Nepal - Relieving poverty in a resource-scarce economy (Vol. 2 of 2) : Annexes

Report No. 8635-NEP Nepal Relieving Poverty in a Resource-Scarce Economy (In Two Volumes) Volume Il: Annexes August 15, 1990 A Joint World Bank/UNDP Study Country Department I Asia Region FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  Currency Eouivalents Year US$1.00 Equivalent (Average) 1984185 Rs. 17.8 1988/89 Rs. 25.6 Fiscal Year July 16 - July 15 List of Abbreviations and Acronyms Used ADB Asian Development Bank ADB/N Agricultural Development Bank of Nepal AIC Agricultural Inputs Corporation APROSC Agricultural Projects Services Center CBS Central Bureau of Statistics CSI Cottage and Small Industries DFAMS Department of Food and Agricultural Marketing Services EGS Employment Guarantee Scheme ha. hectare HMG His Majesty's Government of Nepal IBP Intensive Banking Program IDS Integrated Development Systems ILO International Labour Organization kg kilogram JNSP Joint Nutrition Support Programme K-BIRD Karnali-Eheri Integrated Rural Development Project KHARDEP Kosi Hill Area Development Project LF Labour Force LRMP Land Resources Mapping Project MPHBS Multi-Purpose Household Budget Survey mt. metric tons NCHS National Center for Health Statistics NFC National Food Corporation NFP Nutritious Food Programme NGO Non-Governmental Organizations NPC National Planning Commission NRB Nepal Rastra Bank p.a. per annum PCRW Production Credit for Rural Women Rs. Rupees SCF/UK Save the Children Foundation/United Kingdom SFDP Small Farmers' Development Program SLC School Leaving Certificate SPWP Special Public Works Project SSNCC Social Services National Coordinating Committee SSSR Social Sector Strategy Review T and T Trade and Transit Dispute UNICEF United Nations Children's Fund WHO World Health Organization  FOR OMCIAL USE ONLY NEPAL - RELIEVING POVERTY IN A RESOURCE-SCARCE ECONOMY Volume II Contents Pag ANNEX I Sectoral Papers ............................... 1 1.1 Land Tenure and Tenancy ............................... 1 1.2 Labor Force Issues ....................... 15 1.3 Debt and Indebtedness ...................... 32 1.4 Agriculture and Poverty in Nepal . ......................... 52 1.5 The Formal Sector ........................ 65 1.6 Informal Sector Employment and Incomes . . 80 1.7 Food Security ............................................ 97 ANNEX II Evaluation of Poverty Related Programs.-...... .......... 114 11.1 Food and Feeding Programs . . ............................... 114 II.2 Employment Creation Programs . . . . . i27 II.3 Integrated Rural Development Programs . . . 136 II.4 Targetted Credit . . ...................................... 150 11.5 Skills Training Programs .. ................................ 168 ANNEX III Statistical Appendices .................................... 173 ANNEX IV Bibliography ........................................... 197 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  ANNEX I - SECTORAL PAPERS Contents Page ANNEX 1.1 Land Tenure and Tenancy ....................... ...... 1 A. Introduction .........................1.... B. Background ...............1.............. C. Present Situation ............................. 7 D. Conclusions ............................. 12 ANNEX 1.2 Labor Force Issues ................... .......... 15 A. The Labor Force ............................. 15 B. Employment Arrangements ...................... 20 C. Wage Levels ............................. 23 D. Labor SupplylDemand Issues ............................ 25 E. Migration ............................. 27 F. Further Work ............................. 31 ANNEX 1.3 Debt and Indebtedness ...................... 32 A. Introduction ............................. 32 B. The Informal Credit System ............................ 32 C. The Impact of the Formal Credit System on the Poor ............................. 40 ANNEX I.4 Agriculture and Poverty in Nepal ..................... ...... 52 ANNEX 1.5 The Formal Sector ............................. 65 A. Introduction ............................. 65 B. Sectoral Employment and Incomes of the Poor ........... 65 C. The Potential Role of the Formal Sector in Poverty Alleviation . . ........................... 78 ANNEX I.6 Informal Sector Employment and Incomes in Nepal ........... 80 A. Present Status of the Informal Sector ................ 80 B. Informal Sector Activities ............................ 85 C. Evolution of the Informal Sector- Dynamics and Constraints ........................... . 88 D. Prospects for Informal Self-Employment ................ 94 ANNEX 1.7 Food Security .......................... 97 A. Food Expenditure and Food Insecurity .... .... 97 B. National and Regional Food Balances ................... 99 C. Agricultural Production and Transitory Food Insecurity ............................. . 103 D. Food Prines. .. ....... ... ...*. .... . 104 S. Trausportatian and Food Security ......... .. .60106 F. Food Consumption and Nutrition ...................... .107  ANNEX I.1 - LAND ?TNURt AND TENANCY A. Introduction 1. The issue of land tenure in Nepal is largely one of insufficient land for a burgeoning population, rather than the inegalitarian distribution of existing land. Average holding sizes tend to be small (just over 1 ha. in the terai, and just under 1 ha. in the hills) - and while there are some large landholders, the concentration of land ownership is nowhere near as extreme as it is in many developing countries. In the terai, a large proportion of households (about 50-60Z) are landless or tenants and many of the most important land tenure issues deal with the incentives goverr.ing tenancy arrangements. These are determined in large part by the actual and perceived effects of earlier land reform efforts. 2. This Annex recounts the history of land reform and the background to the current system of land tenure. It goes on to describe the present situation with respect to land distribution and tenancy arrangements, the problems of equity and efficiency which result; and proposes a possible solution. B. Background 11 3. From the end of the Rana regime in 1951 through to 1964, there was a series of agrarian reform measures directed primarily at replacing feudal-type land ownership and tenure systems with ones more appropriate to a modernizing economy. These measures culminated in the 1964 Land Reform Law. 4. This reform was introduced in three stages, from 1964 to 1966. In 1964 the land reform program was enforced in 16 districts of the Katbmandu Valley and Terai. It was extended to another 25 in November 1965 and then to the last 34 districts in 1966.21 Legislation took the first area under agrarian reform by some surprise and from it came the largest amount of 'excess lands'. Actually, publicity on the impending reform (which many expected would not be executed) pre-dated its first phase by several years. A land reform commission was appointed by the Xing in 1961. In that year, the Nepal Gazette reported that 'the ownership of land is concentrated in a very small number of people, while the majority of the cultivators are exploited," and it recognized that this was inimical to both economic development and democracy.3/ 11 This section draws heavily from a background paper by William Thiesenhusen. 2/ A.N. Seth, Report on Land Reforms in Nepal, (FAO Regional Office for Asia and the Far East, Bangkok, Thailand, December 1967) p.6. 3/ Reported in Mahesh C. Regmi, Landownership in Nepal, (Berkeley: University of California Press, 1976), P. 201.  - 2 - 5. The ceilings on landholdings imposed by the 1964 law are: Terai - 18.4 hectares Kathmandu Valley - 3.1 hectares Hill Regior. - 4.9 hectares The acts provide that comper.sation be paid to the landowner on the basis of the land revenue at the following rates: Cultivated land: 10 to 30 times the land revenue; Grass and bush land: 5 to 10 times the land revenue; Waste land: one times the land revenue. Ten percent was payable to the landlords in cash and the remainder in bonds. The excess land was to have been distributed, in order of priority, to: 1) Sitting tenants; 2) Farmers with adjacent land; 3) Other tillers in the area; 4) Landless area citizens. The distributed allotment was to have been made up to the following regional ceilings: Te-ai - 2.7 hectares Kathmandu Valley - 0.5 hectares Hills - 1.0 hectare 6. Landless peasants, apparently a growing group though not shown as such by official statistics, were so far down on the priority list they didn't benefit. In sum, only about 32,000 ha. were distributed (about 1.5 percent of the cultivated land in the country in the early 19708) among some 10,000 peasant families. However, it should be noted that this number is not universally agreed upon; this particular figure came from the Director General of the Department of Land Reforms. For contrast, see Regmi's estimates. 4/ Host land distributed was not, needless to say, land of the highest quality. Choice of land to declare 'excess" was usually at the discretion of the landowner. 4/ Regmi, Landownership in Nepal, P. 202.  7. Shifting land relations in favor of tenant security was a more important goal of the 1964 reform. In the 1950s, the government had attempted several times to modify the tenancy system (most notably in 1957) and to stabilize the relationship between landlord and tenant, but to little avail. Those intermediaries who acquired land from landlords and re-rented it to tenants who would work it were eliminated. 8. A matter of priority in the adminstration of the 1964 law was that the government sent teams to compile lists of tenants currently farming throughout the kingdom who were, therefore, eligible for secure land rights. 'Tenants' were recognized as those who acquired land from an owner and, through their own labor or that of their families, cultivated it. Existing tenants or those who raised the main crop at least once were considered to be entitled to permanent tenancy rights over that land, which could not be abrogated by the landlord unless the tenant did not cultivate the property for one year, defaulted upon the payment of rent (it was this provision which caused tenants to pay their rents to the district land reform office which, in turn, reimbursed th'e landlord), or did anything to reduce the value or the productivity of the land. For residential uses, the landlord could resume his land upon payment of 25 perce-nt of the value of the land so resumed. Individual resumption of agricultural land by landlords or relinquishment of tenancy by tenants needed to be officially sanctioned. 9. The act made provision for special tribunals to deal with disputes, but in 1966 this authority was turned over to district land reform offices who seem to spend the bulk of their time on settling landlord or tenant grievances.5/ Tenancy rights could not be sold or alienated, but they could be inherited. They could not, according to law, be subdivided. 10. In the process of tenant identification, almost 1.8 million individuals were issued identification slips, but certifications of tenancy were awarded to only 300,000. It was estimated that 40 percent of the total number eligible were left out during the idertification process.6/ Initial omission was not the only problem: 'many of the tenants who actually did receive certificates of tenancy or identification slips during the implementation stage, could not be protected because of a lack of adequate record-keeping systems in the offices".71 5/ Bhumi N. Shrestha, Evaluation of Land Reform Programme in Nepal (sic), Center for Economic Development and Administration, (Tribhuvan University, Kiritipur, Kathmandu, Nepal, 1978). 6/ Asian Development Bank and His Majesty's Government of Nepal, Nepal Agriculture Sector Strategy Study, Vol. II, (Kathmandu, December 1982), p. 102.0 7/ Agricultural Project Services Centre (APROSC),'An Evaluation of the Fixed Rentals System' (in Nepali), 1977, quoted in Asian Development Bank and His Majesty's Government of Nepal, op. cit., p. 103.  - 4 - 11. Of the enumeration one authoritative study claimed, mThe Lands office in Morang District carried out an on-the-spot survey in the four panchayats of the district and found the actual number of tenants far exceeds the number recorded in the tenancy register.'8/ 12. Therefore, no precise estimate of the size of the tenancy problem for the country as a whole was re. 'ly available in 1964, and the situation is little changed today. In 1967 nsere was a general feeling that 'about 50 percent of the area in the valley, 60 percent of the eastern terai and 70 percent in the western terai was tenant cultivated.' As regards the Hills, not much area, it seems, is tenant cultivated (except in small pockets here and there).9/ 13. An enormous problem with the 1964 law was that it provided for 'voluntary' surrenders by tenants; this provision was exploited to 'ease' out tenants. "Such surrenders and relinquishments in favor of landlords are anything but voluntary and it is necessary that they ...be outlawed.10/ 14. Rent-control legislation accompanied the tenancy 'security.' (If it had not, the landlord would have had every incentive to increase rents to a point where the peasant would have to abandon the land.) Indeed, beginning with the 1957 Land Act, the landlord was prohibited from charging rents in excess of 50 percent of the total produce in cash or in kind. The 1964 act ret&ined this provisic:. for all parts of the country except the Kathmandu Valley where it fixed renti according to land quality. In 1968, the rents were reduced to 50 percent of the value of the main agricultural crop, a figure which the government retained the right to determine. 15. The government took other measures to improve the terms of rental of tenants early in the reform years, and in some areas fixed the rent in crop quantities to remove arbitrariness from the rent-setting process. For example, in 1971 and 1973, the government fixed rents payable by tenants in absolute amounts of grain in twelve districts of the central and eastern terai. In the Kathmandu valley rents had been set in physical quantities of cereal. The fixed rents were based on average production on a land classification scheme which divided all land broadly into lowlands and uplands, with the lowlands further subdivided into four different grades and the upland into two categories (making a total of six grades of land). Gradations were based on condition of the soil, fertility of the land and its access to irrigation. The fixed rents ranged from 358 kilos to 827 kilos of rice per hectare according to the grade of the land. Dissatisfaction with the imprecision of the grading system has prevented its use in all districts of the country.l1/ 8/ M.A. Zamen, Land Reform and Land Administration, No. TA 3310, FAQ, 9/ Seth, op. cit., p. 3. 10/ Seth, op. cit., p. 14. _1/ ADB and HMG - Nepal Agricultural Sector Review; 1982, p. 102  16. The agrarian reform program which began in 1964 probably did change the system of landownership in Nepal more decisively than had any other government interventior. to date, but writers and commentators disagree on its extent and on whether the law caused more change in its breach or in its observance. The land market was critically affected since tenants could theoretically not be evicted if the land was sold against their will. Since the enactment of the 1964 law, the tenacious among the tenants - those who learned their rights and stuck by them - were made more secure and the payments they had to make to landlords were reduced. On the other hand, the law imposed no new obligations on landlords other than as collectors of the tenants' rent, so the problem of absenteeism was not markedly reduced. Furthermore, the landlords were left with the overwhelming incentive to rid their farms of tenants and resume cultivation with the use of hired workers. Realistically, however, the manager-hired labor combination was appropriate only upon a farm's reaching a certain size. On smaller farms and plots it was simply uneconomic to move to cultivation by hired laborers. Smaller plots more often needed the labor- entrepreneur-manager function to be combined in one person or in one family. It required a tenant or the landlord to farm his property personally. 17. It is possible that the Land Reform Law of 1964 did not make the agrarian structure of the country much more egalitarian either, though data and opinion on the matter are also open to dispute. In a recent study, ARTEP notes: TThe conclusion . . . seems inescapable that despite the reforms of the 1960s there has been no reduction in the degree of inequality in the distrlbution of landholdings in rural Nepal. It is probable that there has been a gradual worsening of the situation during the last two decades.l21 The IDS report is somewhat less bleak in its assessment of change in equity of landholdings between 1962 and 1981, and the data on which it is based are somewhat more complete. The IDS study shows larger representation in terms of families and area in the middle- sized (l-to-3 hectare) farms in 1962 than in 1981, but the smallest farms in 1962 had become even smaller by 1981. Large farms over 19 hectares dropped somewhat in percent of families and land area, though in absolute terms the average size for those over ten hectare farms was almost 19 hectares in 1962 and nearly 22 hectares in 1981.13/ 18. When reviewing the accomplishments of reform, an important point must be understood: the agrarian reform was not designed to be excessively harmful to the landlord just as it was not meant to be a boon to the cultivator. Ex post analysis seems to show that rents were fixed so that the landlord was receiving roughly what he might have obtained from 121 ARTEP Employment and Development in Nepal, Bangkok: 1982, p. 41. 13/ Integrated Data Systems, The Land Tenure System in Nepal (Kathmandu: IDS, 1966), pp. 27-28.  -6- investing his capital elsewhere.14/ 19. The land reform has had little effect in diverting capital and skilled and talented labor away from fanming to other sectors of the economy in Nepal, as had orginally been hoped. Indeed, there was little outside of agriculture in which a landowner could invest. Land reform did give some impulse to expulsion of tenants and removal of those tenants who could be bought off by landlords. Surely, many tenants were displaced in the process, but because of the lack of census materials there is no way of knowing how many tenants met that fate. Also there is no way to know how many unregistered tenants exist today as persons who are indistinguishable from tenants, except for the nomenclature by which they are described. 20. It is known that when tenants exerted their rights, they were difficult indeed for landlords to deal with or to manage effectively; sometimes they demanded one-fourth, more usually one-third, and, rarely, even up to one-half of the property in the ever.t that the landlord wished to sell his holding free and clear. It was not usual for landlords to sell property with secured tenants resident on it; buyers were difficult to find for a property so encumbered. The Land Reform Law of 1964 did, therefore, enicourage de iure landownership transfer into smaller units; whether de facto control changed much or not will never be completely known because of disarray and category changes between the recent agricultural censuses. 21. The land reform's major accomplishment was to give tenants a security that they did not previously have. To the extent this was done successfully, the agricultural system was rendered less flexible and the land market was made less viable. The exercise did point up to the government the necessity of completing the cadastral survey and its accompanying registration system, however. In Parsa and a number of other districts, the system is working reasonably well. Land transactions are indeed being recorded. Fraudulent transactions are discouragLu because all exchanges on sales must be made using the citizen's national identity card. The cadaster is essentially complete on a panchayat-by-panchayat basis in district land registries in 53 districts, in 9 districts it is in progress, and in the remaining 13 districts, mainly in the mountains, the cadaster is yet to be accomplished. One serious problem: land registries sometimes demand bribes, and transfers are made only upon payment of enormous extra- legal fees. Another problem is that if a landowner has land in several panchayats or in several districts, it is very difficult, if not impossible, for authorities to find out the extent of his total holdings from the hand-written, panchayat-specific books of records that are kept in each district office. 14/ Ram Bahadur, 4Assessment of Agricultural Rentsw ("Kut Nirdharan"), Gorkhapatra (Rathmandu), 2028 (1971); and M.A. Zaman, 'Report to the Government of Nepal on Land Reform and Land Adminstration," NEP/68/0l1 (Rome: FAO, 1974), p.13.  22. Some look upon the results of the 1964 agrarian reform with great circumspection. For example, IDS criticized, 'For all practical purposes the tenants have receded into oblivion where they are left to sink or swim according to their own devices. In fact, one wonders whether the tenants would not have been better off without the land reforms. Results from the present field surveys reveal that yields of informal tenants are higher than those of the recorded tenants, mainly because the landowners provide resources to the informal tenants or sharecroppers for cultivation, whereas the formal tenants are left without any resources."15/ 23. The tenancy system as it exists today, has "also resulted in an under-cultivation of land by many absentee 'owner-tillers.' They lack the ability to till the land themselves and at the same time are reluctant to rent out to other able tillers for fear that they may claim tenancy rights later. Nor are such owners willing to dispose of their land to invest in other sectors... _161 C. Present Situation 24. While agrarian reform activities under the 1964 Land Reform Act lost momentum, the issue of land reform has not entirely disappeared in Nepal. For example, the IDS produced its study in 1986.17/ A White Paper was produced by HMG during 1987 and 1988, but no action came of it, and it received little publicity or circulation. Similarly, the 1987 Basic Needs Program of HEG includes a reference to the need for agrarian reforms, but no specifics are provided. It is of interest to note that landowners and landlords appear fully cognizant of the possibilities of land reform activity. For example, Bank mission members visiting a rural land registration office in the terai in 1988 were advised that there had been a sudden upsurge of land transfer activities (including those to family members) at about the time the 1988 Land Reform White Paper was being completed in Kathmandut 15/ IDS, The Land Tenure System of Nepal, p. 43. 16/ ibid., p. 34. 17/ ibid., 1986.  _ R - Table 1.1-1: Size Distribution of Operational Holdings Size of Terai Hills Holding Average Percent of Average Percent of Hectares Size Holdinas Land Size Holdings Land <0.5 0.09 45.4? 2.8? 0.18 54.12 11.62 0.5-1.0 0.74 13.0? 6.5? 0.75 18.7? 16.32 1.0-3.0 1.77 27.7? 33.12 1.49 22.1? 38.3Z >3.0 6.15 13.8? 57.72 5.67 5.1? 33.8? Total 1 17 100.02 100.0? 0.86 100.0 z 100.0? Sources IDS, The Land Tenure Systems of Nepal, 1986. 