity for individuals to dem-

onstrate, to signal, to the world thatan important time has arrived, animportant transition completed.Aspublic events, they also provide anopportunity for actions to be scruti-nized and commented upon and for

judgments to be made. Therefore,celebrations can be

thoughtof as

mechanisms where social and eco-

nomic status is demonstrated, rei-fied, and often enhanced.How do public rituals shape iden-

tity and affect preferences? To under-stand this, it is necessary to under-

at Faculty of Social Sciences on November 20, 2012ann.sagepub.comDownloaded from
http://ann.sagepub.com/http://ann.sagepub.com/http://ann.sagepub.com/http://ann.sagepub.com/


4/21

87

stand anthropological conceptions ofhow Indians define their identities

and shape their perceptions of them-selves. In Homo Hierarchicus, a sem-inal work in Indian anthropology,Louis Dumont (1970) argued thatHindu Indians are fundamentallydifferentfrom Western individuals in

that they see themselves within thecontext of the caste system as part ofa strict ritual hierarchy. Their sense

of individuality is subservient totheir identification with their family,caste, and village. Marriot and Inden

(1977) modified this to say that anIndian is not an individual as much

as a &dquo;dividual.&dquo;A dividuals identityis made up of different transferable

&dquo;substances&dquo; that they give andreceive in their interactions with oth-

ers. These substances then cometogether in a dynamic, negotiated,interactive manner to achieve a

sense of self. Thus personhood isdefined entirely in terms of ones

relationships to others.In more recent work, Mines (1994)

takes issue with this to argue that

Indians have a well-defined sense of

individuality but that this individu-ality is quite different from the West-ern prototype. Indian individuality,according to Mines, is more exterioror civic than personal and is cruciallyconditioned by how others estimate

and evaluate the person. What kind

ofperson is he? How influential is he?

Furthermore, this &dquo;civic individual-

ity&dquo;is determined within

thecontext

to which the individual belongs: his

caste, religion, and community. It isalso spatially defined by the &dquo;size and

locality of the constituencies thatform a persons social contexts&dquo; (22).

Thus ones sense of self is fundamen-

tally influenced by the people whom

one knows and interacts with. Minesfurther argues that it is an individu-

ality of inequality; that is, people of

higher status and rank have individ-ualities that are more stronglyemphasized-individuality is gradedand ranked. Taking all of this intoconsideration, we see that Indiansare strongly driven by status and

rank, and the status and rank are notsimply a matter of individual or rela-tive wealth but are derived from the

size and influence of their familial

and social networks and from the

public demonstration of access tothese networks.4 The size and qualityof ones networks thus have an inher-

ent value, whatAppadurai (1990)

defines as &dquo;the prime value ofsociality&dquo; (188), and are not only ameans to an end.

An Indian, therefore, while not

entirely Homo hierarchicus, is also

quite different from Homoeconomicus. What differentiates

Indian individuality from the West-

ern, more than anything else, is Indi-

ans deep anxiety about what otherswill think of them.And what othersthink of them is fundamentallyshaped by their civic identity: howmany important people they know;how closely they are adhering to thebehaviors expected from someone intheir sociocultural category-that is,are they being good fathers, mothers,

sons, headmen,or

priests? Thus,unlike the caricature of a rationalindividual embodied in most eco-

nomic models, Indians do not live byconsumption alone but fundamen-tally by the perceptions of others.



5/21

88

This perception shapes their iden-

tity, their sense of themselves, andtheir sense of others sense of them.

Their status and role, within the con-

text of their caste and group, there-

fore drives their behavior. If they donot satisfy the obligations assignedto them by their place within society,they will be penalized with a loss ofstatus or rank and looked upon with

less respect, perhaps even withshame. This does not mean that Indi-

ans are not rational but merely thattheir rationality is conditioned bystructures and practices derivedfrom their social and cultural con-

text, just as much as it is subject toeconomic motives and constraints.5

5

Status and rank require constantmaintenance (Goffman 1959). Inorder to participate in social relation-

ships, people have to know if othersare good individuals who act in accor-dance with the their socially delim-ited roles. Thus the maintenance of

rank and stature requires publicdemonstrations of actions fitting the

expected behaviors of ones rankwithin ones social group. Note that

the effort spent in maintaining rank

does not preclude individuals frommoving up. Wealth, a prestigious job,or the acquisition of a rich husbandor son-in-law is desired because theyare preconditions to the acquisitionof greater public regard and respect.But in order to acquire the highersocial rank usually associated with

greater assets, these attributes of

mobility must not just be acquiredbut demonstrated by behavior thataccords with the class of people who

possess such assets. Otherwise,while one may achieve economic

mobility, it will not translate into a

higher social rank. Non-Indians mayalso value rank and social status, but

Indians face powerful incentives to

regularly demonstrate that status,because their behavior is under con-

stant, intimate, and structured

scrutiny.Networks have an inherent value,

but in the context of the extremelyrisky environment of Indian villagelife, the web ofobligations that definethe extent of ones networks are also

