SPECIAL ARTICLEfebruary 23, 2008 EPW Economic & Political Weekly58Nature of Household Dependence on Common Pool Resources: An Empirical StudyKanchan Chopra, Purnamita Dasgupta The authors wish to thank the National Sample Survey Organisation for access to household level data collected in the 54th round in 1998. Participants at the Institute of Economic Growth faculty seminar provided insightful comments to improve the paper. The usual disclaimer applies.Kanchan Chopra (kanchan@iegindia.org) is at the Institute of Economic Growth, Delhi and Purnamita Dasgupta (pdasgupta@icrier.res.in) is at the Indian Council for Research on International Economic Relations, New Delhi. Some recent studies on forest-based common pool resources have interpreted situations in which households choose to spend time on collection from the forest commons for sale and value addition as an income enhancing activity that is independent of the common’s role as a safety net. This paper tests for the hypothesis by distinguishing between non-timber forest produce for sale and for self-consumptions, using the National Sample Survey Organisation data for a sample of 78,000households in Bihar, Karnataka, Madhya Pradesh and Maharashtra. It finds that households collecting only for sale purposes are not likely to be income poor. They may collect because they have more secure property rights, greater access to forests and/or markets and may even beasset rich.Common pool resources (CPRs) are defined as resources that are non-exclusive, to which rights of use are distri-buted among a number of co-owners, often identified by their membership of a community or a village. In certain instances, well-defined property regimes do not exist, although free or open access does not prevail either. The de facto access may be limited to some groups and legitimised by law, conven-tion, customary rights or traditional practices. In the continuum of property rights “common pool resources” lie between common property resources and openly accessible resources. They are resources with varying degrees of access on which multiple and often overlapping property rights and regulatory regimes exist. Such rights of access include those defined on different catego-ries of government forests. Common property resources, on the other hand, are defined in the literature as “private property for a group”.1Given the different modes through which access to them is possible, it has not been easy to determine the magnitude of CPRs in India. Different methodologies have been used to estimate it and estimates have varied.2 There is general agreement however, that in large parts of the country,CPRs provide a source of consumption and income, and therefore, utility augmentation for households that have access to them. In large parts of the country, mainly in the arid and semi-arid regions, CPRs are of a magnitude large enough to impact the decision-making of large numbers of households with respect to their labour-time alloca-tion. Households do allocate time to collection from the commons and for the management of the commons, wherever it is feasible to do so.1 Alternative Approaches and HypothesesWhat is the nature of household dependence on CPRs in develop-ing countries and how does it change with a rise in incomes over time? The early literature on CPRs, in the developing country context postulates that they supplement rural livelihoods and act as the safety nets for the poor, seasonally or specially in times of agricultural crises. Godoy and Bawa (1993), Cavendish (2000) and Hecht et al (1988), among others, have highlighted the supplemental role of non-timber forest product (NTFP) collection for the poorest households. This can be alternatively character-ised as the “substitution” between the CPR based means of liveli-hood and other primary sources of rural livelihood, eg, agricul-tural income. Several micro-studies in different parts of India also provide evidence to support this proposition. The important role ofCPRs in reducing income disparities in the rural areas,

SPECIAL ARTICLEEconomic & Political Weekly EPW february 23, 200859when other sources of livelihood fail, has been noted by different authors [CWS 2001]. The landmark study of Jodha (1986), found that common pool resources add 15 to 23 per cent to poor people’s income. Several others [Pasha 1992, Beck and Ghosh 2000] found that household income of the rural poor was augmented by 12 to 15 per cent from these resources. A more recent study [Dasgupta 2006] finds that theCPRs continue to make significant contribu-tions to the livelihoods of the poorest households, in rare cases up to 40 per cent of their annual incomes.Another strand of literature points out the complementarity between agricultural output and the use of CPRs as inputs to agriculture. A large number of agricultural inputs such as fodder, grazing grounds and irrigation water are made available through the conservation of CPRs. This has been cited in some studies as the raison d’etre for households of varying socio-economic status to come together to protect and conserve the commons. Chopra et al (1990) focus on the complementarity between agricultural and livestock incomes and protection of upper catchments for fodder collection and common water resources for irrigation. This implies that cultivator households get substantial benefits from CPRs. Singh et al (1996) for instance, find that for eight villages in Punjab, the annual income fromCPRs, for cultivator households was greater than that for the landless households. In such situations, a part of the dependence on CPRs would continue with increased incomes.It is important to note that forests constitute a large part of CPRs3 and the literature on CPR dependence is essentially centred on forest dependence. The nature of household dependence on forests, and the factors, which