25. It is particularly difficult to obtain accurate information concerning the size distribution of holdings or the extent of tenancies of various types, partly because large landholders and landlords have reacted defensively to the threats to their interests contained in the 1964 Latd Reform Act and in subsequent discussions of impending agrarian reform. Land has been transferred to nominee holders, not always family members, and there are many wunofficialw or unregistered tenancies, as well as those registered under the Act. Further, there are very limited data on the distribution of land ownership, data being collected mainly on operational holdings. Table 1.1-2: Distribution of Tenanted Land Tenant Families Tcnant Cultivated as Percent of Lands as Percent of Source and Date Total Farm Families Total Cultivated Land Agricultural Census of 1961 40.4? 25.5? According to Progress of Land Reforms During 1964-66 28.3t 4.9? Agricultural Census of 1971 18.9g 15.8? Agricultural Census of 1981 8.2? 4.5? Source: Ministry of Land Reforms, HKG/N  - 9 - 26. With these caveats, the distribution of operational holdings is presented in Table I.1-1 and the distribution of tenure status in Table I.1-2. The interpretation of the land holdings data is made more difficult in Nepal by differing definitions of what constitutes an operational farm and by contradictory evidence on landlessness from different sources. 27. The Central Bureau of Statistics (CBS) defines an operational farm as an area of more than 0.01 ha (as an area in which at least 0.25 kg of maize or 1 kg of rice seeds are sown, or where two cows or buffaloes or where five sheep or goats, or 20 poultry are kept). With this definition, the CBS census data shows a declining percentage of landless households in Nepal; being 1.42 in 1961, 1.02 in 1971 and 0.42 in 1981. By contrast, the National Planning Commission provides quite different evidence, as does a sLmple survey conducted by CBS in 1986/87, as cited in Table I.1-3. Table 1.1-3: Percentage of Landless Households Region Percent Landless NPC DSS Mountains 3.7? 11.31 Hills 2.22 7.3? Terai 18.3Z 35.0? All Nepal 10.4% 1?.02 Sources: NPC, HMG, A Survey of Employment, Income Distribution and Consumption Pattern in Nepal, 1983. CBS, HMG Demographic Sample Survey, mimeo, 1987188 DSS data refer to holdings of cultivated lands by rural households. Other observers come to conclusions similar to those of the NPC.18/ It seems clear that the 45.5 percent of holdings in the size class smaller than 0.5 ha in the Terai shown in the land distribution in Table I.1-1, and having only 2.8 percent of the land and an average size of 0.09 ha, include very many effectively landless households. All the evidence, however, points to a low level of landlessness (in the sense of no agricultural land at all) among rural households in the hills. Many, of course, have very small holdings where some cere.Rls can be grown, but much less than the requirements of the family. As an approximation which is consistent with all the data, it would be safe to assume that almost all hill families have some agricultural land, but by no means enough to produce their subsistence requirements, and about 20 percent of terai rural households are effectively landless, in the sense of net being able to grow a significant 18/ As cited in New Era, 1989.  - 10 - portion of their cereal needs.19/ 28. Making this correction to the terai land distribution raises the mean level of holding size there from 1.17 ha to 1.46 ha, compared with 0.86 in the Hills. This accords more with the comments of informed observers than the raw data in Table 1-1.1. It also points to a salient fact about Nepal. With its limited base of cultivable land and high population growth rate, Nepal already has a significant share of its rural population which is landless. With continuing growth in population pressure in both hills and terai, both the numbers and the proportion of the landless rural population are bound to increase. 29. Even with the population in 1981 and the distribution of holdings in that year, there was very little land to re-distribute. An arithmetic example illustrates this point. Thus, the 1964 Land Reform Act proposed a ceiling on land holdings sizes in the terai of 18.4 ha. In its 1986 Report, the IDS suggested an upper limit of 5.44 ha, a much more radical proposal. If all land in excess even of this IDS ceiling were re- distributed, just exactly Otopping up' the land holdings of those holding less than 1 ha, then there would be sufficient 'excess' land to bring only 47% of those in the lowest size class up to 1 ha, 53Z receiving no benefits at all. Most significantly the non-beneficiaries would represent 24% of the total rural population, more than the total number estimated above who are currently landless. Thus, the landless would not stand to gain at all from the land reform. Note that these people would not only not gain: it seems likely that they would actually ' :t.. The reason is that a redistribution of this type would almost certainly lead to a substitution of family labor for hired labor so that the demand for labor services of the currently landless would likely decline. 30. Of course, other variants of a land redistribution are, in principle, possible, including smaller plot sizes for beneficiaries. However, the basic problem is insufficient land for an already large and growing population. As an alternative, consideration is sometimes given to variants of schemes which would involve small farmers in purchasing land from large, who are encouraged 'voluntarily' to comply with land-holding ceilings by some target date. Such schemes are plausibly attractive, because they nold the promise of 'land reform without tears', in which there are no losers. There are, however, three main problems with such proposals. The first is that land transfers, to get under the ceiling, need not be to small farmers; they may well be to family members of the owners of large holdings or to other non-poor people. The second problem is that the poor usually simply cannot afford to pay the purchase price for land, even amortized over many years. For example, in 1986/87 the market price of rice lands (khet) averaged Rs. 124,000 per ha in the hills and Rs. 82,000 per ha in the Terai. Unirrigated (bari) lands averaged Rs. 42,000 19/ The figure of 18 percent for the NPC in Table A-3 relates to 1981, and the population must have increased since that year. Therefore 20 percent seems a conservative estimate.  - 11 - and Rs. 40,000 respectively.20/ At the same time, the net income from paddy, the highest value crop in the Tarai, was Rs. 6,800 per ha, and the annual cash income for a small farmer family was about Rs. 800 in 1984-85 prices. With these land prices and these returns from land there is no way that small farmers could service a mortgage and thus purchase the land on credit.21/ Third, at Nepal's stage of development there are few other reliable income-producing assets to absorb the investable resources of landowners. Similarly, Nepal lacks social security or effective retirement benefits, so that it is common for land to be held as a safety net. Public servants often hold land, as absentee owners, as security against their leaving public service or upon retirement. The absence of alternative investment instruments and social security measures act as a powerful distincentive both for landlords to sell land or for a land reform. 31. The tenancy data contained in Table I.1-2 almost certainly overstate the decline in tenancy. These data show the proportion of farm families who are tenants declining from over 40? in 1961, to about 19? in 1971, and to 8Z in 1981. A few tenants may have gained their land as a result of the 1964 Land Reform; many more were expelled by landlords concerned about creating permanent tenancy rights. There is no way of knowing how many unregistered tenants remain nor how many tenant-like tenures survive. It is known that, in addition to tenant expulsion, some farmers are shifting their tenants from plot to plot each year to prevent their gaining a permanent tenancy right to the land. 32. While official tenancy tended to drop after the Land Reform Law of 1964, some localized evidence reveals that informal tenancy went underground and flourished. Findings from a survey conducted by IDS in two apparently quite representative districts in the terai and two in the hills found the actual incidence of tenancy much higher than that reported in the census. In the hills, the sample found that a total of somewhat over 21? of farm families were tenants; in the terai district studies the figure was about 39Z, for an average of 31? in the four sampled districts. Tenanted land in the hills was 9? of all land, and in the terai nearly 30?, for a four-district average of 25?. The difference between the IDS figures and the "official' White Paper figures was due to "informal" and, hence, uncertified or unreported tenancy. Informal tenancy amounted to about one half of the reported tenancy in both the hills and the terai. Holding size of these tenanted units was about 0.24 hectares in the hills and 1.19 hectares in the terai. Informal tenants operated 68 percent of tenanted land.221 201 Ministry of Agriculture, Main Report on National Farm Management Study, 1983-85. 21/ It is not unusual for farm land values to exceed the productivity of land by a considerable margin, particularly where high inflation rates are expected. 22! IDS, The Land Tenure System in Nepal, p. 50.  - 12 - 33. The IDS sample survey showed that 96X of tenants thought their main agricultural problem to be lack of access to resources while only 25? of pure owners saw that to be the key problem. Informal tenancy is apparently more participatory than certified tenancy. landlords make more contributions to farm and entrepreneurial inputs in informal than in formal tenancy, so prodiction tends to be higher. Seed tends to be provided to informal tenants by landlords in the hills while the costs of harvesting are shared with landlords in the terai. Costs of chemical fertilizers and irrigation are usually shared in both the hills and the terai. Subsistence and some production credit for tenants tend to come from landlords and not from governmental sources. Rent payable to the landlord was usually half of gross produce, though in some areas of the Terai with a regional labor shcrtage, rents were as low as one-third or one-fourth of total produce.231 34. A fixed rent system geared to land quality went into effect in some accessible provinces in the early 1970s. The idea was that tenants would come to regard rent as a fixed cost each year, and whatever they produced over that amount would be a net addition to their profit; fixed rents were incentive rents. However, the IDS survey showed that tenants often thought that rents were arbitrary and unjust, because they regarded the land classification system as not really scientific. Certified tenants were apt to have a much higher percentage of fixed rent and their rent tended to be based on the principal crop; informal tenants were likely to be paying more of their produce in rent to the landlord. 35. More than government regulation, however, labour supply seemed to be a major determinant of the prevailing landlord-tenant relationship. The IDS study notes: "As long as the man/land ratio remains high, even the most sophisticated machinery to enforce rent regulation will meet with very little success. The landlord in such a situation will always be in a stronger position to control the price of his land in the form of rent. Any attempt to regulate rent against the demand and supply equation is destined to be confined to the statutes". 24/ If labor is especially scarce in an area, landlords will court it with larger shares of inputs and lower rental shares; if plentiful, the reverse is true. Where the peasantry is not organized and in situations where the regulatory mechanism of the gcvernment is weak in its reach to the community level, the ability of policy to abolish tenure types or regulate wages or shares is probably ineffectual at best. D. Conclusions 36. It is evident that there are many problems with the present agrarian structure. First, there is unfinished business left from the 1964 Land Reform Act, with registered tenants holding and operating land at low efficiency levels, as shown by the IDS Survey. Second, the perceived threat of agrarian reform and the possibility of giving security of tenure 23/ Ibid, p. 53. 24/ Ibid, p. 56.  - 13 - for life to tenants has led to defensive action on the part of landowners. This has taken the form of rotating plots for tenants, who therefore have no incentives to increase land productivity. Third, there is already a large and growing class of landless, many of whom might become tenants if the laws were changed. Fourth, there are innovations, such as shallow tube-wells, which hold the potential for substantial production increases, but which are not reaching small farmers on account of tenure and collateral difficulties. 37. There are also some special problems of land management in the vicinity of rapidly-growing urban areas, such as in the Kathmandu valley. In such areas, land values for urban purposes are increasing rapidly, but land transactions are made more complex and unwieldy by existing tenancy rights on land. These issues have not been studied in the context of the present report, so it is not possible to suggest appropriate means of resolving the problems. However, it is recommended that these land policy issues on urban fringes be treated as a separate problem from policy towards rural land proper. In this way, issues of agrarian reform can be considered in their own right and separate from issues of urbanization policy. S8. For the reasons discussed earlier, a redistributive land and agrarian reform may not currently be a prospect in Nepal. In the absence of such a reform as a first-best solution, a second-best solution needs to be sought. Elements of such a reform, which should be implemented as a matter of urgency by HMG 25/ include the following: i) Completion of the 1964 Reform. At present there are about 200,000 registered tenants under the 1964 reform who have security of tenure, whose presence is blocking land transactions by landlords, and who are utilizing the land even less productively than unregistered tenants.26/ The IDS has suggested that such tenants should receive one half of the land they cultivate free of charge, leaving the remainder to the landlord. It is suggested that the latter would gain from this transaction by being able to sell his remaining land for a higher total return than for the larger original parcel of land partially encumbered by tenancy. The evidence upon this matter is not conclusive. However, the principle seems correct, and it would permit a substantial portion of land to return to full production, 25/ The urgency derives from the fact that population pressure on the land proceeds unabated, and any delays in implementing reforms make doing so even more difficult. 26/ The reason for the inefficient land use by registered tenants is that the latter usually supply all the inputs, but share half the output with the landlord. They thus tend to equate the full marginal cost of inputs to one-half the marginal return, and use input levels much lower than socially optimal ones.  - 14 - to be used as collateral for loans, and to gain again the flexibility of use associated with normal land transactions. (ii) Legalize alternative tenure arrangements other than full lifetime security of tenancy either on the basis of a fixed rental (which may be cash or kind) or of share cropping, which, for efficiency, must provide for the same sharing between landlord and tenant of cash inputs (such as fertilizer) as of output. In this connection, it is of interest to note that the IDS survey found that it is more common to share both inputs and outputs between tenant and landlord in informal than in registered tenancies.271 Freeing up the tenure market would permit some of the now landless greatly to improve their well-being by becoming tenants rather than being solely landless laborers. Such flexibility would also allow greater accommodation of the increasing pressure of people upon the land, which is inevitable for many years to come. It would be essential to expand credit facilities to groups of new tenants and other small farmers in the terai under such group credit arrangements as the SFDP, to permit group purchase and operation of small tube-wells. (iii) Bring the 1964 Land Reform to a clear and unambiguous close. In this way landowners would receive the assurance needed to take on more tenants and thus work the land more intensively, and to embark upon investments in production - enhancing land improvements. This would go a long way to overcoming the major problem which exists at present: landlord, tenants, and would-be landlords are all using land inefficiently and inflexibly as a defensive reaction to a land jeform which is perceived as a threatening possibility but is, in fact, never carried to completion. (iv) Streamline the procedures and lower the costs for registering and storing information on land transactions. Technical assistance should be sought on how best these improvements can be implemented in Nepal. The package of changes suggested here will increase landowners' willingness to register land transactions. Doing so and clarifying title will permit land to be used more readily as collateral to support borrowing for such land-improving investments as tubevells. 271 IDS op. cit., p. 51-52  - 15 - ANEX 1.2 - LABOUR FORCE ISSUES A. The Labour Force 1. The two most notable features of the labour force are its rapid growth, and the increasing participation of females. Table 1.2-1: Nepal Labour Force 1981 and 1989 (millions) 1981 1989 (est.) Male Female Total Male Female Total Population 7.7 7.3 15.0 9.4 9.0 18.4 Economically Active 4.5 2.4 6.9 5.7 3.4 9.1 (over age 10) Estimated LF - Participation Ratest 832 46? 652 83Z 571 70S (aged 10 and over) Source: 1981 Census; 1989 mission estimates. 2. In the intercensal period 1977-1981 the economically active population grew by 3.32 p.a. - the result primarily of a 52 p.a. growth rate in female participants. This is clearly illustrated by the changing female LF participation (from 352 in 1971 to 46? in 1981). This increase has been fueled by changes in the terai - where female LF participation rates doubled from 16? to 31? in ten years. Labour Force Composition 3. According to the 1981 census, about 86Z of the work force were self-employed, of whom 972 were engaged in agricultural pursuits. It is difficult to break down the sectoral composition with any certainty, since there are wide variations in census and survey data, due largely to discrepancies in the way in which primary occupations are reported.l/ None the less the following broad breakdown emerges from a number of sources. 1/ Many of the economically active are farmers with exceedingly small plots, who of necessity spend a substantial minority of their time engaged in petty trading, portering, and construction labour - and thus, depending on the phrasing of the question may classify themselves as either farmers or other kinds of workers.  - 16 - Table 1.2-2: Estimated Sectoral Breakdown of the Labour Force Agriculture 862 Manufacturing 2? Construction, Utilities. 22 Transport Commerce 3Z Services 5S Others 2? Source: Mission estimates from various sources. The Multi-Purpose Household Budget Survey (1984185) reports a much higher incidence of non-agricultural employment (just over 20Z) than does the 1981 Census (about 1O?). Two explanations are possible: either there has been a substantial sh.ft out of agriculture (a plausible hypothesis given the saturation of land and the rapid subdivision of farms), or the disparity is accounted for by differences in recording occupations. The truth probably lies between the two - with the MPHBS more accurately reflecting the distribution of secondary activities, and the census probably overstating the proportion of farmers. There probably has been a shift into non-farm activities (although at nowhere near the rate suggested by the difference between the two sources). 4. It is perhaps more instructive to look at time use and the relative contribution of various activities to income. Data from one survey suggests the following broad breakdown of time use on economic activities: Proportion of Time Allocation Subsistence: Agriculture 53S Kitchen Gardening lS Animal Husbandry 8? Hunting & Gathering 5? Food Processing 14t Household Manufacture lS Non-Subsistence: Trading 5? Wages & Salaries lit Source: Acharya and Bennett (1981).  - 17 - That is, subsistence activities accounted for about 802 of economic activity, and wage employment for only a bit more than 10X. A similar picture emerges from the occupational breakdown in the Multi- Purpose Household Budget Survey (NPHBS) viz.: Table 1.2-3t Classification of LF by Employment Status La Self-Employed 29Z Unpaid Family Member 44Z Employer 1? Employee - Government 32 - Private 232 a/ Percentage of economically active population 10 years or over. When one starts to decompose employment by income class, a picture emerges of the poor being relatively more dependent on wage income. One sample survey in the hills2/ found that while both poor and non-poor were largely dependent on subsistence activities, the poor were engaged in portering or other wage activities to provide almost a quarter of their income. whereas the relatively wealthy relied on employment for only a twentieth of their income. Similarly, the MPHBS data indicates that the rural poor rely on wage employment for between 232 and 32Z of their income, as opposed to only 10-15S among the non-poor. 21 Rasuwa-Nuvakot Rural Development Program.  - 18 - Table 1.2-4: Occupations of the Rural Poor (Proportion of Economically Active Within Categories) Rural Hills Rural Terai Male Female Male Female Professional, Technical and Managerial 1.52 0.4? 1.5Z - Office Workers 1.9? 0.5? - Sales and Service 3.62 0.8? 5.7? 3.6? Agriculture 72.0? 87.0Z 78.3? 87.4Z Of which: Farm Labourers (4.9) (6.0) (27.7) (28.0) Production Workers 5.9? 3.2? 3.82 2.4Z Construction, Transport and Communications 15.12 8.7? 10.2? 6.6? Of which: Casual Labourers (13.6) (8.7) (8.9) (6.6) Sources MPHBS special tabulations - see Statistical Appendix for details. Again the data here may overstate the proportions engaged in full-time off- farm employment, because of the multiplicity of casual employment in secondary occupations. 5. Not surprisingly, the poor are relatively less engaged in formal sector employment than the non-poor, and to the extent that they are working for renumeration, it is overwhelmingly as agricultural labourers in the case of the terai, and generally as casual labourers in the case of the hills. To the extent that gender differences arise, poor women are concentrated much more in subsistence agriculture, and to the extent that they do participate off-farm, it is almost exclusively as labourers. Unemployment, Underemployment, and Participation 6. Estimates of unemployment vary, but generally lie in the range of 6X. To the extent that any clear trends emerge, unemployment appears to decline with farm size (reflecting a lower dependence on off-farm employment, which may often not be available). However it is not very meaningful to talk of unemployment when the vast majority of the work force are subsistence farmers, and most labour market relations take place outside of the cash economy.  - 19 - Table I.2-5: Proportion of Classified Working Days Spent in Non-Farm Wage Employment Landless 27? 0.0 - 1.0 ha. 17X 1.0 - 2.0 ha. 13? 2.0 - 10.0 ha. 72 10 ha. + 52 Source: Acharya: A Study of Rural Labour Markets in Nepal, 1987. 7. Time use estimates of underemployment are perhaps more intructive. Studies vary greatly in their estimates, generally in the range of 45Z - 652, and most are open to question methodologically - but a.1 indicate that the degree of underemployment is high. A number of not very surprising conclusions emerge; that: - underemployment is higher in rural than urban areas; - it is generally higher in the hills than the ter.i; - the estimated underemployment rate is higher for females than for males (although this may be based on misinterpretation of female time use); - underemployment of female workers is highest in the rural terai; - and that of male workers is highest in the rural hills and lowest in the urban terai. 8. These conclusions are all consistent with the relative availability of off-farm work (higher in urban areas and the terai); variations in the intensity and seasonality of cropping (leading to more days worked per year in the terai); and to cultural constraints (which, for example, prevent women from participating as fully in employment outside the home in the terai and in urban areas). 9. The relatively low rates of labour force participation, and high reported incidence of underemployment tend to disguise a lot of economically useful activity - especially among women, and especially in thp hills. It is estimated, for example, that 102 of waking hours in the hills are spent in collecting fuel, fodder, and water. The following table from the MPHBS illustrates this point:  - 20 - Table 1.2-6: Rural Time Use (Hours per day) Hills Terai Male Female Total Male Female Total Conventional Economic 4.2 2.7 3.4 5.1 1.7 3.4 Agriculture 2.6 2.2 2.4 4.0 1.4 2.7 Other 1.6 0.5 1.0 1.1 0.3 0.7 Subsistence Economic 1.6 2.5 2.1 1.0 1.9 1.4 Gathering Fuel, Water, etc. 1.3 2.O 1.7 0.7 1.9 1.1 Food Processing & Other \a 0.3 0.5 0.4 0.3 0.5 0.3 Domestic Work \b 2.1 5.4 3.9 1.8 5.8 3.8 Other Activities \c 2.4 1.2 1.6 2.1 1.5 1.9 Self Transport 1.8 1.3 1.6 1.6 1.3 1.4 Leisure 3.9 2.9 3.4 4.4 3.8 4.1 16.0 16.0 16.0 16.0 16.0 16.0 Source: MPHBS. aI Includes hunting and gathering, house repair, etc. b/ Includes cooking, cleaning, laundry, child care, shopping, etc. c Includes education, personal care, religious and social activities. 10. The reported average work burden of 9 1/3 hours per day raises questions as to whether underemployment rates are really as high as reported. On the other hand there is no doubt that there is substantial seasonal underemployment of males in subsistence agriculture; that the proiuctivity of a lot of labour is very low; and that in the absence of alternative productive uses of time, a lot of these tasks are probably performed in a leisurely fashion. There is certainly no question that except in some areas, at some times of the year, there is very little in the way of off-farm employment opportunities. One survey found, for example, that with wide seasonal variations in labour demand, casual agricultural labourers are able to find employment only 68Z of the time that they seek it.31 B. Employment Arrangements 11. To the extent that they are employed outside of subsistence agriculture, the poor are engaged overwhelmingly in agricultural labour (30Z of the economically active poor in the terai), or as casual labourers - mostly in construct~'n (about 0l in the hills). 3/ B. Pokharel and G. Shivakoti. Impact of Development Efforts on Agricultural Wage Labor, WINROCK, 1986.  - 21 - 12. The types of agricultural employment relations, and the methods and levels of payment vary greatly across and within geographical zones. They are particularly different in the hills and terai, as a result of populatin -i pressure (and hence land pressure, degree of landlessness, and land tenurt- arrangements), caste and ethnic factors, and the availabilitv of off-fatrm employment opportunities. Basically, in the hills agricultural labour is exchanged between families, whereas in the terai it is employed for paywmI, in grain, usually under long-term contract, but also on a daily wage bas iF. 13. In the hills one-to-one labour exchange is the usual means ot supplementing farm labour. Under such exchanges (known as perma) days of work are exchanged on an equal basis between families of more or less equal farm size and social standing, and those providing labour are provided A?.itl a mid-day meal. This system is a function of a long-established set of social relations, combined with the lack of a cash economy, and the low mobility of labour (due to difficult access) - in short it functions in place of a wage labour market. The system works since farms sizes are fairly uniform, periods of peak work do not necessarily coincide, and it allows economies of scale in labour use on various farm activities. Somne anecdotal evidence suggests, however, that poorer farmers do not participate as actively in perma exchange as do the moderately poor. 14. Long-term labour contracts are more common among large landowners, both in the hills and the terai; although there are relatively few large landowners in the hills, where the contracting of labour is limited mostly to Brahmins (who cannot plow the land themselves due to caste constraints), and to plowing activities. 15. In the terai there are basically two types of long-term contracts: - attached labour which works exclusively for one landlord; and - semi-attached labour which must work for the contracting landlord when he needs them, but is free to work for others at other times. Both are usually contracted for a period of one year at an agreed price. 16. Attached labour may be either residential or non-residential. Residential workers are provided with some food, clothing, occasionally shelter, and a fixed amount of grain. Non-residential workers are paid a larger amount of grain, plue salt, oil, and other foodstuffs. Both are required to work for one employer only. Such employees are often provided with a loan sufficient to carry them through the agricultural year (reported at about Rs. 500). In the western terai there exists a form of bondage whereby such loans may bind the poor inescapably to 8 given landlord. 17. In the eastern terai semi-attached arrangements are more common. Labourers are attached to a particular employer, but are paid a daily wage only for those days worked. The difference between these and casual workers is that they get access to credit, sometimes use of livestock, and often use of a very small piece of land on a sharecropping basis (which one study  - 22 - estimates may contribute up to 25Z of the family's income). Semi-attached labourers must work for their employer when needed, and are free to seek work elsewhere (although in practice such work may seldom be available). 18. The implicit level of wage earnings across types of contracts seems to have settled at an equilibrium level, once one accounts for the value of food and other resources consumed. The imputed value of the daily wage is estimated by one study to be about 30X lower than the agricultural wage in the same village. However the access to a piece of land (however small) provided under most attached labour contracts, combined with the opportunity to use other family members to work with the employer's livestock, appear to be strong incentives for the landless poor to seek out these contracts. 19. Casual agricultural labour, whereby labourers work for a daily wage or at a piece-rate, occur throughout the country, although more commonly in the terai than the hills. Wages are paid on a daily basis, in either grain or cash, along with a meal. The mission could find no reliable indication of the relative preponderance of casual as opposed to long-term 1-Dour contracting. 20. A final form of long-term employment arrangement is kami, -hereby an annual payment is made by landowners to a member of the occupatinal castes (eg. a blacksmith) for services to be provided throughout the year (eg. production and repair of agricultural implements). Anecdotal evidence suggests that this form of contracting is becoming less prevalent with the spread of the cash economy and manufactured goods - often with very serious income impacts on households from the occupational castes. 21. Non-agricultural employment of the poor is generally in construction or (to a much lesser extent) cottage industries. There is little documentation of the labour arrangements under which such labour is engaged. However anecdotal evidence suggests that a number of practices prevail which reduce the flow of potential benefits to the poor from this sort of employment. 22. Most of these involve a form of contractor 'squeeze' which in one way or another reduces the proportion of the amount spent on the works going to unskilled labour. It is reported that on some large contracts (eg. for road works) labourers are either not paid the wages owing to them, or are "advanced" their pay in the form of subsistence meals, for which they are then charged the full amount of their wages - so that they are indebted to the contractor and effectively caught in a form of bondage - the meals being provided sometimes insufficient to avoid malnutrition. The lack of political power on the part of the poor reportedly prevents them from effectively demanding the wages due to them. 23. Corruption and kickbacks in contracting reportedly amount to between 102 and 302 of total contract costs. Apart from representing an inefficient use of public resources, these funds constitute a margin going to 'profit', broadly-defined, which could be available as labour incomes - either in the form of more man-days of work for the same public works budget, or higher wages. The local organization of small works (eg. under various NGO, IRDP, and local government schemes) has had some success at capturing this additional margin.  - 23 - 24. One common way in wvhich such margins are squeezed is by contractors being awarded jobs costed on the basis of a "District Wage", which is supposadly paid on all public works, and then either paying the local agricultural wage (usually 25? less) and pocketing the difference - or employing women and children at lower wage rates, for works costed on the basis of inputs by able-bodied men. Projects such as the Special Public Works Program (see Annex II.2) have overcome this problem by more intensive inspection and supervision, along with the use of very small "petty- contractors", who come from the district involved. 25. Finally, contractors often reportedly use gangs of Indian labourers - who have been organized as a group by a labour contractor and imported from a low-wage area of India. They apparently work for lower wages, are more easy to manage (as the labour contractor provides a degree of group management), and of course lack any local political base (however weak) from which to demand improved conditions. While this makes perfect sense from the contractors' point of view, the net effect is to reduce the potential employment effects of construction reaching the local poor, and to further bid down local unskilled wage rates. C. Wage Levels 26. Wage is a complex phenomenon in Nepal, since there are many different modes of payment. Labourers are generally paid in kind with grain and enter into a range of land and credit relations with their employers. Although there exist great variations in wage rates (especially between the hills and terai), evidence suggests that the market works fairly well at equalizing wages locally across various forms of labour and contractual arrangements. Ignoring for a moment the vast range of variations, all of the evidence suggests that average wage levels are currently about Rs.23 per day in the hills, and about Rs. 17 per day in the terai.41 27. The following tables provide a good indication of the current level of wage rates: Table 1.2-7: Some Selected Wage Rates (1988) Rs./day Agricultural Wage Labour: Rs. 10-15 Construction & Public Works: Rs. 15-30 Cottage/Small Industries: Rs. 15-30 Agricultural Processing: Rs. 25-50 General Portering: Rs. 20-50 Source: ERL - from Interviews 41 These are an average of daily wage rates for agricultural labour and unskilled construction work - the relevant rates for the poor.  - 24 - Sufficiency 28. There is a remarkable uniformity across surveys of wage rates at around 4.25 kg. of paddy equivalent. If one translates this into calories, adjusting for days worked, dependency ratios, and conversion to consumable food 5/ it comes to about 1850 calories daily per capita - about 15? below the estimated minimum requirement of 2250 per day. 29. In terms of cash income, these rates are equivalent to per capita incomes6/ of Rs. 2130 per annum in the hills ($76 p.a. in 1989 US dollars) and Rs. 1575 in the terai ($56 p.a.); that is, Rs. 178 per capita per month in 1988/89 for the hills (relative to a poverty line of Rs. 210 per month - see Chapter II), and Rs. 131 in the terai (poverty line of Rs. 197 per month). To get a perspective on this - even if we assume that the poor were employed a full 210 days a year (which they almost never are), this would translate into a per capita income of only $5.30 per month in the terai and $7.85 in the hills. Wage Trends 30. Data on wages is patchy and inconsistent, so it is difficult to accurately assess movements over time. One study (Winrock: 1987) documents a real decline of 50Z over the 1970's (and of between 20? and 33? over the period 1976/77-1984185) using wage data from government agricultural stations and neighboring villages. Another (ILO/ARTEP) estimated that in rice equivalent terms wages fell by 30? over the period 1969-1977. A good micro study of one village in the terai found a once-off increase in 1980 when there was substantial political pressure because of the impending referendum, and fear of reform had landlords temporarily on the defensive but it was reportedly the only increase anyone could remember. In the terai there has also been a reduction in the effective payment to semi-attached labourers by reductions in the size of garden plots made available to them for use on a share-cropping basis (by about 302). All of the studies agree that the tendency to shift from payment in kind to cash has resulted in an erosion of wages - which used at least to be paid in a fixed amount of grain. Our own work compiling data from a variety of sources suggests that the average wage was about Rs.5/day in the early to mid-70's, and about Rs. 10/day in the mid 1980's equivalent to about Rs. 22 and Rs. 15 per day respectively (in constant 1987/88 prices, compared to about Rs. 20 today). Variations 31. Nominal wage rates in the hills seem to be consistently about a third higher than in the terai. This is accounted for to some extent - but not entirely - by differences in the price structure - under which a comparable basket of goods costs about 30? more in the hills. The persistence of such a consistent difference at first appears puzzling, in that one would expect the levels of wages to be driven by labour supply and 5/ Assuming 185 days worked per year, a dependency ratio of 2.0, 502 conversion to food, and 3450 callkg. of rice 6/ Making the same adjustment for days worked and dependency ratios.  - 25 - demand relations, which may or may not correspond to the difference in price structure (which is determined mostly by transport cost differences). Such a variation would however be consistent with a truly subsistence wage which has in both regions been bid down to the minimum. 32. As would be expected in a rainfed agricultural economy there are significant seasonal variations in the wage rate, reflecting periods of labour scarcity. During the peak periods daily wage rates may increase by 251 or more (although the intensity of labour, and hours worked may also both be greater). Seasonal variations are reportedly most often reflected in the food component of wages, with the cash component remaining fixed. 33. The nominal rate for female workers on average appears to be 15-207 below that for men. However women in many instances work one or two hours shorter, and there is often a division of labour which allocates less heavy tasks to female workers. Reportedly when piece work rates are paid there is no difference between male and female rates.7/ A case study of agricultural labourers in the terai reports the men as working about 9 hours a day for a total 4.5 kgs. of paddy (including two meals provided equal to 1.5 kg.) resulting in an average wage rate of 0.47 kg./hr.; while women typically worked 6.5 hours for 3 kgs. of paddy - with no meals provided - yielding a rate of 0.46 kg./hour. However in other areas observers suggest that the differential nominal wage rates are not offset by differences in intensity of work. From the available evidence it is not possible to categorically determine whether women are paid real lower wage rates from men, however these examples suggest that one should be careful about generalizing from the lower nominal wage rates without good information on the adjustments for hours and type of work. D. Labour Supply/Demand Issues 34. There are no good data on the demand for labour in Nepal, although crude estimates (below) suggest that in aggregate the available supply substantially exceeds the demand for agricultural labour. Table 1.2-8: Rural Labour Supply and Demand (million person-days p.a.) Estimated Rural Labour Agricultural Supply /a Labour Demand /b Hills 359 261 Terai 398 204 al Supply based on active 1990 labour force, net of hours spent on domestic tasks (average 62Z), and assuming 200 work days per year - this is a conservative estimate of supply. b/ Demand based on labour intensity per ha. by area cropped to each major crop. 7/ S.P. Sharma, Agrarian Change and Agricultural Labour Relations Winrock, 1987.  - 26 - Certainly the high estimates of underemployment, the fact that labour force growth has exceeded GDP growth, the reported declines in wage rates, and village-level anthropological studies, all suggest that labour is available in surplus. 35. As pointed out earlier, the labour force is growing at about 3.3? p.a. on average - although it appears to be growing at a rate of about 5.52 in the terai, and at over 20? p.a. in urban areas. Some of this labour is being absorbed in agriculture; however a number of contradictory forces influence the rate of absorption: (i) the reduction in farm-size increases labour intensity per unit of land, (although the smaller farms tend to be worked by family labour); (ii) the shift to irrigation and high yielding varieties, and the use of fertilizer, result in more intensive use of labour. Even though these inputs are under-utilized, there has nonetheless been a substantial effect. Some localized studies in the terai estimate the demand for labour to have increased by 50 as the result of intensified cropping. (While the addition of a wheat crop in some parts of the terai has increased cropping intensity and thus the demand for labour and total labour income, this was generally found to come about through working longer hours or more days rather than increased wage rates); (iii) as a consequence of the 1964 land reform, landholders have reverted to self-cultivation, thus requiring more labour, while at the same time tenants have become wage labourers. 36. The most dramatic effect of labour force expansion is dispossession and permanent out-migration. With increasing population and decreasing landholdings subsistence farmers go deeper into debt, undertake longer term migration (eg. to India), and enter into debt-peonage relations. Smaller holding sizes also reduce the capacity of patrons to pay for traditional caste services. Consequently lower caste families, who previously provided services and labour, are forced to migrate permanently to the terai, where they have the least access to opportunities of new in-migrants. The net result is that in the hills out-migration has balanced labour force growth; but in the terai a class of landless agricultural labourers has emerged (as a result of in-migration from both the hills and India, and increased landlessness in the terai itself) which has far exceeded tne demand for labour. 37. The apparent surplus of labour is inconsistent with persistent stories of labour shortages in the terai - both in agriculture and in construction. Many studies (New Era, the Winrock series, Acharya) suggest that almost all households (even those with very small holdings) hire in labour at some stage, and that there are labour supply constraints on the expansion of agricultural output. Most likely it is the very high peaks in labour demand associated with rainfed agriculture which are being observed. Such peaks last from 24 hours (for rice replanting) to 2-to-3 months (harvesting), and they almost certainly correspond to the lowest point in the labour supply cycle - since these tend to be the times when seasonal migrants are fully occupied closer to home, in either the hills or in northern India. 38. Also, there are some demand-side rigidities in the use of labour - for example numerous micro-level studies report that landholders in the terai are relatively unwilling to use Nepali labourers who may claim  - 27 - tenancy rights, and eventually lay some claim against their ownership of land under agrarian reform measures. Similarly it is claimed that managers prefer the use of Indian contract labour rather than risk political and managerial problems with indigenous workers. 39. Analysis of household labour supply decisions81 shows that while most households participate in the labour market (69? of sample households) the proportion of time hired out is low (192 for men, and only 6S for women). Determinants of participation (as expected) include landholding and asset structure, family composition, and cultural factors such as caste and ethnic group. Land holdings appear to play a more important role in participation decis4ons than wage rates, suggesting a marked preference for on-farm work; and the influence of non-economic factors (eg. ethnic group membership, family size) is as strong as land ownership or wage rates in determining labour force participation - especially by women. The conclusion is that there are distinct rigidities in labor supply, mostly cultural, which probably hinder flexibility. 