The Rural Non-farm Economy, Livelihood Strategies and Household Welfare in Rural Pakistan

July 31, 2014 The Rural Non-farm Economy, Livelihood Strategies and Household Welfare in Rural Pakistan Technical Paper for 2nd ADB-Asian Think Tank Development Forum To be held on November 20-21, 2014 Seoul, Republic of Korea By: Dr. Shujaat Farooq Research Economist Pakistan Institute of Development Economics (PIDE), Islamabad Cell no: +92-332-8306825, Email: shujaat@pide.org.pk 0

The Rural Non-farm Economy, Livelihood Strategies and Household Welfare in Rural Pakistan Abstract In this study, an attempt is made to analyze the role of rural non-farm enterprises in employment provision and household welfare in rural Pakistan. The study has used various micro datasets including PPHS-2010, HIES, 2010 and PSLM-2010. The results reveal that majority of the non-farm enterprises are micro-enterprises and or informal with limited representation from the production activities. Majority of the non-farm employment accumulate in paid employment and self employment categories. Non-farm economy is one of the major sources of employment and income for poorer households. The higher non-farm income sources for the poorer reflects the equity enhancing in Pakistan. Though there is more rural poverty among non-farm households; however, it is not worst for paid non-farm worker than paid agricultural worker. Non-farm economy may not have a significant impact to reduce rural poverty, but it has a significant positive impact on child school enrollment in rural Pakistan. Key words: Non-farm sector, rural development, employment, household welfare, poverty reduction JEL Codes: I32, J21, J43, O14, Q10, R11 I. Introduction Traditionally, the structural movement of labor from farm to off-farm sector has widely been acknowledged around the globe, especially in developing world for economic growth and rapid industrialization (Lewis 1954). Rural non-farm economy has mostly been remained unobserved in developing world; however, in recent it has gained considerable attention due to rising risks of poverty and vulnerability. The developing world is increasingly coming to realize the contribution of rural non-farm activities for economic development directly because of its size and its responsiveness to growing agricultural output, lowering pressure on urban migration and establishing export markets, and indirectly through provision of financing, processing, marketing and other 1

support services that accelerate agricultural and non-agricultural growth. 1 Sitting in overcrowded urban capitals, policymakers view the rural non-farm sector as a potential breakwater to control rural-urban migration and reducing pressure on overstretched urban service delivery systems. Sitting in rural compounds, households consider non-farm activities as means of diversifying incomes and reducing seasonal and inter-annual consumption risks. Amid growing landlessness, poor households increasingly depend on non-farm earnings for their survival (Haggblade et al. 2007). Traditional theories have linked up the rural development primarily with agricultural growth, due to its predominance in rural life. During 1980s and onward various socio-economic households surveys conducted in developing countries have revealed the rising reliance of rural economy on non-farm sector (Malik 2008). This emergence cropped up due to globalization and liberalization starting from the late 1980s and early 90s in various developing countries which led new economic opportunities for the private sector and foreign direct investment to expand domestic markets and access new markets. As a result, large exporters, agribusiness firms, and supermarket chains increasingly penetrate in the rural economies, altering the scale and structure of rural supply chains as they do. The enormous increase in the availability of information and communication technology greatly facilitated this potential boon (Haggblade et al. 2007). The dominant growth-centric development paradigm in Pakistan has long been looking to the farm sector for rural poverty alleviation while the non-agricultural activities have widely been ignored. The importance of rural non-farm economy in Pakistan cannot be ignored due to multiple reasons. First, around two-third of the total population and three-fourth of the poor are residing in rural areas where half of the rural households are non-agricultural (GoP, 2010a). Second, agriculture is the main livelihood source in rural Pakistan, however, the 2000 Agricultural Census reveals that only 37 percent of the rural households owned land, and 61 percent of these owned fewer than 5 acres. The overall Gini coefficient of land ownership in Pakistan is 0.66; if rural landless households are included, the Gini coefficient is 0.86. Third, growth in agricultural production in Pakistan is not sustained over time and even higher agricultural growth has minor impacts on the incomes of agricultural laborers and landless households. In the 1 Weersink et al. 1998, Escobal 2001, Lamb 2003, Joliffe 2004, Malik 2008. 2

presence of skewed land distribution and fluctuated agricultural growth, rural non-farm sector is an important pathway for poverty reduction and employment provision. Presently more than half of the rural Pakistani laborers are employed in non-farm activities. The non-farm sector includes all economic activities in rural areas except agriculture, livestock, fishing and hunting. Like other developing countries, rural nonfarm sector in Pakistan is comprises of heterogeneous activities. However it is not performing well due to its small manufacturing base. Non-farm enterprises are generally quite small in terms of assets size, employment level, low human capital, lack of access to finance, limited forward and backward linkages, un-experienced managers and higher closure rates (Sur and Jian 2006). A considerable body of literature has discussed the issues of agriculture and poverty in Pakistan; however, majority of the studies have ignored the role of rural nonfarm economy in poverty alleviation and resources diversification. Few studies, however, have analyzed but in limited range. For example, Adams and Janes (1995) examined sources of non-farm income inequality, and Nasir (1999) analyzed the link of poverty with employment. Arif et al. (2000) viewed the level of poverty among the various farm and non-farm groups. Sur and Jian (2006), World Bank (2007) and Malik (2008) have analyzed the structure of rural non farm economy; however, no comprehensive study has been prepared to analyze its structure, employment provision, labor diversification and contribution in household welfare. In view of the growing importance of rural non-farm activities, this examination is essential for policy formulation to eradicate rural poverty. The present study aims to fill this gap by examining the structure of rural non-farm enterprises in terms of business and employment provision, household livelihood strategies in terms of labor diversification and the impact of household livelihood strategies on household welfare. Poverty status of household and child school enrollment status has been taken as the measure of household welfare. The paper is divided in 7 sections. Section 2 presents the theoretical considerations of dynamics of rural non-farm economy, followed by data description and methodology in section 3. Profile of Pakistani rural non-farm enterprises is given in section 4 while the role of non-farm economy in employment provision and labor 3

