Rural - urban growth linkages in India

Policy, Research, and Externual Atfairs WORKING PAPERS Agricultural Policies Agriculture and Rural Development Department The World Bank May 1990 WPS 430 Rural-Urban Growth Linkages in India Peter B. Hazell and Steven Haggblade Improving the nonfarm response to growing demand from agri- culture calls for appropriate growth in agricultural technology, adequate investments in rural infrastructure, and the avoidance of policies that discriminate against small, labor-initensive busi- nesses in favor of their larger, capital-intensive couisins. The Policy, Rescarch, and F 'xemal Affairn Cor(plex distrbhutis PR[ Working Papsrv to dissczntnalc Lht hlztdirngs osI or pin grsb and to aecourage the exchange of ideas among lank staff and all others inerensted in desclolment issue.s I hec,ie pals :AM tLe taries of the authors, reflect only their views, and should h. Used and cinit accordingl) I hc findings, intcrprrtaton, and conctm,sulls ate thie authors' own. They should not he atILnuted to the World lBank. is Board of )irectors. is managernent, or anr of iLs mernNer coulntnes  Policy, Research, and Extemal Affairs _e1710 a,. Agricultural Policies This paper, a backgrounld paper for the l;Y 1991 Comntry Ecoiinomiic Memorandum l'or Inidia, is a product ol tihe Agricultural Plolicies Division, Agriculture and Rural Development Department and is part of a larger effort in FIRE to leani mlore about the ind(irect ef'l'ccts of agricultural growth on the rural nontarm economy, and how the valuc of'the incomc and employment benefits can tv cbnhanccd for the poor. Copies are availabic 1ree from the World Bank, 1818 H Sirect NW, Washington DC 20433. Pl.ase contact Cicely Spooner, room N8-039, extension 3(464 (79 palges with tables). Using two models - an economtcric analysis of' sustain rapid growth in Indiia's manufacturing cross-sectional data on states and tlistricts and a sector. An agricultural growth rate of 2.4 semi-input-output model f-itted to a nationl; pevrcent a year will generate only 1.8 percent (if' input-output table for 1979/80 -- Hazel] and irrigated agriculture) to 1.9 percenit (if rainfed Haggblade analyze thc relationship between agriculture) growth in national manutacturing agricultural growth and growth in the rural output. Even 6 percent growth in agriculture nonifarnn economy. They conclude that. will geneiate only atb)ut 5.5 percent growth in manu facturing OulpUt. Bccause of strong links to agricultural growth, rural nonftaim income and employmcnt Agricultural growth will lead to expansion in will both grow faster than thcir agricultural high-\alluC agricultural output, especially counterparts. A sustained agricultural growth livestock and torticulturall products. Increased rate of 2.4 percent (the past trcnd) will leIad to production of thcse labor-intensivc products 3.() pcrccnt groAllh in nonfarm iieomiic in irural slhol(d especiall hbenefit the poor. areas and towns and 2.8 pCercent growth in nonliIrmn cnlmplo\mcnt 11 aer'I'icIUltr -to' 4 'I'hic si;c of Ihc ariculiur1al incomC nIUltipli- perce:nt, thesc rates imnrc. to 5 _( pe t. IIaI ;1nd crs decpndLs prim nrila on the l\c\l of' pr capita 4.0 per:ent. re'spec't'Iive'! ila ricu0li il i nonle. bUt plulic poliCy Call alffeCt tlheir in agn-MitUdC. They' are posititclv related to Continued LroAth in atcriultural output is tlec dej8 lopllolllt of suLchi rural infra-MI'tUtre as unlikcl Ito provide the g(rownth in product i\c e- roads electri fea ,ion and banking serviCes. pl'oniv nt re(qlair bto lIsorh projected Thyredtolnrcases I'le re' st ronger un(dcr irrigated than rain fed in tte nrural Ilzbor force 'I' emc priplovni-cnt gap agricLultural girov Ih and larger ftor sii alll- toi wi ill incr.asC if irrigation pla\s a dCcreasing role mCdi um- simz'e larni than f0(r largcr fanns in a'ni ulltri pra gro lh Second: Ir\ lounds 0o gro'lth in tlhc rural non farlil CLe0n0m \ cOuldL Impro\ irig tihe niol'anif rcsl)nse to growing bridgc this gap givcn modcratt. agricultural deniand i'rom agriculture calls for appropriate gro\A th. growth in agricultural technology, adequate investments in rural infrastructure, well-devel. Export and domestic urban demand Imlust oped rural towns. and thc avoidance of tax, play ani imipo)rtant role if m1an U faCtIUring is. to regul atory, or licenisinlg policies that discriminiate continiuc to groA 8 percent a \car. Despite thc against sinall, laNor-intcnsi\c businesses in favor strength1 ol the' rural-urban linkages, agriCultural ol iheir larocr. Calpitll-inltnSiVe COUSinS. growth aiorne cannot provi(de enouoh 1miarket to AThc I'1 k>Wrk ic) AnIPl .9kr< ci --, tI' !:ld T; ' I c! \!rk III)JOt \A !rl l.I llk IT II.\nI , POi<.. IC TL r 1hl IuI kI rl.Ml p I,' Th! t 1 F 1 1 tI i> T i IIL.'T! !h I I T I I'lLI' 'I'l .l: I" Ir t )is,' IInnnI'I (c r) I IrI I't( III, I'd 1 t!%<C I' R J [)l>IIIC! !1I TlAt 11 II ( CTr! t L  RURAL-URBAN GROWTH LINKAGES IN INDIA Table of Contents Page EXECUTNE SIRY .............................. I L INTRODUCTION ................................ 1 IL DESCRIPTIVE PROFILE OF NONFAM ACTIVITY ....... ..... 2 A. hIportance ....2.......... 2 B. Location .6 C. Compositon .8 D. Fe-male Participaton .10 E. Trends .12 IIL AGRICULTUE AND GROWTH OF THE NONFAR ECONOY .. ..... 15 A. Key Linkages ........................... 15 B. , ior Evidence .......................... 16 C. Comparisons Across States .................. 18 IV. ESTIMATING RURAL A(;RICULTURAL GROWTH MULTPLIERS .. ..... 26 A. Econometric Model ........................ 26 B. Income Multipliers ........................ 29 C. Employment Multipliers ..................... 31 D. Projections ............................ 