Nepal is witnessing a transition from a predominantly agrarian economy to one where nonagricultural income opportunities are increasingly important. The significance of wage labor in (semi)industrial production and in the service sector is growing rapidly. Along with these changes, there is increasing labor migration as part of the population seeks work in labor markets in urban areas, both in Nepal and India or further abroad. Within the Nepalese urban labor market, carpet production evolved as one of the most prominent industries during the early 1990s, experiencing high annual growth rates. Yet this “gold rush” was followed by a dramatic decline in the mid and late 1990s that severely affected the industry and its workers. In order to analyze these processes, the author conducted an empirical study from 1996 through 1999 consisting of interviews with business people, workers, and officials from governmental and nongovernmental organizations. The study showed that declining income opportunities have drastically affected savings and remittances. As a consequence, those (male) workers with access to more attractive labor markets sought new destinations, particularly in the Gulf states, whereas women workers have been left behind.
The Nepalese economy is characterized by the overwhelming importance of agriculture (NESAC/UNDP 1998: 99) and has one of the world's highest rates of employment in this sector, estimated at about 93% by international sources (UNDP 1997: 183). Official national figures (HMG/NPC/CBS 1993a, 1997) are lower (81%) but still document the predominance of agriculture. Yet most farmers have increasing difficulty ensuring a sustainable livelihood. They have to cope not only with declining yields but also with shrinking landholdings due to partition of inheritance and occasionally to debt clearance. Moreover, the majority of the Nepalese population is not particularly well equipped for work outside the agricultural sector. Illiteracy is still very high (55.5%; HMG/NPC/CBS 1999: 23). This limits access to labor markets and comprehension of terms of trade, as labor markets for (semi)skilled workers usually require a primary, if not a secondary, education. Whereas the (urban) upper strata of society usually have fairly secure access to both education and employment, the rural majority, ethnic minorities, and low-caste groups still have little or no education (Graner 1998) and uncertain access to unskilled or semiskilled work. Although national economic policies aim at promoting labor-intensive industries such as carpet production in order to reduce negative trade balances and increase employment opportunities and purchasing capacities, their success so far has been limited.
To analyze the role of carpet production in labor markets, a study was carried out, consisting of a quantitative survey of workers in Kathmandu (n = 995) and a qualitative survey of workers in Kathmandu and Jhapa District villages (n = 350). The quantitative survey focused on geographic, demographic (gender, age, family structure), and socioeconomic aspects (education, food security) as well as working histories (year in which work was begun in Kathmandu, networks of access). The qualitative survey included a detailed list of family members (age, education level, work migration) as well as coverage of all workplaces in Kathmandu, incomes, reasons for changing workplace, expenses, and remittances to the household of origin. During the village studies, information collected in Kathmandu was supplemented by including migrants to other destinations such as India or the Gulf states.
Agriculture, wage labor, and migration
Despite the predominance of agriculture, landholdings are extremely small for the majority of the population (Table 1; HMG/NPC/CBS 1993b). Larger holdings are found only among (1) those who received a land grant from the aristocracy at some point in history for religious (Brahmin), military, or administrative services rendered (upper strata of either Chetri castes or ethnic groups, such as Magar or Gurung), or (2) those who were involved in collecting taxes based on scarce agricultural profits (Regmi 1978; Graner 1997: 74ff). For the large majority of farmers, renting of (additional) land and/or wage labor, both within and outside agriculture, has become increasingly necessary. Thus, about 20% of all people active in the economic sector are engaged in wage labor, as documented in the Nepal Living Standard Survey (HMG/NPC/CBS 1997). At the same time, there are great variations between daily wages in the different regions as well as between agricultural and nonagricultural wages (Table 1).
A significant proportion of those engaged in wage labor are involved in labor migration and work far from home (Dixit 1997). Labor migration in Nepal has a long history. Its relevance for household budgets has grown and is increasingly important. During the mid-1990s, an average of 23% of all households received remittances from at least one source (HMG/NPC/CBS 1997: 63), with pronounced regional differences (ranging between 13.6 and 30.8% of all households; Table 1). Remittance values are also highly disparate. They range from a regional average of 8000 NRs for the Western Terai to more than 30,000 NRs in Kathmandu. There is a high rate of domestic remittances, indicating the importance of internal migration (58.4% = 32.6% plus 25.8%; Table 1). However, domestic remittances account for only 44% of the total value. Countries other than India account for an extraordinary proportion (almost 80,000 NRs, compared with 5000–15,000 NRs from all other regions). These countries are also the source of a mere 6% of remittances but account for 23% of the total value. Nonetheless, these figures are much lower than the approximate 35–69 billion NRs estimated by Seddon et al (1998: 4, 6), which corresponds to about 13–25% of the GDP.