bring about changes in it are there-fore relevant in this context. Godoy et al (1992 and 2002) have argued that increases in income and modernisation of economies lead to changes in the mix of forest activities on which people depend with an implication that forest extraction opportunities decrease with an increase in incomes from cultivation. Gunatileke and Chakravorty (2003) show for Sri Lanka that even when culti-vation cannot be extended to forest land, agricultural income shows a statistically significant negative relationship with the level of forest extraction. The relationship works through compet-ing labour allocation between the two activities. Households at times make a deliberate choice of spending labour time available on forest extraction as against agricultural activities. Byron and Arnold (1999) emphasise that there exists, “the need to distin-guish between those uses of the forest which reflect actual dependency on the forest in the sense that the users would be left seriously worse off in their absence and those uses which reflect choice and the presence of adequate alternatives”.Angelson and Wunder (2003) also maintain that forests have both potentials and limitations for improving human welfare. Pattanayak and Sills (2001) find, for instance, thatNTFP collec-tion from forests reduces risk, acting as consumption and income smoothing mechanisms, even for the relatively wealthier house-holds. The distinction that Fisher (2004) makes between low return forest activity (LRFA) and high return forest activity (HRFA) is also significant in this context. Fisher concentrates on the significance of these two kinds of activities in determining the role of forests in either “prevention” of poverty (associated with LRFAs which supplement income and may also buffer adverse shocks) or “reduction” of poverty (associated withHRFAs which are market oriented). Another way of stating the same is to note that the above studies distinguish between household depend-ence on forests in the presence of choice and in situations without choice. Such a distinction has implications for human well-being and policymaking. This analysis can clearly be seen to be a restatement of Sen’s (1992 and 1999) assertion of the inherent significance of free choice in assessing developmental activities. Of late, a large number of empirical studies have highlighted the place that individuals give to freedom in assessing development projects [Alkire 2002]. The central issue in the context of household dependence on CPRs, in particular, forestCPRs, is whether it is a dependence based on choice or one arising out of lack of choices? This paper postulates that if the dependence can be linked to market demand for high value products independent of self- consumption, the likelihood of its being of the first kind is high. This has important implications for the rate at which such demand grows in the future and the impact it has on sustainable use of forest products. It is precisely for this reason important to go beyond simple assertions of substitutability or complementa-rity of income from forests with the income from other kinds of economic activity and examine the nature of household demand for forest products with reference to the existence or otherwise of alternative options.In examining collection fromCPRs, our analysis shall distin-guish between different products collected as also between the collection for self-consumption and for sale. The hypotheses stated above shall be tested with household level data from four states in India. The econometric analysis of a large cross sectional data set is rooted in a simple static household decision-making model, which provides a conceptual framework to the study and is described in Section 2. The data and its descriptive analysis are presented in Section 3. For purposes of econometric estimation, we focus on NTFPs, collected both exclusively for sale and sometimes also for self-consumption. Multinomial logit and logit frameworks are used for analysis of household decision-making. Section 4 presents the results obtained while Section 5 puts forth a few concluding observations. 2 Model and Econometric SpecificationWe first discuss the model and then the econometric structure.2.1 HouseholdModelWithin the framework of a one period labour allocation model, with the stock of the resource and the annual flow of products and services assumed to be given, households typically divide time between collection from the commons, working for a wage income and leisure. They may also have some non-labour incomes. We formulate a model to capture the household behaviour with regard to labour allocation, with differential returns to time spent in collection and time spent in other wage employment.The household derives utility from the consumption of non-collected goods as well as from the direct consumption of goods it collects for this purpose from the CPR.

SPECIAL ARTICLEfebruary 23, 2008 EPW Economic & Political Weekly60The standard approach in the literature typically assumes the existence of surplus labour. A distinction is then made between the returns toCPR-based collection activity and the returns to labour in an alternative industrial/agricultural employment. An alternative possibility for modelling such an empirical context while explicitly bringing in the options for sale of collected products proceeds as follows. Let w be the returns to a given employment of time To (where To is thus effectively an employed labour constraint). Assume that the rest of the time (T-To) is distributed between leisure, collections for consumption and collections for sale. With pm as the returns per unit time for collection, the utility function is given by: U=f(X, L, C)Ux, UL, Uc > 0; Uxx, ULL, Ucc < 0 ...(1)and C = F(αc, Tc) ...