31 V. PROJECTIONS OF NATIONAL MANUFACTURING AND TERTIARY DEMAND RESULTING FROM AGRKILTURAL GROWTH ..... . . . . . 40 A. Semi-Input-Output Modeling of Growth Multipliers . . . . 40 B. Results ......... . .. . .. . . .. . .. . .. . . .. . 43 Vl CONCLUSIONS .......... .. ............ .. .. .. . . 47 REFERENCES ........... .. .. .. .. .. .. .. .. .. .. . . 49 APPENDIX A. SUPLEMNTARY TABLES APPENDIX B: A FIVE-SECTOR SEMI-INPUT-OUTPUT W)OEL Background paper prepared for the 1991 Country Economic Memorandum for India  Ust of Tables Table 1: Nonfarm Share of Total Employment. 4 Table 2: Nonfarm Share of Rural and Urban Income. 5 Table 3: Income Distribution by Fuhctbnal Classification. 7 Table 4: Nonfarm Employment Donsity by Size of Settleent ... 8 Table 5: Composition of Nonfarm Activity by Size of Locality ... 9 Table : Women's Share I Total Empoyment. 11 Table 7: Trends In Measured Female Participation Rates .12 Table 8: Changig Composition of Nonfarm Activity .14 Table 3: Agricultural Productivity Ranking of Major Indian States 20 Table 10: Density of NonFarm Activity Across States with Differhg Agricultural income .21 Table 11: Correlation Between Agricultural Productivity, Population and Locatlon of Nonfarm Activity .23 Table 12: Composition of Nonfarm Activity Across States with Differi Agricultural Incomes .25 Table 13: Rural Income Multipilers Across States wlth Dlfferhg Agricultural Incomes .30 Table 14a: Projections of Rural Plus Rural Town Income and Employmwnt Through the Year 2040 (2.4 percent agricultural growth rate) .36 Table 14b: Projections of Rural Plus Rural Town Income and Employment Through the Year 2040 (3.5 percent agricultural growth rate) .37 Table 14c: Projectlons of Rural Plus Rural Town income and Employment Through the Year 2040: Case of High Agricultural Growth (4 percent) .38 Table 15: Projected Employment Growth Rates In Rural Areas Pius Towns ......... .. . .. .. . .. . .. . .. .. . .. . . 39 Table 16: National Agricultural Output and Income MultiPllers for Irrigated versus Rainfed Agriculture .44 Table 17: Compound Annual Growth Rates In Nonfarm Output and Nontradable Agricultural Output Undor Altornative Agricultural Growth Rates. 46  LIST OF APPENDIX TABLES APPENDIX A Table A.1 Changes In Population, Workforce, and Nonfarm Employment In India, Table A.2: Income Distribution by Farm Size, Rural India Table A.3: Income Distrlbution by Farm Slze, Rural India 1970/71 and 1980/81 Table A.4: Locational Changes In Population, Workforce and Nontarm Employinent In India, 1961-81 Table A.5: Measured Female Employment Density by Activity, India 1961-81 Table A.6: Urban Population Distrlbution Across States with Differing Agricultural Incomes: India 1981 Table A.7: Rurai-Urban Differences In Nonfarm Activity in States with Differing Agricultural Incomes, India 1981 Table A.8: Changing Composition of Nonfarm Activity Across States with Differing Agricultural Incomes, India 1971-81 Table A.9: Employment Density by Activity In Six Indian States, 1971-81 Table A. 1 0: Varlations In Male Agricultural Wages a/ Across Regions, India 1970/71-1984/85 Table A.1 1: Composition of Change In Rural Income for a Panel of Households Indla 1970/71 - 1981/82 Table A. 1 2: Rural Consumption Profile by Expenditure Class, India 1983 Table A. 1 3: Regression Coefficients for Estimating Income Multipliers District Data Table A.14: Data, Income and Wage Multipliers for Regions of Varying Agricultural Productivity Table A.15: Regression CoeffIclents for Estkiating Income Multipliers State Data Table A.16: Regression Coefficlents for EsthnatUng Empioyment Multipliers Table A. 17: Changes In Rural Consumption Patterns, India 1967/68 - 1983  List of Appndix Tables - 2 Table A.13: Rural Consumption Parameters Across States With Dlffering Agricultural Incomes, India 1983 20 Table A.19: Semi-input-Output Parameters for India, 1979/80 21 APPENDIX B The authors are grateful to Ataman Aksoy, Pamela Cox, Kelth Obiltas, Frank Place, Toja Raparlu, and Roger Slade for helpful comments on earlier drafts.  EXECUTIVE SUMMARY The rural nonfarm economy accounts for 20 percent of full-the employment In Indla's rural economy and for 30 percent of rural heome. Expanding the definition of rural to Include rAral tmns up to 1930,000 h size, these shares Increase to 25 percent and 35 percent, respectively. Nontarm activlties are espeially huportant to the poor; landless laborers and mall farmers tyolcally obtain half or more of ther Income from nonagricultural actIvitIes. Women are also active participants, particularly In food procossig and household marutacturing activities. Sew In thIs context, It Is clear that the rural nonfarm econouy will play a key role In determining future prospects for employment growth and poverty allevlatW In India. II. Dominating the rural nonfarm economy are commerce, service and small-scale manufacturing activitlis that cater largely to agrcultural atid rural consumer demands. As such, tveir fate Is Intimately linked to agricultural performance. This can be seen In the shifting structural composition of the Indian economy over the. The nonfarm sharos of national empoyment and Income remahid ahlost constant for many decades, but showed a sizeabos Increase In the 1970s. This chango coincided wlth a perlod of rapid agricultural growth associated with the green revolution. Ill. The relationship can also be seen by comparing states of varying agricultural performance. High performance states, such as PunJab and Haryana, have a greater density of rural nonfarm activity, a greater density of rural towns, and proportlonally more commerce, service and factory manufacturing than states wlth poorer records of agricultural productivlty and growth. Iv. More