Contemporary South Asia (2001), 10(3), 319–342
‘Like an act of God’: land, water and social power in northern Pakistan1 A. HAROON AKRAM-LODHI
This article uses a case study of the Salinity Control and Reclamation Project of Mardan, northern Pakistan, to critically examine how the impact of a rural infrastructure project designed to improve water delivery can be compromised by the distribution of land ownership. The article demonstrates that the positive impact of infrastructural investment on farm yields can be translated into much more limited gains in income when farmers must rent land in order to establish viable farms. The design of infrastructural investment must therefore integrate an understanding of the social and economic environment within which the investment is made. In particular, the ownership and control of land, and hence the structure of social power, must be evaluated. ABSTRACT
The World Development Report 1994 highlighted the importance of infrastructure in development. The Report argued that the growth of infrastructural capacity matched the growth of gross domestic product and that, more particularly, ‘the growth of farm productivity … is closely linked to infrastructure provision’.2 However, despite infrastructural growth that ‘has in some respects been spectacular’, overall ‘performance has been disappointing’.3 Three speci c causes of poor performance were discussed: First, the delivery of infrastructure services usually takes place in a market structure with one dominating characteristic: the absence of competition. Second, those charged with responsibility for delivering infrastructural services are rarely given the managerial and nancial autonomy they need to work properly. Third, the users of infrastructure—both actual and potential—are not well positioned to make their demands felt.4
Consider the third cause of poor performance, which implies that enhancing ef ciency in the delivery of infrastructure services requires responding to the demands of consumers. The Report prioritizes prices as being the best way of signalling demand: ‘(t)hrough the price mechanism, consumers can in uence investment and production decisions in line with their preferences’.5 Corrrespondence: A. Haroon Akram-Lodhi, Institute of Social Studies, PO Box 29776, 2502 LT The Hague, The Netherlands. E-mail:
[email protected] ISSN 0958-4935 print/ISSN 1469-364X online/01/030319-24 Ó 2001 Taylor & Francis Ltd DOI: 10.1080/09584930120109540
A. HAROON AKRAM-LODHI
However, contrary to the argument of the Report, prices may not be an effective means of signalling demand because economic agents enter into market relationships with differential endowments.6 Agents with large quantities of assets may enter markets from a position where they seek to monopolitically ‘regulate’ both markets and prices in their own interests. Alternatively, agents that enter markets with limited assets may themselves be ‘regulated’ by the market.7 Thus markets may be ‘interlinked’, and the inability to signal demand may be a function of the distribution of assets. As a consequence, in addition to the three causes of poor performance in the delivery of infrastructure services identi ed by the World Bank, it may be possible to identify a fourth cause: the distribution of assets. Indeed, the distribution of assets may have an affect on both managerial autonomy and market structure, which implies that the causes of poor performance may be hierarchically ordered. This article uses a case study of the Salinity Control and Reclamation Project (SCARP) of Mardan, northern Pakistan, to critically evaluate the role played by the distribution of assets in explaining the impact of a rural infrastructure project. SCARP Mardan was a US$169 million drainage and irrigation project nanced by the World Bank, the Canadian International Development Agency, and the Government of Pakistan. The article argues that it is not possible to divorce issues of performance from broader concerns over the structure of agrarian assets. For farmers to capture the bene ts of improved rural infrastructure, they must be able to allocate their resources as they see t. In turn, this implies that farmers have some degree of control over key productive resources. However, many farmers lack effective control over the most important productive asset in the countryside, land. Moreover, many farmers lack effective control over other, non-land productive assets. Lacking assets, resource allocation decisions may not re ect the choices of farmers, but rather the structure of the asset markets within which they are entwined. This may compromise project impact as the structure of assets may affect the pattern of bene ts derived from the use of infrastructure. Thus, the capacity of a project intervention such as SCARP Mardan to meet its objectives may be intimately connected to the structure of agrarian assets. The article proceeds as follows. The next section introduces SCARP Mardan, and in so doing elaborates both the setting and the data used in the article. The impact of SCARP Mardan is then examined with reference to crop yields, crop mixes, incomes and poverty. It is demonstrated that the pattern of bene ts accruing to farmers appears to be less than that expected by project planners. The nal section investigates the structure of agrarian assets, and shows that the impact of SCARP Mardan was compromised by such a structure. The article concludes by questioning the ef cacy of project interventions that do not factor into their planning the structure of assets. SCARP Mardan Pakistan has the most extensive and complex water delivery system in the world. However, as the water delivery system has developed, attention has focused on increasing capacity to meet increasing demand. This focus has neglected the 320
LAND, WATER AND SOCIAL POWER IN PAKISTAN
ef ciency of delivery and, as a result, has had ecological consequences. Poor water management techniques have fostered rising water tables and salinization, putting stress on water delivery systems, soil capacities and, ultimately, on agricultural production. SCARP Mardan was designed to reduce these stresses and thus improve agricultural performance by remodelling existing surface drains, by installing a network of subsurface pipe drains, and by widening the canals that delivered water. Engineering work commenced in 1984 and was completed in 1992. The physical impact of SCARP Mardan has been signi cant. Provincial Irrigation Department (PID) data demonstrate that the supply of water has in many cases almost doubled since the completion of the remodelling work. There is now ‘no period of the year when supply falls signi cantly below requirements’.8 However, although rules for operating the water delivery system have not changed, variance in peak ow rates and in month-to-month access to water has increased. Increased variance is a function of project design: xed water outlets linking secondary canals to watercourses were widely replaced by pre-cast concrete barrels with gated outlets. The diameter of the barrels was limited and not adjusted to re ect the area to be irrigated. The amount of water linking secondary canals to watercourses is therefore now determined by the size of barrels. Along with increased water supply, SCARP Mardan had a major affect on waterlogging. Prior to 1981, watertable depth maps indicated that in many areas watertable depths were 3 feet or less from the surface and, in some parts, the watertable depth was less than 1 foot. Watertables have been signi cantly lowered and ooding is no longer a serious threat as a consequence of SCARP Mardan. Moreover, water in ltration rates have markedly improved.9 While many farmers continue to raise issues of concern, the popular perception is that SCARP Mardan has had an impact on livelihood security through its effect on incomes. A widely shared sentiment is that SCARP Mardan has been, to quote one farmer, ‘like an act of God’. However, popular sentiment and actual outcomes may not be the same. It is therefore necessary to assess the economic impact of SCARP Mardan, using the farm economics data collected for it. The farm economics data SCARP Mardan’s farm economics data was collected in two iterations, between 1984 and 1986 and in 1996.10 Data collection followed the principles of experimental design, dividing the sampling frame into impact and control samples. For the rst iteration, 15 separate, primarily closed-question surveys lasting from between 1 and 9 months and conducted over a 14-month period were undertaken by enumerators resident in the sample villages. The rst iteration completed surveys on 1.6% of the project area’s population, or 485 randomly selected and sampled impact households. Some 154 households in the control villages selected were randomly sampled. The baseline sample for the rst iteration thus consisted of 639 households. The total population was 5564 321
A. HAROON AKRAM-LODHI