(2)where X: Consumption of non-collected goods with a price of unityL: leisure timeC: Consumption of goods collected from the commonsαc = Tcc/Tc where Tcc + Tcs = TcDistinguishing between time spent in collection for consumption (Tcc) and time spent on collection for sale (Tcs), implies : Tc = Tcc + Tcs where Tc is total time spent on collection activities. Letαc be the proportion of time spent in collection for self-consumption out of the total time spent in collection activities (Tc). Collection is proportional to time spent; hence consumption C can be assumed to depend on αc and Tc, the total time spent on collection. Therefore,αc is treated as the decision variable with the time constraint being such that: T = Tc + To + L ...(3)Further, given a labour employment constraint of To, the household allocates the rest of its time (T – To) between leisure (L) and collections from commons (Tc), whether for sale or consumption. Thus, the returns to To are evaluated at w while thereturns to time spent on collections and leisure are evalu-ated at pm.Therefore, the full income budget constraint is defined in terms of a time constraint as: I + pm (T – To) + w To = pm L + X + αc pm Tc ...(4) In this constraint I represents non-labour income (for instance, remittances, interest income on assets). The Lagrangean is set up and the optimality conditions are derived. £=U (X, L, C)+λ {I + pm (T–To) + w To – pm L–X–αc pm Tc} ...(5)The first order conditions on the decision variables imply: £x = Ux – λ = 0 …………….(a)£L = U L – λpm = 0 …………(b)£ αc = Uc C αc – λ pm Tc = 0 …………(c)£Tc = Uc CTc – αcλpm = 0 …………(d)£λ= {I + pm (T – To) + w To – pm L – X – αc pm Tc }…..(e)from (a) and (b) : Ux/U L = 1/pm λ = Ux = U L / pm from (c) and (d): UcC αc – λpm Tc = Uc CTc –αcλpm Since, UcC αc / λpm = Tc, αcλpm = UcCTc– Uc C αc + λ pm Uc Cαc / λ pm ...(6)Orαcλpm = Uc CTc ...(7)In the above condition, the left hand side gives the gain in terms of the proportion of time spent in collection for consump-tion, evaluated at the market rate of return. This is equated to the (right hand side) gain in utility (in terms of X) from a unit increase in Tc through an increased access to marketed commodities arising from the returns (sale value) out of collection time (Tc ) that the household is able to sell. This system would solve for an optimal αc where, αc = αc (pm, w,To, I) ...(8)And, (1– αc), the time spent on collection for sale is, Ω* = Ω* ( pm, w,To, I) ...(9)Note that in such a formulation the returns to time spent on CPR collection differ from w. We model an empirical context where we do not assume pm and w to be equal because of imper-fections in the market for labour caused by access related varia-bles. Segmented labour markets coexist at a point in time, with pm the price of the marketed product, being different from the wage rate, w. Additional amounts of labour spent on collection are valued at w, whereas the price available in the market for sale of collected goods is pm, and with perfect labour mobility pm would equate to w. 2.2 EconometricStructureThus pm and w need not equilibrate or be equal and household responses would be dependent on the constraints faced. House-holds differ with respect to their capability to access product and labour markets and earn income at a rate of return of w, or to sell products at pm. This access depends on household characteristics and forms the basis for the differences in the nature of their dependence on forests. The categorisation of households follows from the recognition of this fact that households can or cannot respond optimally because of the constraints faced by them. Hence in the empirical investigation of household behaviour with respect toCPRs, a set of characteristics reflecting both exoge-nously determined factors and household characteristics are to be taken into account. The econometric specification is as follows:Hi (Tc, Ω) = x βi + ε ...(10)where Hi (Tc, Ω) stands for household category with respect to collection and sale of CPR products. In order to capture differential access, the econometric model distinguishes between three such categories of households. βi represents household characteristics as specified by a set of independent variables, x.Based on the available dataset, three categories of households are defined for the analysis as follows: – The household collects and consumes but does not sellCPR products: this category of households typifies dependence for survival at low income levels, ie,αc = 1.– The household collects and sells, but does not consume CPR products: these households collections are market-driven as they do not consume these products at all, i e,αc= 0.– The household collects for both sale and self-consumption: thiscategory is typically a mix of the first two kinds of house-holds, i e, 1>αc >0.

SPECIAL ARTICLEEconomic & Political Weekly EPW february 23, 200861Table 1: All-India Summary FindingsI Size of common property land resources (CPLR) Percentage of CPLR in total geographical area (%) 15CPLR per household (hectare) 0.31CPLR per capita (hectare) 0.06Reduction in CPLR during last five years (per 1,000 hectares) 19II CollectionsfromCPR Households reporting collection of any material from CPRs (%) 48Average value of annual collections per household (Rs) 693 Ratio of average value of collection to average value of consumption expenditure (%) 3.02III Nature of use of CPRs (data per household) Share of firewood in value of collection from CPRs (%) 58Average quantity of firewood collected from CPRs annually (kg) 500Average quantity of fodder collected from CPRs annually (kg) 275Source: NSSO, 54th Round, 1999.Table 2: Distribution of Households Collecting from Commons (number and percentage of households)State FirewoodFodder NTFP State TotalIndia 24,744(36) 6,450(9)9,365(14) 67,674(100)Bihar 2,977(40)1,117(15)582(7)7,482(100)Karnataka 1,666(53)539(17)304(10)3,161(100)Madhya Pradesh 3,184(55) 516(9) 1,408(24) 