formally, the relationship between agricultural growth and growth in the rural nonfarm economy can be analyzed wlth the aid of models. Two approaches are used in thl paper. The first Is an econometric analysis of cross-sectiona! state- and district-level data. The second Is based on a sem-Input- output model fitted to a - I -  national kput-output table for 1979/80. Both approaches provide eotknbtes of the agrIcultural income multipiler, defined as the increase In value added In the nonfarm sector attrlbutable to a one rupee Increase In agricultural value added. v. The sconometrli analysis leads to an estbiated Income multipiler of Rs 0.64, distrlbuted as Rs 0.39 In the rural towns and Rs 0.25 In rural areas. The multiplier Is largest In high-Income agricultural regions; Rs 0.93 In Punjab and Haryana versus Rs 0.46 In low productivity agricultural states such as Madhya Pradesh and Blhar. vI. In contrast, the semi-input-output model leads to an agricultural Income multiplier of Rs 1.35. This Is twice as large as the econometric esthiate because the semi-Input-output multipller pertains to the national, not just the rural, economy and assumes highly elastic supplies of nonagricultural outputs. The seml-Input-Gutput model also provides disaggregated multipliers for different types of agricultural growth. The income multiplier for Irrigated agriculture is Rs 1.56 but It is only Rs 1.23 for rainfed agriculture. vil. The models are used to project, through the year 2040, the volume of nonfarm Income, employment and rural demand for manufactured goods that will result from agricultural growth. They project nonfarm Increases under a range of agricultural growtt. rates and separately for Irrigated versus rainfed expansion. The following conclusions emerge: 0 Because of strong agrc'iltural growth linkages, rural nonfarm income and employment will both grow faster than their agricultural counterparts. A sustalned agricultural growth rate of 2.4 percent (equal to past trend) will lead to 3.0 percent growth in nonfarm income In rural areas plus rural towns, and 2.8 percent growth in nonfarm emipioyment. These growth rates Increase to 5.8 percent band 4.0 percent, resrectively, If agriculture grows at the Planning Commisslon's target of 4 percent. - 11 -  By itself, continued trend growth In agricultural output Is unikely to provide the growth In productive employment that Is required to absorb projocted Increases in the rural labor force. The employment gap will Increase further If IrrIgatIon plays a reduced role in future agricuWtural growth. However, secondary rounds of growth Induced In the rural nonfarm economy will brldge this employment gap given moderate rates of agricultural growth. ThIs will be especially Important to the rural poor who depend on nonfarm activity for sizeable shares of their total Incomes. Consequently, the rural nonfarm economy merits close attentlon In Indla's rural development strategy. 0 Despite the strength of the rural-urban llnkages, a(ricultural growth alone cannot provide the necessary market to sustain rapid growth in India's manufacturIng sector. An agricultural growth rate of 2.4 percent per year will generate 1.8 to 1.9 percent growth In national manufacturlng output; the higher figure relates to Irrigated agriculture, the lower figure to rainfed agriculture. Even a six percent agricultural growth rate will sustain only about 5.5 percent growth In manufacturlng output. Clearly, export and domestic urban demand will have to play an kitportant role If manufacturing Is to continue to grow at elght percent or more per annum. o Agricultural growth will lead to strong expansion In high-value agricultural output, especially livestock and horticultural products. For example, an agricultural growth rate of 2.4 percent will gcnerate nearly four percent growth In the output of these products. Given their I'gh labor Intensity. increases In the productlon of tiese products should be especially beneficlal to the poor. _ 111 -  vill. Although the sizo of thj agricultural hicome multipliers depond prbnarlly on the leve' of per capita agricultural hcome, public policy can Influence their magnitude. The multiplers are positively relato'i to the development of rural Infrastructure (roads, electrification, banking servicos, etc.). They are stronger under irrigated rather than rainfed agrlcultural growth and larger for amall- to medium-sized farms than for very large farms. Hence appropriate regional and farm targeting of agricultural technolgy. supported by adequate lnfrastructural Investments, will be especially knportant for kIprovhng the nonfarm response to growing demand from agriculture. Moreover, other studles have shown that well-developed rural towns foster stronger rural-urban growth lincagis, and that government polceas towards small nonfarm businesses are Important. It Is particularly hportant to &vold tax, regulatory or llconcing pollcies that discriminate against small, labor-Intensive businesses In favor of thelr larger, capital Intensive cousins. These relatlonships provide a rich agenda for Identifying government policies that will strengthen the size and distribution of the Indirect benefits emanating from agricultural growth. - iv -  RURAL-URBAN GROWTH LINKAGES IN INDIA L INTROOICTION 1. The rural nonfarm economy accounts for one-quarter of all fulwtimte employmont h rural India and for nearly one-third of rural Income. It Is also the backbone of the economy of numerous small towns scattered tl%oughout the countryside as well as the primary source of Income and emp,oyment for many of Indla's poor. Seen In this lIght, the rural nonfarm economy will play a key role In determining future prospects for mployment growth and poverty allevIatlon In India. 