people, of whom 52.7% were male and 47.3% female. For the second iteration, three closed-question survey protocols were developed that sought to preserve the original sample by surveying as many households that had participated in the rst iteration as was possible. In situations where households were unable to be surveyed for the second iteration, replacement respondents were selected on the basis of having a socioeconomic pro le similar to the missing household. In addition, households were added to ensure that 1.6% of the project area’s population was once again surveyed. In total, the second iteration completed surveys on 568 households within the area where the project had had an impact and 170 households within the control villages, for a total sample of 738 households. The total population was 7991 people, of whom 52.7% were male and 47.3% female. Some 69.7% of the households surveyed in the second iteration had been surveyed in the rst iteration. One area where the farm economics data is incomplete, however, is in its coverage of large landlords. Unfortunately, large landlords failed to co-operate with the administration of the surveys used to gather the data in either iteration. In common with much eld research in the South Asia, they refused to be interviewed.11 As a consequence, no farms in the 1996 data had an operational holding of more than 50 acres. To compensate for this shortcoming, where appropriate the article will use agricultural census data. While not ideal, statistical tests consistently demonstrate that the data for both iterations is representative.12 Moreover, once transformed, the data is normally distributed.13 The economic impact of SCARP Mardan Expected outcome According to its initial log frame analysis, SCARP Mardan had three principal goals. The rst was to increase net annual incomes by improving yields for all major crops in the project area. The second was to reduce the rural unemployment arising out of inadequate agrarian productivity. The third was to ensure that the bene ts arising out of increased employment and farm incomes accrued to families in all income categories. In so doing, it was hoped that SCARP Mardan could at least stabilize and at best reduce disparities in incomes among the rural population of the area. The simple analytics of the implicit logic underpinning SCARP Mardan can be presented graphically, using the quadrant diagram displayed in Figure 1. Quadrant I presents a choice of technique diagram in which units of labour are arrayed on the vertical axis and units of water are depicted on the horizontal axis. Other inputs are held constant, and prices are exogenous. Assuming the farmer can use the resources at their disposal to obtain labour or acquire a quantity of water, the budget constraint (aa) emerges, representing the possible combinations of labour and water that would totally exhaust the resources available to the farm. The slope of the budget constraint is the ratio of the relative price of labour to water. 322
LAND, WATER AND SOCIAL POWER IN PAKISTAN
Figure 1. The simple analytics of implicit project logic. Output (I0) is produced using (0L1) labour and (0W1) water at relative prices (aa). The application of (0W1) water to the production function (0Qc) generates cash crop output (0Q1). With the relative price for cash crops and subsistence crops (cc) and a production possibility frontier of (Y1Y1), the production of cash crop output (0Q1) results in the production of subsistence crops (0S1). SCARP Mardan was to increase water availability, altering the relative price of labour and water to (ab), resulting in output (I1) using (0L2) labour and (0W2) water. Increased water in the production function (0Qc) generates cash crop output (0Q2). It also shifts the production possibility frontier to (Y2Y2), and with a relative price for subsistence crops and cash crops of (dd), the production of subsistence crops rises to (0S2).
Levels of farm output that can be produced using quantities of labour and water are depicted using isoquants that represent the technological possibilities facing an output-maximizing farmer. Given the constraints depicted in Quadrant 323
A. HAROON AKRAM-LODHI
I, the output-maximizing farmer will produce output (I0) using (0W1) water and (0L1) labour. Changes in the quantity and quality of water available to farmers resulting from the implementation of SCARP Mardan would then be represented by a shift in the budget constraint from (aa) to (ab). The improved water delivery system would deliver more water per unit of labour, and thus the quantities of labour and water capable of being used by a farm with a given set of resources would alter because of a shift in relative prices. The output-maximizing farmer would increase production to output (I1) using (0W2) water and (0L2) labour. Quadrant II depicts a production function (0Qc) where cash crop output on the vertical axis is a function of water usage on the horizontal axis. Prior to SCARP Mardan, levels of water (0W1) permitted the production of cash crops (0Q1). Improvements in the water delivery system result in a more effective use of greater quantities of water, which is conceptually equivalent to an increase in the quantity of the input from (0W1) to (0W2). The result is an increase in the production of cash crops from (0Q1) to (0Q2). The impact of increased cash crop output can be depicted using the production possibility frontier depicted in Quadrant III, where the farmer can produce either cash crops or subsistence crops. The possible alternative combinations of cash crops and subsistence crops are governed by resources, technology and relative prices; the latter two are exogenously determined and thus outside the control of the farmer. The maximum level of production is depicted by production possibility frontier (Y1Y1) and the relative prices facing the farmer for the crops are depicted by curve (cc). The pro t-maximizing farmer produces at point (e3), producing (0S1) subsistence crops and (0Q1) cash crops. The remodelling of the water delivery system causes a shift in the production possibility frontier to (Y2Y2) as improved water delivery makes possible increased production of both cash crops and subsistence crops. The expansion of the production possibility frontier establishes a new point of tangency with relative price curve (dd) at point (e6). The pro t-maximizing farmer increases production of subsistence crops to (0S2) and increases production of cash crops to (0Q2). If, as is common in many agrarian economies, the wage rate is determined by the value of subsistence crops, the implication is that wage rates rise. In combination with the increased production of cash crops, it is likely that incomes increase. The cumulative impact of SCARP Mardan should thus have been witnessed in the interlinkages between the choice of technique chosen by farmer, the level of production, and the impact of production on the subsistence wage. Theoretically, on the completion of the remodelling of the water delivery system and the installation of subsurface drains, farmers should have increased their use of water and labour. As a result, the production of both cash crops and of subsistence crops should have increased. The increased production of subsistence crops should, in turn, have boosted the wage rate which, along with the increased production of cash crops, should have led to an increase in incomes. 324
LAND, WATER AND SOCIAL POWER IN PAKISTAN
Actual crop yields The simple analytics presented in Figure 1 can now be examined with reference to farm economics data. The increase in water delivery was discussed earlier. The impact of the engineering work on production can be gauged in part from Table 1, which presents crop yields for wheat and maize, the two most important food security crops in the project area. Table 1 demonstrates that in 1986 an acre of land produced 722 kg wheat, while in 1996 an acre of land produced 1111 kg wheat, an increase of 54%. The situation with maize, if anything, is more dramatic. Between 1986 and 1996, maize yields increased by almost 250%. Sugarcane is the most important cash crop in the project area. Table 2 presents data on sugarcane yields. However, some qualifying remarks must be made about the data. In 1996, two distinct events occurred in speci c locations that together served to dramatically reduce sugarcane yields. The rst was the onset of a major problem with termites in several villages. The second event was the maturation of maintenance problems in the water delivery system, which resulted in ooding in a small number of hamlets. These factors have necessitated a cautious analysis of sugarcane yields. Table 2 provides sugarcane yields for three different ‘scenarios’. The rst scenario is the sugarcane yield per acre obtained from all reporting sugarcane producers. The result of the rst scenario con rms the crop failure, with yields declining by over 20%. Therefore, the second scenario is based on an assessment of what constitutes a crop failure. Structured discussions with farmers, eld researchers, agricultural economists and agronomists have led to the conclusion that sugarcane yields that fell by more than 40% below the yield level obtained in 1986
Table 1. Wheat and maize yields in kilograms per acre, by farm size Farm size (acres) , 1 1 to , 2.5 2.5 to , 5 5 to , 7.5 7.5 to , 12.5 12.5 to , 25 25 1 Average
1986 Baseline sample
1996 Impact sample
Wheat
Maize
Wheat
Maize
488.00 783.00 522.00 1170.00 556.00 782.00 1280.00
199.00 248.00 184.00 139.00 177.00 109.00 220.00
1308.27 794.52 923.46 1545.72 1174.25 1252.06 1368.89
481.19 585.10 423.16 423.16 372.87 443.92 517.65
722.00
180.00
1111.87
445.63
Source: A.H. Akram-Lodhi, ‘The farm economy’ in J. Freedman and A.H. Akram-Lodhi (eds), Water, Pipes and People: The Social and Economic Impact of the Salinity Control and Reclamation Project in Mardan, Northern Pakistan (Lahore: Vanguard Press, 2001).