diversification has been detailed in section 5. The penultimate section discussed the role of non-farm economy in equity enhancement and household welfare. Conclusions and policy recommendation have been reported in last section. II. Dynamics of Rural Non-Farm Economy and Household Welfare: Some Theoretical Considerations The absence of land or poor land endowments are the key push factors to work in non-farm activities. On the other hand, higher wages in non-farm sector could be the major pull factors. These pull factors would be more strong if farm income is not sufficient to fulfill family needs (Barrett et al. 2001). The pull factors may also become more powerful when higher agricultural productivity stimulates employment generation in non-farm sector through its linkage effects (Haggblade and Hazell 1989). Traditional rural insight, developed in colonial era has notion the non-farm sector as a low productivity sector. Hymer and Resnick ( 1969) contended that peasants produced only food and some basic non-food goods called Z goods, to serve their needs. With rising world economy, labor was induced to move from non-food items to cash crops for exports. As a result, there was a decline of rural non-food goods and an expansion of both exports and imports. The later studies not support this idea due to its non-applicability to the post-colonial era. Ranis and Stewart (1993) revealed that both the micro and macro policies will determine the future path of the economy whether it should produce Z goods or balance sectoral growth or Z goods will be displaced by imported goods or by subsidized urban domestic goods. During late 1980s and early 1990s, the novel liberalization paradigm has launched new opportunities for the private sector in rural areas by lowering government involvement in the production sector and relaxing control on foreign exchange. As a result, massive foreign investment was witnessed in Asia, Africa and Latin America with rising rural non-farm activities. This new era endowed new economic opportunities for some rural suppliers to access new markets in the non-farm sector. They were also exposed others to new threats of competition from cheap manufactured imports and by imposing quantity quality standards that risk excluding undercapitalized rural enterprises on which the rural poor often depend (Haggblade et al. 2007). 4

The empirical evidences from South East and some South Asian countries signify that higher agricultural production spur the expansion of rural non-farm economy, commencing usually near cities, spread eventually to include a broad spectrum of rural economy (Yusuf and Kumar 1996). For example, the late 1970s agricultural reforms in China gave much freedom to farmers to diversify their production strategies. In parallel the restricted urban migration called hukou 2 and massive public investment led to establish Township and Village Enterprises (TVEs) and specialized households (Ravallion 2009). It led two important consequences for the Chinese labor market: first, it absorbed surplus labor and facilitated industrialization without relying on migration, and second, the free entry of TVEs increased competition in the market and created pressure for SOEs reform (ADB 2007). Overseas remittances also stimulate rural economy by raising rural investment, construction activities and agricultural inputs. In some cases, migrants contribute to common funds for local public investments thus creating jobs for local masons and carpenters (Adams 1998, Ellis and Freeman 2004). In Pakistan, the return migrants from Middle East have been establishing their small level businesses by utilizing the gained experienced and saved money at abroad. There are two major types of rural non-farm activities: high labor productivity and lower labor productivity. A common view is that rural off-farm employment is a low productivity sector producing low quality goods (Lanjouw and Lanjouw 2001). From a social welfare perspective, employment creation is important even in lower labor productivity sector, especially when fast growing rural labor force cannot be much absorbed in over-crowded agriculture sector or agricultural employment is not an option for certain sub-groups (Lanjouw 1999, Arif et al. 2000). The evidence suggests that nonfarm activities have the potential to improve rural well-being by absorbing surplus labor, especially for the poor agrarian economies (Barrett et al. 2001). In Pakistan, high agricultural growth has benefited mainly to large and medium farmers, therefore, the incidences of poverty are much higher in pure agricultural zones i.e. Southern Punjab, Rural Sindh (Malik 2005, Arif and Shujaat 2012). It is expected that rising landlessness 2 Under the hukou system, a rural migrant cannot avail urban services without obtaining registration there, which can be difficult and costly particularly for the poor (Ravallion 2009) 5

along with bulk of potential working youth in rural Pakistan has generated low productive rural non-farm activities; however, the low return may also enhance household income and consequent rural welfare. In short, the contribution of rural nonfarm economy is highly appropriate for those agrarian economies that have high risk of poverty, vulnerability and unfavorable labor-land-ratio (Admas and He 1995, Stifel 2010). Finding part-time local non-farm employment is vital for the welfare of small farmers and their families. The average share of non-farm economy in total rural household income is varying across the continents; it is 37 percent in Africa, 47 percent in Latin America and 51 percent in Asia. Primary employment shares, the most widely available indicator of rural non-farm activity, are about 30 percent in Asia and Latin America, 20 percent in West Asia and North Africa, and 10 percent in Africa (see Appendix 1). Various studies found negative correlation between the share of non-farm income and poverty in a number of countries. In case of Ethiopia, farm and non-farm diversification not only offers higher income but also better child nutrition (Barett et al. 2001). In case of Tanzania, it has a positive association with per capita food consumptions (Lanjouw and Feder 2001). Zhu and Luo (2006) found that non-farm income was an important source to reduce income inequality in rural China. A rising trend of rural economy including manufacturing, trade and services can be seen in South Asian countries. It implies that not only the links between agriculture and rural poverty should be examined, but also the rural non-farm sector should receive attention. A dynamic labor-intensive agriculture combining with a modernizing nonagricultural sector in Pakistan can provide diversified employment opportunities to the rural households, resulting rapid growth, classless distribution, diminishing rural unemployment and underemployment and lowering the pressure on rural-urban migration. Special policy orientated attention is required to eradicate rural poverty and hunger by promoting non-farm activities in rural Pakistan. The ongoing paper explores the linkages between non-farm activities and rural welfare in Pakistan. 6