5,812(100)Maharashtra 3,222(60)679(13)514(9)5,374(100)Figures in parentheses denote percentage of households in each category.Table 3: Average Annual Value of Collections by Households (Rs per household per annum)State FirewoodFodderNTFPIndia 1,191.391,339.951,936.85Bihar 846.681,108.031,479.67Karnataka 775.68944.901,218.45Madhya Pradesh 1,022.29 1,633.04 1,150.59Maharashtra 835.041,406.083,047.41Further, we hypothesise that, in the household context, the decisions to collect/not collect and consume/sell are taken simul-taneously at a point in time rather than in a sequential manner. Each household, faced with a set of conditions defining its income and asset situation and its market and CPR access situation, takes decisions which place it in one of the three categories. The multi-nomial logit framework (MNL) specification is thus appropriate for analysing the data. The data on collections is for selectedNTFPs from CPRs and includes fruits, roots, tubers, spinach, gums and resins, honey, medicinal/herbs, fish, leaves, weeds, grass, cane, bamboo, etc.Households that collect only for sale (HS) are critically a dif-ferent group. For these households, sale of collected goods is an income enhancing activity, in the absence of self-consumption requirements. This paper focuses on such households and by high-lighting their characteristics, investigates the conditions under (and the extent to) which forest-based CPR dependence is an activ-ity undertaken in the presence of other options and represents market-linked expansion of choices. The mixed category of house-holds, those that collect for sale and consumption is difficult to identify as the belonging to either the subsistence or the commer-cial sale categories. They represent parts of the process of change from collection for self use to sale of surpluses in the market, a continuum which as hypothesised above is not extendedto include those who collect with the express intention of sale. The variable, proportion of sales to collection was generated for the mixed category and the mean value as well as variation looked into. This simple diag-nostic investigation on the households in the mixed category was done to cap-ture a variation within this category in terms of the proportion of sales out of total collections. The results are on the whole reassuring in terms of the low standard deviations. They indicate that households tend to cluster around the mean with low variation. This indicates that households that collect only for sale are indeed located in a cluster dis-tinct from those that collect for con-sumption and sale. We consider a situation with three outcomes (y – 1, 2, and 3) and a vector of explanatory variables X. The three outcomes are unordered. In the MNL model, we estimate a set of coefficients β(1), β(2), β(3) corresponding to each outcome category. The model, how-ever, is unidentified in the sense that there is more than one solution to β(1), β(2) andβ(3) that leads to the same prob-abilities for y=1, y=2, and y=3. To identify the model, we set β(1)=0. The remaining coefficients,β(2) andβ(3) would measure the change relative to the (y=1) group. The relative probability of y = 2 to the base category is Pr (y=2)/Pr (y= 1) = eXβ(2) The coefficient numbers from the estimation exercise are not interpretable in themselves. The signs and significance levels of these coefficients are therefore interpreted relative to the base category of households, ie, the group of households that do collect from the CPRs for consumption only. It was considered appropriate to treat 1 as the base category since it is a “pure” case of households that collect for consumption only. The results shall be interpreted accordingly. Marginal effects have been calculated and interpreted in arriving at conclusions. In some cases where one category or more have very few observations, logit estima-tion is conducted. 3 DescriptiveDataAnalysisThis section provides analysis of the data generated by all-India surveys.3.1 Household Dependence on CPRsThe following are all-India findings: All-India Findings: The data used in this paper is taken from the 54th round survey conducted in 1998 by the National Sample Survey Organisation (NSSO). It provides for the first time in India a comprehensive state and national level database on the size, utilisation and contribution of CPRs. It also provides disaggre-gated information at the state level in terms of agro-climatic zones. The survey aims at an assessment ofCPRs in terms of their contribution to the lives of the rural population. The role of CPRs in providing biomass, fuel, irrigation water, fodder for livestock and other forms of economic suste-nance has been the main focus of the survey. The results are based on a comprehensive survey of 78,990 rural households in 10,978 villages across the country.TheNSSO defines common property resources as the resources that are accessible to and collectively owned/held/managed by an identifiable com-munity and on which no individual has exclusive property rights. Two differ-ent concepts have been used to deter-mine the size and access to CPRs in this report. The de jure approach was used for collection of data on the size of CPRs. In this approach only those resources were treated asCPRs which were within the boundary of the vil-lage and were formally held (by legal sanction or official assignment) by the village panchayat or a community of the village. The second approach, the