325
A. HAROON AKRAM-LODHI
Table 2. Sugarcane yields in kilograms per acre, by farm size 1996 Impact sample 1986 Baseline sample
Scenario 1: all reporting farms
Scenario 2: excluding crop failures
, 1 1 to , 2.5 2.5 to , 5 5 to , 7.5 7.5 to , 12.5 12.5 to , 25 25 1
208.00 14,741.00 13,662.00 14,021.00 16,355.00 10,947.00 15,585.00
9599.54 7545.00 11,164.51 11,769.95 11,902.30 9379.11 640.00
17,360.00 12,960.00 20,016.85 17,492.38 15,626.40 16,338.18 0.00
0.00 12,000.00 24,084.44 20,755.66 16,927.71 0.00 0.00
Average
13,748.00
10,781.48
17,426.41
19,582.23
Farm size (acres)
Scenario 3: canal head only
Source: Akram-Lodhi, op cit, Table 1.
could be considered as examples of severe crop failure. The second sugarcane yield scenario excludes such farms. Excluding crop failures produces an average yield of sugarcane of 17,426 kg per acre, an increase of almost 27% over the 1986 sugarcane yield. An increase in sugarcane yields of 27% is respectable but not spectacular. Therefore, a third scenario has been developed. The head of the watercourse is the optimal location to grow water-intensive crops such as sugarcane. Therefore, the third scenario estimates sugarcane yields for those farms located at the head of the watercourse. Using this approach, sugarcane yields within the project area would reach 19,580 kg per acre, an increase of over 42% over the 1986 yield. Participatory research indicates that the second scenario already described most closely approximates the yields that would have occurred in 1996 had the crop not failed. Farmers, in particular, have consistently stated that yield increases in sugarcane have been moderate. So too have several prominent agronomists. In this light, it would appear that, while within the sample there have been impressive productivity gains in wheat and maize, productivity gains in sugarcane appear more modest. Actual crop mix The productivity of land will re ect and effect how it is used. Table 3 presents the structure of the cropped area for wheat, maize and sugarcane. The table demonstrates that the period since 1986 has witnessed a dramatic rise from 29% to 42% of the cropped area being devoted to sugarcane. By way of contrast, there has been a reduction from 21% to 16% in the area devoted to wheat. Table 3 also demonstrates a very slight increase between 1986 and 1996 in the area devoted to maize. Maize presents an interesting story, for three reasons. 326
26.00 26.00 31.00 24.00 22.00 22.00 5.00
21.00
, 1 1 to , 2.5 2.5 to , 5 5 to , 7.5 7.5 to , 12.5 12.5 to , 25 25 1
All farms
0.67
16.39
16.45 15.98 21.04 15.00 14.80 15.54 12.48
1996
29.00
18.00 28.00 25.00 25.00 30.00 28.00 34.00
1986
1.71
41.80
31.52 32.42 35.98 41.86 44.00 49.91 52.45
1996
Sugarcane
12.00
19.00 17.00 16.00 13.00 12.00 11.00 5.00
1986
0.49
12.06
10.97 12.47 12.64 12.60 12.59 11.63 5.05
1996
Maize
Note: ‘Other’ includes fruit, vegetables, rice, tobacco and fodder crops. Source: Akram-Lodhi, op cit, Table 1.
Average cropped area
1986
Farm size (acres)
Wheat
Table 3. Structure of crop mix
38.00
37.00 29.00 28.00 38.00 36.00 39.00 56.00
1986
1.12
29.75
41.06 39.86 30.34 30.54 28.61 22.92 30.02
1996
Other
100.00
100.00 100.00 100.00 100.00 100.00 100.00 100.00
1986
3.13
100.00
100.00 100.00 100.00 100.00 100.00 100.00 100.00
1996
Total
LAND, WATER AND SOCIAL POWER IN PAKISTAN
327
A. HAROON AKRAM-LODHI
First, it is a water-intensive crop, and thus the limited change in the overall area devoted to maize is interesting given the efforts to improve the water delivery system. Second, maize is a staple food, and it might be thought that improvements in the water delivery system would lead to an increase in the area devoted to maize. This has not occurred. However, unlike wheat, which is popularly thought of as a ‘superior’ food grain, maize is widely considered to be an ‘inferior’ food grain. In these circumstances, it might be expected that improvements in the water delivery system would lead to an increase in wheat and a decrease in maize, as the two staples are food grain substitutes. However, this too has not occurred: the area devoted to wheat production has declined, while the area devoted to maize production has remained static. Third, maize is capable of being used as a fodder crop for animals. However, fodder production has dramatically declined. The decline in fodder production re ects a fall in the amount of livestock held by farms of the area, a decline that, if anything, reduces the need for maize if it is to be used as fodder. It also re ects land that was previously being used for fodder being shifted into the production of higher value-added crops. The combination of these factors means that it is unclear why maize production has not witnessed some change in the share of the cropped area. Differential wheat, maize and sugarcane productivity records may help explain the lack of change in maize production. Along with the increased yields for food security crops demonstrated in Table 1, Table 2 shows that there has only been a correspondingly moderate increase in yields for the most important cash crop. It this light, it is not surprising that farms have taken land out of food crop production and shifted into cash crop production. Such a reallocation is a rational response to differential productivity when combined with differential cash returns to productive activity. This also would explain why farms have maintained maize production: they have been reallocating areas previously devoted to wheat and other crops to sugarcane in an effort to enhance cash income secure in the knowledge that increased productivity in key food crops, and in particular maize, will mean an overall improvement in the economic security of the household. In light of this possibility, it is useful to examine the evolution of rural incomes between 1986 and 1996. This is done below. Actual income Table 4 depicts the distribution of real monthly income in 1986 prices. The table demonstrates that, in constant terms over 10 years, the incomes of owner-tenants had risen by only 38% and that of tenants by only 24%. The income of owner operators had, in constant terms, fallen. Thus, the growth in productivity recorded in the previous subsection has failed to be translated into an equivalent growth in real incomes. Moreover, it can be noted that there has been only modest movements between income categories since 1986.14 In light of the modest growth in real incomes, it is useful to construct a headcount index based on consumption norms and thus assess the extent of 328
LAND, WATER AND SOCIAL POWER IN PAKISTAN
Table 4. Distribution of real monthly income by tenure Income in rupees , 300 300 to , 600 to , 1200 to 1800 to 3000 to 4000 to 6000 to 9000 to 12,000 1 ,
,
,
,
,
,
600 1200 1800 3000 4000 6000 9000 12,000
All groups Average income per household
Owner
Owner-tenant
Tenant
Total
0.0 2.1 9.90 15.50 25.40 10.60 12.70 9.20 7.70 7.00
0.00 0.00 4.00 4.00 16.80 19.80 21.80 15.80 9.90 7.90
0.00 9.00 5.50 9.70 24.40 15.70 23.00 13.80 2.80 4.10
0.0 1.10 6.50 10.20 23.00 15.00 19.60 12.80 5.90 5.90
100.00 4728.83
100.00 6153.66
100.00 4625.79
100.00 4879.67
Source: Akram-Lodhi, op cit, Table 1.