III. Data Sources and Methodology A. Data Sources To analyze the structure of rural non-farm economy, employment diversification and household welfare, the study has used multiple data sources as no single national dataset details all the relevant information to fulfill these objectives. Various rounds of Pakistan Labor Force Survey (LFS) and national statistics have been utilized; however, the study has mostly used three cross-sectional micro datasets which are Pakistan Panel Household Survey (PPHS)-2010, Pakistan Social and Living Measurement Survey (PSLM)-2010 and Household Integrated Economic Survey (HIES)-2010. PPHS 2010 is the third round of panel dataset carried out by the Pakistan Institute of Development Economics (PIDE) where its first two rounds were conducted in 2001 and 2004. The 2010 round covers 4,142 households from all the four provinces with more representation from poor and rural regions (for detail, see Arif and Shujaat (2012)). The present study has used PPHS 2010 to examine the profile of non-farm enterprises as only this dataset provides the latest information. To analyze rural employment, labor diversification and its determinants, Pakistan Social and Living Standards Measurement (PSLM)-2010 has been used. PSLM 2010 is the district and provincial level representative dataset conducted by Pakistan Bureau of Statistics (PBS). Various modules of PSLM collect a wide range of information on household s socio-demographic and economic characteristics including education, health, employment, assets etc. The 2010 PSLM has covered 76,546 households from all the 4 provinces and 114 districts of Pakistan including the urban regions. The study has used HIES 2010 for poverty estimates. HIES is the subset of PSLM survey and representative at provincial level. It is used to estimate official headcount poverty in Pakistan as it has detailed consumption modules covering all aspects of consumption including food and non-food expenditures. The 2010 HIES covers 16,341 households from all the four provinces of Pakistan. B. Methodology It is worth to mention here that the present analysis is carried out only in rural Pakistan. Before explaining methodology, clarification on three concepts is necessary which are non-farm, rural and poverty. Rural non-farm activities lie on or between 7

the boundaries of usual rural-urban and agricultural and non-agricultural categories. The ongoing study has followed the 2007 official industrial classification where agriculture including the livestock has been considered as the farm activities while the non-farm activities include all other employment except agriculture. Regarding rural clarification, both the PSLM and HIES follow rural-urban definition of 1998 census in which the rural towns fall under administrative status are treated as the urban areas, therefore, these towns are not included in present analysis. Regarding poverty measurement, the study has followed the official methodology as defined by The Planning Commission of Pakistan which can be called as Cost of Basic Needs approach. The basket of basic needs consists of food, education, clothing, health, housing, transportation and recreation. The cost of this basket is poverty line (Rs. 723.4 for 2000-01 year), as defined to impart 2,350 calorie intake per adult per equivalent per day with an adjustment of nonfood minimal requirement. The study has inflated the official poverty line for 2010 period by using the Consumer Price Index and applied it on HIES 2010 dataset to measure headcount poverty. The study has defined labor diversification at individual and at household level. At individual level, we construct five categories of workers by using their primary and secondary occupations, as detailed in below table. At household level, the households have been divided into three categories by using the primary nature of work activities of their employed household members, which are; farm households, non-farm households and mix households. Activity Status Main Agriculture Main Non-farm Only Agriculture Only Non-farm Mix activity Definition In which the primary sector of activity is agriculture In which the primary sector of employment is non-farm activity It which primary sector of employment is agriculture and have no secondary sector of employment or the secondary is also agriculture It which primary sector of employment is non-farm and have no secondary sector of employment or the secondary is also non-farm In which the primary sector of employment and secondary sector of employment are varying 8

The following equation has been estimated to find out the determinants of rural labor diversification; Act_cat i = α 0 + α1 I i + α2 Hhi + α3 Rg i + µ i (1) The dependent variable Act_cat represents the employment status of rural worker i with three outcomes; only agriculture, only non-agriculture and mix activities which has been constructed by using the primary and secondary work activities. On right hand side, Ii is vector of independent variables measuring individual characteristics for worker i: gender, age and education, vector Hhi measures household characteristics including household size, land ownership and irrigation status. Vector Rgi measures the community and regional characteristics including the distance to middle school, distance to roads and province. Since the dependent variable has three outcomes, therefore, the multinomial logistic regression has been applied. Headcount poverty rate and child school enrollment status (age 5-15 years) have been taken as the household welfare indicators in this study. The following two equations have been estimated to analyze the impact of labor diversification on household s welfare; Povi = α 0 + α1 Hh_diversificationi + α2i i + α3 Hhi + α4 Rg i + µ i (2) Eni = α 0 + α1 Hh_diversificationi + α2i i + α3 Hhi + α4 Rg i + µ i (3) In equation 2, the dependent variable Povi is the headcount poverty status of rural household i. On right hand side of equation 2, Ii is a vector of independent variables for the head of household characteristics i.e. education, age and gender, vector Hhi is comprises of household characteristics: land ownership and dependency ratio, 3 vector Rgi contains the provincial characteristics. In equation 3, vector Ii represents the individual characteristics of children i: age and gender, vector Hhi is comprises of parental and household characteristics including the education of mother, father, household size and presence of TV, and vector Rgi contains the provincial dummies. The key independent variable is the household diversification with three outcomes; farm households, non-farm households and mix households. In equation 2, the dependent variable poverty status has two outcomes; poor and non-poor while in equation 3 the school enrollment status has 3 Dependency ratio is number of dependent (below age 15 and above age 64) divide by number of independent (age 15-64). 9