40. Other indications, however, suggest the labour market is more elastic then is commonly supposed. Once the decision is made to participate, women show a marked response to wages in determining the hours that they will work - suggesting conscious trade-offs are being made between the returns to household work and off-farm work. Surprisingly the number of children below the age of nine has a positive effect on the number of hours females were willing to work outside the home - probably because younger children contribute to household work thus releasing women for market employment. Equally surprisingly, the number of children ten or older was found to have a negative impact on hours supplied - possibly because market work is shared with older children. The findings suggest that the determinants of female labour force behavior are more complex then usually supposed, and caution should be exercised in proposing simple linkages between, for example, fertility or education and labour force participation. E. Migration 41. Seasonal migration for trade has always taken place; however, large-scale seasonal out-migration is a relatively recent phenomenon. Traditionally one or more family members would migrate briefly to earn the small amounts of cash necessary to purchase cloth, salt, etc.. Recently short-term migration has taken on much more of a distress nature - as remaining plots of land cannot provide adequate food to support families, and thus the males outmigrate in the slack season so that both they and the family members left behind can adequately feed themselves. Anecdotal evidence suggest that remittances from such migrants are very small or non- existent. 42. For at least four months of the year there is little agricultural work in the hills, and during this period male family members migrate in search of casual work while females stay behind to look after the house and farm. It is estimated that in some parts of the hills as many as 70? of the 8 M M. Acharya: Rural Labour Markets in Nepal. University of Wisconsin, Ph.d. Thesis 1987.  - 28 - males migrate seasonally. (Village-level surveys typically find between 30-50 of households with at least one seasonal migrant).9/ Those migrating are overwhelmingly male, and most frequently in the 25-30 age range. Since harvesting takes place slightly later in the terai, some of them find agricultural work, but most are engaged in forestry operations, road, and other construction. Some (higher caste) also migrate seasonally to work land they own in the terai, but these are a very small proportion, and available evidence suggests that seasonal migrants belong to the ezc-iomically and socially disadvantaged groups. There is no good in,7ormotion on the consumption impacts on those migrating or on those left bohind - but evidence suggests that the migrants lead little more than a ha.A- to-mouth existence. 43. Long-term temporary migrants (over 6 months) go overwhelmingly to Jndia, where they work in the cities as watchmen, or in Assam, Himchal Pradesh, and the border areas as construction labourers, porters at coal depots, or in forestry. Long-term migrants do bring back cash savings, as well as cloth and other consumer goods. Such migration may be once-off or recurrent. In general the low caste (and poor) migrate more frequently, to: (i) repay debts; (ii) fill their needs for cash to buy animals and cloth, and (iii) make up food deficits. High caste men, on the other hand, may make a single trip to India at the time they leave their parents' home, in order to save money to start their own household. 44. The relative importance of seasonal vs. longer-term migration dopends on the landholding and labour market characteristics of the place of out-migration - especially the degree of seasonal variation of labour demand due to cropping patterns and such factors as the availability of irrigation. McDougallO/ contends that although long-term absence is indirectly related to land ownership status and caste, the critical factor seemed to be household size. Although Caplanll/ points out that increased opportunities for wage labour around a village does not seem to affect the numbers going to India. Villagers claim that local work is sporadic, and that cash earned on a daily basis tends to melt away on consumption. Work in India pays slightly better, always pays in cash rather than kind, and is also paid on a weekly or monthly basis, making it easier to save. 45. There exists no good aggregate data on the importance of remittances in incomes of the poor. In financial terms it seems that migration is more of a coping strategy than one for poverty-alleviation. The Rastra Bank estimated that only 8% of the cash income of rural hills households came from remittances; and one localized survey suggests that the relevance of remittances is lower for the poor than the non-poor: 91, ERL. 10/ C. McDougal: Village and Household Economy in Far Western Nepal, Tribhuvan University, 1968. 11! P. Caplan: Priests and Cobbers: A Study of Social Change in a Hindu Village in Western Nepal, 1972.  - 29 - Table I.2-9: Sample Estimates of Remittances by Migrants Remittances as Average a Proportion of Income Remittances Low Income 2-52 Rs. 120-210 Medium Income 5-llS Rs. 340-680 High Income 18-241 Rs. 1190-2490 SOURCE: Report on Rasuwa-Nuwakot Rural Development Program (1982) as reFt'rted in New Era, op. cit.. Temporary migration is none the less an important coping strategy - especially in some areas, and for some families. One sample of 3 Districts found 25-30S of households to be "solely dependent on seasonal migration for supplementary income' (i.e. presumably meaning that they had no other source of off-farm income). Savings, however, especially from seasonal migration, are used for current expenditure, and are hardly ever invested. Caplan emphasizes that in the villages she studied, temporary migration (except to join foreign armies) does not enable a family to raise its economic standing substantially. 46. Army pensions and income are very important within given ethnic groups (and thus in particular areas), but are not as relevant overall to the incomes of the poor. Within these groups (Gurung, Tamangs, Magars, Limbu, and Rai) army remittances have traditionally been the major source of cash income, although their relative importance is declining where there exist roads and bazaars. In the households affected, of course, it makes a staggering difference in income. (One study, for example, estimated that without army pay, households in one Gurung village would have had average cash incomes of Rs. 70 per month instead of Rs. 462 (1976)). The second round effect is generally felt in their propensity for agricultural innovation, inclination to hire labour, and capacity to serve as moneylenders. Since these are often tightly-knit social groups, the spin- off effects may not be very large. 47. Permanent Migration. The terai was originally land-abundant and labour short (in the 1960's). and consequently the initial hill settlers encouraged further migration to provide labour. (It is claimed, however, that many of these came from India.) It is difficult to accurately quantify the scale of internal permanent migration, but estimates based on the differences between natural and actual rates of population increase suggest out-migration from the hills at an annual rate of about 1.2Z (i.e. half of the marginal population growth) during the 1971-81 intercensal period. 48. It is not clear how the income effects of permanent migration are distributed. Studies suggest (not surprisingly) that it is the relatively poorer who out-migrate, but that once they arrive in the terai it appears most end up at the bottom of the economic heap - as landless labourers  - 30 - existing on the margins of survival. Thus they go from a completely untenable situation (in the hills), to a not much better one in the terai - an improvement, but certainly not a poverty-alleviation strategy. Admittedl- these conclusions are based on a few micro-level studies, and this is an area where further research would be useful. In aggregate, given the current state of development, the terai is already saturated, although estimates suggest (see Chapter 3) that there exists substantial unexploited potential. If this is the case then the marginal existence of new immigrants may be a temporary phenomenon, which can be overcome with, for example, more effective irrigation and the opening up of new areas. 49. Data on the scale of international migration is inconclusive, although there is no doubt that the labour market in the terai (and elsewhere in Nepal) is affected by substantial inflows of Indian labour. The only quantitative analysis (based on census data) suggests that net international migration is approximately zero, but there is probably very substantial under-reporting of both in- and out-migrants. At any rate the pattern of international migration is undoubtedly complex. 50. The 1964 land reform inadvertently encourages landholders to either resume cultivation themselves, or to look for tenants who would not claim rights to the land. Both of these responses to land reform created incentives to use Indian labourers and tenants rather than hill people. This has resulted in a paradox - that hill people go to India, while Indians migrate into the terai. There is thus a sort of swap going on: Nepalis tend to go to Bombay, Delhi, and north India (where the pay is probably higher), while Indians migrating into the terai tend to come from across the border (UP, Bihar) where unemployment is high and wage rates low. 51. The free flow of labour has been a factor at least since the days of British India, when it was consciously encouraged by both parties. During the twentieth century Nepal has attempted to reduce the inflow, but to little effect - both because restrictions are physically difficult to enforce, and because there exist strong economic incentives for non- compliance. It is difficult to see what Nepal can do in this regard. The use of obvious restrictive measures which are successful in other countries (eg. work permits) invite counter-vailing measures from India. The regularization of tenancy arrangements (see Annex I.1) may go some ways towards removing incentives, but would be unlikely to have a dramatic impact. 52. The Goverunent has a number of options to reduce the impact of the inflow of Indian labour. It could close the border, require work permits, visas, and/or the registration of Indian workers. All of these have been tried in varying degree, with little success. Closing the border is physically difficult, the other measures are difficult to police, and all invite counter-measures from India. A number of measures could have a limited impact by reducing the incentives to use imported labour: eg. regularizing land tenure in the terai (to remove the fear of land transfer to Nepali workers), improving skills training and organization of Nepali labourers (eg. through measures to encourage local labour contractors).  - 31 - F. Further Work 53. The analysis has suggested a number of other areas for further investigation, they include: (1) the quantitative effects of changes in agricultural intensity on labour absorption (what is the best we can hope for in increased labour absorption?, which improvements have the greatest labour-utilization impact?), and the mechanics of labour absorption in agriculture (i.e. how the poor as opposed to the non-poor are affected); (2) the consumption effects of seasonal migration - that is, are those left behind (the women and children) faring substantially worse than those who leave?; what is the scale of remittances?; are there measures which can be taken to effect the sharing of benefits?; and, (3) the distributional effects of permanent migration - the question here is to what extent is permanent migration really a poverty-alleviation strategys how well do migrants fare in income terms? relative to their pre-migration status? relative to the society (urban or terai) which they enter? What are the barriers they face, what (if anything) can be done to remove them? All of these are areas where further research could pay off.  - 32 - ARMOX 1.3 - DEBT AND INDEBTEDNESS A. Introduction 1. Prior to 1949 Nepal had no organized system of agricultural credit, and rural households depended exclusively on informal _redit sources. Although these informal systems varied widely between (and even within) localities, the general impression from both historical and ethnographic accounts is that most worked against the poor. Indeed, one rarely encounters references to *moneylenders" in analyses of Nepal's agrarian system without connotations of exploitativeness. 2. To what extent is this characterization accurate in contemporary rural Nepal? Does indebtedness within the informal system result in loss of assets - or sometimes even bonded labor - among significant numbers of the poor? And, where the consequences of indebtedness are less extreme, to what extent does it nevertheless place serious constraints on the political and economic responses of Lsie poor, thereby limiting the potential for equitable growth? Has the introduction of formal credit provided viable alternatives to the poor? Are they getting formal credit? How are they using it? And to what extent does debt service payment (whether to formal or informal sources) reduce the income/consumption levels of poor households? 3. Definitive answers to these questions -- especially the first three -- are not possible here. However, a review of available ethnographic, historical and legal literature permits a better understanding of the complexities of traditional informal credit systems and some tentative answers to the first three questions. Review of macro-level data and special surveys conducted by the Nepal Rastra Bank and a number of other sources does permit us to answer with some confidence the last set of questions. B. Informal Credit Systems in Nepal Sapat Loans: The Traditional Safety Net 4. Historical sources, linguistic usage and Nepal's rich ethnographic literature provide some important clues on how to conceptualize the country's extremely complex and varied informal credit sector. A good starting point is the important distinction between sapat and rin which prevails throughout Nepal. A sapat is an interest-free loan between friends and neighbors or relatives, while rin is an interest bearing loan. If grain is borrowed as sapat, it is returned in kind, in the same measure. Any additional payment or *interest" (byaa;) is up to the borrower. Sapat loans are truly informal as no bonds are exchanged, written records are not kept and the repayment schedule is flexible -- though it is generally expected to be short term. The consequences of non-payment are limited to the souring of social relations, loss of future borrowing power with the creditor and a lowering of prestige in the community. 5. Though difficult to document, sapat networks are extensive in rural Nepal. One of the few sources of systematic information on these networks is a study of household decision-making patterns in eight villages in different  - 33 - parts of Nepal (Acharya and Bennett, 1981). Interestingly, the study revealed that women play a dominant role in decisions on small loans of cash (less than Rs. 15 for the survey) and produce which are continually circulated between households. In some communities these small exchanges are almost the exclusive preserve of women. For more unusual sapat loans involving larger amounts of cash, men generally assume the dominant role, as they do for most rin transactions and certainly for transactions with the formal credit system (discussed below). Although it hardly provides an adequate 'safety net' for poor families, the existence of this 'invisible, network of small cash and kind exchanges transacted largely between women of different households does provide some measure of security for households facing temporary economic hardships. Rin Loans: A Continuum from Cooperation to Exploitation 6. In contrast to sapat, a rin bears interest ranging from lOZ to 1502 and up -- though according to numerous micro studies, the most common rate seems to range between 30 and 402. 1/ However, it is extremely difficult to generalize about interest rates for traditional or informal credit because rates often vary by season2/ and by the economic status of the borrower and their consequent ability to pledge land or jewelry as collateral. Moreover, interest on a single loan may be collected in many different forms. In part this is due to the relative scarcity of cash in Nepal's rural economy. But it is also related to the fact that the old Mulki Ain (National Legal Code) set the 'just and legal' interest rate at 102 on cash and 20 or 25? on loans in kind. While written bonds could not commit to anything higher than this, there were many other systems which evolved for collecting higher actual rates. For example, often an extra five to ten percent called ghiukhane (literally, 'eating butter oil") was discounted or deducted from the capital at the time of the loan.3/ 1/ According to personal communication with the former Governor of the Nepal Rastra Bank, this is a realistic rate considering the cost of capital and the risk of default. 21 Especially in food deficit areas in the Western Mountain regions interest on in-kind grain loans may vary by as much as 252 between the harvest season in the fall when grain is plentiful and late winter when grain stocks are low (Levine 1987). 3/ National level data from the Nepal Rastra Bank's Agricultural Credit Review Survey (1980:179) indicate that charging a fixed transaction fee is still common practice in the informal credit sector -- and that the rates charged for small and marginal farmers are higher (ranging from 10 to 25Z) than those charged to medium and large farmers (from 5 to lO).  - 34 - 7. Other means of collecting additional interest which are still widely prevalent include payment in labor service4I (or in common property goods such as firewood, birch bark, honey, grass, etc. whose primary (private) cost is the labor involved in collecting them). But perhaps the type of still current unofficial "interest' on informal credit which has had the most detrimental long term effect on the poor is the practice of turning over the usufruct rights on the debtor's land to the creditor until the debt and legally permitted interest are paid. 8. Anthropological accounts from various parts of Nepal reveal that in addition to the 'friendly' sapat loans (which occur to som6 extent in every community), debtor/creditor relationships involved in traditional rin transactions span a continuum which ranges froms (i) temporary debtor/creditor relationships between social and economic equals -- a kind of "revolving' indebtedness in a relatively closed and socially homogeneous community; (ii) systems which express and maintain economic (and sometimes social) inequality and may involve expropriation of labor, grain and sometimes the use of land; and (iii) systems which result in a permanent net transfer of assets (e.g. land), sometimes leading to bonded labor or forced migration out of the community for the debtor. 9. Although some geographic and ethnic patterns emerge from the literature, as is evident in the discussion vhich follows, these patterns must admit to many exceptions. It is probably more accurate to think of multiple systems (spanning at least the first two types on the continuum) existing in any one community. Most examples of the first, essentially reciprocal, type of system have been recorded in the more remote hill and mountain communities, generally among Tibeto-burman speaking groups. McDougal for example, writes that among the Kulunge Rai in the eastern hills, 4/ Hitchcock (1966) in his study of the Kagars reports one case where the debtor made a trip to a distant bazaar to bring back salt for his creditor where the unpaid labor time involved actually amounted to 920Z of the loan (though one presumes there must have been some additional utility in the trip for the unfortunate debtort). Levine (1987) reports that for debtors in Humla, labor service is commonly attached to loans and the interest rate is reduced proportionately by the amount of labor service pledged. Thus, the base rate of 50? is reduced to 40Z for debtors who promise one day of labor service for every khal of grain they borrow and the rate drops to 25? for promises of two to three days of labor service per khal (a khal is equivalent to somewhat more than half a bushel and a khal of buckwheat or millet cost 100 to 120 rupees in 1983 when one U.S. dollar was equal to Re. 14).  - 35 - 'except for a few Sherpas who are marginally included, the network of credit relatioas is completely within the Rai community. There is no class of money lenders and the distinction between creditor and debtor is a situational one. ...The characteristic picture which emerges is that of a rather reluctant creditor, a man who happened to have some money on hand which people knew about, and which he would have preferred to use for another purpose." (1979:44) 10. Mcfarlane (1976) reports that the Gurungs he studied borrowed only from each other and not from Brahmans or Newari storekeepers outside the community (both renowned as moneylenders). Pignede (1966) also describes the credit system of the Gurungs whom he studied as fairly closed, and Messerschmidt (1976). studying a Gurung coamunity in another location, reports that some of the lower statts Gurungs did borrow from outside moneylenders, and a few had actually lost land to them. Dhikur: Recent Developments in Traditional Rotating Credit Associations 11. Perhaps the most striking example of the first reciprocal or co- operative type of informal credit system on our continuum is the dhikur or rotating credit and savings association. The word, derived from the Tibetan terms referring to "grain rotating turn by turn", is pronounced differently in different ethnic groups and also known by the Nepali term dbikuti which means "storehouse, barn or treasure chest". Both of these terms suggest that the practice originated with in-kind grain exchanges and indeed, though most are now based on cash, both livestock and grain-based dhikurs still exist. The most detailed account of the dhikur as an institution comes from Messerschmidt (1978), who describes it as: "a rotating credit association of shareholders who pay regular installments at a predetermined rate and frequency and who each received the combined installments (the fund) once. This is the share to invest or spend as one wishes. Installments are the same for every shareholder in a particular turn, but increase each time by a type of simple incremental interest. In effect, each shareholder pays interest for the privilege of using the fund." (1978:142) 12. Most of the dhikurs documented by Messerschmidt had between 10 and 30 members -- though he notes one of 105 members. The average installment was around Rs. 300 per member which would have meant an initial fund (usually taken by the organizer who initiated the dhikur to meet his own need for capital) of between Rs. 3,000 and Rs. 9,000 -- though Messerschmidt reports that the fund for one dhikur involving the wealthiest traders began at Rs. 50,000 and ended at Rs. 79.000. The dhikur fund revolved either annually or, more often, semi-annually and the receiver of the fund was determined either on the basis of need or more commonly, by drawing lots or secret bidding. The usual interest increment was 102 per annum (though rates of 14 and 162 were also reported) which would mean that for an average installment beginning at Rs. 300 per member, the second semi- annual deposit would be Rs. 315, the third, Rs. 330 and so fourth.  - 36 - 13. Although most dhikurs are comprised of adult male members, there is r.o ban on women members if they have sufficient means or if they are sponsored by a male relative. Though the installments are generally much lower, Messerschmidt reports the existence of all-female dhikurs -- formed in some cases to raise funds for their dowries. 