Export-oriented carpet production and labor markets
The woolen carpet industry has been one of the most important labor markets in Nepal, especially during the (early) 1990s. Carpet manufacturing in Nepal (Figure 1) received a major impetus when Tibetan refugees fled south in the early 1960s. International donor agencies, for instance, the International Red Cross Society and later the Swiss Agency for Technical Assistance (SATA; now called Swiss Agency for Development and Cooperation, SDC), designed and financed programs aimed at securing the livelihoods of these refugees. Thus, carpet production was established at the 3 largest Tibetan refugee camps in Nepal (Hagen 1975; Gombo 1985: 98ff; ISD 1997; Graner 1999). In the 1970s, production bottlenecks encouraged business people from Nepal or of Tibetan origin to found new manufactories. These new companies were not located within the camps, and labor force recruitment was not confined to the relatively small group of refugees. By 1989, carpet exports reached 1 million m2. During this period, the Rupee was (again) devalued, especially as a result of implementation of the IMF/World Bank's Structural Adjustment Program. Thus, growth rates in financial terms ranged from 31 to 93% in the national currency (NRs) and 17 to 86% in US$ (Figure 2).
These impressive figures, however, must be read with caution. In order to save the economy from downward pressure on prices, the Nepalese government introduced a so-called floor price for all export items in the 1970s. In the case of washed carpets, this minimum rate was 54 US$/m2 in the early 1990s. But many export deals took place and still take place at unofficial rates (Khadka 1998). Exporters agree to reimburse their partners with a certain amount of money via informal channels (the so-called hundi system; unpublished conference paper by Gurung, October 2000, Kathmandu). Therefore, no data are available on the actual financial value of carpet exports (Graner 1999). In addition to dramatically reduced prices, trade volume declined after 1993–1994 so that national statistics recorded a decrease in export values that has continued, with intermittent years of recovery, until today. The reasons were manifold, including an oversaturated European market (due to overproduction in Nepal) and cases of quality decline (use of recycled wool wastage). Above all, however, public debate on working conditions and especially on child labor had a devastating impact on the image of Nepalese carpets. A previously exotic, natural product was now stained by the sweat and tears of innocent children, bringing the “misery behind the looms”—as proclaimed by one of the most active NGOs fighting child labor (CWIN 1993)—straight into living rooms all over Europe.
During the export boom of the early 1990s, carpet manufactories sprang up like mushrooms throughout Kathmandu Valley and demand for laborers, both skilled and unskilled, was enormously high. When demand increased (to more than 3 million m2), production was often subcontracted to smaller manufactories. Many exporters provided financial support to staff who had previously been in charge of their company's production (so-called loom masters). These could then establish their own (small) manufactories. Thus, production spread to small- and smallest-scale manufactories and even private households, where overhead costs were much lower (O'Neill 1997: 139; Graner 1999: 207ff). Many other subcontractors entered the industry, without direct enticement from exporters, by raising their own capital. Yet all of these production units were still linked, directly or indirectly, to export markets.
During this time of increasing demand, even the smallest production units were in a favorable bargaining position vis-à-vis their contractors because orders were usually tightly scheduled and had to be finished in time, possibly at higher rates. Local-market rates rose from 1150–1500 NRs/m2 during 1990–1991 to 1350–1750 NRs/m2 within 1 or 2 years while production costs increased less and at a later time. Hence, this was a time of comparatively high profits (100–300 NRs/m2; Graner 1999: 209). This multilayered structure, however, had extremely negative effects when both demand and export prices plummeted during the mid-1990s. Many of the larger exporting companies concentrated production in their own factories. Overcapacity, coupled with a drastic decline in demand from exporters, meant that small-scale enterprises in particular were no longer in a position to bargain for rates that could guarantee them a good profit margin (of only 20–150 NRs). Thus, declining export volumes and reduced rates had a devastating effect on local carpet production; this was slightly buffered by the devaluation of the Rupee when inflation set in later on.