2. The rural nonfarm economy Is also Intimately linked to agrIculture. For example, a substantial share of rural xiufacturing Involves agroproc4aslng and the production, repair and supply of farm Inputs. Moreover, t;he dominant sectors In the rural nonfarm economy consist of trade and service establishments that cater largely to rural consumer demand. The prospects for growth In the rural nonfarm economy will, therefore, hinge on future agricultural performance. 3. Increases In farm Income stImulate demand for consumer goods and services (Meilor 1976). Likewise, a growing agriculture demands production Inputs and supplies raw materials to transport, processing and marketing firms. In additIon to stimulating natlonal economIc growth, these production and consumptlon linkages affect poverty and spatlal growth patterns. Because most of the resultant growth In nonfarmn activIty Is located In rural areas and small towns, It can contribute to the contalnment of excessive rural-to-urban migration. Moreover, when agricultural growth Is focused on small and modIum-sized farms, the resulting demand patterns typically favor products produced by small, labor-intensive ' erprises whose growth can contribute to hwreased employment opportunitlos for the poor (Johnston and Kllby 1975).  -2- 4. This paper examines the knportance of these rural-urban growth linkages In Indla. It kiims to assess the imiact of agricultural growth on natlonal demand for nonfarm products. In additlon, because growing land scarcity ralses concerns about prospects for rural labor absorptlon, the paper highlights the Impact of agricultural growth on rural nonfarm Incomes and employment. S. Four major sections address these objectives. The fIrst provides a descriptive overvlew of nonfarm activity In Indla. It examines the Importance, compositlon and location of nonfarm activity as well as general trends over the past 30 years. The second explores the relationship between agriculture and changes In nonfarm activity. After reviewing previous growth linkage studles, It compares nonfarm activity In high- and low-productivity agricultural states cross-sectlonally and over time. The third sectlrt estimates the volume of rural nonfarm Income and employment generated by agricultural growth, while the fourth projects patterns of demand for nonfarm goods emanating from alternative agricultural growth scenarlos. IL DESCRPTIVE PROFLE OF NONFAM ACTrvrrY A. kmportance 6. Nonfarm enterprises account for one-th' d of all full-tUne employment In India (Table 1). In large citles, nonagricultural pursuits occupy 95 percent of the workforce, In rural towns, 75 percent, and In rural areas, 20 percent. These proportlons have remained roughly corstant since 1961. 7. Part-time and seasonal employment frequently increase the knportance of nonfarm activity. lnaeed, some nonfarm undertakings In Indla are highly seasonal; 20- 50 percent of rural manufacturlng enterprlses ooerate only part-time or seasonally (National Sample Survey, 1969). But In the aggregate, measured secondary employment appoars very small. According to the 1981 population census, only 2.4 percent of  -3- Indla's economically active population finds secondary employment in nonfarm sectors.1 This holds true in both rural and urban areas. Natlonal Sample Survey data place the rural figure e Xi lower (National Sample Survey 1961). Of course, standard labor force definitions, because of their emphasis on usual employment and Inability to fully capture f*e,ale participation, may obscure the extent of seasonallty and part-time nonfarm employment, as the following Income figures suggest. 8. Income data reveal a larger role for nonfarm activity, Indicating that It contributes about two-thirds of national Income compared to a one-third employment share. Similarly In rural areas, excluding rural towns, nonagriculture normally contributes 25-35 percent of total income In contrast with Its 20-25 percent share of employment (Table 2). The higher Income than employment share Implies either greater returns to labor In nonfarm activity or considerable part-time and seasonal nonfarm pursuits uncaptured in the emplo,.. ent statistics. 1 So India (1988), Tabl* B-6.  -4- Table 1: Nontfau Share of Total Employment: hWd 1961/81 Rural Areas Plus Rural Towns Rural Rural Total Large Total Year Areas Towns a/ RRT Urban bl Natlonal (Percent of Total Full-Time Workers) 1 961 1 8.0 79.4 22.8 96.2 27.7 1971 15.2 76.5 20.4 95.4 27.9 1981 18.9 77.4 24.3 95.8 33.3 a/ Rural towns are defined as urban areas under 100,000 in population. They are settlements of betwen 5,000 and 100,000 people. b/ Large urban setlements are all those with population exceeding 100,000. Total urban figures reported in the censuses equal the sum of what have been partitioned here into rural towns and large urban settements. Scurce: Population consus of 1961, 1971, and 1981. See India (1961a-d, 17a-b and 1981b). Raw data are reproduced in Appendix Table A.1.  Tabe 2: Nonfarm Share of Rural and Urban hcolr V: hdba 1067/68 to 19B1/2 R U R A L Urban Source of Income 1967/68 1968/69 1969/70 1970/71 1975/76 1981/82 1975/76 Own harm 62.8 54.9 61.2 60.5 55.8 53.3 4.7 Wage iaboL (11.7) 19.9 17.7 17.4 13.7 (16.5- 0.5 1 1.9) Totd agri- culture (74.5) 74.8 78.9 77.9 69.5 (69.8- (5.2) 65.2) Nonfarm Selfmploym nt 10.3 8.4 7.6 9.0 8.1 26.4 Wage IaboP (2.0) 3.4 2.1 3.0 8.6 (2.9 - 10.6 7.5) Salary 10.5 8.0 5.9 5.8 9.8 - 49.1 Rent & dividends 2.7 5.4 5.5 4.? 