central elements of coping strategiesagainst risk and poverty. The recent

development economics literaturehas focused on empirical investiga-tions of informal insurance, measur-

ing the extent to which families areable to smooth consumption by rely-ing on village networks (for example,Townsend 1994; Udry 1994). It is

clear that, in the absence of net-

works, life would be unimaginablyhard. When crops fail or jobs are lostor the main wage-earner dies (all

quite routine events in the harsh

reality of rural life), one relies onones friends and relatives to cope.The greater the level of respect and

regard one receives from friends and

relatives, the easier it is to ask fortheir help. Networks, however, areuseful not only for unanticipatedshocks but also for more everydayforms of serious but predictable prob-lems. Households withbetter connec-

tions may be able to get access to

better jobs (Collier and Garg 1999) orfind ways of getting lower prices for

food, or they may be able to get betterinformation on potential marriagepartners for their children. There-

fore, respect and regard do notmerely have direct utility; they arealso central elements in poverty



6/21

89

alleviation strategies. What others

think of a person is not importantjust for its own sake-it canmean the

difference between life and death.

Given this, celebrations, which are

a time of intense public scrutiny,become arenas where reputationsare managed and enhanced. Life-

cycle events become theaters where

public reputations are maintained,and stadiums where people competein games ofstatus competition, going

beyond their role as milestones. Therules of the game are determined bykinship systems, ritual calendars,the distribution of wealth and caste,and other structural endowments

that are slow to change and thereforecan be considered as exogenous con-

straints in the short term. Because

these structures provide rules for

what is considered appropriatebehavior, they determine the criteria

by which people are judged. They are

part of the capability set (Sen 1988)available for individuals and families

to maximize their preferences. Thus,culturally determined preferencesand constraints interact with eco-

nomic motives to create powerful

incentives for expenditures on publiccelebrations, particularly for individ-uals seeking social mobility.The format of the rest of this arti-

cle is as follows: I present an investi-

gation of wedding celebrations in thenext section, followed by an analysisof festivals in section 3. Section 4 con-

cludes the article and presents some

policy implications.

ONWEDDING CELEBRATIONS

Wedding celebrations in ruralIndia can seem extremely lavish to

an outsider, especially in contrast to

the extreme poverty of rural Indian

life, with large numbers of peopleinvited for feasts and ceremonies

that can go on for several days.66

Many weddings seem less influenced

by norms in the village than by pat-terns in cities, and celebrations of

poor families imitate the more

extravagant patterns common inricher families. Take the case of a

wedding of the daughter of a small

agriculturist with two acres of landthat I observed. The teenage daugh-ter was marrying a young man whohad finished his B.A. and had a job as

a low-level clerk in the city. The

groom was dressed in a crisp graysuit, the bride in a silk sari.A largenumber of guests, including a(minor) local politician, had been

invited, and the newlyweds weredriven away at the end of the cere-

mony in a large whiteAmbassadorcar. When asked why he had spent so

much money on a wedding that was

obviously well beyond his means, thefather said that his daughter hadmarried into a &dquo;good family&dquo; (the

grooms father was a relativelywealthy landowner from another vil-lage), and he wanted to have a&dquo;show.&dquo;

Thus wedding celebrations have alot to do with social status and pres-

tige. Mobility within a village is oftenachieved by imitating the behaviorsof families of higher social orders(Srinivas 1966). For the parents of a

daughter in particular, a marriage ispotentially the most importantsource of mobility since marryinginto a good family can greatlyenhance how a family is viewed by itspeers, and a prestigious match is an



7/21

90

occasion for great celebration and

status displays. This may explain

why some weddings are particularlylavish. When a family marries into arich family, it is in the interest of the

former to demonstrate this to the

rest of the village, particularly if therest of the village does not know the

new in-laws. The most effective wayof signaling a familys newfound

affinity-derived status is to have as

lavisha

weddingas

theycan

possiblyafford. On the flip side, if a familymarries into a poor local family-onewell known to everyone in the

village-this may also be an occasionfor celebration, but lavish displaysare no longer necessary since notmuch can be gained by signaling.

To betterunderstand the nature of

marriage expendituresin India, it

might help to outline the basic na-ture of Indian marriage markets:

7

1. Marriage is restricted to

endogamous groups; that is, peopleare permitted to marry only within awell-defined set of families who

make up their subcaste.

2. Marriage is patrilocal-bridesleave their parents home to live withtheir husbands. Marriages are ar-

ranged for both grooms and brides bytheir parents, and the preferences ofthe parents usually drive all thechoices and arrangements.