two outcomes; whether child of age 5-15 years currently enrolled or not enrolled. Since both the dependent variables have two outcomes, therefore, the Binary Logistic Regression has been applied to estimate the impact of household labor diversification on household s welfare. IV. Profile of Rural Non-Farm Economy in Pakistan There is no precise number of rural non-farm enterprises in Pakistan but the extrapolation from PSLM 2010 survey suggests that there are about 5 million rural nonfarm enterprises in Pakistan. The survey indicates that 23 percent of the Pakistani rural households are the proprietor or partner of non-farm enterprises. Across the provinces, rural households in Punjab own 26 percent, followed by 24 percent in Khyber Pakhtunkah (KP), 18 percent in Balochistan and 15 percent in Sindh. Like other South Asian countries, majority of the non-farm enterprises are micro-enterprises (96.6%), run by only single person, while 2.8 percent of the enterprises are employing less than 10 persons and rest of are employing 10 and above workers. Rural enterprises are predominantly related to trade and services activities. Production enterprises account for only 12 percent of the enterprises, trade for 50 percent and services for the remaining 38 percent. The share of production enterprises is quietly small in Sindh and KP provinces (Figure 1). This share is also less than the other South Asian countries; it is 27 percent in Bangladesh and 40 percent in Sri Lanka (ADB and World Bank 2005). 70 60 50 40 30 20 10 0 Figure 1: Sectoral Distribution of Rural Nonfarm Enterprises (in %) 61 56 46 41 35 31 19 13 15 8 9 12 Punjab sindh KP Balochistan Overall 66 50 38 Production Trade Services Source: Authors estimation from PSLM 2010 micro dataset 10

PPHS 2010 survey discloses that most of the rural enterprises in Pakistan can be considered to be informal businesses, not only do they have less than 10 workers, but very few of them pay taxes (11%). Majority of the enterprises are fairly young as more than half of the enterprises have been in operation for less than 10 years. Services activities tend to be older (13.7 years) than the manufacturing/production (9.4 years) and trading enterprises (10.5 years). Majority of the non-farm enterprises (94%) are owned and operated by men. This number is not surprising as female labor force participation in Pakistan for non-agricultural activities is only 14 percent, the lowest in South Asia. About one-fifth of the managers have completed secondary education while more than one-third had no schooling (Table 1). Table 1: Province-wise Profile of Rural Non-Farm Enterprises in Pakistan Overall Punjab Sindh KP Balochistan Enterprise Profile Sole-proprietorships (%) 95.1 95.9 93.2 96.0 91.7 Pay any tax to govt. (%) 10.8 11.2 8.2 16.0 9.1 Average Age of enterprise (years) 11.3 12.0 7.4 16.8 12.7 Manager Profile Manager is male (%) 94.1 92.4 97.3 96 100.0 Manager Education (in grades) by category Illiterate 35.8 38.8 29.7 20.0 58.3 1-5 grade 24.8 22.5 35.1 20.0 8.3 6-9 grade 20.9 25.5 10.8 20.0 8.3 10 and above 18.6 13.3 24.3 40.0 25.0 Source: Authors estimation from the PPHS 2010 micro dataset Like other developing countries, rural non-farm enterprises primarily operate as sole-proprietorships in Pakistan with more family workers captivation. On average, these enterprises hire 1.3 full-time and 0.9 part-time workers including the paid and family workers, thus totaling to 2.3 workers, on average. More than one-third of the rural enterprises have hired family workers either full time or part time, while near to one-fifth of rural enterprises have reported to hire paid workers. More than three-fourth of the enterprises hire only one worker, either paid or unpaid, while a minor percentage of only 2.6 percent employee more than 5 workers. There is no representation of such enterprises from KP and Balochistan provinces (Table 2). 11

Table 2: Province-wise Employment Size of Rural Non-Farm Enterprises in Pakistan Employment Distribution Overall Punjab Sindh KP Balochistan Average number of workers 2.3 2.6 1.7 1.5 1.4 Average full time workers only (in numbers) 1.3 1.3 1.6 1.3 0.8 Average part time workers only (in numbers) 0.9 1.3 1.9 0.3 0.7 Employment size distribution of enterprise Less than 2 workers 77.9 81.6 77.0 60.0 58.3 2-5 workers 19.5 15.3 20.3 40.0 41.7 More than 5 workers 2.6 3.1 2.7 0.0 0.0 Enterprise hiring full time paid workers (%) 15.3 10.7 21.6 36.0 8.3 Enterprise hiring full time family Paid workers (%) 26.1 18.9 32.4 52.0 50.0 Enterprise hiring part time paid workers (%) 4.2 4.1 2.7 8.0 8.3 Enterprise hiring par time family Paid workers (%) 13.4 14.3 10.8 4.0 33.3 Note: Manager is not included in employment calculation Source: Authors estimation from the PPHS 2010 micro dataset Labor productivity in these enterprises varies widely across activities and gender. A comparison of rural non-agricultural daily wages in Table 3 reveals that services sector has highest wages while manufacturing sector has the lowest wages in both Pakistan and Sudan. Pakistani male, on average, earns 2 to 3 times more wages than the females. More than half of the rural non-farm enterprises are located in homes either inside or outside the residence, with a minor percentage at road side, main commercial area or industrial sides (Figure 2). PPHS survey reveals that 87 percent of the enterprises sell their products in the same village/town, followed by 6.9 percent to cities, 5.3 percent to other villages and only 0.7 percent to other provinces and countries. Table 3: Average Daily Wage in Rural Non-farm Sector by Type of Activity Type of Industry Pakistan (in Rupees) Sudan Both sexes Male Female (in Sudani Pounds) Agriculture 327 357 90 - Manufacturing 232 238 89 21-23 Trade/Commerce 321 323 177 75-80 Services 350 365 202 150-180 Source: Authors estimation from PSLM 2010 for Pakistan, Haggblade et al. (2007) for Sudan 12