4.0 - 5.6 Total nonanrm (25.5) 25.2 21.1 22.1 30.5 (30.2 - - 34.8) TorM hoome 100.0 100.0 10 10.0 100.0 100.0 100.0 a/ Excludes trasfer income from 1967/68 and 1975/76 to make income definition comparable with owt years. b/ In some yes, published asults fail to disaggregate farm and nonfarm wages. Figures in parenthese partition wages based on farm and nonfarm shares prevailing in other years. For 1967/68, the estimated wages breakdown takes nonfarm share of total wages at 15 percent, the I1 prevailing in 1968/691970/71. Since nonfarm share of wages appear to have risen over time, Xt 1961/82 stimate offers a range. The lower bound takes nonfarm wage share at 15 percent, the upper bound puts it at 38.7 percent, the level prevailing in 1975/6. Not applicable. Wages disaggregated in those years. Breakdown not available Source. National Council for Applied Economic Research (1972, 1975, 1980, 1986a).  9. Although many labor-intensive nonfarm activities provide work opportunities for the very poor, the aggregate data suggest a mixed Impact on Income distrlbution. In rural and urban areas alike, nonfarm Income constitutes the largest share of Income among both the very poor and the very rich (Table 3). For the wealthy, salarles, business profits and rents are most Important, while the poor depend most heavily on wage Income, both farm and nonfarm.2 10. By landholdlng, no such ambigulty arises. The smallest landholders routinely depend more heavily on nonfarm earnings than do families with larger holdings (Tables A.2 and A.3). Moreover, the recent NCAER panel study of rural households Indlcates that over time the smallest rural landholders, like all rural households, have become Increasingly dependent on nonfarm earnings (Table A.3). B. Location 11. Not surprisingly, the density of nonfarm activity Increases dramatically In urban areas and with town size. In rural settlements, about 50 people per thousand work In nonfarm occupations (Table 4). Yet In even the smallest rural towns, of 10,000 to 20,000, that flgure quadruples. A second Jump In nonfarm activity occurs as town size Increases to 50,000 and then 100,000. 1 2. Currently, about 20 percent of Indla's nonfarm employment Is based In rural towns, defined In this paper as localitles between 5,000 and 100,000 In populatlon.3 A further 35 percent reside In large citles over 100,000, while rural areas house the remaining 45 percent (Table A.4). 2 Unfortunately, as In Tables 2 and 3, most studies fail to disoggregate between farm and nonfarm wages when presenting income distribution prof I I-_. But aggregate income figuroe indicate that nonfarm wagso account for about 40 percent of rural wages and 95 percent of urban wages. Applying these percentages to the wage data In Table 3 indicatoe that under any conceivable form-nonfarm distribution, nonftrm wages account for a far higher share of incom for the poor then for the rich. Pal and Quison (1983, Table 13) corroborate this with NCAER data from 1970/71. J Tho remaining urban settlements, those over 100,000 are roforred to as largo cities.  -7- Table 3: hemo Distrbutlon by Funtiotnal Classification, bidl 1075/76 Rent, Income Dlvidends, Level Farming Wages Salary Business Transfers Total (Rs/Household) (Percent) Rural Areas Less than 3,600 40.1 45.0 2.3 6.0 6.3 100 3,601- 7,500 58.5 16.2 11.7 7.0 6.8 100 7,501-15,000 64.5 2.1 18.8 8.4 6.2 100 15.001-30,000 74.5 .2 10.0 9.8 5.5 100 Over 30,000 40.5 0 7.5 38.8 13.2 100 Urban Areas Less than 3,600 4.7 54.6 16.4 17.3 7.0 100 3,601- 7,500 5.0 15.6 50.5 21.1 7.8 100 7,501-15,000 4.7 2.1 56.5 26.9 9.8 100 15,001-30,000 3.8 .1 57.5 29.4 9.0 100 Over 30.000 6.4 0 41.0 44.1 8.5 100 a/ Includes farm and nonfarm wages together. Nonfarm wages account for roughly 40 percent of rural wage. and 95 percent of urban wages. Source: NCAER, 1980.  -8- Tab 4: Nonfwa Wlloy nt Density by Size of Settement kdla, 1971 Nonfarm Agricultural Size of Locality Employment Employment (Employment per 1,000 Population) 100,000 phls 284 14 50,000-99,999 243 39 20,000-4P,000 224 61 10,000-19,999 1 99 93 5,000- 9,999 198 96 Under 5,000 51 287 Ss.us Ihndia (1971a-b) C. Compodton 13. The composition of nonfarm activity differs considerably across locality sizes. in rural areas, services and household manufacturing dominate nonagricultural pursuits (Table 5). But In the move to rural towns, commerce and services lead the dramatic surge In nonfarm activity. Similarly, factory manufacturing and transport Increase substantlally. Even the prevalence of household manufacturing Increases In rural towns, although It declines In Importance In large urban centers. In the largest urban localities, factory manufacturing emerges as the dominant nonfarm activity.4 4 The date In Table 6 co_ froe 1971 because the 1961 breakdown of rural towns is not yot available.  Table 6: C_botn of Nonfam Actlvfty by Stte of Localty hda 197! Non-Agriculture House. NorH4ouse- Hold Hcd OOsr AQrIcLi5Jrs TotWl Mhg MIfr. Mfr. CondS. Co4mn. Trans. SenAoe TOTAL (Full-Tlme Employment per 1,000 Population) Total Employment Rural 287 51 1 11 8 2 8 3 18 338 Rural Town se 68 220 4 19 43 10 56 23 65 287 Large Urban b/ 14 284 2 11 86 10 61 34 79 298 Males Rural 230 45 1 9 7 2 8 3 16 274 Rural Town 54 198 4 1S 39 9 54 22 55 252 Large Urban 12 260 2 9 82 10 58 33 68 271 Females Rural 57 7 0 2 1 0 1 0 2 64 Rural Town 14 22 0 4 3 1 2 1 10 36 Large Urban 2 25 0 2 4 1 3 1 13 27 a/ Rural towns ae al urban are under 100,000 in population. b/ Large urban wra ae all above 10,000 in population. Souce: India (1971a-b)  - 10 - 14. These differences, at least as measured by employment statistics, reflect almost exclusively changes In the level and compositlon of male employment. Measured female participatlon5 remalns minor in all locallty sizes and activities and Increases perceptibly only In services In rural towns. D. Femal Partlipation 15. Females account for about 15 percent of national employment, 20 percent In agriculture and 10 percent In nonfarm act!vlty (Table 6). Among nonfarm occupations, women are most prevalent In household Industries, where they account for 20 percent of all workers. Since household manufacturfing declines precipitously In large cities, the female share of nonfarm and total employment drops to Its lowest level In these major centers. 