absolute poverty in the area. Using the 1990/1991 Pakistan Household Income and Expenditure survey,15 the World Bank estimated an adult equivalent poverty line of Rs 296 per person per month. In rural NWFP, it was estimated that 40.6% of the population could be classi ed as poor. In ating the poverty line to 1996 prices and comparing it with the SCARP Mardan data, it would appear that 44.9% of the study households could be classi ed as poor.16 Thus, despite infrastructural investment, there may be a larger percentage of poor people in the project area than in rural NWFP as a whole. Curiously, this nding sits alongside the nding that mean consumption in rupees per month for the study area was, at Rs 872 per adult equivalent, greater than the mean consumption gure of Rs 709 per adult equivalent for rural NWFP estimated by the World Bank when such gures are in ated to 1996 prices. The reason for this nding is explained below. SCARP Mardan appears to have brought forth some impressive increases in productivity. For many peasant farmers, these productivity improvements seem to be ‘like an act of God’. However, real incomes have only grown modestly. Moreover, the extent of poverty is either greater in central NWFP as compared with rural NWFP as a whole or the extent of poverty in rural NWFP is greater than that claimed by the World Bank.17 In either instance, SCARP Mardan has not been able to signi cantly increase incomes or enhance equity. It is therefore necessary to explain this outcome. Assessing SCARP Mardan Water and social power From a purely technical point of view, SCARP Mardan was not a failure. The 1996 estimated economic rate of return for SCARP Mardan was 12.3%.18 While less than the 18% estimated at the design stage, it is nonetheless in excess of the 329
A. HAROON AKRAM-LODHI
11.3% interest rate prevailing on public and private debt in Pakistan in 1979 when the project was developed. The principal reason for the difference between the estimated economic rate of return and the actual economic rate of return is twofold. First, the cost of the project greatly exceeded expectations. Second, the impact of SCARP Mardan on yields and on production has not been as great as had been anticipated. Nonetheless, SCARP Mardan did generate a positive return on the initial investment and, in so doing, theoretically contributed to a rise in aggregate economic welfare. If the project was not a failure, why did it not raise incomes? To answer this question, Figure 1 is helpful. In Quadrant I, the remodelling of the water supply system and the installation of subsurface drains makes more water available for a given level of on-farm resources. As a consequence, the budget constraint shifts to (ab), and the relative price for labour and water alters. However, if the relative prices for labour and water do not change, the budget constraint does not shift. This implies that lesser output would be produced than that which would have been expected. These effects would then feed through the other quadrants: lesser levels of production in Quadrant II; a different combination of cash crop and subsistence crop production, and at a lesser overall volume of production, than would have been projected in Quadrant III; and a correspondingly smaller affect on incomes. The impact of SCARP Mardan would fall short of that which its designers had hypothesized. The key to the capacity of SCARP Mardan to deliver its expected outcome was thus the movement in relative prices surrounding the choice of technique. However, few irrigation projects, either in Pakistan or elsewhere, have been able to shift the relative price of water. Ostensibly, the reason is managerial: a shift in the relative price of water requires a fundamental change in water delivery practices. In Pakistan, a form of supply-based delivery practice termed warabandi is used. In the warabandi system, the supply of water is determined at the head of the watercourse and is subject to seasonal variation. The allocation of water uses a xed periodic sequential rotation based on the size of the area served. The allocation mechanism is factored into a water charge termed abiana, which also is based on a calculation of the cultivated area, the season, and the types of crops being grown. In such a structure, water charges are xed. Therefore, to alter the relative price of water, the warabandi system had to be replaced by a demand-based system of water delivery. In a demand-based system, the farmer estimates water requirements and thus determines how much and when water is needed. The farmer thus becomes capable of varying the volume of water received by adjusting the length and frequency of deliveries. The World Bank favoured a shift to a demand-based system and this was, in part, the reason behind seeking to install pre-cast concrete barrels with gated outlets as part of the engineering works because the gated outlets could be adjusted according to farmer demand and upstream water levels. Moreover, gated outlets can be metered, and the more precise identi cation of water delivery patterns can allow the establishment of a water charge based on supply and demand. 330
LAND, WATER AND SOCIAL POWER IN PAKISTAN
However, despite the installation of gated outlets, the water delivery system did not become demand based. In seeking to explain this, W. Quarry emphasizes a selective attention to complex inter- and intra-institutional issues on the part of the project planners, a related neglect of the internal and external constraints facing the project, and the importance attached by engineers to physical outputs rather than process.19 While this assessment is undoubtedly correct, it can be simpli ed. In central NWFP, water is more than the leading agricultural input:20 it is a critical source of social power. Although the mechanics of social power in central NWFP will be discussed in the following section, it is necessary to note here that a shift to a demandbased delivery system threatened to have an affect on the structure of social power, for two reasons. The rst has to do with the nature of the warabandi system. While the warabandi system ensures a reliable water supply, it offers ample scope for rent-seeking. Abiana assessments are made by local of cials of the PID called patwaris. The collection of abiana is the responsibility of a village malik, who retains a 5% commission. The structure leaves ample scope for corruption—in order to secure a longer turn in the xed periodic rotation, have a water outlet enlarged, or prevent a neighbour obtaining water. However, while it is in the main poorly paid of cials who are rent-seekers, it is important not to underestimate the primary source of social power personi ed by the large landlords. They both make the largest payments to rent-seeking of cials, while at the same time ensuring that those who occupy key bureaucratic positions within the water delivery system will support the interests of landlords. Indeed, it is common for landlords to arrange the placement of family members in key bureaucratic positions, thus sustaining a close relationship with the state’s rural functionaries. The shift to a new and less understood water delivery system thus had implications for both the landlords and state functionaries who comprise the apex of rural social power. The second reason why a demand-based water delivery system threatened to alter the structure of social power has to do with the possible implications of such a shift for collective action. To construct a demand-based system, farmers must have a means of expressing demand. The most common way of institutionalizing the capacity of farmers to express demand is by forming water-user associations. 21 However, organizing farmers into units capable of taking collective action has social implications, especially in an area where farmers have challenged the power of landlords. Indeed, when gated outlets were initially installed at the head of the watercourse prior to the increase in water levels, and farmers at the tail end of the watercourse started to receive less water, they immediately started to collectively tamper with the gated outlets. Local resistance was used by the PID in its continuing, tenacious rearguard action to prevent the shift to a demand-based system. Table 5, which provides information on farm location and water availability, contains two pieces of evidence that suggest water may interlink with other aspects of social power in central NWFP. Examining rst the comparative location of farms, it is peculiar that, while in 1986, 45% of households were 331
A. HAROON AKRAM-LODHI
Table 5. Farm location and water availability
Location on watercourse Head Middle Tail Tubewell Rain-fed land only
1986 baseline sample (%)
1996 impact sample (%)
45.0 23.0 32.0 0.0 0.0 100.0
14.5 35.1 49.3 1.1 0.0 100.0
Water availability 25% of requirement 50% of requirement 75% of requirement 100% of requirement Rain-fed land only
25.2 34.4 7.1 32.2 1.1 100.0
Source: Akram-Lodhi, op cit, Table 1.