{{{{{{{{{{{{ Figure 2: Place of Business of Rural Non-farm Enterprises (%) mobile other fixed place roadside baazar (like weekly)/baazar mandi main commercial area local market/ baazar industrial site home, outside residence home, inside residence 1 2 2 2 7 9 0 10 20 30 40 21 26 30 Source: Authors estimation from the PPHS 2010 micro dataset Table 4 shows that rural non-farm enterprises in Pakistan are generally quite in terms of asset size and business turnover. Annual median sale revenues range from Rs. 110 thousand to Rs. 201 thousand across the provinces with an average of Rs. 130 thousand at national level. Annual profit seems to be low with a median value of Rs. 60 thousand only. Median value of assets including the fixed assets and property is very low. Lower asset values reflect the lower productivity of workers in these enterprises. Table 4: Sale and Assets of Non-Farm Enterprises in Pakistan (in 000 Rs) Overall Punjab Sindh KP Balochistan Annual Sales revenues (mean) 443 343 500 113 278 Annual Sales revenues (median) 130 120 110 125 201 Annual net profit (median) 60 60 60 90 80 Value of property (Land/Building) (median) 25 20 18 130 120 Value of assets (median) 2 2.5-4 - Asset per worker (median) 1.2 1.5-3.2 - Source: Authors estimation from the PPHS 2010 micro dataset Remote locations frequently mandate long and tortuous supply chains linking rural producers with final markets. Rural infrastructure seems to be one of the major obstacles for non-farm enterprises in Pakistan to their operation and growth including the access to finance, human capital, physical capital and access to markets. The ongoing research has plotted both the soft rural infrastructure and physical rural infrastructure at 13

district level in which the district literacy rate (%) and average distance to secondary school (in km) has been taken for the former and percentage of villages who have access to metallic road with less than 1 km and average distance to commercial bank (in km) has been taken for the later. As shown in Figure 3, on y-axis percentage of rural households who own non-farm enterprises has been plotted with the soft and physical infrastructure indicators. The trend reveals that rural households in those districts, who have, on average, higher literacy and less distance to secondary schools, own more non-farm enterprises. Similarly, households with easy access to metallic road and commercial banks own more non-farm businesses in their districts. Figure 3: District level Rural Infrastructure & Households own Non-farm Enterprise (in %) 0 0 10 20 30 40 10 20 30 40 20 40 60 80 literacy_rate Fitted values non_farm 0 20 40 60 distance_high_school Fitted values non_farm 0 0 10 20 30 40 10 20 30 40 0 20 40 60 80 100 access_metal_road Fitted values non_farm 0 50 100 150 200 distance_commerical_bank Fitted values non_farm Source: District level rural infrastructure data has been taken from Mouza Statistics, GoP (2008) Very few of them use the modern business services i.e. accounting, marketing, insurance and information technology (see appendix 2). The profile of these enterprises as presented above suggests that these rural enterprises are poorly equipped to provide the catalytic higher rural growth, decent job generations and poverty reduction as required by these enterprises to cope with the rising risks and competitions of associated 14

with globalizations. Relatively fewer shares of production enterprises in Pakistan highlight the missed potential for value addition. There seems absence of essential agricultural support services and linkages, necessary to stimulate the growth of non-farm sector. With poor equipment including the human, physical and financial margins along with regional disparities, often restrict low income households to run low productivity enterprises with higher labor intensity and lower financial returns. V. Employment Provision and Labor Diversification in Rural Non-farm Economy This section analyzes the role of non-farm sector in employment provision and household s livelihood strategies that how they individually and collectively allocate labor supply between farm and off-farm activities. Functionally, rural non-farm economy is supposed to play a central role in the process of structural transformation, during which the agricultural share in national income declines and transfers to manufacturing and services sector. Similar structural transformation can be perceived in Pakistan by transferring agricultural share to services sector and stagnant share of industrial sector over the last four decades (see appendix 3). In parallel, inter-temporal labor movement also took place with declining share of agricultural employment, but it is still a dominant source of rural employment. An impressive rural employment growth in trade and construction activities can be seen overtime with some marginal increase in services sector, especially the professional services (see appendix 4). One major realization, however in Pakistan is that the share of labor associated with agriculture has not declined at the same pace as the share of agriculture has declined overtime. Within non-farm employment, four subsectors manufacturing, construction, commerce and service are the more important for employment provision in rural Pakistan (Figure 4). 15