16. Employment data, though they afford the most comprehensive framework for viewing changes in nonfarm activity and differences across reglons, are least reliable In evaluating women's economic role. The Invisibillty of women working In the home coupled with a restricted definition of economically galnful actIvity compromise our efforts to fully measure the economic participation of women. Table 7 Illustrates the potentlal magnitude of the undercounting, what some refer to as the 'statistical purdah' (World Bank, 1989). Beginning In 1971, the population census adopted a more restrictive employment definitlon than that used In 1961,6 and nonfarm activitles In which women predominate bore the brunt of the discounting. Measured female nonfarm and total participation rates dropped by over 60 percent between the two 6 S.o World Dank (1989) for a good discussion of how dramstically conventional labor force definitions undercount fem_le participation. 6 Notorojan (1962) and Jacob (1966) provide a good review of the problems with changing labor force definitions. rable A.4 illustrate- some of the anomalioe resulting from the changing definition. Notice the meosured decline in total and nonform workers between 1961 and 1971 even in tho face of a 20 percent incroose in population.  - 11 - Table 6: Wome's Share In Total Employment, India 1971 Rural Rural Large Total Activity Areas Towns - Urban National (percent) Agriculture 19.9 20.6 14.3 19.8 Nonfarm Mining 15.2 10.4 10.0 13.4 Household Industry 21.0 22.7 18.5 21.0 Nonhousehold manufacturlng 12.8 7.8 5.1 8.1 Construction 9.8 7.9 9.1 9.2 Commerce 7.8 4.3 4.3 5.5 Transport 3.2 3.4 3.4 3.3 Other services 12.3 15.1 16.5 14.1 Total nonfarm 12.9 9.9 8.6 10.8 Total full-time Workers 18.8 12.4 8.9 17.4 a/ Rural towns are all urban areas under 100,000 in population. b/ Large urban areas are all setliemwns above 100,000 in population. Source: India (1971a-b).  - 12 - TOM 7: Trends hI MStea d Female Partkipation Rat". hdla 1961-81 Female Female Female Workers Workers Nonfarm Workers as a percent per 1,000 per 1,000 of all Populatlon Population Fulltbiie Workers Rural 1961 154 22 34.2% 1971 64 7 18.9% 1981 78 9 22.5% Total National 1961 135 25 31.4X 1971 57 10 17.3% 1981 68 13 20.2X ~ hndis (Iak)i 1971a-b, 1581c). See Appendix Table A.5 for a detailed disaggregation. our understanding of poverty given the low-return, part-tkie, labor-intensive, often home-based activities In which women predomInate. E. Trend 17. The years nice 1961 have witnessed several changes In the Indlan nonfarm economy. As Table 1 Indlcates, the decade of the 19709 represents a key turning point. Up until 1970, India's agricultural share of natlonal employment remained constant, constant for a century or more accordIng to some.7 Not until the 1981 census did the first ovidence emerge of an Increase In the national share of nonfarm omployment; It rose from 28 percent In 1971 to 33 percent In 1981. The Income profiles tracked by NCAER researchers likewise Identify the first substantial boost In nrral nonfarm icomes durin the 1970s (Table 2). Thls coincides with the widespread ' See Sinha (1962), Vynn and Mathei (1978) and D..hpande and D.ehpand. (1985).  - 13 - employment; It rose from 28 percent In 1971 to 33 percert In 1981. The Income profiles tracked by NCAER researchers likewise Identify the first substantial boost In rural nonfarm Incomes durlng the 19709 (Table 2). This colncides with the widespread adoption of green revolutlon wheat and rice varietles and provldes at least circumstantial evidence linking the big spurt In agricultural growth with erlargement of the nonfarm economy. 18. Urbanization has accompanied the rlsing prominence of nonfsrm activity. The nonfarm employment densities have remained constant in rural areas, rural towns and In large urban centers (Table 1). This, coupled with the rlsing natlonal share of nonfarm In total employment, can only be possible If rural towns and large cities Increase In relative size, as Indeed they have. 19. Yet nationally, the composition of nonfarm activity has shifted only slightly durng Its decade of growth. Because of changing labor force deflnitions, reported differences between 1961 and later years do not reliably track changes in economic activity. Between 1971 and 1981, w.an definitions remained roughly comparable, the only perceptible change has been a slight Increase In the prominence of nonhousehold manufacturing In both rural and urban areas (Table 8). Dlsaggregatlon at the state level, however, does reveal a more substantial shift, as Sectlon 1I1. C. will discuss.  - 14 - Tr 8: ChUang CoumpoItlon of Nontam ActMity, hd. 1961 to 1981 Non-Agriculture HouSe Houser Hold Hdod oaer AWfI Tota Mwhn MfV. M1r. Conw" Conrnm Trans Sess TOTAL (Full-Time Employment per 1,000 Population) Rural Employment 1981 290 57 1 11 12 4 10 4 17 346 1971 287 51 1 11 8 2 8 3 18 338 1961 al 369 81 - 28 7 3 9 2 32 450 Urban Employment 1981 38 253 2 14 72 12 58 27 as 289 1971 37 256 3 15 67 10 59 29 ;3 293 1961 al 42 293 - 26 70 12 55 27 102 335 a/ Labor fnrce definion in t981 not comparabl with those in 1971 and 1961. Source: India (1961a-d, 1971a-b, 1981c).  