SPECIAL ARTICLEfebruary 23, 2008 EPW Economic & Political Weekly62cent of the value of collections. The proportion of sales in the total collections is uniformly higher for NTFPs, indicating a higher level of commercial activity for this product. Other studies also indicate that the collection of NTFPs from CPRs is more market driven than that of firewood and fodder. Note also that micro evidence reveals thatNTFPs which have a significant market demand are collected by almost all types of households, irrespec-tive of community or economic class [Ravindranath et al 2000; Rao 2000]. Findings for Selected States: Four states were selected for detailed empirical analysis in this study. The states were selected on the basis of the significance of CPRs in the economies of the states.CPRs are concentrated in the central plateau region and in arid and semi-arid tracts of the country. The states of Karnataka, Madhya Pradesh, Maharashtra and Bihar were selected for this study. As per the NSS data base, CPRs constitute 22 per cent of the land area in Madhya Pradesh and 11 per cent, 10 per cent and 8 per cent in Maharashtra, Karnataka and Bihar, respectively. The selected states are therefore a fairly representative sample of states with varying levels ofCPR dependence. Table 2 (p 61) indicates the nature of dependence of house-holds at the all-India level and in the selected states on CPRs as measured by numbers collecting each of three commodities, firewood, fodder and NTFPs. The NTFP collection involves 7 to 24 per cent of households in the four states, Madhya Pradesh having the highest percentage of 24 per cent. Large numbers of house-holds collect firewood from the commons in all four states, the percentage varying from 40 to 60, while the average for India is 36 per cent. The percentage of households collecting fodder is lower, ranging from 9 per cent for India to 17 per cent for Karna-taka. It is lowest in Madhya Pradesh at 9 per cent. Table 3 (p 61) gives the value of annual collections by house-holds as reported in the NSSO study. On average the numbers seem lower than those reported by micro-studies. Note that value of NTFP collection at Rs 1,936 per household for India is higher than that of firewood or fodder. Maharashtra households with a value of approximately Rs 3,047 collect NTFPs of highest value. Table 4 gives a pic-ture of the range of NTFPs collected in the states and in the coun-try as a whole. At the all India level, leaves, weeds, cane grass, bamboo constitutes a large part of total col-lections. Fruit and fish follow these. Maha-rashtra and Bihar fol-low the same pattern but in Karnataka, cane grass and bamboo are more significant than Table 7: MNL Results on Pooled Data for Four StatesHousehold Type 2: Sale Only MNL Marginal CoefficientsEffects Asset poverty 0.19 -0.007Irrigation 0.112 0.079Net sown area 0.022 -0.007Distance from forest -0.01 0.00Distance from metalled road 0.375* 0.033Patta 0.91*0.223Constant 0.01 Household Type 3: Sale and Consumption Asset poverty 0.302 0.048 Irrigation -0.49* -0.13Net sown area 0.0 83 0.017Distance from forest -0.013 -0.002Distance from metalled road 0.23 0.007Patta -0.57*-0.252Constant term -0.62* Prob chi2 0.000 No of observations 1144* indicates significance of the coefficient at 5 per cent level.de facto approach, was adopted for collecting information on use of CPRs. For this purpose CPRs were extended to include all resources which were in use by the community by convention irrespective of ownership, and even if they were located outside the boundary of the village. The size of CPRs was therefore based on a stricter de jure definition while the “use” data took into account the actual position with regard to access. Government forests, which have been classified into three categories in India as per their legal status: reserved forests allowing restricted access, protected forests allowing access to locals and, unclassed forests (all other) have also been treated in this manner, thereby distinguishing between the conceptual basis for defining size and use. In this paper, we use data on collections, which is based on the de facto approach to CPRs in the dataset. This is consistent with our definition of CPRs as the resources to which access is defined through varied legal and conventional sanctions. TheNSSO study (1999) is based on a substantially larger sample as compared to the micro-studies on which earlier evidence with respect toCPR dependence has been reported. However, it is interesting to note that the proportion of CPR area in total geographical area falls in the same range as reported from the micro-studies. On average, the NSSO reports lower percentages for the value of collection to consumption expenditure. The relative dependence of the poor is more than of the non-poor, the latter are particularly dependent onCPRs for firewood. These findings are similar to the evidence gained from micro-studies. Table 1 (p 61) provides country level summary statistics on CPRs as estimated by the NSS. It becomes clear from the table that CPRs form a substantial part of the total geographical area (15 per cent for the country) and that a large percentage (48 per cent) of rural households reports collection from CPRs. Additionally, the share of firewood in collection from commons amounts to 58 per Table 4: NTFPs Collected and Percentage Distribution of Collected Items(%)State/NTFP IndiaBiharKarnatakaMadhyaPradeshMaharashtra1 Fruits 17.86 15.48 31.03 28.21 25.292 Roots, tubers, spinach, etc 9.10 14.22 1.59 7.93 0.333 Gums and resins 0.61 – – 1.65 0.164 Honey 2.96 1.83 9.28 2.21 2.305 Medicinal/herbs 2.72 0.69 2.38 0.496 Fish 16.93 19.72 11.41 4.96 16.917 Leaves 26.51 28.78 10.34 43.28 29.728 Weeds, grass, cane, bamboo 23.31 19.27 36.34 9.38 24.79Column totals equal 100.Table 5: NTFP Collection, Consumption and Sale: Distribution of Households(%)Household Type India Bihar Karnataka Madhya Pradesh MaharashtraCollection for consumption 45.88 55.15 59.29 11.51 37.67Collection for sale 31.72 26.80 32.24 67.83 45.83Collection for sale and consumption 22.38 18.54 8.55 20.67 16.50Total number of households 9377 582 304 1408 Table 6: Descriptive Statistics for the Variables Variable Mean Value No of Observations Asset poverty (=1 if poor) 0.199 2809Net sown area ( in hectares) 1.147 2039Irrigation (=1 if access to irrigation and/or mechanisation) 0.307 1616Distance from forests (in kms) 3.353 1792Distance from metalled road (in kms) 2.518 2757Patta (=1 if access to patta) 0.016 2807