located at the head of the watercourse, by 1996 only 14.5% of households were so positioned. By way of contrast, while in 1986, 32% of households were located at the tail of the watercourse, by 1996 almost one-half of all households were similarly positioned. Given that 70% of the baseline sample formed part of the impact sample, this data implies that, over a 10-year period, the location of many farms shifted from the head of the watercourse to the middle and the tail. Structured conversations with farmers reveals that relocation was a means by which landlords sustained the structure of social power. Relocation occurred as a result of landlord-led alterations in tenurial contracts: the cessation of lease renewals on existing parcels of land; the introduction of new; harsher tenurial terms and conditions that encouraged farmers not to renew leases on existing parcels of land; and, of course, forcible eviction. The capacity of landlords to exercise this power is discussed in the following section. Here, it can be stressed that the relocation of farms has important implications for agrarian productivity. Farms at the tail of the watercourse have often complained of receiving inadequate amounts of water, either as a consequence of leakage or deliberate diversion. It may well be that the disappointing performance of SCARP Mardan, in part, is a function of the relocation of farms. Table 5 also demonstrates that, despite the impressive improvement in water quality, many farms continue to receive quantities of water that are deemed insuf cient to meet productive requirements. With the failure to shift to a demand-based system, farmers still follow warabandi schedules. However, the ef ciency of the drainage system means that maintaining soil moisture requires the more frequent application of water. For many, however, an increase in water access has not occurred. Comparable data from the baseline sample is not 332
LAND, WATER AND SOCIAL POWER IN PAKISTAN
available, but in 1996 some 59% of the impact sample claimed that the water they received met only one-half or less of the water that they required. Only 32% of farms were satis ed that the water they received met all of their requirements. In part, this may be due to the relocation of farms already elaborated. At the same time, when the frequency of delivery becomes a vital means of enhancing productivity, it seems unlikely that those landlords with the resources to allow themselves more frequent access to and ef cient use of water would not avail themselves of such opportunities. In so doing, their comparative social power would be enhanced. Implicit in the logic underpinning the design of SCARP Mardan was the idea that a shift in relative prices was necessary if it was to have its anticipated economic impact. However, contrary to the objectives of the World Bank, there was not a shift to a demand-based system, and relative prices thus did not change. There is good reason to believe that a possible explanation as to why prices did not change lies in collective action by the rural elite to maintain their ability to use water as a source of social power. Clearly, prices are formed in particular social and economic settings, and such prices may not re ect marginal principles, but rather the capacity of select market participants to use interlinking markets to exercise social power.22 In this light, it becomes necessary to better understand the structure of social power in central NWFP. This can be achieved by examining the material basis of social power; that is, the structure of agrarian assets. Indeed, such an examination is necessary because, if the arguments developed in this article so far are correct, it may well be that it is the structure of agrarian assets that resulted in actual outcomes deviating from expected outcomes. Land and social power in central NWFP Land is intricately linked to social and cultural identity in central NWFP. The predominant ethnic group in the area are the Pakhtun, and they strive to follow an ‘ideal’ code of behaviour known as Pakhtunwali.23 In a Pakhtun’s ‘perfect’ world, strictly following this code of behaviour would ensure the respectability, reputation and honour of the household and its patriarch. However, the world is not perfect, and Pakhtuns seek to follow as best they can Pakhtunwali in order to be seen by society as striving to ‘be’ Pakhtun. For the Pakhtun, a key means by which respectability, reputation and honour can be accrued is through the capacity to demonstrate autonomy and, hence, self-reliance. For the Pakhtun, the most important freedom is not having to recognize another person as a superior.24 Thus, the three basic tenets of Pakhtunwali—revenge, hospitality and refuge—are all ways in which a Pakhtun can show a basic mastery of the circumstances that they confront.25 The need to demonstrate autonomy and self-reliance explains why, for the Pakhtun, agriculture and combat are deemed to be the only two occupations that bring honour.26 As a consequence, one who does not have land is not deemed to ‘be’ Pakhtun. Similarly, working for another suggests that a person is not 333
A. HAROON AKRAM-LODHI
Pakhtun. Moreover, in that it deepens the capacity to demonstrate autonomy and self-reliance, ‘efforts to increase one’s own holdings forms the leitmotif of Pakhtun life’.27 The sale of land is dishonourable, and those who do so, or those who give up land in some other way, are not ‘real’ Pakhtuns. For the Paktuns, then, control of land is central to their personal and social identity. This means, in turn, that identity is shaped by a material factor; the ownership and control of land. Here, inequality is the norm. Both in popular perception and in successive agricultural censuses, large landlords dominate the agrarian structure of central NWFP. For example, in the village of Sarfaraz in 1985, 40% of households owned 3.3% of the operated area and 63% of farm households were tenants. Similarly, in the village of Platoo in 1985, 40% of households owned 12.7% of the operated area and 90% of the farm households were tenants.28 Such concentration had eased to a very limited extent in the sample, and in central NWFP as a whole, by the early 1990s. Thus, in the 1980 Agricultural Census, farms of 25 acres or greater constituted just over 5% of all farms but controlled over 48% of the owned area.29 In the 1990 Agricultural Census, farms of 25 acres or greater constituted less than 4% of all farms and controlled just over 37% of the owned area.30 Thus, large farms witnessed a drop in area owned. At the other end of the spectrum, in the 1980 Agricultural Census, farms of less than 2.5 acres constituted over 42% of all farms and controlled just over 6% of the owned area.31 By 1990, farms of less than 2.5 acres constituted 48% of all farms and controlled over 9% of the area owned.32 A modest redistribution of land ownership had thus occurred. The reason for this modest redistribution was demographic: inheritance-based partition had reduced the holdings of large farms, transforming them into middle-sized farms, and reduced the number of previously existing middle-sized farms, transforming them into small farms. On aggregate, the absolute numbers and area of small farms increased. Nonetheless, concentration remained signi cant. The