15 - Ill. AGRICULTURE AND GROWTH OF THE NOIARM ECONOMY A. Key Lhkages 20. Why does nontarm activity vary over time and across regions? Certainly resource endowments, location, ethnicity, historical happenstance, and government policy all play a role. Yet agriculture, because of Its size, must be added to the list ,f key suspects. 21. Agriculture can Influence nonfarm activity in at least three ways: through productlon, consumption and labor market linkages. On the productlon side, a growing agriculture requires Inputs -- of fertilizer, seeds, herbicides, pumps, sprayers, equipment and repalr services -- elther produced or distrlbuted by nonfarm enterprises. Moreover, Increased agricultural output stimulates forward productlon llnkages by providing raw materials that require milling, processing and distributlon by nonf arm firms. Consumption linkages arise when growing farmer Incomes boost demand for basic consumer goods; these typically Increase over time as rising per capita income Induces diversification of consumptlon spending Into nonfoods. Much of the overall Increase In demand -- for Inputs, services, distrlbution and many bAsic consumer goods -- can be serviced by firms In rural areas, and rural towns. Yet the heavy productlon Inputs and consumer durables are more likely to be produced In large cities or abroad. 22. Although productlon and consumptlon llnkages have attracted most of tha Initial Interest In agricultural growth llnicages (Mellor and Lele, 1972; Johnstor and Kllbe, 1975), more recent investigat0ns highlight a third Important link, the labor market Interactions. In rural areas, In particular, rising agricultural wages raise the opportunity cost of labor !n nonfarm activitles. This Induces a shift In the composition of nonfarm activity out of very labor-intensive, low-return activitles and Into more skilled, higher Investment, high-return activitles (Hossaln, 1988; Ahmed and Hossain,  - 16 - 1988). Thus rising agrkcultural productivity may be Instrumental In ducng a structural transformation of the rural nonfarm economy. S. Prior EvIdence 23. Not all analysts have expressed ccnfidence In the prospects for agricultural- led growth. In a provocative and often-cited revIow, Vyas and Mathal (1978) argue that agricultural growth has not In fact stknulated development of the rural nonfarm economy. Using the population census data reproduced In Table 1, they point out that the nonfarm share of rural and urban employfment remained unchangod between 1961 and 1971. In their viow, skewed Income gains In agriculture lhited consumption linkages, while Inadequate rural Infrastructure lknlted the ability of rural firms to supply the modest Increases In Input and consumer demands. 24. Yet most subsequent analyses -- based on longer the series or disaggregated at the state of district level -- disputo Vyas and Mathals posshulstlc concluslon.8 Tkne sories ovidence from fast-growing agricultural states document the strongest connections between agriculture and the nonfarm economy (Chadha, 1986a). Studies of the Punjab (Chadha, 1986b; Bhalla, et al. 1989) and Haryana (Bhalla, 1981) all highlight the knportance of rising demand for consumer goods and agricultural Inputs as the result of Increased agrIcultural production. Chadha, In particular, emphasIzes the Importance of farm machinery and other Input supply In the Punjab. He notes that whilo state manufacturlng grew at the same rate as agriculture In the 19609. It grew much faster than agriculture In the 19709. Because of first-mover advantages, he believes, machinery manufacture established to supply Its own state agricultur6 stood poised to export to other states In the 1970s. If so, these spilovers caution that IdentifyIng the spatial distributlon of agricultural growth llnkages may be complex. a The most direct ebuttal, based on longer time _rio, come fron Doehpand. and Le.hpand. (1965).  - 17 - 25. These studies likewise corroborate ANed and H4ossaln's (1988) Initial evidence on labor market llnkages. In Haryana and the Punjab, Increased demand for agricultural labor has resulted In the highost farm wages In India. By raising the opportunity cost of labor In nonfarm pursuits, this has led to a decline In very low-return household manufacturlng and a parallel rlse In higher-return modern small factories and services. 26. Other tito-serles evidence comes from the moderately prosperous agricultural region of North Arcot. Using a sknulatlon model for the region, Hazell and Ramasamy (1989) have esthiated domand multipliers emanating from agricultural growth over the 1970s. They estimate that as a result of productlon and consumption linkages every 100 Rs. Increase In agricultural Income Induced an additional 82 Rs. In Income In other sectors of the rural economy. Production llnkages accounted for about half of the increase and consumption linkages the other half. 27. To date, cross-section comparisons across districts and states have produced similar, although less robust, correlations between agriculture and nonfarm activity. In part, this arises because so many Important factors other than agriculture vary across areas, and they also Influence the level of nonfarm activIty. Raw material availability varies across regions; consequently leatherworking industries predominate In Rajastan, while wood processing Is largest In well-forested states like Bihar (Papola, 1985). Moreover, trad;tlon, caste, historical accident,9 and Indla's elaborate system of subsidies and policy protectlon for small and village industries10 complicate cross- section comparisons. 9 Popola (19 6), for example, describes one district in Utter Pradesh specializing In th, production of plastic jewelry which it supplies to such of the rest of Indin. Despite an absence of local raw materials or market, the activity flourishes, for no obvious reason. He surmiseo that tradition or historical accident must *xplain this puzzle. 