SPECIAL ARTICLEEconomic & Political Weekly EPW february 23, 200863leaves. Fruit also constitutes a larger part of total collections. In Madhya Pradesh where 24 per cent of households are engaged in NTFP collection, 43.28 per cent of the collections consists of leaves(possibly tendu leaves for bidi making contribute signifi-cantly to this). NTFP Collection, Consumption and Sales Behaviour of Households: We next examine collection, consumption and sales behaviour of households (Table 5, p 62) with respect to NTFPs for all India and for the selected states, ie,Bihar, Karna-taka, Madhya Pradesh and Maharashtra. The NTFP collection data is available for three categories. The number of households involved is larger than that in fodder collec-tion, though way below that in firewood collection. The commercialised nature of this activity is evident from the fact that collection “for sale only” is a large per-centage with 31.72 per cent at the all-India level falling in this cat-egory. The percentage increases in the states of Karnataka and Maharashtra to 32 and 45 per cent approxi-mately. It is highest in Madhya Pradesh at about 68 per cent. Explanatory Varia-bles for Econometric Estimation: We esti-mate the multinomial logit and logit models for two alternative for-mulations: (a) for all four states as a single composite model; (b) for each of the selected four states using household level data. The multinomial logit and logit models seek to explain the varying dependence of households onNTFPs Table 9: NTFP Collection and Sale in BiharHousehold Type 2: Sale Only MNL Marginal CoefficientsEffectsIrrigation -0.36-0.014Net sown area 1.39* 0.178Asset Poverty 1.77* 0.427Distance from forest 0.47* 0.023Distance from metalled road 0.3 0.099Patta 24.81*2.306Constant term -3.29* Household Type 3: Sale and Consumption Irrigation -0.66-0.113Net sown area 0.89** 0.027Asset poverty -0.32 -0.324Distance from forest 0.57* 0.075Distance from metalled road -0.25 -0.103Patta 22.082.001Constant term -1.64* Prob > chi2 0.000 No of observations 180 ** indicates significance at 1 per cent level. * indicates significance of the coefficient at 5 per cent level.Table 8: MNL Results on Pooled Data for Four States with State DummiesHousehold Type 2: Sale Only MNL Marginal CoefficientsEffectsAsset poverty 0.151 -0.012Irrigation 0.196 0.079Net sown area -0.0154* -0.020Distance from forest -0.046* -0.002Distance from metalled road 0.054 0.022Patta 1.794*0.349Dummy MP 2.258* Dummy Karnataka -0.747* Dummy Bihar -0.007 Constant 0.01 Household type 3: Sale and Consumption Asset poverty 0.291 0.051 Irrigation -0.373 -0.114Net sown area -0400. 0.010Distance from forest -0.044 -0.005Distance from metalled road -0.104 -0.032Patta -0.416-0.326Dummy MP 2.54* Dummy Karnataka -2.65* Dummy Bihar 1.23** Constant term -0.987** Prob chi2 0.000 No of observations 1144 ** indicates significance at 1 per cent level. * indicates significance of the coefficient at 5 per cent level.through variations in household characteristics. The character-istics considered relevant for determining collection, consump-tion and sales behaviour with respect toCPRs can be placed in four distinct groups reflecting economic status, access to CPRs, access to markets and institutional arrange-ments for governing collection.Economic Status: We use the data from the NSS 1998 survey for generating three indi-cators of “household economic status”. (i) Asset Poverty: Households that do not own livestock and possess less than one hectare of land are characterised as asset poor. This is gen-erated from the avail-able data on livestock and landownership. While the former is a categorical variable indicating ownership or otherwise, land-ownership is specified inthe data set in terms of number of hec-tares owned by the household. (ii) Net Area Sown: This is an indicator of agricultural income in the current year. (iii) Access to Mechanised and Irrigated Agriculture: This is an indicator of a higher level of agricultural income from land sown. These three indicators reflect different aspects of poverty – asset poverty (comparative) social deprivation due to access factors or disempowerment and income poverty. Other Characteristics – Ease of physical access to CPRs, the source of supply forNTFPs, is approximated by the variable “distance from forests”.– Ease of access to markets, the source for demand, is approxi-mated by the explanatory variable “distance from metalled road”.– Existence (or otherwise) of institutions for CPR management is defined in terms of two kinds of institutions: (a) existence of forest-based institutions such as joint forest management groups and/or van panchayats and (b) existence of tree patta schemes which are in essence rights to the products of trees in existence in some parts of the country. Table 6 (p 62) presents descriptive statistics for each of the explanatory variables defined above.4 Results from MNL and Logit Models The following are estimates from pooled data.Table 10: NTFP Collection and Sale in KarnatakaHousehold Type 2: Sale Only MNL Marginal CoefficientsEffectsIrrigation 1.12**22.54Net sown area --2.18* --1.18Distance from forest -0.73** – 3.91Asset poverty -2.87* –2.03Distance from metalled road -0.71* –7.86Patta -38.56 -15.42Constant term 3.09* Household Type 3: Sale and Consumption Irrigation 1.0822.88Net sown area 3.38 1.18Distance from forest 14.38 3.23Asset poverty 6.67 2.03Distance from metalled road 36.59 7.97Patta 33.35 15.26Constant term -2.95 Prob > chi2 0.000 No of observations 94 ** indicates significance at 1 per cent level. * indicates significance of the coefficient at 5 per cent level.