14. Even though these rotating ^redit associations remain informal in the sense that they are outside both the banking and the court systems, they are far from casual. The dhikur has a written constitution and each member must si&r. a contract for the security pledged against default (Messerschmidt, 1978:146). Pines are levied for late payment of installments or for missing a meeting, and default is extremely rare (Messerschmidt 1978; Seibel and Shrestha, 1988). Although this high level of discipline is to some extent explained by the fact that traditional dhikurs comprised a fairly homogeneous group bound together by multiple social and economic ties, the transparency created by the written records (and the implied possibility of recourse to the formal court system) has probably also been important. A less subtle means of insuring low default was simply to prevail upon any would-be member with a poor reputation in financial matters to take the last turn. As Messerschmidt observes, fthe feeling is that if a member is likely to default, the temptation would be greater immediately after taking the fund. By taking the last and largest fund there is every incentive to live up to the obligation all along the way'( 1978:146). 15. At the time of Messerschmidt's observations, membership in a dhikur was primarily confined to a single ethnic group and the practice itse.f almost exclusively to Tibeto-burman speaking peoples. More recent stidies show that they are now undertaken by many ethnic groups.5/ 16. Although it is difficult to establish because of the lack of earlier documentation, the practice seems to have spread remarkably in the last decade -at least in the western hill and mountain region for which we have data. Not only are there more dhikurs, but many have fairly large funds requiring substantial monthly contributions. In the 25 member multi- ethnic group mentioned in the footnote above, the beginning monthly contribution was Rs. 2,000. Seibel and Shrestha report that during the 25 month period over which they observed the fund a total of Rs. 1,070,000 (equivalent to $48,000 at the current rate of exchange) had been raised. This was not an unusually large fund and they note that for some of the dhikurs, the total turnover during the life of the association is significantly greater than $US 1 million (1988s188). 5I Seibel and Shrestha (1988) surveyed 13 groups in the western hill region with membership open across ethnic groups. The dhikur recorded in the greatest detail was founded by a Newar and among its 25 members included both Indo-Aryan (Brahmans and Chetris) as well as Tibeto- burman (Newari and Thakali) groups; it should be noted that all these are both educationally and economically dominant groups in Nepal. The authors make no mention of the occupational castes as also being able to join such multi-ethnic dhikur associations -- or forming their own.  - 37 - 17. This suggests that there is a high level of demand for credit -- and for alternatives to the individual moneylender as a source of credit -- in rural Nepal which is not currently being met by the formal credit system. There has been no detailed documentation of how the dhikur funds are used -- how much goes to finance agricultural investments, bow much to small scale manufacturing and agro-processing, local businesses or long distance trading, and how much to weddings, medical costs or consumer durables. There are no regulations and members are free to use the fund however they wish when their turn comes. However, while some portion is probably used for consumption, most of it seems to be used for business investment. Unlike the traditional Thakali and Bhotiya dhikurs which were primarily formed to provide capital for long distance trading ventures, most of the members of these "new generation' dhikurs in the middle hills are likely to be small businessmen running cloth shops or hotels, or farmers, often with a secondary occupation. For many of them, observe Seibel and Shrestha, the dhikurs "have been their only major source of external finance' (1988:193). And since membership in a dhikur provides credit only once to each member when the fund revolves to him/her, most small entrepreneurs will belong to 10 or 12 different associations in order to meet the continuing need for capital. Sosaitees: The Emergence of Permanent Savings and Credit Groups 18. As is evident from the foregoing discussion, the installments required by most of the dhikurs documented are at a level which puts this form of informal savings and credit institution well beyond the reach of households below the poverty line. Although the references to women's and even children's dhikurs with much lower installments indicate that there is no structural reason to prevent the formation of dhikurs with mor^ modest stakes being organized among the poor, there does not seem to be much evidence of this. In fact, Seibel and Shrestha point out that in many other countries, "rotating savings and credit associations have become a predominant source of small enterprise finance, while non-rotating groups with a permanent loan fund have become the main informal financial institution of the poorer sections of the population, particularly small farmers" (1988:184). They note that the common pattern whereby the rotating funds evolve into a permanent loan fund does not seem to have occurred in rural Nepal. Such permanent savings and credit associations are found in various parts of Nepal, but they seem to be a new and parallel development. 19. The major source of information on these permanent associations, usually known as "sosaitees", is a study carried out by the Nepal Rastra Bank in 1988. Although the existence of such groups is reported from all parts of Nepal, this initial study covered only the Western and Far Western development regions. It documents an impressive range of "self-help groups", some formed specifically for financial purposes and others for non- financial tasks. The latter include groups organized for water management, temple maintenance, marketing, labor exchange or community development -- though in many cases groups initiated for such non-financial purposes have subsequently expanded to offer credit and savings opportunities to their members.  - 38 - 20. Unlike the dhikurs which are fairly uniform in their structure and function, the case studies of non-rotating savings and credit groups reveal many differences in organization and purpose. There is also a wide variation in the size of the group (from 22 to over 1,000 households) and the amount of capital involved (from Rs. 4,000 to Rs. 165,000). But in general, both the total funds and the individual contributions are much smaller in the non-rotating sosaitees than in the dhikurs. This makes them more accessible to the poor. Several of the sosaitees allow two kinds of membership: 'founding* or 'life" shares that involve a substantial contribution and ordinary memberships which require only a nominal fee or simply registration. Such groups attempt to encompass both rich and the poor households and are often founded by school teachers or other literate members of the community out of a sense of social duty. Much closer study would be required to understand the internal dynamics of such economically heterogeneous groups and to assess their actual role in the local political economy. 21. There are however, some sosaitees formed by disadvantaged communities for their own members. In one case from Kanchanpur 30 low caste Lohars (metal workers) have organized themselves into a group with total assets of Rs. 38,000 including Rs. 20,000 in outstanding loans to its own members. There is also a very large group organized by Tharus around an irrigation system in the western inner Terai. A very idealistic group in Syangjya district called the Janakalyan Sosaitee (People's Welfare Society) was set up explicitly to help the poor by providing an alternative to moneylenders and also to encourage women to earn and save. Although this group does encompass members from different economic strata, the membership fee is the same for all: Rs. 100 initially and Rs. 10 per month. With 100 members a substantial capital fund has been collected. Like several other groups the Janakalyan Sosaitee charges 24Z interest.6/ The loans can be used for consumption as well as production purposes and range in size from Rs. 100 to 15,900. The Negative Consequences of Debt in Traditional Informal Credit Systems 22. From these accounts it would appear that the first type of credit relationship on our continuum -- the co-operative, non-exploitative type -- is far more likely to occur as a group phenomenon, within the institutional framework of the dhikur or sosaitee, than between individuals. Certainly the literature offers very few examples -- in any ethnic community -- where rin transactions between individuals conform to the rather idealized non- exploitative type described by McDougal for the Kulange Rai. Indeed, a recent field observation of a Rai community in the upper Arun valley provides evidence of considerable deterioration in the traditionally egalitarian patterns. The Pradhan Panch (himself a Rai whe returned from 6/ In two of tho case studies the interest rate charged to group members was 60?. The fact that this was still considered attractive to the borrowers suggests that the combination of interest rates and transaction costs involved in other alternatives must be even higher.  - 39 - Burma with considerable personal savings) now owns roughly 50? of the local farm land -- some of which he acquired when the owners could not repay their debts to him. In general, it is clear that in this community over the last few decades the distribution of land has shifted in favor of a few wealthier families as a number of small and marginal farm families have been driven to forced land sales to repay their debts. 23. The phenomenon of the more sophisticated and educated getting control over land from the original owners through the process of debt is widespread. The most well-known is the loss or encumbrance of the kipat or communal clan lands held by members of the Rai and Limbu communities who became indebted to Nepali-speaking Brahmans and Chetris. This phenomenon, widely documented, is associated with the migration of caste Hindus into the eastern parts of Nepal, the increasing value of land (particularly rice growing land), and the growing power of the Kathmandu-based Nepali government vis a vis the traditional tribal or clan organizations over the last century. As Sagant put it: "Debt has undermined the traditional tribal structures and brought them to the crossroads of Hinduization. The passage from member of the tribe to citizen of Nepal has occurred through the process of debt' (1980:280). 24. But the process was not a simple one, and the BrahmanlChetri immigrants were not the only ones who benefited. In Sagant's detailed study of credit and land mortgage transactions in eastern Nepal from 1900 to 1970, he reports, in fact, that there was not a single case of forcible seizure of debtor's property for the realization of debt. However, the prevailing system of land mortgage involved the transfer of usufructuary rights over the land to the creditor. And although the creditors might appoint the debtor as tenant, they also had the right to appoint a new tenant. To strengthen their own positions locally, immigrant Nepali-speaking creditors often appointed members of their own family or community as tenants. Thus, while the title of the kipat land remained with the Limbu debtor, he risked losing access to it until he could repay the debt. For most Limbu farmers recovery was extremely difficult. Sagant reports that only one in ten was able to repay the debt and regain usufructuary rights. The main value of the land title was that it permitted the original Limbu owners to take out second and third mortgages -- which creditors were apparently willing to give because it made repayment and redemption of the land even more unlikely and increased the security of their own control over the land. 25. The of loss of usufructuary rights over land in the Mewa Khola where Sagant conducted his study was extensive. He reports that in 1966, 572 of tba land plots on record were mortgaged. He also notes however, that Nepali-speaking outsiders were not the only creditors. Nearly half of the mortgages were transacted within the Limbu community -- though this moderated the effect of the debt because the creditor was far less likely to bring in an outside, non-Limbu tenant. 26. In addition to the well known case of the erosion of Limbu control over kipat land , many different ethnographic accounts from other parts of Nepal lend support to the stereotype of the high caste Brahman (and also to some extent the Chetri) land owners, and in the Terai, the Mahajan grain  - 40 - merchants who create ties of dependency by lending cash or grain to less sophisticated tribal peoples. However, as in the Limbu case, a more important factor than ethnicity here may be literacy, familiarity with the legal system, involvement in the cash economy and of course, sufficient surplus to lend. Although high caste Hindus have in many parts of Nepal had advantages in these areas, there are also innumerable examples of poor Brahmans and Chetris indebted to wealthy Sherpas, Gurungs or Thakalis, or to Newari traders and shop keepers -- all of whom were far more cosmopolitan and well connected than they. 27. Another commonly held view is that the third and most exploitative type of credit relationship on our continuum -- that leading to permanent loss of assets by the debtor, to bonded labor or forced migration -- is more frequently encountered in the Terai than in the Hills. This view appears to be supported by much of the ethnographic literature and the work on land reform. Harsher, more impersonal relations between creditor and debtor are more likely to obtain in areas like the Terai where the economy is more open and market oriented and the population more heterogeneous and less bound by the multiple, cross-cutting economic and social ties which tend to characterize hill and mountain communities. 28. Although enslavement for debt was forbidden in the 1853 code, bondage was not.71 Over the next century the number of ethnic groups/castes who could be legally placed in bondage were progressively reduced until by 1935 only certain tribal groups and untouchable castes were still vulnerable. But not until 1951 when bonded labor was legally abolished was the principal established that the debtor's body (or that of his wife or offspring) could not be seized in payment of debt. Despite its illegality the practice of bonded labour persists, in the hills as well as the Terai. As recently as 1984 a government sponsored survey of 10 village panchayats in the hills and 19 in the Terai uncovered 89 bonded laborers. C. The Impact of the Formal Credit System on the Poor 29. It is against this background of the more negative aspects of Nepal's varied and complex traditional credit systems that the government's efforts to establish the formal banking sector and extend its reach to the rural areas and in more recent years, specifically to the poor, must be understood. To most policy makers in Nepal the informal credit sector has been seen as at best inadequate to the needs of an increasingly monetized economy, and at worst, part and parcel of traditional caste and class-based systems of exploitation. The government's largely unsuccessful attempt to establish a network of village cooperatives, its creation of the Agricultural Development Bank (ADB) and subsequently within it, the Small Farmer Development Program (SFDP), and the pressure placed on commercial 7/ One of the key distinctions here is that bondage extended only until the debtor died (or was able to pay off the principal and interest of his debt) while enslavement extended beyond his own life and affected the status of his offspring as well.  - 41 - banks to open rural branches and channel a growing percentage of their lending through these branches to agriculture, services and cottage industries and to households below the poverty line through the Intensive Banking Program (IBP), have all been efforts to put in place some alternative to these traditional credit systems. In this section we will review a number of national-level surveys as well as some micro-level studies to assess the extent to which the rural poor have gained access to the formal banking system and to get some sense of their current patterns of credit use. Volume of Borrowing: Increasing Use of Credit-Especially among Large Farmers 30. Among the best sources of macro-level data on credit use in rural Nepal are the two agricultural credit surveys carried out by the Nepal Rastra Bank (NRB) in 1969/70 and in 1976/77 and the NRB evaluation of the impact of SFDP on credit use among small farmers in 1982. 8/ Although there is no comparable recent data, the two credit surveys allow us to make some observations on trends up to the early 1980's, most of which are confirmed by more recent sample surveys. As would be expected in a country like Nepal in the process of moving to a more monetized economy from one based largely on subsistence agricultural production, overall credit use (from both formal and informal sources) has been increasing sharply. Table 1.3-1 shows that the proportion of families borrowing increased from 38? in 1969/70 to 51? in 1976177, and the average amount borro'red per family increased from Rs. 132 to Rs. 586. Table I.3-1: Growth in Borrowing by Farm Size 1969/70 - 1976/77 All Families Large Medium Small 1969/70 1976/77 1969/70 1976/77 1969/70 1976/77 1969/70 1976/77 Proportion of Borrowing Families (Percent) Region Hills 37? 42? 25? 512 35Z 432 39Z 52? Terai 39? 55? 262 622 352 502 43Z 552 Overall 38? 512 262 55Z 35? 472 412 552 Average Borrowing Per Family (Amount in Rs.) Hills 136 380 201 559 171 285 117 242 Terai 130 670 332 3,098 136 620 94 405 Overall 132 586 287 1,434 146 488 101 380 Source: NRB (1980), Vol. I, Tables 6 and 7, p. 138. 8/ NRB, Agricultural Credit Survey of Nepal 1972 and, 1980. NRB. An Evaluation Study of Small Farmers Development Projects of Nepal, 1982.  - 42 - 31. However, when these figures are disaggregated by size of landholding, it is clear that the increased volume of borrowing for that period was much greater for large farmers than for small farmers. An examination of these data by region shows that both the proportion of families borrowing and the average loan sizes (for all land holding groups) was higher in the more highly monetized Terai. Sources of Borrowing Table 1.3-2: Proportion of Farm Families Borrowing from Different Sources by Farm Size Institutional Non-Institutional Size Group \a 1969/70 1976/77 1969/70 1976/77 Overall 19? 24? 82? 76? Large 18? 34? 72 14? Medium 22? 21X 14? 15? small 641 452 79? 71? a/ Large Medium Small Hills above 1.02 ha. 0.51-1.02 ha. Up to 0.5 ha. Terai above 5.42 ha. 2.71-5.42 ha. Up to 2.71 ha. Source: Derived from NRB (1980), Vol. I, Chapter 12, Table 2. 32. Although informal sources still predominate, the NRB surveys (Table I.3-2) show that the proportion of households using institutional credit increased from 19? to 242 in the period between 1969/70 and 1976/70. And in the hills the proportion of credit from institutional sources rose from 5? to 33Z of overall borrowings (Table I.3-3). This would seem at first glance to indicate that the government's efforts to provide an alternative to traditional informal credit systems was, at least in the hill areas, gaining momentum.  - 43 - Table I.3-3: Comparative Data on Average Borrowings by Region * Hills Terai 1976177 1976177 Credit Agency Amount Percent Amount Percent Institutional 297 33.1 532 43.6 Cooperatives and Sajhas 42 4.7 173 14.2 Agricultural Development Bank 203 22.6 271 22.2 Comercial Bank 52 5.8 88 7.2 Private 600 66.9 689 56.4 Village Moneylenders 327 36.5 245 20.1 Professional Moneylenders 19 2.1 48 3.9 Landlords 25 2.8 30 2.5 Agricultural Traders 42 4.7 77 6.3 Friends and Relatives 187 20.9 270 22.1 Others - - 19 1.6 Total 897 100.0 1,221 100.0 * Nepal Rastra Bank (1980), Table 12, Vol. I. 33. Once again, disaggregation of the data by size of land-holding reveals the unsurprising fact that the larger land owners have captured the major share of institutional credit. Although the proportion of borrowing families is highest among small farmers (for both formal and informal sources) in both surveys, the proportion of small farm families using institutional credit actually declined. The 1980 NRB survey found that medium and large farmers together captured 92Z of the total institutional credit in 1969/70, and by 1976/77 their share had risen to 97Z. Data from a later NRB survey9/ show that between 1976 and 1982 both the amount of credit taken from institutional sources and the proportion of institutional credit in the total borrowing of small farm families had declined. The differential access of large holders to forma' credit sources is confirmed by other studies. A micro-level study carries out by IDS in 1986 found that only 6X of the tenants had borrowed from institutional sources compared to 44? of the owner tillers. 91 NRB (1982), An Evaluation Study of Small Farmers Development Projects in Nepal. (Table 3, page 42). The non-project control group for this study are comparable to the small farm families enumerated in the two earlier NRB surveys.  - 44 - 34. The drop in small farmer involvement in the formal banking system -- and the consequent increase in their dependence on non-institutional sources -- was particularly sharp in the terai. While 56% of small farmer borrowing in the terai came from the informal system in 1976177, in the 1982 survey 912 came from non-institutional sources. Given our earlier observation that the harsher, ltbs egalitarian/co-operative forms of the informal sector lending seem more prevalent in the Terai, the decreasing share of institutional credit going to small farmers is not a welcome trend there. 35. On the whole however, it is important to keep in mind that all non- institutional sources of credit are not the same. As we have seen, the informal credit sector in Nepal includes both potentially exploitative sources (e.g. money lenders, agricultural traders and landlords) and presumably more cooperative sources (e.g. friends and neighbors, relatives and self-help groups) with more equitable terms. Moreover, the formal banking system is not without its own forms of exploitation. Numerous informants -- from clients (both rich and poor) at the village level to high ranking bank management staff -- concede that n most cases there is a 'consideration' of about 1O% of the loan capi al which must be paid by the client to the bank staff. This, in addition o the high transaction costs of multiple visits to the bank, make formal ector credit considerably more expensive than the official controlled rate of between 8 and 18Z. In fact, the Agricultural Sector Review reports that the "non-interest cost of borrowing from the formal sector has been estimated at 82Z to 128Z per annum on a six month NRs 1,000 loan -- so money lender rates of between 35% and 150% are attractive'. 