3. Marriage is considered finaland, while there are cases of separa-tion, divorce is not an option.

4. The burden of celebrating amarriage almost always falls on theparents of the bride.

There is a considerable amount of

variation between subcastes in

terms of marriage and kinship pat-terns (Karve 1965). While most pro-hibit marriage outside the subcaste,the size of the group can vary a greatdeal. Some communities in South In-

dia prefer marriage within a circle ofclose relatives, usually between aman and the daughter of his eldersister or between the children of

brothers. Suchconsanguineous

mar-

riages are becoming increasinglyrare because suitable grooms are dif-

ficult to find.Another characteristic

of Indian kinship systems is the vari-ation in their prohibition againstmarriage between partners who be-

long to the same village. In the lan-

guage of kinship studies, such com-munities whose members marryoutside their own village are called

&dquo;village exogamous.&dquo; In Karnataka

state, where the data for the presentarticle are from, the northern dis-tricts show a tendency toward villageexogamy, while in the southern dis-

tricts the majority do not prefer itand may even have prohibitionsagainst it. Such preferences and pro-hibitions are prescribed by the cus-toms of the community and are exog-enous to the choices of households,who may suffer social sanctions by vi-olating them.

Getting ones daughter married,within these constraints, is consid-ered an Indian parents primaryduty; to have an older unmarried

daughter is a tremendous misfortunewith large social and economic costs.The costs of getting a daughter mar-



8/21

91

ried, however, have been steadily ris-

ing in real terms across the Indian

subcontinent. In Karnataka state,dowries have been increasing sub-

stantially for several decades

(Caldwell, Reddy, and Caldwell

1988), and dowries in the presentdata average six times the annual

income of a family. There are several

possible reasons why dowries inIndia may have increased, but it is

beyond the scopeof this article to

examine them.8 My focus here is on

wedding celebrations, and in particu-lar on the narrow question ofwhether wedding celebrations are, atleast partly, driven by a desire for

higher social rpnk.Every wedding requires a mini-

mum level of expenditure that

depends uponnorms in the commu-

nity.A basic celebration may helpmaintain ones stature within the vil-

lage, but a lavish wedding representssomething else. Spending more thanwhat is expected provides people inthe village with new informationthat helps them update their percep-tion of ones social status. In particu-

lar,if the bride is

marryinga

presti-gious groom-prestigious because heis either rich or well educated-then

this marriage alliance has statusbenefits for the wifes family. How-

ever, a lavish wedding without somereal change in status may simplygive people the impression that the

spender is extravagant.Thus, if families are rational, then

this extra money should be spentonly when there is new informationto communicate. If there is nothing toshow off, it would not be rational for a

father to spend anything beyond theminimum (maintenance-level)

expenditure required to fulfill hissocial obligations. Furthermore,even ifthe bride were marrying a rich

groom, there would be no new infor-

mation to communicate to the villageif everyone in the village alreadyknew that the groom was rich. The

very existence of the marriage alli-ance provides enough informationfor the

villageto

updateits

opinionon the social standing of the brides

family. Therefore, lavish weddingswould only make sense when thebride is marrying a prestigiousgroom and when the village has noinformation about the groom and his

family.The variations in village exogamy

provideus with a

wayto translate

this point into a testable hypothesis.When marriage rules prohibit part-ners from the same village from get-ting married, then the bride familyshome village will not know muchabout the grooms family. On theother hand, if brides and groomsfrom the same village customarilymarry each

other,then the

villageshould have very good informationon the grooms family. Thus, if lavish

weddings serve as a means of

increasing status, it is only whenthe groom is a high-status &dquo;catch&dquo;from another village that weshould observe particularly lavish

weddings.This hypothesis will be tested by

estimating a multivariate regressionof wedding celebration expenses onthe characteristics of the groom and

bride and their families, the number



9/21

92

of alternate partners in the marriagemarket, and whether the subcaste

that the family belongs to practicesvillage exogamy.9 I will interact thevillage exogamy variable with theeducation and wealth of the groom. If

these interaction terms are positive,we can infer that it is only when

grooms are educated and wealthy(that is, prestigious) and also from anoutside village-resulting in a lack ofinformation about them in the brides

village-that marriage expenses are

significantly higher. This would sup-port the hypothesis that lavish cele-brations are a signal of improvedsocial status, communicating newinformation about mobility via a

costly action.Wealth will be measured by

whether the families possess any

land, since about 40 percent of thehouseholds are landless. To measure

the human capital of the husbandand wife, I use their years of school-

ing and their age at marriage.Alter-natives in the marriage market aremeasured by the date of the marriage(to capture exogenous trends) andthe ratio of the number of women to

the number of men at marriageableages (defined as women aged 10-19and men aged 20-29) measured atthe year of the marriage for thebrides home district. In addition to

these variables, I also include dum-mies for whether the family is Mus-lim or belongs to a disadvantaged orscheduled caste.