50 40 30 20 10 0 21 Figure 4: Share of Major Sectors in GDP and in Employment, 2013-14 44 17 14 Agriculture Manu. & mining Construction Wholesale & Trade 2 7 19 14 13 5 Transport & Communication 2 1 Electricity, gas & water 27 15 others Share in GDP (%) Share in Eemployment (%) Note: others include finance & insurance, housing, private and government services Source: GoP, Government of Pakistan, 2014 There are several reasons for this structural shift. Overseas migration and return migration seems to be one of the major factors in some rural regions, especially in north and central Punjab and north KP. Arif and Irfan (1997) have examined occupational shifts among migrants returned from the Middle East and found a clear move out of the production sector into business activities, with a decline in production sector employment by 40 percent. Unequal land distribution could be another reason as only 37 percent of the rural households own land. Though three land reforms have so far been carried out in Pakistan; however, they remained unsuccessful for right allocation. The 2000 Agricultural Census shows that near to two-third of land owner households are small farmers (own less than 5 acres) who own only 15 percent of the cultivated land while only 2 percent large farmers (own 50 acres or more) own 30 percent of the total cultivated land. In Pakistan, land is distributed far more unevenly than income (Adams, 1995; Hirashima, 2009). Other reasons of establishing the non-farm activities could be the education, public investment in rural areas, and higher agricultural income for landholder households. A. Rural Non-farm Labor Diversification in Pakistan Both the micro and macro factors determine the livelihood strategies of households to allocate labor in farm and off-farm activities. These factors vary across the households and regions as according to the available opportunities and risks. Non-farm 16

employment activities can be classified into four major categories; employer, selfemployed, paid employed and unpaid family helper. Table 5 shows that non-agricultural income accumulates to rural households primarily come through the paid and self employment; both these categories constitute around 93 percent of all non-farm workers while unpaid family worker and employer contribute only 6 percent and 0.8 percent of their share. There is also marked difference in the nature of activities as near to two-third of the self employed workers are engaged in trade and transport services. Service and manufacturing are other important sectors for this category. Service sector and the manufacturing sector are the other important sectors to provide the livelihood to rural workers. For wage employees, services and construction activities account for as much as two-thirds of rural non-farm employment. Transport and manufacturing were the other important sectors of employment for wage employees. Government employees, especially in educational and health activities account for significant proportion of rural services sector. Table 5: Rural Non-farm Employment in Pakistan by Employment Type and Status (in %) Type of Industry All nonfarm workers 1996-97 2010-11 Self employed Paid employees All non-farm workers Self employed Paid employees Mining 0.5 0.4 2.9 1.7 0.4 1.7 Manufacturing 13.4 13.2 11.9 11.3 8.7 11.4 Electricity gas & water 1.5 0.1 1.9 1.2 0.3 1.6 Construction 24.1 2.5 31.7 22.4 2.0 31.4 Whole sale & retail trade 19.3 55.7 6.4 23.1 54.7 9.6 Transport &communication 12.1 12.4 12.8 11.7 14.7 11.1 Professional services 0.9 0.5 1.0 1.03 1.2 1.0 Social & personal services 28.3 15.2 31.4 27.6 18.2 32.2 % share 100 20.2 73.6 100 24.3 68.9 Total 100 100 100 100 100 100 Source: Authors estimation from the PSLM 2010 micro dataset, Arif et al. (2000) for 1996-97 numbers Though results not given in table, the distribution of females in rural non-farm activities is quite different with a significant share of unpaid family helper (18%), while 17

64 percent are the self-employed and 18 percent fall in paid category. Only the services (social and personal) and manufacturing sector contributes to 83 percent in non-farm employment with their shares of 58 percent and 25 percent, respectively. Both the production (20%) and services (71%) only contributes to 91 percent employment for paid employed females while production (26%), trade (20%) and services (53%) are the main employment sources for self-employed females. PSLM survey provides the information on both the primary and secondary nature of work activities. Males have almost equal distribution in farm and off-farm activities on primary activities than females, while both the sexes are mostly involved in farm activities in their secondary occupations (appendix 5). As detailed in methodology, the ongoing study has estimated the labor diversification at individual and household level, where the former diversification has been classified into five categories on the basis of primary and secondary nature of work activity while the later has three categories. As given in table 5, the individual level employment diversification shows that majority of the rural workers have only primary sector of employment, either agriculture (53%) or non-farm (40%) while only 3 percent of the workers have also reported their secondary occupations. Only 5 percent of the workers are differ over their primary and secondary work activities called mix activity. The female s distribution is highly skewed toward only the farm activities with a very lower percentage in off-farm and mix activities. Table 6: Distribution of Rural Employed Worker by Employment Diversification (in %) Activity Type Male Female Both Sexes Only Agriculture 46.7 81.2 52.9 Only Non-farm 44.9 15.5 39.6 Both Agriculture 1.8 2.5 2.0 Both Non-farm 0.7 0.3 0.6 Mix activity 5.9 0.6 5.0 Total 100 100 100 Source: Author s calculation from PSLM, 2010 micro dataset The household level employment diversification presents an interesting picture as given in table 7. There are more pure non-farm households than the farm households at national level with a significant representation of mix households (20%) as well. Around 6 percent of the households are not involved in any employment activity. Heterogeneity prevails across the provinces with more agricultural households in Punjab and Sindh, 18