36. In the terai and particularly in the hills the importance of the village moneylender appeared to have declined between the two NRB surveys. Professional moneylenders and agricultural traders however, had become more important as had landlords in the hills. There is no further break-up of the category of 'friends and relatives' so we do not know what proportion of this borrowing was in the form of interest-free sapat loans, what was in the form of individual rin transactions and what represents loans from dhikur or sosaitee associations. It is reasonable to assume however, that the terms for this category of credit were more favorable than those of the money lenders, traders and landlords -- and in some cases even more favorable than the banks. This category of "friendly" credit appears to have accounted for a declining share of borrowing in the hills and an increasing share in the terai. 37. Most of the available data -- from both micro and macro-level surveys --suffer from lack of specificity about informal credit sources. One of the few studies that does offer a more detailed breakdown is Molnar's (1981) work on the Kham Magars of the far western hills. Like most studies, she found that only about 20% of the borrowing in this community was from institutional sources. Of the rest about half were interest free sapat loans and the remainder was borrowed at 30% per annum. Data from the most recent national level survey (CEDA 1988) show that the formal banking system still  - 45 - provides only about 20S of rural credit -- slightly less than the share (22Z) of private money lenders (Table I.3-4). This leaves friends and neighbors3 as the source of over 502 of the borrowing in rural Nepal. One could hazard that the "friends' category here (and probably the category 'relatives' where it is used in other surveys) probably coincides for the most part with sanat loans while the category *neighbors' presumably covers more "business-liken arrangements with individuals as well as group credit and savings associations. Table I.3-4s Source of Loan by Ecological Region (Percentage of household) Region Bank Cooperative Neighbor Moneylender Friends Others Mountain 11.7 0.4 47.1 24.3 8.6 2.5 Hil.ls 18.8 1.0 48.3 17.4 6.5 2.0 Terai 20.6 2.7 39.5 27.7 5.8 0.6 National Average 19.1 1.4 44.3 22.5 6.7 1.4 Source: CEDA, Sixth Plan Impact Evaluation Study, 1987/88. (Note: rows do not sum to 100Z in source). 38. Data aggregated from eight coordinated micro-level studies in different parts of Nepal (Acharya and Bennett, 1981) reveal a pattern roughly similar to the macro level data cited above: institutions provided 162 of the credit, money lenders 22Z, relatives 29Z and friends and neighbors 33?. These aggregate data however, mask important local variations in both the volume and sources of credit. For the Maithili community in the Terai 712 of the loans were from institutions, while for two of the more remote communities in the hills (Tamang) and mountains (Baragaonle) there had been no dealings with banks. For some groups relatives were the most important source of credit; for others friends and neighbors were more important. In only one community, the Tibetan-speaking Baragaonle, was the moneylender reported as the major source of credit. Although, like most studies this one does not give adequate detail about the actual terms prevailing for these different informal credit sources, such data -- especially when considered in the context of the macro-level finding that 'friends and neighbors ' provide 502 of rural credit -- suggest that the proverbial 'exploitative' moneylender does not actually dominate the informal credit sector in all parts of rural Nepal.  - 46 - Use of Credit: The Poor are Increasingly Borrowing for Consumption 39. The two NRB surveys show a reduction in the propcrtion of consumption loans between 1969170 and 1976/77 (from 44Z to 282 of total borrowing) with a corresponding increase in loans for farming expenditure and especially capital expenditure. However the shift away from consumption borrowing was mainly among large farmers. Although the reported proportion of consumption loans for small farmers fell between 1969170 and 1976177, the NRB survey of small farmers In 1982 showed that 68? of their borrowing was for consumption -- and there was also a noticable drop in small farmer borrowing for capital expenditure and farming (Table 1.3-5). Table I.3-5: Purpose of Borrowing for Small Farm Families (Percent) Purpose 1976177 1 1982 /2 Farming Expenditure 17.89 8.54 Capital Expenditure 36.06 23.54 Consumption Expenditure 42.37 36.32 others 3.68 31.58 TOTAL 100.0 100.0 Note: The "controlled groups" small farmers of 1982 survey were similar to the small farm group of 1976/77 survey. Sources: Agricultural Credit Review Survey Nepal, Derived. 1/ NRB (1980), Table 17, p. 100. 2' NRB (1982), Table 8, p. 17. 40. These findings are confinmed by surveys in five districts carried out by the Department of Food and Agricultural Marketing (1982:184-86). Yadev's (1984) study of a sample of terai villages also found that 862 of the loans to households in the lowest economic stratum were for consumption expenditure and only 6X for farming expenditure (Table 1.3-6).  - 47 - Table 1.3-6: Purpose of Borrowing in Sampled Terai Village by Economic Status, 1984 (Percent) Economic Status Lower Purposes Low Middle Middle Higher Total Farming Expenditure 5.6 13.8 20.9 53.6 27.2 Consumption Expenditure 86.1 60.3 39.5 4.4 41.7 Social Ceremonies/ Miscellaneous Expenditures 8.3 25.9 39.5 42.0 31.1 TOTAL 100.0 100.0 100.0 100.0 100.0 Sources Yadav, (1984). Derived from Table 5.12, p. 99. Debt Service as a Proportion of Income 41. It is clear that the poor are borrowing primarily to meet consumption needs rather than for productive investments. Since little of the borrowing goes to increase their earning capacity, there is concern that in poor households debt payments will further reduce household consumption levels. Or if income is so low that meeting basic survival needs leaves nothing over and pressure to repay is high, they may be forced to sell what few productive assets they have, entering the 'debt spiral' into even greater poverty. 42. The NRB's Multipurpose Household Survey gives information on borrowings as a proportion of monthly income and repayment as a proportion of monthly espenditure. At the all-Nepal level for rural families in the lowest income decile, almost 20? of monthly income was borrowed. In the rural hills this figure reached 261. Repayment however, accounted for only about 2? and 3? respectively of expenditures. Since the latter figures a * the amount of income consumed in debt service, it does not appear at fir&- glance to be an overwhelming 'burden' on the household budget that would seriously reduce consumption levels. However, the imbalance between borrowing and reported repayment is disturbingly high: rural families in the lowest decile group are accumulating Rs. 90 of unpaid debt per month (about 20X of gross income). If these data are accurate, then the very poor in Nepal are in a precarious position indeed. Even the friendliest creditor could be expected to grow weary of providing a continual 'dole' that was never paid back -- and a more business-like creditor might after a long enough period, insist on liquidating the creditor's assets or on extracting bonded labor to redeem the debt.  - 48 - Monthly Borrowing to income & Income to Repayment Ratios In Rural Nepal, by Income Deciles Borrowing/Income Repayment/Borrowing 20%~~~~~~~~~~ ~~200% 1150% 10%1 4_: 100% 5% ! X- -50%0 0% ~ 0% I If III IV V VI VIl Vill IX X Income Deciles -Rural Nepal I Rural Terai Rural Hills ' Rural Mountains LJ Borrowing/income 43. Is this in fact the case? Although the MPHBS data are in general considered quite accurate, collecting reliable household level information on debt is extremely difficult since informants from different economic strata may distort their responses in different ways. Overall however, the data on debt by income decile reveal a fairly consistent U-shaped curve with high levels of borrowing among the two lowest deciles (which would include the ultra poor who are presumably borrowing primarily for consumption/survival), much lower levels among the middle deciles (who tend to avoid as much as possible the risk associated with taking credit), and rising levels of borrowing again among the top four deciles (who are probably borrowing for productive investments).10/ These data are also consistent with what other sources reveal about the continuing predominance of informal credit sources for all income levels, but especially for the poor. For the all-Nepal data, it is not until the 6th percentile that any significant amount of formal sector borrowing occursll/ 101 There is however a somewhat anomolous jump in the borrowing levels of those in the 7th decile group for both the Terai and the Hills. 11/ Again there are anomolies here with a sudden jump in use of formal sector credit among the 6th percentile group and then a sharp fall in the next three groups.  - 49 - 44. Overall, households are borrowing more each month than they are repaying and always the gap is largest for the lowest income deciles and decreases more or less steadily as income increases.12/ Data by farm size show the same pattern in all three ecological regions. It should also be noted however, that the pattern of higher loan default by the poor which emerges from these data is inconsistent with the fact that the repayment rates for the Small Farmer Development Program which targets credit to the poor have generally been better than repayment rates for agricultural credit in general which goes almost entirely to larger farmers. 45. Despite some inconsistencies in the data, the MPHBS does seem to indicate that the poor are rapidly accumulating an unsustainable burden of debt. The finding noted above, that borrowing accounts for roughly 20? of income for families in the lowest income decile is supported by data from an another NRB study which found that small farm families had debts equivalent to 32Z of their gross farm production (1982X36-37). If maturities were two years and the average market rate of interest is 40t, about 25? of income would have to go to debt payments. 46. Levine's (1987) in-depth anthropological study of indebtedness in 6 villages in the Humla district of Par Western Nepal documents the reality of the 'debt trap' into which many resource-poor families falll3/ . In this food deficit region most loans are taken in the form of grain for consumption. The study found wide variation between villages, with a large proportion in many villages reporting no grain debts, and a substantial minority reporting major indebtedness. In Rongphu, the poorest village surveyed, 81? of the households reported having made payments on existing grain loans during 1982. Despite this, 83? reported that they still had debts left to pay -- though most of these debts were clustered at the lower end of the range (Rs. 2,000 or less). The real impact of these figures on the consumption levels of poor families is evident when we translate the rupee repayment figures back into grain. Households making interest payments worth Rs. 1000 per year were sacrificing 10 khal of buckwheat or millet which is almost enough to feed one adult member for a year. 12/ However, in addition to the inconsistencies noted in the data within regions (see previous footnotes), there are also some significant differences between regions. Thus, in the terai monthly repayments are still less than half of borrowings even in the top income decile. In contrast, in both the Hill and Mountain regions the top income decile households are paying back more than they borrow each month. 13/ This section relies on the account of Levine's study provided in the IDS report 'Review of Studies on Poverty', October, 1989, pp. 154-161.  - 50 - Table 1.3-7: Distribution of Grain Debts and Payments by Value Proportion of Families (1) Making Interest (2) With Outstanding Value and Capital Payment Grain Debts Re. Equivalent (1982183) (1983) 0 47? 58t 1 - 1,000 27? 242 1,000 - 2,000 17? 11l 2,000 - 3,000 6? 3? 3,000 - 4,000 2? 3? 4,000 - 6.500 1? 1? Total 1OOZ 1OOZ 47. In the village of Kalunga where all but one of the 25 households had grain debts, Levine carried out an in-depth study of 5 of the most severely indebteded households. These households were so deeply in deficit they had no hope of paying off their past debts, and yet all expected to take out new loans that year which would further add to their debt burden. Table I.3-8 shows the annual production. loan repayment and outstanding debt for these households. The average rate of interest for these loans was 34.42 -- in addition to the days of labor service and payments of goods such as firewood, honey, etc. mentioned earlier as customary 'extra interest. This means that the average household owed 14.1 khal of grain in interest -- an amount equal .o nearly one quarter of their annual grain production. In fact, on an average these households were able to pay only 12.2 khal or about 20? of their grain production for 1982 plus the labor service, etc.. The next year they would owe 34.4S interest on their enlarged loans in addition to new loans they would be forced to take to meet their consumption needs over the winter and spring.  - 51 - Table 1.3-8: Indebtedness in Five Khalanpa Households, 1982-83 Total Grain Payment on Outstanding Number of Production Grain Loans Grain Debts Households Members (in Khal) (in Khal) (in Khal) 1 11 80.0 13.5 50.0 2 3 32.5 6.2 17.5 3 3 46.2 15.0 14.0 4 5 85.0 14.2 61.0 5 10 56.2 65.0 Mean 6.4 59.9 12.2 41.5 Source: Levine (1987), Ibid, Table 9, p. 43. 48. Clearly in Humla, households with large stores of grain can profit enormously from grain lending, while households forced to borrow grain can become entrapped in a vicious cycle of debt where interest payments alone consume a major share of their income. Since Humla, where Levine's study was carried out, is one of the poorest and most remote areas of Nepal, the level and nature of indebtedness documented there cannot be generalized for the entire country. Nevertheless, it does suggest that the patterns of unsustainable debt among the rural poor which emerge from the all Nepal MPHBS data need to be taken seriously. 49. The evidence reviewed here suggests that our initial question as to whether or not the existing informal credit systems are primarily 'exploitative' may not have been the best way to frame the problem of rural indebtedness in Nepal. It is clear that there is exploitation in many of the traditional local credit systems. But the deeper problem seems to be how to escape from the low equilibrium systems where borrowing is used for short term survival rather than to increase productivity. Although usurous interest rates accelerate the debt spiral for poor families, it is probably more often the lack of productive investment opportunities and entrepreneural skills that make debt unsustainable for the poor. (In India where there is the right mix of skill and opportunity, the poor have shown themselves able to generate a profit with interest rates of up to 36Z.) This has important imp'ications for government policies and programs aimed at providing alternative sources of credit to the poor through the formal banking system. Above all it suggests that the 'minimalist" or 'credit only' approach to targetted credit will not be effective in Nepal. The provJsion of credit will probably have an impact only if accompanied by program mechanisms to link the poor borrower to the range of production support services needed to make the loan enterprise profitable. Ultimately the growth and expansion of these small enterprises depends also on the provision of adequate physical and institutional infrastructure. What credit programs need to deliver to the poor Nepal is not cheap credit, but credit accompanied by and linked to services to make it productive.  - 52 - ANNEX I* 4 - AGRICULTURE AND POVERTY IN NEPAL The following tables present the background information used for the discussion in Chapter 4. All production estimates come from models developed in the Master Plan for Irrigation in Nepal (1989). Typical cropping patterns (Table 1.4-5), yields (1.4-6) and production costs were estimated for rainfed and irrigated conditions in both east and west Nepal. In addition, cropping patterns and yields were predicted assuming that certain foreseeable improvements in irrigation would occur. For this analysis these models were adapted for the typical farm sizes found in the Multi-Purpose Household Budget Survey. It is important to note that cropping intensity is not assumed to change with farm size. However, numerous case studies show that small farmers crop more intensively than large farmers under the same farming conditions, so the income estimates presented here may underestimate the income small farmers earn from agriculture. In all cases, achieving the estimated potential income from field crops depends on improvements in accessibility, input supply and the research and extension system discussed in Chapter 4. Table 1.4-1: Indexe, of Cultivated Areas. SUDPIY of Inputs. Production and Yield Scale of Input and Aroe Covored -Scale of Return Irrigation HYV', Total Annual Yield Cultivatod Annual Irrigated Area Sale of Produc- Col.6/Col.1 Year Area Invostment Arca Covered Fertilizer tion .100 1 2 8 6 6 6 1976 100.00 100.00 100.00 100.00 100. 00 100.00 100.00 1976 102.24 131.66 112.78 100.68 85.62 108.44 101.17 1977 102.80 170.98 18.608 110.41 104.06 98.04 K5.87 1976 108.78 1983.14 211.84 165.21 124.58 94.89 91.48 1979 108.73 830.70 291.08 144.00 125.38 96.61 98.14 1980 102.80 810.18 486.72 166.86 187.97 865.18 62.86 1981 106.16 B84.48 469.60 164.67 148.61 101.86 95.47 1982 108.88 479.16 520.6G 218.70 157.99 106.43 97.30 1983 112.65 649.21 698.41 192.17 202.77 88.67 76.71 1984 114.80 726.12 655.98 182.07 289.05 118.68 98.89 1985 119.98 866.07 798.75 188.00 275.86 111.48 92.90 1986 124.60 1125.20 884.52 204.78 281.08 117.44 94.26 1987 124.93 1128.26 9s3.67 210.82 290.84 109.68 86.95 1988 184.88 1187.99 1085.57 286.58 840.05 127.18 94.81 * High yielding variety includes Paddy, Wheat and Malze. Source: Column 1 to 4 and 6 to 8 HMG, Npal, 1989 Column S AIC, Nepol, 2046 (Pouch).  - 53 - Table 1.4-2t Incidence of Irrigation by Farm Size (Percent of Area Irrigated) 1/ Source and Data Large Medium Small Marginal NRB 1976177 Terai 18.1 16.8 10.9 10.6 Hills 7.7 7.3 3.9 7.0 MPHBS 1989 Terai - 39.7 35.5 - Hills - 36.8 27.4 - NPC 1983la Terai 40.2 32.0 27.6 24.4 Hills 21.9 25.3 23.1 20.4 /a Percentage of farm households with half or more of their land irrigated. Table I.4-3: Institutional Borrowing by Farm Size 1976-1977 Source of Borrowing (percent) Share of Institutional Proportion of Farm Credit by Farm Land in Size Institutional Private Size Group Size Class (I) (X) Large 42.8 57.2 73.6 42 Medium 30.4 69.6 15.6 24 Small 23.1 76.9 8.0 26 Marginal 9.9 90.1 2.8 8 Overall 24.0 76.0 100.0 100 Sources NRB 1980. 1/ Land size categories (ha.) NRB and NPCs MPHBS: (Average farm size in category) Terai Hills Terai Hills Large 5.1+ 1.02+ Medium 2.4-5.1 0.51-1.02 3.1 0.52 Small 1.02-2.4 0.20-0.51 1.1 0.31 Marginal <1.02 <0.20  - 54 - Table 1.4-4s Characteristics of Farmer Groups in the Hills 'Large- *Medium' 'Small' 'Marginal' (Poor) (Very Poor) Ave. family size 8 6 5 4 Holding size (ha.) >3.0 1.0-3.0 0.5-1.0 <0.5 Z of households 5 22 19 54 Ave. holding size (ha) 5.7 1.7 0.75 0.18 Total cultivated area (X) 34 38 16 12 Household income from agriculture (Z) 76 70 58 47 Family labor used on farm (2) 77 68 56 31 Total household income per month (Rs.) 1284 841 635 462 Total per capita income per month (Rs.) 162 139 128 108 Household agricultural income per month (Rs.) 973 591 372 217 Monthly agricultural income per capita (Rs.) 122 99 74 54 Share of 1985 poverty line income derived from agriculture (Z) 76 61 46 34 Source: IDS (1986), The Land Tenure System in Nepal and MPHBS (1989).  Table 1.4-5: Cropping Sequences and Cropping Patterns Farming TeraI/Dun Valtey-East milt/Atn. valley-East Hilt/Mtn. Slope-East Type Cropping Proportion Cropping Proportion Cropping Proportion Intensity of Area Intensity of Ares Intensity of Area Paddy-FaLtow 100 0.58 Paddy-Fattow 100 0.27 Paddy-Fatlow 100 0.07 Rainfed Paddy-I.C. reLay 120 0.31 Maize-Miltet-Fattow 150 0.37 Maize-MiLLet-FaLLow 120 0.68 Maize-MitLet-Fallow 150 0.04 Maize-Mustard 130 0.17 Maize-Nustard 120 0.04 Maize- Mstard 130 0.07 Upland Paddy-Fallow 100 0.2 Maize-Cereal 110 0.08 Upland Paddy-faLLow 100 0.03 Maize-Potato 110 0.1 Paddy-Oilseed 130 0.06 Paddy-Cereal 160 0.7 Paddy-Cereal 160 0.4 Paddy-W.C. 125 0.21 Naize-Paddy-Fallow 150 0.28 Maize-Paddy-Fatlow 150 0.6 Monsoon Paddy-Cereal 160 0.28 Maize-Potato 150 0.02 Irrigation Jute-Paddy-FaLLow 130 0.21 Maize-Paddy-FaLLow 150 0.2 Maize-Potato 150 0.04 Paddy-Paddy-Faltow 200 0.15 Paddy-Paddy-Fallow 200 0.19 Paddy-Cereal 200 0.85 Paddy-Paddy-Oilseed/ Paddy-Cereal 200 0.59 Maize-Paddy-Faltow 150 0.15 Pulse (r) 225 0.07 Paddy-Paddy-Cereal 300 0.1 Year Round Paddy-Cereal 200 0.45 Kaize-Paddy-U.C. 175 0.1 Irrigated Paddy-Paddy-CereaL 300 0.16 Paddy-Potato 200 0.02 Naize-Paddy-h.C. 175 0.07 U' Sugarcane 200 0.08 Paddy-Potato 200 0.02 Paddy-Paddy-FatLow 200 0.1 Paddy-Paddy-Otlseed/ Paddy-Cereal 200 0.95 Paddy-Paddy-OlLseed/ Pulse 225 0.05 Maize-Paddy-FaLLow 150 0.05 Year Round Pulse 225 0.07 Paddy-Cereal 200 0.59 Irrigated Paddy-Cereal 200 0.5 Paddy-Paddy-Cereal 300 0.15 Future Paddy-Paddy-Cereal 300 0.2 Maize-Paddy-W.C. 175 0.13 Nung-Paddy-Wiheat 300 0.02 Paddy-Potato 200 0.02 Sugarcane 200 0.06 Paddy-Potato 200 0.02 Notes: Cereal- normaLly wheat W.C.(main) = Winter crop - Wheat in the Teral and barley in N/M valleys W.C.(retay)= Winter Crop - Pulse/oltseed (often linseed mixed) Source: Master Plan for Irrigation Oevetopment in Nepal (1989) Cycle 1: Annexes Vol 2 pp.28-29  Table 1.4-5: Cropping Sequences and Cropping Patterns (continued) Farming Terai/Dun Valley-West Hill/Mtn. Valley-West Hill/Mtn. Slope-West Type Cropping Proportion Cropping Proportion Cropping Proportion Intensity of Area Intensity of Area Intensity of Area Paddy-Fallow 100 0.66 Paddy-Fallow 100 0.42 Paddy-Fallow 100 0.06 Maize-Millet-Fallow 150 0.02 Maize-Millet-Fallow 150 0.18 Maize-Millet-Fallow 120 0.37 Rainfed Maize-Mustard 130 0.2 Mafze-Mustard 130 0.11 Maize-Mustard 120 0.4 Maize-Cereal 120 0.1 Maize-Cereal 120 0.21 Maize-Cereal 110 0.44 Upland Paddy-Fallow 100 0.02 Maize-Pulse 120 0.08 Maize-Pulse 110 0.03 Upland Paddy-FaLlow 100 0.05 Potato-Fallow 100 0.02 Paddy-Olfseed 130 0.21 Paddy-Cereal 160 0.75 Paddy-Cereal 160 0.65 Mansoon Paddy-W.C. 125 0.12 Maize-Paddy-Fallow 150 0.24 Maize-Paddy-Fallow 150 0.35 Irrigation Paddy-Cereal 160 0.4 Maize-Potato 150 0.01 Paddy-Pulse 130 0.25 Maize-Potato 150 0.2 Paddy-Cereal 200 0.87 Paddy-Cereal 200 0.7 Paddy-CereaL 200 0.75 Year Round Maize-Paddy-W.C. 175 0.02 Paddy-Paddy-Cereal 300 0.1 Maize-Paddy-Fallow 150 0.25 Irrigated Sugarcane 200 0.08 Maize-Paddy-W.C. 175 0.18 Paddy-Potato 200 0.03 Paddy-Potato 200 0.02 Paddy-Paddy-Oltseed/ Paddy-Paddy-Oilseed/ Paddy-Cereal 200 0.75 Pulse 225 0.05 Pulse 225 0.05 Maize-Paddy-Fallow 150 0.25 Year Round Paddy-Cereal 200 0.75 Paddy-Cereal 200 0.7 Irrigated Paddy-Paddy-Cereal 300 0.05 Paddy-Paddy-Cereal 300 0.1 future Maize-Paddy-W.C. 175 0.02 Maize-Paddy-W.C. 175 0.13 Mung-Paddy-Wheat 300 0.05 Paddy-Potato 200 0.02 Sugarcane 200 0.06 Paddy-Potato 200 0.02 Notes: Cerealt noratlly wheat W.C.