Data and results

The sample consists of 800 house-holds randomly chosen from five dis-

tricts spread across north and southKarnataka state. Seven villages from

each district were randomly chosen,and 20-30 households were then ran-

domly selected from each village. The

marriage data that are employed inthis article were collected retrospec-tively from the marriedwomen in the

sample.About a third of the house-holds in the sample did not have anyever-married women of reproductive

ageand therefore did not answer the

womens questions. Ofthose that didhave eligible women, many did not

provide complete answers to the ret-

rospective marriage questions,requiring that they be dropped fromthe analysis.After eliminating about10 outliers, we are left with a sampleof about 300 women spread acrossthe five districts. There are no signif-icant differences between this

subsample and the complete samplefor those questions that wereanswered by all the households.

Table 1 reports summary statis-tics from the survey data. The aver-

age wedding celebration expensesamount to about 3000 rupees, 11 per-cent as large as the average dowry,which is 26,000 rupees.Averageschooling levels are 4.3 years for menand 2.9 years for women. The aver-

age age at marriage is about 25 yearsfor men and 17 years for women. The

mean marriage took place in 1980,with a standard deviation of 8.4,

showing that the marriages span afairly long time. Over this period, the

average marriage squeeze ratio is 1.13,indicating a surplus ofwomen in the

marriage market.About 9 percent ofthe sample is Muslim, and 22 percentbelong to scheduled castes. Finally,



10/21

93

TABLE 1

WEDDING CELEBRATION: MEANSAND STANDARD DEVIATIONS

68 percent of the households custom-

arily practice village exogamy.Table 2 reports estimates for the

wedding celebration ordinary leastsquares (OLS) regressions. Lookingat the regression with no interac-tions, we see that grooms with landhave celebrations that cost a statisti-

cally significant 1394 rupees morethan those of landless grooms. Fam-

ilies that belong to subcastes that

practice village exogamy have mar-

riage expenses about 1500 rupeeshigher than those who marry within

the village. The results ofthe interac-tion effects between village exogamyand the wealth and education of the

husband are also consistent with the

hypothesis presented above. Notealso that village exogamy interactswith whether the grooms family hasland such that wedding expensesincrease by 2600 rupees, an increase

that is significant at the 5 percentlevel. Similarly, an additional year of

grooms schooling in families that

practice exogamy increases marriageexpenses by about 280 rupees. Thusgroom quality significantly matters

only for those families practicing vil-

lage exogamy, where the brides vil-

lage has no knowledge ofthe grooms

family.10 These results are thereforeconsistent with the notion that wed-ding celebrations are driven by adesire to provide information about



11/21

94

TABLE 2

WEDDING CELEBRATION EXPENSES:

OLS REGRESSIONS (ROBUST STANDARD ERRORS)

the bride familys enhanced socialstatus.

ON FESTIVALS

Festivals are also centrally impor-tant to the lives of Indians (Fuller

1992).11 Weddings may cost more, butthey occur only two or three times inthe lifetime of a household head,while festivals take place every fewmonths. Festivals are different from

marriages in that they are collective

behaviors where everyone celebratesan event simultaneously.A description of the first time Isaw a village festival might beinstructive. Myteam and I had spentseveral days in a village interviewing

poor families from the pottercaste. On the sixth day, we werewarned that it would be &dquo;difficult to

find people in their homes [the next

day] because they will all be atMariammas12 festival.&dquo; When wearrived the next morning, the villagehad been transformed beyond recog-nition. The women had put on goldand silver jewelry and had jasmineflowers in their hair. There were col-

orful fragrant garlands strewn onpoles around the village and on the

pillars of the local temple, along withfestive decorations and balloons. Twoloudspeakers tied to the temples roofblared film songs.And then, at about10 in the morning, the processionbegan. It was led by a group of



12/21

95

traditional musicians. They were fol-lowed a brightly painted bullock cart

carrying an image of the goddess,swathed in silk, gold, and flowers andtended by two priests. Immediatelybehind the bullock cart was a groupof about 10 important-looking menand women, followed by hordes of

dancing children. I asked some peoplein the crowd who these important-looking people were and was toldthat they were from the families whohad made the largest contributionsto the festival and had helped orga-nize the celebrations. The festival

was obviously a very expensive event

entirely financed by voluntary con-

tributions ; and all the contributions,whether small or big, were made

public at an auspicious time whenthe Temple Committee, which orga-nized the event, announced them on

loudspeakers.Sociologists since Durkheim

( [1912] 1965) have argued that col-lective celebrations serve an impor-tant function by providing occasionswhen communities reify their groupidentity. Turner (1982), for instance,describes festivals as &dquo;generally con-

nected with expectable culturallyshared events.&dquo; He suggests that,when a social group celebrates a par-ticular event, it &dquo;celebrates itself&dquo; by&dquo;manifesting in symbolic form whatit conceives to be its essential life.&dquo;

Thus festivals may serve to build

social cohesion by reinforcing tieswithin a community. Furthermore,

by providing a specific time and placewithin which families can demon-

strate their commitment to beingresponsible members of the villagecommunity and also compete for sta-tus with others, festivals provide a

socially sanctioned arena for publiclyobservable action. By providing a

space where everyone can view

everyone elses behavior, they also

generate &dquo;common knowledge&dquo;13 andhelp solve the coordination problemsinherent in collective action (Chwe1998). In this sense, they help buildthe capacity for collective actions. 14

Thus, at the village level, festivalsenhance social cohesion and build

trust while providing an arena inwhich families can maintain and

enhance their social status.