while more mix households in Punjab and KPK, reflecting more resource diversification in KP and Punjab than the other provinces. However, more households in KP are doing nothing as compared to other provinces which might be due to the worse law and order situation in this province. Table 7: Distribution of Rural Households by Employment Diversification (in %) Province Only Farm Only Non-farm Mix Doing Total Households Households Households Nothing National 36.3 38.1 19.9 5.7 100 Punjab 35.0 34.0 24.0 7.0 100 Sindh 44.9 34.7 18.9 1.5 100 KPK 20.3 42.0 24.4 13.3 100 Baluchistan 44.2 46.5 8.5 0.7 100 Source: Author s calculation from PSLM, 2010 micro dataset B. The Determinants of Rural Employment Diversification This section details results over the determinants of rural employment diversification where the dependent variable has three outcomes on the basis of reported primary and secondary nature of work activities of employed workers: only agriculture, only non-agriculture and mix activities. Two models have been estimated: in first model, all the households are included while in second model only land holder households have been analyzed to observe the impact of irrigated land on employment diversification. The results are given in table 8 with reported relative risk ratios (RRR) of multinomial logistic model in which farm activity serve as the reference category. Regarding individual characteristics of rural workers, male are more likely to be employed in non-farm and mix activities as compared to females working in farm activities. The positive impact of age on non-farm and mix activities suggests that with gaining experience, rural workers switch their status of employment to off-farm activities. Education, especially the higher education seems to have a strong impact on workers to work in off-farm and mix activities (Table 8). 19

Table 8: The Determinants of Livelihood Strategies of Rural Workers--Multinomial Logistic Model Model 1 Model 2 (Only land owners) Correlates Only Non-Farm/Farm Mix/Farm Only Non-farm/Farm Mix/Farm RRR St. Error RRR St. Error RRR St. Error RRR St. Error Sex (male=1) 4.541* 0.114 20.375* 2.099 3.136* 0.143 36.375* 6.268 Age (years) 1.096* 0.003 1.252* 0.009 1.062* 0.006 1.279* 0.012 Age square 0.999* 0.000 0.997* 0.000 0.999* 0.000 0.997* 0.000 Education (below primary grade as ref.) 5-9 grades 1.738* 0.032 1.464* 0.058 1.728* 0.061 1.390* 0.072 10-12 grades 3.030* 0.071 2.317* 0.106 3.293* 0.126 2.056* 0.116 14 and above grades 14.621* 0.859 9.069* 0.783 16.956* 1.182 7.011* 0.679 Land ownership (acres) 0.920* 0.002 0.980* 0.002 - - - - Irrigated land (yes=1) - - - - 0.572* 0.020 0.578* 0.029 Middle school (>44 mint) 0.617* 0.016 1.028 0.057 0.702* 0.034 0.874** 0.059 Public transport (>44 mint) 0.751* 0.031 1.143*** 0.087 1.064 0.070 1.236** 0.109 Province (North and Central Punjab as ref.) South Punjab 0.586* 0.015 1.025 0.049 0.729* 0.031 1.117** 0.068 Sindh 0.500* 0.011 0.464* 0.022 0.669* 0.029 0.747* 0.049 KPK 1.465* 0.037 1.988* 0.092 1.658* 0.068 2.463* 0.143 Baluchistan 0.677* 0.016 0.096* 0.009 0.366* 0.019 0.098* 0.013 Constant 0.067* 0.004 0.000* 0.000 0.048* 0.005 0.000* 0.000 Pseudo R2 0.160 0.179 N 92,006 39,327 *significant at 1%, ** significant at 5%, *** significant at 10 % Source: Author s calculation from PSLM, 2010 micro dataset 20

Regarding household characteristics, landownership put away the workers to join non-farm and mix activities. However, as shown in model 2, the quality of land also matters for land owner households as un-irrigated (barani) land induce the households to shift their supply of labor toward the non-farm and mix activities. The two community variables related to soft and physical infrastructure which are access to middle school and public transport show that an increase in distance to these facilities reduces the chances for workers to be participating in off-farm and mix activities. For deeper analysis, the province Punjab has been split into two sub-heads, the southern Punjab and Punjab excluding south. The results show that except KP province, rural employed workers residing in all other provinces including the southern Punjab are less likely to be participating in non-farm and mix activities while comparing these workers to those who are living/working in north and central Punjab (Table 8). VI. Role of Non-farm Economy in Equity Enhancement and Household Welfare As detailed earlier that 23 percent of the rural households own enterprises or some sort of micro businesses in Pakistan. The possession of rural enterprises and share of income sources in households by household expenditure quintiles are given in table 9 in which the household s sources of income has been given by farm-income (agricultural wage and total farm income excluding agricultural wages) and off-farm income (business income and non-agricultural wages). The results reveal that wealthier households are more likely to own non-farm rural enterprises in Pakistan. Regional variation also prevails with more ownership in province Punjab and KP than the other provinces. In all the four provinces, enterprise ownership tends to increase monotonically with the per capita household expenditures/quintile (see appendix 6). Though the richest households own more enterprises and also earn their income from farm sources, the poorer households are reliant for employment in these enterprises and earn significant share of their income from non-farm sources as the share of non-farm wage in household income is 46 percent with an aggregate of 57 percent for the lowest quintile. Since majority of poorer households earn their income from non-farm sources, especially the non-farm wages, therefore, the non-farm income sources for the poorer reflects equity enhancing in Pakistan. In some developing countries, non-farm income 21