(main) a Winter crop - Wheat in the Terai and barley in HIM valleys W.C.(relay)= Winter Crop - Pulse/oilseed (often linseed mixed) Source: Master Plan for Irrigation Development in Nepal (1989) Cycle 1: Annexes Vol 2 pp.28-29  Table 1.4-6 Yield Assumptions for Farm Models* (Metric Tons per Hectare) Terai Nill Valleys Hill Slopes Rainfed Irrigated Rainfed Irrigated Rainfed Irrigated Current Foreseeable Current Foreseeable Current Foreseeable Spring Paddy - 3.00 3.80 3.50 4.00 - - Paddy Main 1.60 2.50 4.10 2.00 3.00 3.80 1.25 1.80 2.50 Follow Paddy - 2.20 3.80 - 2.70 3.50 - - - Upland Paddy 1.20 - 1.25 - - 1.00 - Maize 1.50 2.00 2.70 1.60 2.70 3.20 1.20 1.50 2.25 Wheat 0.90 2.20 2.70 1.60 2.70 3.20 1.20 1.50 2.25 Barley 0.85 1.20 1.50 0.75 1.20 1.50 1.00 - Ut Millet 1.00 - - 1.40 - 0.80 .- Pulse 0.45 1.00 1.50 0.70 1.00 1.50 1.00 - - Hung - 1.00 1.30 - - - - - - oflseed 0.50 0.75 1.00 0.30 0.60 1.00 0.25 - Potato - 10.00 14.00 . 10.00 14.00 5.00 10.00 15.00 *from the Master Plan for Irrigation Development in Nepal 1989  - 58 - Table 1.4-7s Potential Income Per Hectare from Field Crops in the Hills (1988/89 Re. per annum) East Valley Hill Floor Slopes Rs. Rs. Rainfed (Actual) 7,244 4,827 Current year-round irrigation (Potential) 24,569 11,903 Future year-round irrigation (Potential) 27,674 18,520 Sources Based on Master Plan for Irrigation Development in Nepal (1989), Cycle 1, Annexes-Volume 2. Yield and cropping pattern assumptions are shown in Tables 1.4-5 and I.4-6. The calculation of potential income from current year-round irrigation assumes that farmers have reliable access to water and chemical inputs all year. The calculation of potential income from future year-round irrigation assumes that irrigation water control improves and that farmers increase their input use accordingly. Table 1.4-8s Minimum Land Holding Necessary to Earn a Poverty Line Income from Agriculture in the Hills 2/ (hectares) Eastern Nepal Valley Slope Rainfed 1.39 2.08 Half Irrigated 0.63 1.21 (current technology) Half Irrigated 0.58 0.86 (foreseeable technology) All Irrigated 0.41 0.85 (current technology) All Irrigated 0.36 0.54 (foreseeable technology) 2/ Based on Table 1.4-7, assuming a family size of 4.  - 59 - Table 1.4-9s Estimated Household Income to be Derived from Horticultural Production on 0.18 ha. Monthly Proportion of Farm per capita Poverty Line Income Income Income Cropping System Re. Rs. I Bananalpineapple 5.030 105 50 Orangellimelintercrop 8,490 177 84 Orange/lime/pear/intercrop 8,020 167 80 Apple/wheat 9,440 197 94 Source: Based on Nepal Hill Fruit Develogment Proiject (1987). Adapted to a farm of 0.18 ha and household size of four. Table I.4-10: Household Income and Landholdings in the Terai by Level of Poverty Very Poor Poor Non-poor Family Size 6.33 7.14 6.76 Land Owned (ha.) 0.47 0.97 2.85 Land Operated (ha.) 0.52 1.13 2.78 Rhet Land Operated (ha.) 0.24 0.95 2.30 Monthly Income Per Capita (Rs.) 79 156 340 Total Monthly Income (Re.) 497 1113 2299 Total Monthly Agricultural Income (Rs.) 186 557 1668 Monthly Agricultural Income Per Capita (Rs.) 29 79 246 Household Income from Agriculture (X) 37Z SOZ 73S Share of 1985 Poverty Line derived from Agriculture (Z) 151 402 127X Sourcet MPHBS (expressed in 1988189 Rupees).  - 60 - Table 1.4-lls Potential Income per Hectare from Field Crops in the Terai (1988/89 Rs./ha./yr.) East West Rainfed (Actual) 6,376 5,354 Current year-round irrigation (Potential) 18,364 16,976 Future year-round irrigation (Potential) 28,228 23,662 Sources Master Plan for Irrigation (1989). Yield and cropping pattern assumptions shown in Tables I.4-5 and 1.4-6. The calculation of potential income from current year-round irrigation assumes that farmers have reliable access to water and chemical inputs all year. The calculation of potential income from future year-round irrigation assumes that irrigation water control improves and that farmers increase their input use accordingly.  - 61 - Table I.4-12: Estimates of Total Labor Use in Agriculture Estimates of labor use per hectare are based on the Master Plan for Irrigation Development in Nepal (1989) estimates of cropping patterns and labor/hectare/crop for rainfed and irrigated land. Estimates of total irrigated land and potentially irrigated land are based on the same source. The total number of full-time jobs iD. agriculture is calculated using 180 labour days/year which is somewhat lower than most estimates of full-time employment. However, the calculation of labor days/ha/year includes only labor used for individual field crops. It does not include time spent on livestock herding, fodder collection, irrigation maintenance and other activities necessary for agricultural production, but not directly related to any single crop. All three estimates assume increased utilization of inorganic fertilizer and pesticides in areas with non-monsoon irrigation. This, in turn, requires that: (i) the input distribution system improves significantly; (ii) transportation costs in the hills do not make fertilizer use uneconomical; and (iii) farmers will be able to finance the purchase of inputs either through savings or credit. Assumptions Low Estimate: Hills: No increase in areas irrigated. No new irrigation technology. Low input use in inaccessible areas. Two-thirds of irrigated land has only monsoon irrigation. Production improvements come from improved management of existing irrigation systems and increased input use in accessible areas. Terai: No expansion in area cropped. No new irrigation technology. Production improvements come from improved management of existing irrigation systems and increased input use. Medium Estimate: Hills: Irrigation expanded to all valley floor land using current irrigation technology. Improved access to inputs where there is irrigation. Terai: Some expansion in area cropped and area irrigated. No new irrigation technology. High Estimate: Hills: Expanded area irrigated. Improved irrigation technology. Improved access to inputs. Terai: Expanded area cropped and irrigated. Improved irrigation technology.  - 62 - Table I.4-12a: Low Estimate Labor Cropped Days/Hal Total Labor Hectares Year Days/Year (000) Hills Rainfed Valley 316,785 210 66,524 Slope 846,666 161 136,313 Irrigated Valley Accessible 40,000 408 16,320 Inaccessible 78,000 267 20,826 Slope Accessible 23,000 383 8,809 Inaccessible 47,000 264 12,408 Total Hills: 261,200 Terai Rainfed 638,000 131 83,578 Irrigated 721,000 289 208,369 Total Terais 291,947 Total Nepals 553,147 Total number of full-time jobs in agriculture: 3.07 million  - 63 - Table I.4-12b: Medium Estimate Labor Cropped Days/Ha/ Total Labor Hectares Year Days/Year (000) Hills Rainfed Slope 846,666 161 136,313 Irrigated Valley 273,748 408 111,689 Slope 161,202 383 61,740 Total Hills: 309,742 Terai Rainfed 400,000 131 52,400 Irrigated 1,100,000 289 317,900 Total Terai: 370,300 Total Nepal: 680,042 Total pumber of full-time jobs in agriculture: 3.79 million  - 64 - Table I.4-12cs HiSh Estimate Labor Cropped DaysiHa/ Total Labor Hectares Year Days/Year (000) Hills Rainfed Slope 846,666 161 136,313 Irrigated Valley 273,748 476 130,304 Slope 161,202 428 68,994 Total Hills: 335,611 Terai Rainfed 421,000 131 55,151 Irrigated 1,338,000 315 421,470 Total Terai: 476,621 Total Nepal: 812,232 Total number of full-time jobs in agriculture: 4.51 million  _ 65 - ANNEX I.5 - THE FORMAL SECTOR A. Introduction 1. This annex examines the current and potential contribution of the formal sector to poverty alleviation in Nepal. It includes manufacturing, government, construction, utilities, and some proportion of the trade and service subsectors. There is clearly a lot of overlap with the informal sector (Annex I.6), especially in the areas of cottage industries, tourism, and commerce. We have defined the foimal sector as consisting of those establishments with more than 10 employees (consistent with the Ministry of Industry's definition); we have also somewhat arbitrarily incluled construction and tourism under the formal sector, although they clearly straddle both the formal and non-formal sectors. The following sections examine the various subsectors in turn. B. Sectoral Employment and Incomes of the Poor Manufacturing 2. The manufacturing sector employs about 150,000 persons in 9,400 establishments. It accounts for about 6? of GDP, but only 2? of the labour force. Table 1.5-1: Manufacturing Employment and Output - 1986/87 (Employment in numbers, values in current Rs. millions) Gross Value Industry Employment Output Added Structural Clay 41,492 456 275 Grain Mills 21,890 5,295 945 Other Agricultural Processing 12,714 795 360 Textiles and Garments 17,570 860 380 Jute, Spinning and Weaving 15,730 1,000 390 Sawmills and Furniture 6,740 335 130 Printing and Allied 3,400 153 70 Metal Fabrication, etc. 3,250 587 160 Total (includes misc. others)s 152,580 13,540 4,490 Source; CBS, Census of Manufacturers. 3. The manufacturing sector is characterized by very basic activities - mostly brick-making and grain-milling (between them 701 of manufacturing establishments, 40X of employment, and 42? of output). More recently garments, carpets, and some modern fabricating industries have started to emerge, accounting between them for about 15? of both employment and output.  - 66 - 4. There has been sustained healthy growth in manufacturing output (about 7? p.a. over the 10 years to 1986/87, and 122 p.a. over the five years 1981182-86f87);l/ and impressive employment growth (about 10 p.a.), led by growth in textiles and brickmaking - which have increased by three- fold over the last five years, to become the two major employers. (Excepting the large number of small grainmills, which are basically service plants). Table 1.6-2: Key Changes In the Manufacturing Sector 1976 77-195/87 (Employment in numbero, output in constant 1988/97 Ra. o I lIons) 1976/77 1888/87 Averaoe Annual Orowth Gross Gross gross Outptt Employment Output Employment Outout Employment Food and Allied 8790 20,836 4,940 82,124 -56 3/ 4.5X Drink/Tobacco 847 8,887 888 11,195 0 51 Textiles and Weaving 4S1 8,323 2,154 385,75 18X 18X Wood/Paper/Print 878 5,970 805 11,829 5x 7.5X Plastics/Chemicals 41 740 88B 7,S80 8X 25X Non-Metallic 158 7,650 988 45,490 201 18X MetaIIIc Subsector 119 1.890 771 5.810 21X 15X Total 10,978 59,037 13,637 152,679 21 3/ 10o Incl. misc. others) n/ NOTE: Annual variations In food prices can distort gross output changes whon meeaured point to point - the results are not very .oaningful. Source: Annex III, Table 4.1. 5. The impressive (proportional) employment growth is on a very small base; in absolute terms the manufacturing sector has been adding only 9,000 jobs a year, compared to labour force growth of about 250,000 per year. Despite rapid growth, total employment in carpets, rugs, textiles, and garments, for instance, was still only 17,000, according to the 1986/87 Census of Manufacturers. 6. It is unlikely that the growth in manufacturing has had much impact on incomes of the poor. The types of jobs created are not those which employ the poor, with their low levels of education and skills. Only about 20? of manufacturing jobs are unskilled, as can be seen in Table I.5-3. 1/ These rates should be treated with caution - based on GDP figures - unrevised 86/87 figures may still be on the high side, and output has since declined as a consequence of the trade & transit dispute.  - 67 - Table 1.5-3, Composition of Manufacturing Employment by Skill Level and Gender (Sample survey, 1983) Male Female Total Managerial and Technical 1,703 103 1,806 (162) Semi-Skilled 6,378 773 7,151 (642) Unskilled 1,818 453 2,271 (202) Total 9,899 1,329 11,228 (88S) (122) SOURCEs Shrestra: Women's Employment in the Industrial Sector; CEDA, 1983. 7. The location of industrial plants - in Kathmandu and the urban terai - makes it all the more unlikely that they will employ the poor (only 52 of the poor live in urban areas). Almost certainly some of the poor are absorbed in unskilled jobs in, for example, brick-making and textiles. However the competition for formal sector jobs, and the nature of job- selection (with the importance of family connections and caste) make 1bt less likely that the truly poor will be employed. 8. Manufacturing growth could potentially make a difference to the large numbers of landless in-migrants to urban centres in the terai. However the scale of growth required to make a significant difference would be enormous. Total manufacturing employment in the terai is estimated, for example, to be in the order of 50-75,000; whereas such in-migrants are expected at a rate of about 125,000 per annum by 2010. 9. Wage levels for the unskilled in manufacturing are better than agricultural and construction wages, but probably not much better in those types of employment which engage the poor. Interviews and reports in Kathmandu suggest that women working in textile plants receive between Rs. 25 and Rs. 50 per day, and data from a sample of industrial establishments gives the average rate for unskilled workers in brickmaking as Rs. 25 per day in 1986/872t - or roughly equivalent to the agricultural wage. The point is that even a major expansion of jobs at these wage levels - while useful - would not be sufficient to raise the poor out of poverty. 21 Economic Services Centre, as reported in New Era, 1989.  - 68 - 10. Trends in labour productivity reveal a slight increase in output per employee: Gross Output per Employee (in constant 1988/89 Rs.) 1976/77 Rs. 70,000 1981/82 Rs. 85,000 1986/87 Rs. 90,000 That is, productivity is increasing, and labour absorption can be expected to expand slightly less rapidly than output growth. This is consistent with experience in other countries (eg. Mexico, the Philippines), where industrial growth has resulted in less employment creation than was expected or hoped for. 11. Expansion of manufacturing employment will also involve substantial capital requirements. The capital labour ratio is estimated at about Rs. 39,000 in 1986187.3/ Therefore to create 100,000 jobs p.a. would require investment of about Rs. 4 billion annually - or 52 of GDP. (Although there may be some scope for low-cost expansion initially due to underutilized capacity). 12. Expansion of the industrial sector is hindered by a number of constraints which have been well documented elsewhere.41 They include: - Nepal's landlocked location - resulting in high input costs, and uncompetitive outputs (it is estimated, for example, that goods may have to be handled as many as 17 times between Calcutta port and the factory gate); - distortions as a result of its situation vis-a-vis India (see Chapter III) which: (a) make it difficult to follow an independent trade-liberalization policy; and (b) provide financial incentives to invest in rent-seeking activities (especially transhipment of goods and smuggling) at the expense of physical plant and employment-generating activities; - a small and fragmented domestic market; - expensive internal communications and lack of infrastructure; and, 3/ Gross fixed assets per employee, excluding food processing ndustries, in 1988/89 Rs., based on Census of Manufactures data; the vulue c-f assets may be underestimated. 4/ See for example Nepal - Policies for Improving Growth and Alleviating Poverty - The World Bank, 1988.  - 69 - - a poorly educated and unskilled workforce. Despite recent encouraging growth, there is little prospect of the Government (or anyone else) being able to lift these constraints in the foreseeable future enough to allow the kind of industrial expansion which is needed to have a serious impact on the employment and incomes of the poor. Construction 13. Construction is probably the single most important source of non- agricultural incomes for the poor. Many are involved on an intermittent basis, for a few weeks or months at a time, and consequently it is difficult to estimate the total number of full-time equivalent jobs with any accuracy. Banskota estimatesS/ that under the 6th Plan 50-100,000 persons per year were engaged in road building. Various estimates suggest that about 300 person-years of work are created per km. of road constructed - so that at the average rate of addition of 250 km. p.a. this would yield about 78,000 jobs - consistent with the ICIMOD estimate. 14. Public expenditure on construction (which accounts for the majority) amounted to about Rs. 6 billion in 1987/886/ This would translate into between 200,000 and 250,000 jobs.7/ If we assume that the private sector contributes another 50X, then toLal construction employment might be in the neighbourhood of 350,000. 15. Estimates vary considerably, but in aggregate less of construction expenditure goes to labour than one might expect - due to, among other things, the types of works undertaken, high transport costs, and large contractors' margins. 5/ H. Banskota, Hill Agriculture and the Wider Market Economy; ICIMOD, 1989. 6/ Construction Industry Study. GDP figures for 1987188 show total construction activity valued at Re. 5.9 billion. If this is correct then construction employment could be much less than we estimate. Further work is needed to determine the magnitude of construction employment. 71 Assuming that between 301 and 372 of expenditure goes to labour, average wage of Rs. 40 per day, and 220 working days per year.  - 70 - 16. Expenditure on public works is broken down approximately as follows: Approximate Breakdown of HMG Construction Expenditure (1987/88) (Rs. billions) Irrigation 1.38 232 Other Agriculture 0.59 10? Power 1.23 202 Roads 1.63 272 Other Transport 0.13 22 Social Infrastructure 0.87 142 Other 0.18 32 In addition there is private sector activity which consists of housing construction, commercial building in urban areas, and some irrigation works. There is of course a wide range of labour-intensity among the various types of works. Hydro projects employ little labour for example, while roads and housing are probably the most labour-intensive. 17. It is estimated that between a third and half of expenditure on roads goes to labour, although the share to unskilled labour is much lower on large highway projects, and higher on rural roads. Of transport expenditure under the 1988/89 Development budget only about a quarter was for district level and feeder roads. Small irrigation projects are also relatively labour intensive. Drinking water projects on the other hand are generally material-intensive, and not a very significant source of employment (Pradhan estimates that only about 152 of expenditure goes in wages). Trail development is the most labour intensive, with almost 90X of expenditure going to local labour. 18. The proportion of unskilled labour in construction is higher than in any other sector, but still amounts to only about one-third to one-half of labour expenditure. These jobs almost certainly absorb the poor - the wage levels are low enough to be self-targetting, and anthropological studies report the landless, poor, and out-migrants as engaging in construction work. 19. A number of experiments have been tried to increase the employment effects of construction works (See Annex II). The Special Public Works Programme gets about 502 of expenditure to labour on small-scale irrigation works implemented expressly to maximize local labour participation. The Lamosung-Jiri road, also designed to be labour-intensive, achieved an average of only about 312 expenditure on labour. One promising area for expansion may be in maintenance. Local engineers estimate that up to 802 of maintenance expenditure could accrue to unskilled labour, as opposed to only 202 under construction works. Table 1.5-4 shows the cost composition of earthworks contracts on the Lamosung-Jiri Road.  - 71 - Table I.5-4: Cost Composition of Earthworks Contracts Project Area Kathmandu Abroad Salaries and Wages 332 82 - 41X Equipment and Tools 12 - 52 6Z Taxes and Interests - 6X - 62 Overheads 3Z 122 - 15X Contractors Profit 12 162 32 20X Sub-Contractors Profits 9X 22 - lX Total 472 442 82 1002 Earthworks Contract - Piece Work Salaries and Wages 412 82 - 492 Equipment and Materials - 22 30Z 322 Taxes and Interests 22 92 - 112 Overheads 22 12 - 3Z Contractors Profits 52 - _ 52 Total 502 202 30X 1002 20. Major works (dams, highways, bridges, etc.) are executed by large contractors, with large amounts of machinery and skilled labour. They are location-specific, and unskilled labour requirements, while substantial in some cases, are tightly specified, so it is difficult to vary the scope to increase labour absorption. It is the wide range of small rural works which will have the most employment impact on the poor. However careful design is needed to ensure that the potential income benefits actually flow down (including, for example, the use of local contractors, piece-work, monitoring of contractors, etc.). 21. The impact of construction expenditure on the poor is limited by the large proportions flowing out of the project area (typically about two- thirds of contract payments). Pradhan has assembled the following data on the distribution of expenditure:  - 72 - Table 1.5-5: Estimated Financial Flows Under Infrastructure Prolects Within Kathmandu or Outside Project Area Other Urban Area Nepal Hountain Road 32? 272 41? Suspension Bridge 352 45Z 31? Drinking Water 302 20S 50? Irrigation - Large Scale 32Z 27X 41t - Small Scale 60? 20? 20Z - Community Type 9o0 5? 5? Trail Construction 90Z 4? 6? Source: P. Pradhan, Public Works and Employment in Nepal, mimeo. 22. In addition, contractors' margins are very high constituting as much as 30? of contract costs, according to Table I.5-4. There is evidence that these could be substantially reduced by altering contractual and supervision arrangements (see discussion of contractual arrangements in Annex I.2). 23. There is thus scope for increasing the impact of construction expenditures on the poor by: (a) reconsidering the complexion of public works budgets (eg. other things being equal, allocating expenditure in favour of small rural works and maintenance rather than large capital projects; (b) increasing the labour intensity of some types of works; and, (c) following methodologies and contractual arrangements which maximize (without abandoning efficiency considerations) the proportion of funds staying in the local area. In aggregate such measures might potentially result in a substantial increase (maybe 50? ?) of inflows to poorer areas and households at the same level of construction expenditure. However more work is needed to assess the potential costs, benefits, and possible strategies. 24. The relevant questions are: how cost effective is it to re-allocate public works budgets towards more employment-creating activities? and, what is the scope for increasing the labour intensity of construction techniques, at acceptable cost, for the existing package of works?  _ 73 - Such a piece of work should include: - an analysis of existing labour-intensive and employment- generation initiatives in Nepal and elsewhere to determine which are successful and replicable; - assembling a base of data on construction activity, employment, and financial flows; - an assessment of the administrative costs, employment benefits, and efficiency trade-offs, involved in various types of labour intensive techniques; - an examir-tion of the mechanics of contracting arrangements, employment conditions, and proposals for supervision arrangements which would maximize the returns to labour at a given level of expenditure; and, - an analysis of various types of construction works to identify those which yield the greatest combination of employment and project benefits. The result should be a proposed program (for HMG and donor support) specifying the magnitude and type of works to be included in such a program, as well as contractual and administrative arrangements for its implementation, and measures to safeguard the employment conditions of labourers. 