The data I have are not adequateto study the relationship betweenfestivals and social cohesion at the

village level. Therefore, in the statis-tical analysis, I focus on a house-hold-level analysis of some of the

relationships discussed previously.In particular, I look at the determi-nants and effects of festival expendi-tures.Are they driven by privatemotives, or are they better thought ofas altruistic contributions toward a

public good?Are households that

spend more money on festivals heldin higher esteem by their peers? Doesfestival participation have real eco-

nomic returns? It is possible thatfamilies spend money on festivals forno other reason than pure entertain-

ment, particularly in rural areaswith limited access to movie theaters

and television sets. But if festivals

were pure entertainment, expendi-tures on them should not generateany other returns.

To test these hypotheses, I regressa households festival expendituresagainst various household character-istics, including the household heads

age, the number of adults and chil-

dren in the family, the heads educa-



13/21

96

tion, the land owned by the house-hold in the past and the present, and

the households income. I alsoinclude the number of girls in the

family who are of marriageable age.If their presence raises festival

expenditures, it is consistent withthe notion that they are being put on

display in order to find good spouses,thereby indicating that festivals pro-vide an opportunity to communicate

information for private gain. Finally,in order to see if festival expendi-tures are driven by demonstration

effects, I include the festival expendi-ture of the neighboring household asan explanatory variable, as well as aset of village dummy variables.

In addition to analyzing the deter-minants of festival expenditures, I

also examine whether festival expen-ditures generate social and economicreturns. I examine two types ofreturns: the familys social statusand a household-specific index of theunit prices for food faced by the fam-

ily. The social status variable is con-structed by asking each family in thevillage about its sense of the respect

and social standing accorded to otherfamilies in the village. This is codedinto a measure with four levels, ascore of 1 denoting the lowest level ofsocial standing and 4 denoting the

highest.l5The unit food price index requires

some explanation . 16 In the process ofdoing fieldwork, I noticed that therewere

largevariations in the unit

prices charged for food to differentmembers of the village. The primaryreason for this was quantity dis-counts, since many of the familieswere severely liquidity constrainedand were compelled to purchase food

once a week on payday. The fieldworkrevealed that households that had

good social relationshipswere able toget around this by forming shoppingclubs that would buy food in bulk andthen distribute it among the mem-

bers. Some well-regarded individu-als were able to get from the shop-keepers food on credit, which againpermitted them to get around liquid-ity constraints. Thus socially well-

connected households seemed to suf-fer less fromhigh prices. On the otherhand, households that did not havegood social relationships faced mark-

ups of 15-40 percent on their food

purchases. To get a summary mea-sure of the variation in unit pricesacross households, I construct aLaspeyres price index that uses the

pricesand

quantities purchased bythe median household as a base tocalculate price variations, for a bas-ket of goods, for everyone in the sam-

ple. With this method, the medianhouseholds indexnumber is 100, anda household with an index of 200

faces prices that are double the sam-

ple median.The social status and price index

variables will be regressed againstfestival expenditures, 17 the house-hold heads age, the number of adultsand children in the family, the heads

education, the land owned by thehousehold in the past and the pres-ent, and the households income,along with a set of village dummyvariables.

Data and statistical analysis

The data used in this article arefrom a census survey of an

endogamous subcaste of pottersspread across three villages, Halli,



14/21

97

TABLE 3

FESTIVAL EXPENDITURES: MEANSAND STANDARD DEVIATIONS OF VARIABLES

Beedu, and Ooru, 18 located within 70miles of each other, also in the state of

Karnataka. Only 20 percent of theadults in these villages actually prac-tice pottery; most are day laborers,with small-scale farming and silk

rearing the other major activities.Halli and Ooru are multicaste vil-

lageslocated within commutable dis-

tances of large towns. Beedu, a vil-

lage consisting entirely of potters, islocated in a &dquo;forward&dquo; district with

relatively good schools and a longtradition of providing basic socialservices.