SPECIAL ARTICLEfebruary 23, 2008 EPW Economic & Political Weekly644.1 Estimates from Pooled DataThe NTFPs constitute an array of products, collected for self- consumption as well as for their market value. We have informa-tion on three categories of households, and use category 1, those who collect only for consumption as the reference category. The results that follow are to be interpreted as relative to this compar-ison category. The MNL results for data pooled across the four states (Table 7, p 62) show that households that collect for sale are typically those which have access to forests through some institutional arrange-ments such as tree pattas. They are also located at a greater distance from metalled roads. Households collecting for consump-tion and sale have less access to both irrigation and forest institu-tions. They appear to be the less privileged. Further, introducing state level dummies in the estimation (Table 8, p 63) indicates that for households collecting for sale, the dummies are significant for the states of Madhya Pradesh and Karnataka. This means that we expect collection behaviour to be different for these two states. For household type 3, those collect-ing for sale and consumption, the dummies for Madhya Pradesh, Karnataka and Bihar are all significant. This significance of state dummies indicates that household behaviour indeed varies across states. An analysis of state level data is therefore called for in order to have a better understanding of the factors determining the nature of household dependence on NTFPs collected from forest CPRs.4.2 StatewiseEstimatesThe state-specific analysis uses the same set of explanatory variables across all four states. As explained above, the variables stand for current year income from agriculture (as approximated by area sown and access to irrigation and other mechanisation), asset status, existence or otherwise of property rights and physi-cal access to sources of supply (forests) and sources of demand (markets access through roads). (i) Bihar:MNL results indicate that in Bihar (Table 9, p 63), the households are likely to have a larger net sown area, and hence, obtain larger income from agriculture, be at a greater distance from forests and, have greater access to tree pattas (an institu-tional form of access to the product of trees) than those who collect only for self-consumption. The magnitude of the marginal effects is high for the variables net sown area and tree patta. This indicates that existence of secure property rights (through tree pattas) is more important than proximity to forests to enable a household to collect for sale only. Asset poverty is also a signifi-cant explanatory variable. It can be concluded that some asset poor households may also collect for sale purposes.ADVERTISEMENT FOR THE POST OF DIRECTOR, NABAKRUSHNA CHOUDHURY CENTRE FOR DEVELOPMENT STUDIES, BHUBANESWARTELEPHONE / FAX : 0674 – 2300471Applications are invited for the post of Director in the Nabakrushna Choudhury Centre for Development Studies, Bhubaneswar (an autonomous institute of higher research, supported by the State Government of Orissa and the Indian Council of Social Science Research).1. Qualifications and Age Limits The candidates should be scholars of national eminence in any sphere of social sciences, particularly in Economics, Sociology or Anthropology having (a) sufficient organizational and administrative experience, (b) capacity to conduct and guide interdisciplinary research of fundamental as well as applied nature and field studies and (c) publications of high standard. The age of the candidates should be within 55 years. However, the age may be relaxed in exceptional cases.2. Compensation Package(1) Pay:-(a) Rs.16,400 – 22,400/- per month with admissible dearness allowance, for a person still in service + Rs.500/-per month as Director’s allowance. (b) For retired persons, pay may be fixed according to the re-employment rules i.e., pay on re-employment + pension shall not exceed the last pay while in service. Allowances will be paid as admissible under the rules.(2) Accommodation:- The Director is entitled to rent-free accommodation or house rent allowance (Rs.5,000/- per month) in lieu thereof.(3) Other perquisites:- Miscellaneous perquisites as admissible to the Chief Executive of a prestigious Research Institute shall be available to the Director.3. Tenure The tenure of appointment is for three years, extendable in suitable cases. In case of a person still in service, retention of lien in the parent organization shall be allowed and matters relating to leave-and-pension contributions etc. (if any) shall be settled in consultation with the said parent organization.4. Last Date of Application Applications alongwith detailed curriculum vitae (specifying inter-alia, the organizational/administrative experience, the academic and research attainments and the particulars of publications) and superscribing the envelope ‘Application for the post of Director, NCDS’ should reach the Secretary of Nabakrushna Choudhury Centre for Development Studies, Bhubaneswar – 751013 (Orissa) by Speed Post within one month from the date of publication of this advertisement.5. Selection The Selection Committee (constituted in consultation with the Indian Council of Social Science Research) will shortlist the candidates for final selection. It reserves the right to reject the applications without assigning any reason thereof.SECRETARY