25. Public works and construction are not, in the long run, going to be a major, sustainable, source of employment in Nepal or anywhere else. They are still small relative to total needs (eg. 350,000 out of a work force of 8 million), and cannot ultimately grow faster than Government revenues. However: - intensified expenditure on rural works can be an important stop-gap measure to support the rural poor until population programs and agricultural improvements take effect; - some of the adjustments outlined above can increase the poverty alleviation impact per unit expenditure; - these expenditures also create productive infrastructure in rural areas which, if there is careful project selection, will also benefit the rural poor; and, - there is room for a substantial increase in expenditure on maintenance, which could have a particularly beneficial impact on the rural poor. Hydroelectricity 26. Hydro development is often cited as a major potential contributor to growth and to foreign exchange earnings. Nepal has bountiful supplies of water and sites suitable for hydro production for export to India. Large  - 74 - scale exports could be in the order of 1,000 MW within the next twenty years. The magnitude of financial benefits could be in the order of USS 295 million equivalent annually,81 or about a lO increase in GDP - potentially an important source of resources to finance growth and services which could benefit the poor - although it is possible that these funds would only substitute for deficits currently financed by concessional aid. 27. The effectiveness of hydro exports in contributing to poverty alleviation of course depends on their distributional impact. The employment effect is minimal - the National Electricity Authority currently employs only 7,000 persons in total, of which only about 1,600 are unskilled. Even during the construction phase, the impact is limited. The following table shows the average numbers employed over a three year period in construction of the Kulekhani hydro-electric project (the largest built to date): Skilled Unskilled Total Dam 1,210 780 1,985 Powerhouse 520 600 1,120 Total 1,725 1,380 3,105 Source: Bjoners: Socio-Economic Effect of the Rulekhani Hydro Electric Project, 1983. 28. From an economic perspective, hydro development is similar to other natural resource exploitation projects such as oil or mineral ore extraction for export. The two main dissimilarities are: (a) the super- large investment costs of hydro and the consequential major importance of the financing arrangements and (b) the geographically limited market. Oil and ores can, in principle, be exported anywhere in the world; the su-plier of hydro is captive to the (usually only one) customer. This again underlines the importance of financing and contractual arrangements. 29. The main similarities are that both types (oil/ore and hydro) of projects generate very little domestic employment, together with substantial amounts of foreign exchange at low domestic resource cost. 'Dutch disease" type problems are therefore important. That is, the impact of the project tends to lead to a substantial appreciation of the exchange rate above levels which would otherwise be sustainable. In turn, this leads to a shift in the domestic terms of trade in favor of non-tradeables and against tradeables. The main tradeables sectors, agriculture, manufacturing, and tourism are all therefore adversely affected. While the flow of revenues generates investable funds potentially available to expand agriculture and industry, at the same time the production incentives facing these sectors (and the incomes generated by them) are reduced, and it 8/ Assuming export of 1,000 KW at a 502 capacity factor, sold at Indian Rupees 0.90 per kwh (US$0.07/kwh).  - 75 - becomes more profitable to invest in the non-tradeable sectors. Real estate is often the big gainer. In the actual experience of the oil- producing countries, this analysis has been amply borne out in practice. In most of them, the development of agriculture and manufacturing has stalled, if not gone backwards. A significant counter-example is Indonesia, which succeeded in expanding rice production in spite of increased oil revenues. In almost all other countries, the tradeables sectors languished (see Alan Gelb's Oil Windfalls: Blessing or Curse? 1988). There is also, of course, a tendency for a marked deterioration of the income distribution. 30. This diagnosis has disturbing implications for Nepal and for poverty alleviation. Without very strong and effective economic management and imaginative counter measures, it seems possible (if not likely) that the poor will be made worse off by hydro projects on a significant scale. Services, Trade, and Transport 31. There is no reliable data on formal sector employment in services, commerce, or transportation, although in aggregate these constitute the largest part of the formal sector after construction. On the basis of occupational breakdowns (contained in the MPHBS, census, and elsewhere), and some broad assumptions about the proportions falling in the formal and non-formal sectors,9/ it appears that about 180,000 people are employed in these sectors (1990), broken down as follows: Transport 20,000 Services 100,000 Commerce 60,000 These consist of the employees of larger commercial establishments - shops, trading firms, and banks; private sector services such as health and education and consulting firms; and employees of the larger trucking companies and airlines. (Remembering that under our definition the majority of workers in trade and services are in the informal sector - which engages approximately another 500,000 persons). 32. No data is available on the skill breakdown of employees in these sectors. They may be expected to use more unskilled, or at least semi- skilled, labour than the manufacturing sector, but given their scale and location (eg. in urban areas), one must assume that except for peons and guards, they employ relatively few of the poor. (Assuming the same sort of distribution as in government, peons might account for 251 of employment - or 45,000 jobs). 9/ The assumptions are that the labour force is broken down as follows: Services (5?), Commerce (3Z), Transport (0.5?) (see Annex 1.2), and that three-quarters of the employees in trade and services and half of transport employees are in the informal sector (arrived at on the basis of interviews in Nepal).  _ 76 - The Government Sector 33. The public sector is estimated to employ about 130,000, broken down as follows: Public Service: 90,240 \a Parastatal Enterprises: 40,000 \b a/ 1987/88 - does not include temporary employees. b/ 1986/87 - may double-count RNAC and Nepal Electricity Corporation. The government, by the nature of its activities, employs the more highly educated, and will absorb few of the poor. There are about 25,000 peon positions in the public service - which may constitute opportunities for the poor. However access to government jobs is strongly influenced by connections of caste and family, so that even the unskilled jobs are unlikely to go to the poor. 34. These data do not include the army - which is almost certainly a significant employer, although not necessarily of the poor, since entry is highly competitive, and tends to be selective of particular ethnic and tribal groups. 35. The large parastatals consist of utilities (about 1/3 of employment), state-owned manufacturing enterprises, and the balance smaller unspecified entities. Again it is unlikely that these will employ the poor, for the same reasons that manufacturing firms and the government as a whole do not. 36. The main effect of Government expenditure on the poor may be through increased economic activity in small towns and road-head bazaars, where the salaries of growing numbers of public servants have fueled phenomenal local growth. There is no analytical evidence on the scale of these multiplier effects, but the very large infusions of cash, where there previously was none, have resulted in very rapid increases in commercial activity. It is not clear, however, that the poor have benefited. It is claimed that the businesses, teashops, and services catering to the public servants are not operated by the poor, and purchase their inputs largely from formal sector sources and Kathmandu. Tourism 37. Nepal was visited by 265,000 tourists in 1988, who spent an average of about $25 per day each. In total, tourism generated about $70 million (2Z of GDP) and contributed 15Z of foreign exchange earnings. Tourism has been gri-wing at 8Z rer year, and could potentially have a  - 77 - substantial effect on incomes; although less on employment, and the distribution of the income effect depends a lot on the type of tourism. 38. There are two aspects to the tourism sectors the formal component (eg. 'star' ranked hotels, organized tour and trekking companies, etc), and the more informal sector effect of trekkers' expenditures in towns and villages - eg. on guesthouses, meals, and locally-engaged porters. The majority of expenditure is of the first type - and tends to remain in Katbmandu (if in Nepal at all - an earlier Bank report estimated that only 602 of gross earnings stayed in the country). About 70-80? of tourists never leave Kathmandu and Pokhara, 102 go on organized treks, and 15? are backpackers. ERL estimates that only about $4 million equivalent (out of the total of $70 m.) is spent in rural areas. 39. A New Era survey found that about 6,650 persons were employed in tourism businesses in 1988 - of which about 302 appear to be unskilled workers. Adjusting the sample coverage for smaller lodges and restaurants it is likely that about 10,000 are employed in the sector - the overwhelming majority of them in Kathmandu and Pokhara. 40. We estimate that direct expenditure on local goods and services probably creates the equivalent of another 10,000 full-time equivalent person-years of employment in rural areas. These benefits, however, are highly concentrated in specific areas and among specific groups. About 601 of trekkers are backpackers, who spend an estimated US$4-5 per day in the hills. The others (who account for the majority of expenditure) go on organized treks - for which they pay about US$32 per day. US$20 of this goes to pay the Sherpas, and only $3 is spent in the hills - the balance going to Kathmandu and agency profits.101 Porters have been almost exclusively Sherpas drawn from the Khumbu region, although some other groups are now also starting to participate. In addition trekkers are concentrated on a very few very narrow routes: to Jomson, Annapurna, Namche Bazaar - and while tourism income has made a vast difference to those living on these trails, there has been very little spillover effect elsewhere. General 41. The non-wage effects of formal sector growth on incomes of the poor is probably much more limited than it would be in other economies. Certainly there is some employment of domestic servants and trickle-down through local services, but this is concentrated almost entirely in urban areas, and the marginal propensity of both firms and well-off individuals to spend on imports is high. There is a multiplier effect through increased demand for agricultural products, but as we have seen, the poor produce little or no output for sale - so the impact of this demand will affect the poor largely if it results in substantial increases in labour intensity which large farmers cannot satisfy with family labour. 101 Sources ERL.  - 78 - C. The Potential Role of the Formal Sector in Poverty Alleviation 42. In this section we pull together the subsectoral results and make some projections of the likely scale of formal sector employment over the next 20 years. It is not reasonable to attempt to second-guess in which sub-sectors growth might or should take place - however in the light of experience in Nepal and elsewhere, we can place some bounds on what is feasible. 43. The central question is: What is the best we can expect from the formal sector in contributing to poverty alleviation? The answer to this gives us some indication of the extent to which (if our objective is poverty alleviation) we should place our hopes in, and concentrate our efforts on, lifting constraints on formal sector expansion. 44. The projections are based on low, expected, and high scenarios for output growth by sector. The middle scenario corresponds to a continuation of relatively healthy GDP and industrial growth achieved over the last four or five years (without adjusting for the temporary decline due to the trade and transit dispute) - that is, sustained GDP growth of about 3.5Z p.a., and growth in manufacturing and services of 6-72 p.a.. (These are in fact optimistic levels, when one remembers we are talking about sustaining them over 20 years.). The lower scenario corresponds more to historical experience - GDP growth averaging about 2-3Z p.a., and the formal sector slowing somewhat to 4-51 p.a. (even this is not a pessimistic scenario.) The high scenario corresponds to a major transformation of the economy, such as has been experienced in countries like Thailand and Korea with rates of growth of 8-10X in industrial output sustained over the 20 year horizon, and GDP growth in the neighbourhood of 4-5S. Table I.5-6 below summari as the results, details are presented in Annex III, Table 4.2, along with the underlying assumptions. Table I.5-6: Proiected Formal Sector Employment - 2010 Total Formal Sector Employment in 2010 (thousands) Low Medium High - With Construction 1,443 2,065 2,857 Proportion of LF: (10.52) (15X) (212) - Without Construction 1,016 1,433 1,928 Proportion of LF: (7.5Z) (10.5) (14Z) Incremental Formal Sector Employment Growth 2010 (jobs created per annum) - With Construction 45,805 104,737 194,000 Proportion of Incremental LF: (122) (27.5Z) (512) - Without Construction 41,534 85,774 147,570 Proportion of Incremental LFs (11z) (22.52) (392)  - 79 - 45. Even vith very optimistic assumptions, only about a fifth of the labour force could be absorbed by the formal sector (including construction) by 2010. Under more likely assumptions, only 15 of the total labour force would be employed in the formal sector; and new employsaent growth, if sustained at relatively high proportional levels, could absorb perhaps a quarter of the 380,000 persons entering the labour market each year. Even this means adding new jobs at something like two-and-a-half times the current rate. (i.e. at 105,000 p.a. as opposed to about 40,000 p.a. in 1990).  - 80 - ANNEX 1.6 - INPORMAL SECTOR DEPLOYMENT AND INCOMES IN NEPAL A. Present Status of the Informal Sector 1. It is evident from the small farm sizes that agriculture is a subsistence activity necessitating the generation of incomes and employment from off-farm activities. In an economy with a very small organized sector, reliance on informal sector activities for subsistence and reduction of under-employment is inevitable. 2. The availability of economic opportunities to low-income households is determined by control of physical assets - both private (land, labour, livestock, implements) and i~blic (forest and water resources) - as well as of social assets (P.ach as kinship or gender relations). These opportunities are regarded as being in the informal sector when they incorporate such features as small-scale operation, simple and labour-intensive technology, unregulated and competitive markets, low barriers to entry and family ownership of enterprises. As this definition implies, informal sector activities are largely residual in nature and, as a result, are treated in a residual way by most economic studies and data collection exercises. Thus, the availability and quality of information on the sector tends to be extremely limited. 3. The discussion in this Annex pieces together fragments of information from numerous studies with either a macro-economic perspective of the Nepal economy or with a sector or area-specific focus. These studies supplement the primary data collected by the Nepal Rastra Bank's Multi-Purpose Household Budget Survey (MPHBS) in 1984-85 which provides the most comprehensive coverage of employment in Nepal undertaken in recent years. The Extent and Type of Employment in the Informal Sector 4. Table I.6-1 presents one estimate of informal sector employment based on the information obtained from the MPHBS. The survey provides the percentage distribution of economically active persons over 62 activity groups. Breaking down the economically active population by the same proportions, and making some assumptions about which occupations fall in the informal sector, yields estimates of the number of main workers engaged in each activity.l/ l/ The classification of people into activity groups has been based on the concept of "main occupation". The percentages of main workers engaged in these activities have been deflated by the proportions of informal work expected to be undertaken in each activity respectively for rural and urban areas. The proportions used are based on observation and experience of rura' Nepal, as well as both the plains and hill regions of neigbhoring states in India. The resulting distribution of main workers in the rural and urban informal sectors has then been multiplied by the numbers of economically actife persons (above ten years of age).  - 81 - Table I.6-1: Estimated Informal Sector Employment - 1990 (Main Occupations) Rural Urban Total Prop. of Prop. Trop. of Workers Informl Workers Informal Workers Informal Occupational Group X of Total '000 Sector X of Total '000 Sector '000 Sector Economically active population > 10 yre. ('000) - 8261 - - 918 - 9179 Sales Workers 2.81 231 26.6x 8.4 78 26.4X 308 25.98 Service Workers 1.5l 120 18.45 6.2 67 19.61 178 14.91 Production Workers 3.1X 264 28.2s 9.8 88 80.31 343 28.7X Transport Worker. 0.1% 12 1.3X 1.9 16 6.1X 29 2.4X General Labour 3.6x 285 81.6X 6.6 51 17.7X 336 28.2% Total: Informal Sector 10.9X 903 100.01 31.7 291 100.0X 1194 100.0X Agriculture 82.2X 6791 84.9 320 7111 Residual (formal) 6.91 see 83.4 807 874 Using the same proportions for 1990 21 and estimated figures for the economically active populations, the current employment of main workers in the informal sector is estimated at 1.2 millions; with around 900,000 (or 752) rural. This represents less than 112 of the economically active population in rural areas and nearly 32? in urban Nepal. (In Table I.6-1, it is interesting to note that the numbers of main workers in the formal sector - including construction - emerging as a residual after deducting the numbers of workers engaged in the agriculture sector, amount to roughly 870,000 in 1990 - very similar to the estimate arrived at independently in Annex 1.5). 5. Within the informal sector, in rural Nepal production workers account for less than 302 of the total while sales and services workers account for 262 and 13? respectively. The catch-all category of 'general labour' is the largest - with nearly 322 of main workers. Given the detailed categories listed in the MPHBS (Annex III, Table 5.1), however, it is difficult to see what, specifically, this large category could be engaged in. It can only be surmised that this represents a pool of marginal (highly underemployed) labour not engaged sufficiently regularly in any activity to warrant classification. Urban Nepal displays a roughly similar distribution amongst these categories though, as might be expected, the share of general labour (182) is significantly lower. 6. Details of the activities included within the broad categories contained in Table I.6-1 are provided in Annex III, Table 5.1. These show the importance of commercial activities (shopkeepers and street vendors), handloom textiles and tailoring, the production of potable alcohol and, inevitably, agricultural implement production and repair in the rural employment profile. The importance of domestic services and 'other production workers* reinforces the impression of a significant floating labour force involved in marginal employment. Largely the same categories are important in urban Nepal, though here alcohol production is limited 21 On the assumption that, in the absence of dramatic economic events, the pattern of employment undergoes relatively slow changes over time.  - 82 - while hotel workers are relatively important. Also relatively important are transport workers, responsible for moving goods in a country where the lack of transport is a major impediment to economic activity, and jewelry makers. 7. The classification of the economically active population in Nepal into categories of main workers does not provide a complete picture of the extent of informal sector employment. In developing countries, much of household income generation, in rural areas in particular, is undertaken on a multi-activity basis. This activity pattern spreads risk, augments incomes and reduces unemployment. Correcting for the distribution of secondary occupations (on the basis of time-use data) suggests that informal sector employment will account for nearly 1.4 million full employed person years or 15? of the available employment (Annex III, Table 5.3). 8. The regional and gender distribution of main workers in the informal sector is presented in Annex III, Table 3.2. The table reveals the relative lack of off-farm economic opportunity in the mountains. The predominance of female workers in the hill areas (though still significantly lower than the numbers of male workers in the hills) can be attributed to social factors which impinge upon the mobility and freedom of action of terai (plains) women. In a predominantly Hindu social milieu, it is inevitable that informal sector activities - needing considerable interaction outside the home - are the main occupation of a much lower proportion of female than male workers both in rural and urban needs. 9. In rural Nepal women are involved mainly as sales and production workers and in urban areas also as service workers. The data show that in the informal sector, women are particularly active as proprietors of retail outlets, sales persons and as domestic service workers in the rural terai and throughout the hills. As production workers, women are particularly involved in textile and garment production and as alcohol producers in the rural hills. There are virtually no women involved as main workers in laundering, blacksmithy and transport (including portering) which are important categories for men. Informal Sector Incomes and their Impact on Poverty 10. There is only fragmentary evidence on incomes generated from informal sector activities. Most studies present evidence on farm and off- table 1.6-2: Household Informal Sector Incomso X Major Sources (IP4S. 198S) (Re. 1984/85 por mon ) Source Rural Urban NEPAL Rs- S -%;:- x -16- X Monthly Total 1192 100.0 1794 100.0 12O8 100.0 Incomo from Informal Sector Non-agri