way in which the land market operates provides a partial explanation as to why concentration remained signi cant.33 In 1996, the average price of purchased land in the impact sample was Rs255,387 per acre, making it far and away the most valuable asset in central NWFP. The high price of purchased land placed it out of the reach of many, if not most, rural households. However, despite the high monetary value attached to land, it is common for landlords to be unwilling to sell land because it is a source of personal and social identity, and in so being also is the principal source of social power. As P. Bordieu has argued, economic capital—in this instance, land—can be used to ‘capture’ cultural, social and symbolic capital.34 The ‘ eld of relationships’, or structures, within Pakhtun society are thus based on unequal power rooted in control over land. This in turn fosters the establishment of a ‘habitus’, an internalization of domination, in which structures of oppression—in this case, control of land—are transformed into the symbolic representations that allow someone to ‘be’ Pakhtun. The material basis of social power thus gives rise to a social structure shaped by Pakhtunwali and containing a number of extra-economic social and political 334
LAND, WATER AND SOCIAL POWER IN PAKISTAN
obligations, including allegiance to the landlord’s faction, which solidify the capacity to express social power.35 Nonetheless, the basis of social power, ultimately, is material and, for this reason, is rooted in the capacity to express social power through the deployment of coercion. Two anecdotes can illustrate these ‘mechanics’ of social power in central NWFP. The rst takes place in a village that is dominated by a large resident landlord, who has used his control of economic capital to launch a highly successful career as a politician and, in so doing, has diversi ed both his social and symbolic capital. Indeed, the landlord was, at one point, nance minister of NWFP. However, his entry into politics was not done to better the lot of those he ‘represents’, but it was done to reinforce the established doxa36 (‘the unmentioned, undisputed and tacit part of the social order’)37 through the use of the state—if necessary, by being able to deploy coercion without fear of state sanction. In the past, the landlord has prevented the construction of a girl’s school in the village as well as the installation of piped water, barring his own family compound. The day-to-day affairs of the landlord within the village are supervised by an overseer who walks around the village protected by armed men. In 1995, the landlord increased land rents for many of his tenants. In a hamlet in which every family were landless tenants, two male household heads went to the landlord and asked him to rescind the increase in land rent. The landlord responded by immediately dispossessing every family in the hamlet. The hamlet had no means of challenging their dispossession, given the landlord’s capacity to use coercion and his protected position within the state apparatus. The second anecdote of the mechanics of social power in NWFP takes place in a village that is dominated by an absentee landlord. Despite the fact that this landlord controlled far less land than the landlord depicted in the rst vignette, the material basis of his social power still allowed him to express his social power coercively. In 1990, the landlord sought to dramatically raise cash rents. This brought forth some resistance from individual farmers. Resistance was not generalized because Pakhtunwali requires that households maintain their honour by ful lling any commitments that have been undertaken,38 thus rendering contract compliance an issue of honour. In this way, symbolic and cultural capital serves to reinforce economic capital. Those households that did resist were those for whom compliance threatened their livelihood, and thus their degree of self-reliance. For these households, the landlord had an aggressive response: public beatings for some, wounding of others, and, in the case of one tenant, extra-judicial execution by employees of the landlord. The social power of the landlord was thus, despite doxa, ultimately a coercive one. As land is a source of social power, it is, for farmers, scarce. Thus, in 1996, just over 45% of the households in the sample owned some land, while just over 54% of the households were landless. This compared with 46% of the households in the sample in 1986 being landless, and 54% of the households owning some land. Of those households that did possess land in 1996, the average size of land owned was 4.12 acres. In 1986, the average size of land owned was 6.4 acres. 335
A. HAROON AKRAM-LODHI
Being landless does not necessarily mean that a household does not control any land. Farms can also gain usufruct rights to land through land rental, supplementing an inadequate productive base and thus enhancing their livelihood security by creating small yet viable operational holdings. Land rental contracts may take the form of cash leases for a season or a year. Alternatively, rental contracts may be agreed on a sharecropping basis for a season or a year. In central NWFP, sharecropping contracts take the form of an equal division of crop output between landowner and tenant in exchange for the use of the land by the tenant. In addition, the landowner commonly provides a proportion of the variable inputs, such as seed or fertilizer, necessary for farm production to proceed. There are thus three different types of tenurial status in central NWFP: owner-occupiers; owner-tenants, those that both own some land and rent in some additional land; and tenants that solely rely on the renting of land to provide them with a farm. In that only the rst of these categories does not rent any land, the relationship between the ownership and control of land is complex. Table 6 begins to unravel this complexity by presenting data on the distribution of operational holdings and operated area. Table 6 demonstrates that, in 1996, the average size of a farm was 4.67 acres. This compares with an average farm size of 6.2 acres in 1986. The overall distribution of operational holdings in 1996 was apparently biased towards smaller farms. Farms of less than 5 acres accounted for almost 64% of all farms and almost 31% of the operated area. Farms operating holdings of between 5 acres and less than 25 acres constituted 35% of all farms and controlled 64% of the operated area. Granted, part of the reason for the distribution lies in the data that was collected. As noted previously, large landlords failed to co-operate with the administration of the surveys used to gather the data. However, this distribution is consistent with that witnessed in central NWFP as a whole although, of course, the precise parameters of the distribution will be different. Table 6. Distribution of operational holdings and operated area by farm size, 1996 Farm size (acres) , 1 1 to , 2.5 2.5 to , 5 5 to , 7.5 7.5 to , 12.5 12.5 to , 25 25 to , 50 50 1 All farms
Average size
% of operational holdings
% of operated area
0.54 1.62 3.34 6.00 9.35 15.59 31.83 0.00
11.09 22.17 30.65 17.39 13.48 4.57 0.65 0.00
1.28 7.71 21.95 22.34 27.02 15.25 4.45 0.00
4.67
100.00
100.00
Source: Akram-Lodhi, op cit, Table 1.