Table 3 provides some summarystatistics about the sample. Theaverage annual household income is

14,471 Indian rupees-about $1644in purchasing power parity (PPP)-adjusted U.S. dollars ($482 at the

exchange rate prevalent at the time).The average land holding is very low

at 0.59 acres. Education levels are

also low, with the average maximum

schooling of a person within a familyat 4.54 years (the mean number of

years of completed schooling amonghousehold heads is much lower, 1.6

years). Despite the low levels of edu-

cation, the villages have recentlyundergone a fertility transition, withthe average household consisting of3.0 adults and 1.5 children.

Table 4 shows that festival expen-ditures rise with income, indicatingthat they are a normal good. Theyalso rise with the maximum level of

education in the family, the numberofyoung children, and the number of

girls of marriageable age. Thus wecan infer that private incentives are

important determining factors. Theamount spent on festivals by neigh-bors is not significant, indicatingthat demonstration effects are not



15/21

98

important. The level of expendituresdisplays a fair amount of variation

across two of the three villages, withBeedu having significantly lower lev-els of festival expenditures than

Halli, the omitted category. This is

despite the fact that the householdsin all three villages in this samplebelong to the same endogamoussubcaste and therefore observe the

same religious and ritual calendar.

This suggests that the private incen-tives driving festival expendituresmay differ from community to com-

munity for reasons other than socialcustom.

If households derive private bene-fits from festivals, what are they? Toexamine this, consider Table 5, which

presents results of OLS and instru-

mental variable (IV) regressions ofthe price index. Four different sets of

regressions are shown. Column (a)shows an OLS regression with festi-val expenditures included. Column

(c) is the same as (a) but includes an

interaction term between festival

expenditures and income. Columns

(b) and (d) follow the same pattern as

(a) and (c) but present IV estimates.Since these households are below theIndian poverty line and spend 62 per-cent of their incomes on food, includ-

ing income as a regressor should con-trol quantity discount effects. Yetincome does not have significanteffects on any of the four regressions.This is because it is highly correlatedwith the number of adults in the fam-

ily, which in turn is correlated withlower prices because large familiesare compelled to purchase food inlarger quantities.An additionaladult in the family reduces the price

TABLE 4

DETERMINANTS OF FESTIVAL

EXPENDITURES: OLS WITH HUBER-WHITE

STANDARD ERRORS (lfllN PARENTHESES)

index by more than 2 points in theOLS regressions.

Furthermore, an additional acrein the size of the fathers plot of landreduces prices by 0.6 points. How-ever, conditional on the land owned

by the father, current land holdingsdo not affect the

priceindex. Since

there has been a rather large net lossof land in this community over thelast 20 years, this result suggeststhat the reputation of the family-indicated by how much land the fam-



16/21

99

TABLE 5

DETERMINANTS OF PRICE INDEX: OLS WITH

HUBER-WHITE STANDARD ERRORS

(Itl IN PARENTHESES)

ily used to own-is far more impor-tant in accessing social networksthan current land holdings are.

Most important, higher festival

expenditures are negatively corre-

lated with the price index. The effectseems to be stronger for poorerhouseholds. Looking at the interac-tion between income and festival in

column (d), a 1000-rupee increase inannual festival expenditures reduces

the price index by 6.2 points at themedian income. This effect is

reduced to 5.7 points at the seventy-fifth percentile income level, andincreases to 6.67 points for incomes

at the twenty-fifth percentile. This isconsistent with the hypothesis thatfestival expenditures have an effecton prices independent of the othermeasured household attributes byproviding a social return to a family



17/21

100

TABLE 6

DETERMINANTS OF SOCIAL STATUS: ORDERED PROBIT (Izl IN PARENTHESES)

NOTE: Coefficients report effect of unit change in x on standard deviation change in y*.

Thus higher festival expenditures,perhaps achieved by increasing afamilys ability to tap into social net-

works, seem to give a householdgreater access to lower food prices.

Table 6 presents ordered probitestimates of the determinants of

social status. In order to ease inter-

pretation, the results are presentedas the effect of a marginal change ofan independent variable on the

standard deviation of the status vari-

able. Neither current nor past owner-

ship of land has an impact on status.However, controlling for wealth, fam-ilies with more income have higher

social status in the IV specifications.This suggests that a family with

greater liquidity and a steady jobthat produces a high and predictableincome is valued, possibly because itis able to provide a buffer against



18/21

101

risk.Annual festival expendituresare independently associated with

highersocial status, and, once again,

festival expenditures interacted withincome to result in lower status.

Looking at the IV estimates in col-umn (d), at the median income a

1000-rupee increase in festival

expenditures raises the familys sta-tus level by 0.39 standard deviations.Even if one does not believe in the

ability of IV estimates to establish

causal connections, it is clear thatfestival expenditures and social sta-tus are strongly associated with oneanother. This is consistent with the

notion that festivals serve as arenas

where social status is maintained

and enhanced.