11 See Singla *t al. (1988) for evidence on tho connection between government support and level of nonfarm activity.  - 18 - 28. Even so, Radhakrishna et al. (1988), who compared three advanced agricultural districts in Uttar Pradesh with three laggards, found a higher nonfarm employment share In the agriculturally prosperous areas. PaPola (1985), comparing two different districts In the same state, found no correlation. But since he covered only a portlon of rural nonfarm activity -- and the least buoyant at that, household manufacturlng -- the lack of associatlon cannot be considered persuasive. 29. Khandker (1988) has used pooled time serles, cross-section district data to examine the relatlonships among rural employn,ent, wages, agriculture and Infrastructure. He finds both agricultural output and nonfarm employment higher In regions with higher agrocilmatic potential, but he does not attempt to measure the direct connection belween the two. 30. The labor market llnks between agricultural and rural nonfarm activity seem consistently robust In the cross-section studies. All comparlsons to date have confirmed the positive relationship between earnings in agriculture and earnings In rural nonfarm activity. 1 i C. Comparisons Across States 31. To further explore the elfect of agriculture on nonfarm activity, this section undertakes a descriptive analysis using state-level data. After ranking states according to agricultural productivity, It selects six for careful revlew, two high-, two low- and two medlum-productivity states. Initlally, cross-section comparisons examine how the size, compositlon and location of nonfarm activity vary across productivity zones. Then discussion turns to the time-serles evidence from these same six states to see how growth In agriculture affects growth of the nonfarm economy. 32. Table 9 ranks states according to several measures of agricultural productivity. It shows that per capita agricultural income, per capita total income, per capita 11 See Papola (1985), Chadha (1986) and Radhakrishna at al (1988).  - 19 - foodgraln production and growth rate of foodgrain production all provide a broadly sknilar ranking. So from this listing, two high-productivity states (Punjab and Haryana), two low-productivity states (Bihar and Madhva Pradesh) and two middle-productivity states (Karnataka and Gujarat) have been selected for revlew. 33. Consider first the density of nonfarm activity, which Is generally higher In high- Income agricultural states. As Table 10 Indlcates, th's holds true In both rural and urban areas. But as the complete state profile In Figure 1 shows, the generally positive association between nonfarm activity and agricultural Income masks considerable varlation In Individual states. Kerala In particular houses an unusually high proportion of rural nonfarm activity for reasons that appear unrelated to the character of the state's agriculture. Figure 1 further illustrates the stronger farm-nonfarm assoclation when one expands the concept of rural to Include rural towns. 34. Indeed It Is Important to separate out the rural towns. Our six-state comparison suggests a greater predominance of rural towns in the high-income agricultural states. In fact, rural towns In Punjab and Haryana house nearly twice the population share of similar-sized settlements In Bihar and Madhya Pradesh. (Table A.6). 35. Furthermore, the relationship seems to hold more generally across India. States with high farm Income are typically more urbanized and less rural than states with low- productivity agriculture (Table 11). And the big difference in urban structure lies In the predominance of Intermediate-sized towns. This suggests that a growing agricultural sector may Indeed contribute to a dispersed pattern of urbanization, as Mohan (1984) and Wanmall (1988) maintain.  - 20 - Tab 9: Agrklltural Prodactlvty Ranking of Major kidan States Annual Total Rate Income of Growth Agricultural Income Per Foodgramn in Food Per Aaricultural Population Cadta Production Production Average Average Averags 1983/84 1973174- 1984/85- 1961/62- 1982/83 1985/86 1975176 1982/83 1986/87 1981/84 (Rs./Capita) (Rs/Capita) (kg/capital (percent) PunJab_/ 2,764 3,423 1,486 3,484 924 6.3 Haryana a/ 2,357 2,773 1,263 2,798 507 4.6 Kerala 1,347 2,250 1,018 1,447 43 1.0 RaJasthan 1,314 1,651 739 1,574 191 2.5 Maharashtra 1,294 1,623 662 2,525 123 2.0 Andhra Pradesh 1,282 1,440 734 1,536 165 2.4 Karnataka 1/ 1,136 1,501 837 1,559 157 2.6 Gujarat -' 1,116 1,626 718 2,182 98 4.2 Orlssa 1,066 1,256 634 1,308 219 1.5 West Bengal 920 1,813 699 1,595 158 1.4 Uttar Pradesh 902 1,106 503 1,439 247 3.0 Madhya Pradesh 862 1,189 570 1,311 239 1.6 Taml Na/du 731 742 595 1,373 135 .6 Blhar - 599 852 415 995 138 .7 a/ Selcted for comparative anaiyuis. Source: Economic Intelligence Srvice (1988), Economic Monitoring Service (1986).  - 21 - Tabe 10. Deost of Norfm Acvty Ac States with DWfforkg AgrIcutural hobme, hida 1981 States Rural Areas Urban Areas al (Full-tknu Workers per 1,000 Population) High Agricultural Income PunJab 67 263 Haryana 67 258 Medium Agricultural Income Karnataka 61 247 Gujarat 53 258 Low Aoricultural hcome Madhya Pradesh 44 241 Bihar 38 212 a/ Urban inud ail localities over 5,000 in population. It encompasses both rural towns and large urban areas. Source: India (1QS1c).  -22- FIGURE 1 RELATIONSHIP BFETWEEN NONFARM EMPLOYMENT AND AGRICULTURAL INCOME ACROSS INDIAN STATES, 1981 0 Andhro Prodesh a Bihar RURAL AREAS 0 Gujorat 0.6 * Haryam - . Hiwal Prodesh 0 A O Karnataka 0 Kerala i . * Maya Prodesh . * Maharashtra p 0.4 _ * b/~~~~~~~~~~anipur . * ~~~~~~~~~~~~~Orissa 0 o.3 _