336
LAND, WATER AND SOCIAL POWER IN PAKISTAN
Table 7. Structure of operational holdings and operated area by tenure, 1996 Farm size (acres) , 1 1 to , 2.5 2.5 to , 5 5 to , 7.5 7.5 to , 12.5 12.5 to , 25 25 to , 50 50 1 All
All
Owner operated
Owner-tenant operated
Tenant operated
100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00
82.35 30.39 24.11 20.00 22.58 20.00 25.00 0.00 30.87
1.96 15.69 25.53 20.00 33.87 40.00 75.00 0.00 21.96
15.69 53.92 50.35 60.00 43.55 40.00 0.00 0.00 47.17
Source: Akram-Lodhi, op cit, Table 1.
With an average size of operational holding of 4.67 acres and an average size of land ownership of 4.12 acres, land rental is obviously important. Therefore, Table 7 provides the structure of operational holdings by tenurial status for each of the size categories displayed in Table 6. As is demonstrated in Table 7, in 1996 owner-operators constituted approximately 31% of all farms, owner-tenants constituted almost 22% of all farms, while tenants constituted over 47% of all farms. Thus, some 69% of all farms rented land. Moreover, as is clear from Table 7, the importance of land rental to farmers was felt across the range of farm sizes. With a xed supply of land controlled by landlords who undoubtedly have a degree of monopoly in the land market, the terms and conditions governing land rental can be a means of transferring the bene ts of production from the direct producer to the landlord. Therefore, Table 8 details the monetary value of the cash and in kind rents paid by leaseholders and sharecropping tenants. Examining rst the lower portion of Table 8, it is demonstrated that the average monetary value of all cash rents per acre in 1996 was Rs2365. The average monetary value of all in kind rents per acre in 1996 was Rs2891. These two gures do not sum to the total monetary value of rent per acre, which in 1996 was Rs 3256. The reason that rents in cash and rents in kind do not sum is because of the overlapping (in the sense that many tenants pay both cash and in kind rents for a single piece of land) character of tenancy contracts in the area. The dif culty of separating out the cash from the in-kind component of total rents on land that is simultaneously leased and sharecropped means it is not advisable to assess the monetary value of cash and in kind rents on a per-acre basis. A way of dealing with this issue is to assess that the monetary value of cash and in-kind rents is on a per holding basis. This is shown in the upper portion of Table 8, where rents per holding in cash and in kind in 1986 prices are presented. 337
338
2364.95 2891.39 3256.35
Per Acre, 1996 prices In cash In kind Total
1510.14 4440.51 5950.65
Sarfaraz/Platoo, 1996
1396.32 3890.59 5286.92
Sarfaraz/Platoo, 1986
8.15 14.13 12.55
1986–1996 change (%), Sarfaraz/Platoo
Source: A.H. Akram-Lodhi, ‘A bitter pill? Peasants and sugarcane markets in northern Pakistan’, European Journal of Development Research, Vol 12, No 1, 2000, pp 206–228.
2288.05 3167.42 5455.47
Whole sample, 1996
Per holding, 1986 prices In cash In kind Total
Rent
Table 8. Rents by holding and by acre
A. HAROON AKRAM-LODHI
LAND, WATER AND SOCIAL POWER IN PAKISTAN
Per holding, and in 1986 prices, the average real monetary value of all cash and in-kind rents paid by tenants in the sample was Rs5455 in 1996. Of this, Rs2288 was paid by leaseholders in cash and Rs3167 were paid by sharecroppers in kind. An important question that then suggests itself is whether tenancy contracts are becoming more stringent. It is possible to examine trends in the real monetary value of cash and in-kind rents for two villages in the sample, using data from the rst iteration. 39 In 1986, the average monetary value of cash and in-kind rents paid per holding in the villages of Sarfaraz and Platoo amounted to Rs5287. Of this, Rs1396 was paid by leaseholders in cash and Rs3891 was paid by sharecroppers in kind. By 1996, rents in the two villages had increased. In real terms, the average monetary value of all cash and in-kind rents per holding in 1996 was Rs5951. This was some 10% more than that of the sample as a whole. Of this, Rs1510 was paid by leaseholders in cash; an increase of 8.15%. Sharecroppers paid some Rs4440 in kind; an increase of 14%. Overall, the average monetary value of all cash and in-kind rents per holding was 12% higher in 1996 than 1986 in real terms. On aggregate, land rental, whether by itself or in conjunction with the ownership of some land, has become more important in the structure of access to land among farmers in the period since 1986. This difference, in all likelihood, is attributable to SCARP Mardan. The project has had the effect of making land a potentially more productive resource than was previously the case. Certainly, the general perception among rural households of central NWFP is that land has become more valuable because of its productive potential. This perception is re ected in the extremely high price of purchased land. With land becoming a potentially more productive resource, and with the control of land being a pivotal aspect of Pakhtun identity, more households seek to access land to enhance their livelihood security in circumstances of continuing widespread poverty. Indeed, the relocation of farms between 1986 and 1996 may be related to the shift in the structure of access to land and, in particular, the increasing importance of land rental. It is possible that the expansion of land rental has occurred on less favourably located land, the productivity of which has nonetheless been perceived as having improved by those renting the land. Moreover, not only are potential lessees willing to pay higher seasonal fees for land, but they are also ready to accept the authority of the landlord in an extensive network of extra-economic social and political obligations that reinforce the latter’s social power.40 Rental contracts are both an important means of inequitably transferring the bene ts of SCARP Mardan from the direct producer to the landlord and a means of sustaining the social power of landlords. The interlinking of markets also helps explain why mean consumption for the project area is greater than mean consumption for rural NWFP as a whole, even though poverty in the area is greater than poverty in rural NWFP as a whole. The inequitable distribution of agrarian incomes, and thus the mean consumption gure for the area, is higher than the rural provincial average because it is pulled up by those with greater incomes. The evidence for this proposition is clear; the highest 20% of income accounts for 57.2% of all income in the project area, 339
A. HAROON AKRAM-LODHI