CONCLUSION

The evidence presented in thisarticle suggests that publicly observ-able celebrations, such as weddingsand festivals, play an important rolein the lives of the poor by serving asarenas where reputations are main-tained and enhanced. They providean opportunity for families to com-municate information about their

mobility, about their willingness tobe good members of the village, andabout their willingness to participatein webs of obligation. This informa-tion takes on a crucial role in the

Indian context, where social net-works and relationships play a cen-tral role in shaping peoples identi-

ties. They are also extremelyimportant as elements of strategiesto cope with risk and poverty. Thus itis perfectlywithin reason that Indian

villagers spend as much as they do onpublic celebrations.

However, to the extent that thesecelebrations are status competitions,driven

bya need to

signal changesin

mobility, they may be &dquo;wasteful&dquo; inthe sense that they take resources

away from investments in schooling,health, and agriculture. But a judg-ment of waste should be rendered

with care because it may reflect a

model of welfare derived from West-

ern notions of individualism. To the

extent that celebrations and their

expenditures maintain identities,reify social relationships, fulfill obli-

gations, or represent investments insocial capital, they are indeed pro-ductive expenditures within this cul-tural context.

This has important implicationsfor development policy. Some sourcesof economic growth, such as mar-ket-driven improvements in urbanemployment, may reduce a familysdependence upon its traditional net-works. This could reduce the incen-

tives to maintain these networks and

thus possibly reduce participation in

village festivals. This would, in turn,adversely influence the traditionalmechanisms of maintaining socialcohesion within a village, which mayreduce its capacity for collectiveaction and consequently have a nega-tive effect on well-being (Narayanand Pritchett 1999). On the other

hand, as individuals become econom-

ically better off, their need to demon-strate their mobility to their peersmay increase the size of wedding cel-

ebrations and other forms of conspic-uous consumption. Thus economic

development may reduce social cohe-sion and trust while increasing sta-tus displays, moving celebrations

away from their role as participatory



19/21

102

events and toward an emphasis ontheir function as competitive games.

This could indeed be wasteful.One possible way to avoid this out-

come is to construct developmentmechanisms that improve well-beingin a manner that does not privilegeindividual over collective action. If

an intervention is more partic-ipatory, requiring collective decisionsand collective management, it would

maintaincommunity

cohesion and

perhaps have more equitable bene-fits. There are several mechanisms

currently in place that are attempt-ing such strategies-micro-financeprograms and social funds, forinstance-but they have not been

adequately studied to determine if

they retain and perhaps even build a

communityscapacity for collective

action. Further research will have to

be conducted to see whether commu-

nities targeted by such interventionsmove away from conspicuous to whatcould be called cooperative consump-tion : celebrations that are less about

showing off than they are aboutmaintaining links across families,building bonds, and sustaining webs

of obligation.

Notes

1. The data are from two different sam-

ples in Karnataka state.2. For an elaboration on this methodology,

see Rao 1997.

3.I use the word "signal" in the technicalsense of transmitting information from an in-

formed to an uninformed agent via a costly ac-tion (Salanie 1997).4. This helps explain why Westerners,

upon encountering Indians, are often puzzledby their detailed expositions on all the impor-tant people they know or are related to.

5. Of course, this is probably true of hu-man beings in all cultures, with the logic of

socioculturally derived incentives changingaccording to the contextwhile the logic of

economic incentives stays more or less the

same.

6. This section is culled from Bloch, Rao,and Desai 1999.

7. Some of these features have become

less restrictive in educated urban circles, but

they continue to be a defining aspect of villagelife.

8. For more on this subject, see Caldwell,Reddy, and Caldwell 1988; Rao 1993; Raheja1995; Kapadia 1995.

9. Note that the village exogamy variablemeasures the customary practices prevalentin the community to which the respondinghousehold belongs. This can be treated as ex-

ogenous to the dowry and wedding celebrationdecisions since they are unlikely to change inthe short term.

10. The village exogamy effect could alsooccur because husbands from outside the vil-

lage would increase the size of the weddingparty due to the greater number of potentialguests who may have to be invited. However,the interaction effects do not necessarily fol-low from this explanation and are far moreconsistent with a signaling motive.

11. This section of the article is derived

from Rao in press.12.A local village goddess.13.A definition ofa game where all players

know the structure of the game, know that the

other players know it, know that others knowthat they know it, and so on. See Osborne andRubinstein 1994 for a more precise definition.

14. Or what is now often called social capi-tal (Woolcock and Narayan in press).

15. The social status regressions will be es-timated with an ordered probit regression.

16.A detailed analysis of this can be foundin Rao 2000.

17. Festival expenditures will be treated asan endogenous variable using instrumentalvariables (IV), with the number of marriage-able daughters and the expenditures of neigh-bors as excluded variables.