The labor force: Young migrants from across the country
This situation of the local carpet market has parallels on the labor market. The sharp rise in demand for exports during the late 1980s and early 1990s led to enormous demand for laborers, especially for the labor-intensive processes of carpet production, such as spinning of wool and weaving. Official sources estimate the number of workers at less than 100,000 (HMG/NPC/CBS 1994a,b), whereas sources generally quoted (albeit inaccurate) claim that the labor force employed in carpet manufacturing totaled about 250,000–300,000 persons (Shrestha 1991; Pradhan 1993; Sharma 1996). The actual figures are likely to be about 60,000–80,000 weavers with an additional 30,000–40,000 persons for spinning, carding, and dying of wool, cutting and washing of finished carpets, and technical and managerial staff. By 1996, these figures had declined to some 60,000–80,000 workers, about half of them women.
The increase in demand led to an unprecedented wave of migration to Kathmandu Valley. Laborers usually come from rural areas in districts adjoining Kathmandu Valley, some from other eastern districts, and a few from western Nepal (Figure 3). Analysis of age structure reveals that the labor force consists predominantly of adolescent youths: 80% are 25 or younger and 52% are 20 or younger (Figure 4). The percentage of young people 14 and 15 years of age is low (5%), although personal histories indicate that 199 persons (20%) were under the official working age (younger than 14) when they started working. Within the labor force surveyed, no children declared that they were less than 14 years of age; only a negligible number among those aged 14 or 15 looked younger. This information is confirmed by a recent study on child labor that aimed to estimate the ratio, magnitude, and regional pattern of child labor in Nepalese carpet manufactories (CPS/Rugmark 1999). The study identified 391 children under the age of 14 from a total labor force of 23,049 (1.7%). Among these children, 69.4% of all boys and 65.4% of all girls were between 13 and 14 years old, that is, were to come of age within a few months. Although these data show that child labor has not yet been completely eradicated, they document the tremendous success of the fight against child labor in carpet manufactories.
Laborers usually have low levels of education (World Bank 1998: 4). According to recent field data (n = 991), 48.9% did not complete first grade and a total of 73% did not complete primary schooling (ie, less than fifth grade). Among women, the percentage with no schooling was 69%. Educational levels also exhibit a distinctive regional pattern, with slightly higher levels among workers from the Terai (Figure 3).
Recruitment of laborers, especially during the boom period, was done via family/village networks and middlemen. Many people already employed in the production process brought along (younger) family members or other villagers when returning to Kathmandu after their annual trips home. These unskilled people were then usually trained for about 2–3 months, weaving the monochrome inner fields of carpets, while the skilled workers executed the more complicated designs that required the ability to read design sheets (naxa). During times of high demand, many manufactory owners also engaged so-called thekkadars (contractors), people responsible for recruiting laborers from their home towns/villages or elsewhere. For this purpose, many of them asked for advances from manufactory owners for the parents of newly recruited weavers. However, advance payments were usually not handed over to parents since they were not in a position to demand money to have their children join the labor force in Kathmandu. The village studies provided evidence that some thekkadars even asked for money from the parents in order to accept responsibility and provide food for their children.
Wages and remittances
A relative shortage of labor, coupled with a steadily rising demand for laborers, meant that people searching for work were in a relatively good position. Wages therefore steadily increased during this phase (Table 2) and people taking up employment at a new workplace demanded advance payments (so-called peskii) (O'Neill 1997: 180ff). At the same time, workers were changing workplaces frequently, asking manufacturers to hand over advances; sometimes they left before earning these advance payments. Data on duration of employment (n = 994) indicate that 46.6% had been working at their current manufactory for a year or less. The trend was even higher among men (54.8%).
After the onset of recession, and with increasing concentration of production in larger manufactories, workers experienced rapid deterioration of their bargaining power. Dramatic declines in demand have led to an enormous oversupply of laborers, with a concomitant negative impact upon wages. Owners of small manufacturing units selling their carpets at local markets were driven to bankruptcy by increasing production costs and decreasing carpet prices. Manufactories were closed and unemployment increased. Child laborers in particular were simply released, as it was euphemistically phrased in an impact assessment study on implementation of the Labor Act abolishing child labor (ITAD/NASPEC 1994: 2), as they could most easily be replaced by adult workers in search of work. Since the mid-1990s, workers in carpet manufactories have experienced slowly rising nominal wages but highly inflationary costs of living, leading to a steady decrease in real purchasing power. The price index for basic food items (1990–1999) was about 180% (NRB 1999), while the index for wages was about 150%. Thus, whereas the average wages of weavers were sufficient to cover the subsistence needs of 2–3 persons in the early 1990s, by the late 1990s, average wages only met the basic needs of 1–2 persons (Table 2). In addition, infrequent and low-volume orders resulted in underemployment and a further decline in monthly wages.