compared with 44.5% in Pakistan as a whole. The lowest 20% of incomes accounts for 3.9% of all income in the project area, compared with 7.3% in Pakistan as a whole. The concentration of income is so skewed as to imply that the area is subject to poverty processes more severe than the rural provincial norm. Conclusion SCARP Mardan sought to deliver changes in the quantity and quality of water inputs. This was to result in higher levels of agricultural production and agrarian productivity. Enhanced productivity would permit farmers to devote more of their land to cash crops, and thus lead to an improvement in farm incomes. Two reasons why SCARP Mardan failed to meet its explicit objectives can be readily identi ed. The rst concerns the capacity of farmers to allocate their resources as they see t and to capture the bene ts of an improved water delivery system. In central NWFP, land is power; cultural, political and economic. For farmers to be able to effectively allocate their resources, they must control the key production resource; land. Yet, as was clear from the 1986 baseline sample, many farm households did not control land. Some 46% of rural households, many of whom were farmers, had no secure access to land. Many small landowners also relied on access to land rental to create small yet viable operational holdings. By 1996, it was clear that access to suf cient land had diminished in the area over the past 10 years, and that the size of operational holdings used by farm households had declined. In this light, the distribution of land was sure to have an impact on the capacity of SCARP Mardan to meet its goals, yet the project did not undertake any analysis of the distribution of land (and non-land) assets in the districts where it was to have an impact. There is now literature that argues that the distribution of assets has a critical impact on the pattern and rate of economic growth.41 Without an analysis of the distribution of assets, there was no way to know whether SCARP Mardan would engender growth. Granted, the pattern of development witnessed within the project area in 1996 would not have been comprehensible based solely on an analysis of asset distribution. However, the inability of the project to engender sustainable growth at the rates envisaged would have been more comprehensible had an understanding of the structure of asset distribution been in place prior to project design. The second reason that can be identi ed as to why SCARP Mardan failed to meet its objectives has to do with markets. In orthodox economic analysis, prices emerge endogenously from the operation of markets. However, it is clear that prices emerge in the determinate social and economic settings within which markets operate. It is therefore necessary to understand the social and economic setting within which market operations and price formation takes place.42 In this light, the fact that no analysis of market operations was developed prior to the design of SCARP Mardan is important. Lacking such an analysis, planners— whether they were seeking to enhance livelihood security, enhance pro t, or 340
LAND, WATER AND SOCIAL POWER IN PAKISTAN
enhance social control—had no real way of knowing the types of response that SCARP Mardan would produce among farmers subject to increased production. For the bene ciaries of rural infrastructure projects such as SCARP Mardan, the outcome may appear ‘like an act of God’. This does not mean, however, that such projects are capable of delivering what is expected of them. To ful l expectations, it is clear that project design must integrate an understanding of the determinate social and economic environment within which the project is going to impact. In short, project design must integrate issues of social power. Notes and references 1. Earlier versions of this article were presented to the Development Studies Seminar at the School of Development Studies, University of East Anglia, UK, and to the Rural Development Seminar, Institute of Social Studies, The Netherlands. The author thanks the late Abdul Hamid, Catherine Akram-Lodhi, Jim Freedman, Melissa Godwald, Bill Marshall, Wendy Quarry, Ben Rogaly, and an anonymous referee of this journal. The Canadian International Development Agency has nanced the research for this article under contract SEL-95–0075. However, the author is solely responsible for the contents of the article, and any errors and omissions which it may contain. 2. World Bank, World Development Report 1994 (Oxford: Oxford University Press, 1994), p 3. 3. Ibid, p 4. 4. Ibid, pp 6–7. 5. Ibid, p 7. 6. M. Mackintosh, ‘Abstract markets and real needs’, in H. Bernstein, B. Crow, M. Mackintosh and C. Martin (eds), The Food Question: Pro ts versus People (London: Earthscan Publications, 1990). 7. H. Bernstein, ‘The political economy of the maize liere’, in H. Bernstein (ed), The Agrarian Question in South Africa (London: Frank Cass, 1996). 8. W. Quarry, ‘Ef ciency and sustainability of the irrigation system’, in J. Freedman and A.H. Akram-Lodhi (eds), Water, Pipes and People: The Social and Economic Impact of the Salinity Control and Reclamation Project in Mardan, Northern Pakistan (Lahore: Vanguard Press, 2001). 9. Ibid. 10. J. Freedman and A.H. Akram-Lodhi (eds), ‘Research methodology’, Water, Pipes and People: The Social and Economic Impact of the Salinity Control and Reclamation Project in Mardan, Northern Pakistan (Lahore: Vanguard Press, 2001). 11. A. H. Akram-Lodhi, ‘Agrarian classes in Pakistan: an empirical test of Patnaik’s labour-exploitation criterion’, Journal of Peasant Studies, Vol 20, No 4, 1993, pp 557–589. 12. Freedman and Akram-Lodhi, op cit, Ref 10. 13. A.H. Akram-Lodhi, ‘The farm economy’, in J. Freedman and A.H. Akram-Lodhi, op cit, Ref 10. 14. Ibid. 15. World Bank, Performance Audit Report: Salinity Control and Reclamation Project (SCARP) Mardan (Washington, DC: Operations Evaluation Department of the World Bank). 16. J. Freedman (ed), ‘A case for equity’, Transforming Development: Foreign Aid for a Changing World (Toronto: University of Toronto Press, 2000). 17. S. Ali, ‘Poverty assessment: Pakistan’s case’, The Pakistan Development Review, Vol 34, No 1, 1995, pp 43–54. 18. J. Freedman, J. and A.H. Akram-Lodhi, ‘Findings and lessons learned’, in Freedman and Akram-Lodhi, op cit, Ref 10. 19. Quarry, op cit, Ref 8. 20. J. Boyce, Agrarian Impasse in Bengal: Institutional Constraints to Technological Change (Oxford: Oxford University Press, 1987). 21. R. Chambers, Managing Canal Irrigation: Practical Analysis from South Asia (Cambridge: Cambridge University Press, 1988). 22. A.H. Akram-Lodhi, ‘A bitter pill? Peasants and sugarcane markets in northern Pakistan’, European Journal of Development Research, Vol 12, No 1, 2000, pp 206–228. 23. A. Ahmed, Pukhtun Economy and Society (London: Routledge Keegan Paul, 1980). 24. C. Lindholm, Generosity and Jealousy: The Swat Pukhtun of Northern Pakistan (New York: Columbia University Press, 1982), p 205.
341
A. HAROON AKRAM-LODHI 25. Ibid, p 211. 26. S.A. Rittenberg, Ethnicity, Nationalism and the Pakhtuns (Durham, North Carolina: Carolina Academic Press, 1988), p 25. 27. Lindholm, op cit, Ref 24, p 57. 28. Akram-Lodhi, op cit, Ref 11. 29. Government of Pakistan, Pakistan Census of Agriculture 1980 (Lahore: Agricultural Census Organization, 1980). 30. Government of Pakistan, Pakistan Census of Agriculture 1990 (Lahore: Agricultural Census Organization, 1990). 31. Government of Pakistan, op cit, Ref. 29. 32. Government of Pakistan, op cit, Ref. 30. 33. J. Freedman, ‘Income growth, inequalities and project outcomes’, in J. Freedman and A.H. Akram-Lodhi, op cit, Ref 10. 34. P. Bourdieu, Outline of a Theory of Practice (Cambridge: Cambridge University Press, 1977). 35. Freedman, op cit, Ref 16. 36. Bourdieu, op cit, Ref 34. 37. C. Humes, ‘Pierre Bourdieu: re exive practice’, http://www.ac.www.edu/ , curth/papers/bourdieu.html, accessed 30 March 2001. 38. Ahmed, op cit, Ref 23. 39. Akram-Lodhi, op cit, Ref 11. 40. Freedman, op cit, Ref 26. 41. L. Taylor, Varieties of Stablization Experience (Oxford: Clarendon Press, 1988). 42. A.H. Akram-Lodhi, “‘We earn only for you”: peasants and “real” markets in northern Pakistan,’ Capital & Class, No 74, 2001, pp 77–106; and Akram-Lodhi, op cit, Ref 22.
342