sources are inequitable as they have less contribution for the poorer households i.e. Ecuador and Vietnam or neutral equitable i.e. India and Ethiopia. In absolute terms, the poorest rural households in Pakistan earn near to four times as much income from nonagricultural wage employment (Rs. 65,846.7 annually) as from agricultural wage employment (Rs. 17,452.3 annually). The results suggests that even a low return from non-farm enterprises may contribute to enhance household income and consequent increase in the welfare of poorer rural households who are landless and are mostly engaged in low productive agricultural activities. Table 9: Percentage of Rural Households that Own Enterprise and Annual Household Income Shares by Expenditure Quintile Household s Per capita Expenditure Quintile Ownership and Overall income sources Poorest quintile Quintile 2 Quintile 3 Quintile 4 Richest quintile Households own enterprise 18.2 22.3 23.4 27.6 31.8 23.0 Household s source of income Agricultural wages 11.3 8.4 5.3 4.3 1.2 7.3 Total Farm (excl. 31.4 34.6 43.8 48.6 55.5 44.8 agric. wages) Net Business Income 11.1 13.1 14.6 15.6 16.2 13.8 Non-agricultural Wages 46.2 43.9 36.3 31.5 27.1 34.1 Total Non-farm 57.3 57 50.9 47.1 43.3 47.9 Source: Authors estimation from HIES 2010 micro dataset Poverty incidences in Pakistan are considerably higher in rural areas than in urban areas throughout the history of Pakistan, with a gradual shift to rural areas. Except the latest numbers, rural poverty fluctuates across the decades (see appendix 7). The question is how do poverty levels differ across rural population groups engaged in farm and offfarm activities? The present study has established various categories of farm and off-farm activities at individual and at household level on the basis of primary occupation (Table 10). The individual classification shows that poverty incidences are much lower among the workers who are employer in non-farm activities or own cultivator and livestock holder in farm-activities. Paid employee category is challenging with highest incidences of rural poverty, but interestingly paid employees in non-agricultural sector have more than 6 percentage lower incidences of poverty than their counterparts getting wage in 22

agriculture sector. Female workers are facing the higher incidences of poverty among all the categories except the own cultivators. The household level categories show that incidences of poverty are lower among the farm households as compared to the off-farm households; however, the results reveal that the diversified households (both the farm and off-farm) are getting some benefits of diversification since with lower incidences of poverty among these households than the non-farm households. Another interesting finding is that both the pure farm households and mixed households headed by females have the lower incidences of poverty than their relative households headed by males. Table 10: Incidences of Poverty by Activity Status of Rural Employed Workers and Households, 2010-11 Activity type Both sexes Male Female At individual level Employer (non-agriculture) 5.1 5.1 - Self employed (non-agriculture) 13.6 13.1 22.8 Paid employee 20.2 19.2 29.3 - Non-agriculture 18.8 18.4 23.2 - Agriculture 25.3 22.3 39.4 Own cultivator 7.9 8.0 4.8 Share cropper/contract cultivator 15.6 15.7 - Live stock 9.7 9.7 9.7 At household level (by sex of head of household) Only farm households 14.1 14.1 11.6 Only non-farm households 19.3 19.2 22.0 Mix households 16.2 16.3 9.4 Overall poverty (%) 16.0 15.5 19.6 Source: Author s calculation from HIES, 2010 micro dataset Table 11 presents data on the nature of activities of rural non-farm workers by their poverty status. Data on the nature of activities of females are also presented in this table. Activities that are particularly important for the poor include manufacturing, construction, transport and domestic services activities. Such activities might resemble more closely the low productivity options, providing incomes to persons who lack an alternative source of income. For example, wages of construction workers (semi-skilled and unskilled) are very low in the country. In real terms they have declined overtime or best are stagnant. While viewing the poverty incidences among self-employed and paid workers where more than 94 percent of the rural males and females are employed, the results reflect mark differences of poverty between the two groups. Poverty incidences 23

are comparatively smoother and are less among the self-employed workers as compared to the paid employee category. The lower incidences of poverty in services activities for the wage-employee category might be due to the employment of these workers in educational and medical services run by the government. Female workers have higher incidences of poverty than males except mining and transport sector which might be the data representation issue as very few Pakistani females are engaged in these sectors. Table 11: Incidences of Poverty in Rural Non-farm Employment in Pakistan by Employment Type and Status (in %) Overall Female only All nonfarm Type of Industry Self Paid All non-farm Self Paid employed employees workers employed employees workers Mining 13.8 23.5 19.4 6.7 - - Manufacturing 23.3 17.8 24.5 40.8 36.4 43.3 Electricity gas & water 3.5 1.1 3.6 - - - Construction 22.3 9.6 22.4 37.9-42.1 Whole sale & retail trade 16.1 11.6 24.5 31.2 19.3 55.4 Transport & communication 15.0 17.9 15.1 4.4-6.0 Professional services 10.7 10.8 13.0 17.1-17.1 Social & personal services 13.8 15.4 13.5 15.2 16.4 15.0 Overall poverty (%) 17.3 13.7 18.8 23.3 22.8 23.2 Total 100 100 100 100 100 100 Source: Authors estimation from the HIES 2010 micro dataset The above discussion concludes that rural non-farm economy may not have a clear pattern to reduce rural poverty in Pakistan as poverty incidences are higher among the non-farm households compared to the farm households. However, access to non-farm jobs does lead to equity enhancement and to improve the absolute income levels of the poorer laborer as the paid labor in non-farm activities has lower incidences of poverty than the agricultural paid labor. Though majority of the labor in non-farm activities especially in manufacturing sector are engaged in low productive activities and thus, are facing higher incidences of poverty, but even lower productivity may contribute to enhance household income as it is an important source of secondary employment for the small and landless farmers. In figure 5, the district level rural poverty and rural multiple 24