These circumstances have had a drastic negative impact upon remittances. In the late 1990s, many workers, especially families, had to use most of their earnings to cover living expenses in Kathmandu. Annual remittances in 1992–1993 amounted to about 1500–7500 NRs per worker and have declined to about 500–5000 NRs since 1995–1996. At the same time, the percentage of workers remitting has also declined. Thus, carpet production has become a nonremitting or small-scale remitting industry in the national economy, especially when compared with international labor migration (HMG/NPC/CBS 1997: 67; Seddon et al 1998). Workers' attempts to ameliorate wages and working conditions have been carried out through unions. In August 1998, for instance, a list was handed over to the Ministry of Industry, the Central Carpet Industry Association, and the Department of Labor. Among other things, it demanded a 50% wage increase, harmonization of wages in all manufactories, improvement of working conditions and health facilities, and kindergartens and evening classes for basic schooling (Galaicha Majdur Union Samiti 1998; Graner 1999: 213). Although 21 of 26 demands have so far been officially agreed to, implementation is lacking.
The labor market at the end of the 1990s was divided into (1) workers in exporting companies and (2) workers in manufactories dependent on subcontracting orders. While the former have successfully voiced their demands and achieved significant improvements, the latter face more severe difficulties. Their wages are lower due to reduced bargaining capacity resulting from less regular orders (ie, higher underemployment), lower quality products (60-knot carpets, whereas many exporting manufactories produce at least some 80- and 100-knot carpets), and low piece rates (350–400 NRs/m2 for 60-knot carpets, compared with exporting manufactories' 390–450 NRs/m2). Demanding higher wages and/or other social benefits may improve a worker's position in the short term but risks increasing production costs beyond economic viability, leading to further closures and unemployment.
Conclusion: A fragile domestic labor market and competition on foreign labor markets
Income conditions in carpet manufactories have severely deteriorated in the last 5 years. Carpet production has accordingly become much less attractive as a goal of labor migration, as illustrated in Jhapa District (n = 80), where labor migration to Kathmandu Valley carpet manufactories increased steadily from 1984, when the first boy was recruited. Within the next few years, many others were also recruited or followed friends and elder brothers or sisters. This migration peaked in 1991–1992, declined suddenly in 1993, and ceased completely in 1994. By then, the first migrants had left for the Gulf states (Kuwait), which are becoming more and more popular for people from all over Nepal (Seddon et al 1998). Access to labor markets in the Gulf states is almost exclusively controlled by manpower agencies (vertical networks), with fees at about 40,000–80,000 NRs, which corresponds to 2–3 years of a carpet worker's salary. This destination for labor migration also has a distinctive gender dimension, as the Nepalese government prohibited recruitment of women to the Gulf states in 1998, following the rape of a migrant Sherpa woman. Thus, women are forced to stay behind, facing the choice between no employment or far less attractive labor markets such as carpet production.
Pressure on rural households to seek income opportunities outside agriculture is increasing. Many farmers who try to improve their livelihoods have little education and face enormous difficulties in finding decently paid wage labor in the rural or urban labor market. Wherever it is based on export production, the domestic labor market is highly sensitive to changes on the world market, as has been shown for carpet manufacturing. It also faces difficulties integrating growing numbers of migrant workers, with supply again exceeding demand. Although the Nepalese government and workers' unions may institute minimum wages, regulations are not always implemented. Conditions are deteriorating in urban labor markets; purchasing capacity and hence remittances are declining, leading to a low-level remittance economy.
Thus, by the late 1990s, many men previously employed in carpet weaving looked for work in other regions of the world, mainly in the Gulf states. These foreign labor markets, however, also face serious competition from South Asian as well as other countries. Although men engaged in these labor markets receive comparatively high wages, newly recruited workers sometimes have to accept wages below official minimum rates. Besides, it is virtually impossible for external bodies such as the Nepalese government or trade unions to monitor conditions. At the same time, women are excluded from this opportunity and are, once again, left behind in far less attractive local and regional labor markets.
The research project was funded by the DFG (German Research Council). The author would like to thank Siddhi Tumbahanghe and Suresh and Surendra Manandhar for their valuable contributions during data collection. Thanks also go to David Seddon (University of East Anglia, Norwich) for his helpful comments on an earlier draft and to the anonymous reviewers.