18. The names of the villages have beenchanged.

References

Appadurai,Arjun. 1990. Technology andthe Reproduction of Values in Rural



20/21

103

Western India. In Dominating Knowl-

edge : Development, Culture, and Re-

sistance, ed. Frdrique Apffel

Marglin and StephenA. Marglin. Ox-ford : Clarendon Press.

Bloch, Francis, Vijayendra Rao, andSonalde Desai. 1999. Wedding Cele-brations as Conspicuous Consump-tion : Signaling Social Status in RuralIndia. Development Research Group,World Bank. Mimeographed, July.

Caldwell, John C., P. H. Reddy, and PatCaldwell. 1988. The Causes of Mar-

riage Change. In The Causes of Demo-graphic Change: Experimental Re-search in South India. Madison:

University of Wisconsin Press.

Chwe, Michael Suk-Young. 1998. Cul-

ture, Circles and Commercials: Pub-

licity, Common Knowledge and SocialCoordination. Rationality and Society10:47-75.

Collier, Paul andAshish Garg. 1999. On

Kin Groups and Wages in the GhanianLabor Market. Oxford Bulletin of Eco-nomics and Statistics 61(2):133-51.

Dumont, Louis. 1970.Homo Hierar-chicus : The Caste System and Its Im-

plications. Chicago: University of

Chicago Press.

Durkheim, mile. [1912] 1965. The Ele-

mentary Forms of the Religious Life.Trans. J. W. Swain. New York: Free

Press.

Fruzzetti, Lina. 1990. The Gift of a Vir-

gin: Women, Marriage and Ritual in a

Bengali Society. New York: Oxford

University Press.

Fuller, C. J. 1992. The Camphor Flame:

Popular Hinduism and Society in In-dia. Princeton, NJ: Princeton Univer-

sity Press.

Goffman, Erving. 1959. The Presentation

of Selfin Everyday Life. New York:An-chor/Doubleday

Kapadia, Karin. 1995. Siva and Her Sis-ters : Gender, Caste and Class in RuralSouth India. Boulder, CO: WestviewPress.

Karve, Irawati. 1965. Kinship Organiza-tion in India. 2d rev. ed. Bombay:Asia

Publishing House.

Marriot, McKim and Ronald Inden. 1977.Toward an Ethnosociology of SouthAsian Caste Systems. In The NewWind: Changing Identities in South

Asia, ed. Ken David. The Hague:Mouton.

Mines, Mattison. 1994. Public Faces, Pri-vate Voices: Community and Individu-

ality in South India. Berkeley: Uni-

versity of California Press.

Narayan, Deepa and Lant Pritchett.1999. Cents and Sociability: House-hold Income and Social Capital in Ru-ral Tanzania. Economic Developmentand Cultural Change 47(4):871-97.

Osborne, Martin J. andAriel Rubinstein.1994.A Course in Game Theory. Cam-

bridge : MIT Press.

Raheja, Gloria Goodwin. 1995. "CryingWhen Shes Born and Crying When

She GoesAway": Marriage and Idiomof the Gift in Pahansu Song Perfor-mance. In From the Margins of HinduMarriage, ed. Lindsey Harlan andPaul B. Courtright. New York: Oxford

University Press.

Rao, Vijayendra. 1993. The Rising Priceof Husbands:A HedonicAnalysis of

Dowry Inflation in Rural India. Jour-nal of Political EconomyAug.:667-77.

. 1997. Can Economics Mediate

the Link BetweenAnthropology and

Demography? Population and Devel-

opment Review Dec.:833-38.

. 2000. Price Heterogeneity and"Real" Inequality:A Case-Study ofPrices and Poverty in Rural India.Re-view of Income and Wealth 46(2):201-11.

. In press. Celebrations as Social

Investments: Festival Expenditures,Unit Price Variation and Social Status

in Rural India. Journal of Develop-ment Studies.

Salanie, Bernard. 1997. The Economics ofContracts. Cambridge: MIT Press.



21/21

104

Sen,Amartya. 1988. The Standard of

Living, Lectures I and IIThe Tanner

Lectures. In The Standardof Living,ed. Geoffrey Hawthorn. New York:

Cambridge University Press.Srinivas. M. N. 1966. Social Change in

Modern India. Berkeley: University ofCalifornia Press.

Townsend, Robert M. 1994. Risk and In-

surance in Village India. Econo-metrica 62:539-91.

Turner, Victor. 1982. Introduction. InCelebration: Studies in

Festivityand

Ritual, ed. Victor Turner. Washington,DC: Smithsonian Institution Press.

Udry, Christopher.1994. Risk and Insur-

ance in a Rural Credit Market:An

Empirical Investigation in Northern

Nigeria. Review of Economic Studies

61(3):495-536.

Woolcock, Michael and Deepa Narayan.In press. Social Capital: Implicationsfor Development Theory, Researchand Policy. World Bank ResearchObserver.

the annals of the american academy of political and social science-2001-rao-85-104.pdf

Documents