SPECIAL ARTICLEEconomic & Political Weekly EPW february 23, 200865the households are located closer to roads, farther away from forests and cultivate more land than those that collect for self-consumption only. However, since they have a lower level of irri-gation facilities, their incomes may not be higher than that of the other two categories. The access variable (in particular, market access) is significant. The households that collect for sale and consumption are also (significantly) located further away from forests than those that collect for consumption only. It can be deduced that they are located closer to markets. This corroborates the significance of market access in this state.5 ConcludingRemarksAn interesting picture emerges for the four selected states from the above results (Table 13). Income related variables are significant in all states. Asset and access related variables are also significant on average in three out of four states. Better property rights as a distinguishing charac-teristic of the households turn out to be significant only in one state. The households in Karnataka are income and asset rich and collect for sale mainly due to the better opportunities provided by access to markets and nearness to forests. It can be concluded that collection for sale is therefore an option that they choose in a situation of expanded options. The situa-tion is analogous to Fisher’s (2004) high return forest activities. In Bihar, the households are income richand have better defined property rights, though they may be asset poor and have a lower level of access to forests. Households in Maharashtra too are income rich and have better access to markets, though they are not close to sources of supply, ie, forests. Both these states throw up a mixed picture with both asset (and income) rich and poor households collecting for sale purposes, given appro-priate property rights structures and access to markets.Finally, households in Madhya Pradesh are income poor, have less access to markets and are closer in location to Households that collect for sale and consumption are located further away from forests and have more land. (ii) Karnataka: In Karnataka (Table 10, p 63), the households are less asset poor than those who collect for consumption only. They also have higher income from agriculture since theyhave more access to irrigation, though they have less sownarea. They are at a lesser distance from forests. Magni-tudes of marginal effects are also high for the asset poverty, irrigation and landownership variables. In other words, the results indicate that the households are not “poor”, rather theyhave better access to both assets and better lands because of irrigation. The model does not provide good indications of the distin-guishing features of households which collect for both sale and consumption. They do not seem to be substantially different from those households which collect NTFPs for consumption only. This points towards the emergence of a small group of compara-tively better-off households collectingNTFPs for sale. (iii) Madhya Pradesh: The results of the multinomial logit for Madhya Pradesh (Table 11) indicate that the households have less access to irrigation and are at shorter distances from forests than those that collect for consumption only. The households are also likely to be located away from metalled roads. The households have lesser net sown area than households collecting only for self-consumption and are asset poor as well, though these two variables are not significant. The results indicate that NTFP collection, even for sale, in this state continues to be a subsistence activity with households being in all proba-bility, forest villages located in the interior away from roads.The above conclusion receives support from the results with respect to the house-holds which collect both for sale and consumption. These households are similarto the households, have less irriga-tion and are at lesser distance from forests. There is not much difference between these two sets of households indicating that the households have not emerged as a separate group responding to commercial factors.(iv) Maharashtra: Results from Maha-rashtra (Table 12) indicate that the nature of NTFP collection and sale differs in this state. Fewer households collectNTFPs, either for sale or consumption. However, Table 11: NTFP Collection and Sale in Madhya PradeshHousehold Type 2: Sale Only MNL Marginal CoefficientsEffectsIrrigation -15.9*-0.236Net sown area -0.16 -0.011Asset poverty 0.12 -0.03Distance from forest -0.1** 0.009Distance from metalled road 0.26** 0.013Patta 36.37 3.10Constant term 17.94** Household Type 3: Sale and Consumption Irrigation -16.45**-2.48Net sown area -0.01 0.007Asset poverty 0.69 0.135Distance from forest -0.15** -0.025Distance from metalled road -0.11 0.008Patta -7.51 -3.61Constant term 17.61** Prob > chi2 0.000 No of observations 738 ** indicates significance at 1 per cent level. * indicates significance of the coefficient at 5 per cent level.Table 12: NTFP Collection and Sale: MaharashtraHousehold Type 2: Sale Only MNL Marginal CoefficientsEffects Irrigation –0.68 –0.579Net sown area 0.34* -0.098Asset poverty 0.01 0.304Distance from forest 0.26* -0.012Distance from metalled road -0.57* –0.019Patta -1.44 6.008Constant term 17.94* Household Type 3: Sale and Consumption Irrigation 2.1** 0.650Net sown area –0.93 –0.162Asset poverty -1.54 -0.381Distance from forest 0.39** 0.045Distance from metalled road –0.62 -0.041Patta -32.5 -7.722Constant term –0.98 Prob > chi2 0.000 No of observations 132 ** indicates significance at 1 per cent level. * indicates significance of the coefficient at 5 per cent level.Table 13: Summary Characteristics of Households Collecting for Sale Only Variable BiharKarnatakaMadhyaPradeshMaharashtraIncome Income rich (1) Income rich(2) Income poor Income rich(1)Assets Asset poor Asset rich NS NSAccess to markets NS More access Less access More accessAccess to supply source Lower access More access More access Less accessAppropriate property rights Better property rights NS NS NS1 Income rich (1) implies rich by one indicator while Incomerich (2) implies rich by two indicators of economic status. 2 NS implies that the variable is not a significant distinguishing characteristic for the state concerned.

SPECIAL ARTICLEfebruary 23, 2008 EPW Economic & Political Weekly66Notes1 See, for instance, the definition by Bromley (1989).2 The terms “common property” and “common pool” resources have often been used synony-mously in the empirical literature. A distinction is now made with the former being a subset of the latter, relevant only when well laid out rules for entry, use and exit from the group possessing the right exist. Estimations in effect target at common pool resources. In India it is estimated that such land resources are about 70 million hectares, concentrated mainly in the central plateau areas and the arid and semi-arid zones of the country. 3 Chopra and Gulati (2001) estimate that forest department owned common pool land resources are 25.069 million hectares of the total of 70 million hectares. A substantial part of the rest (under the ownership of local bodies and private ownership with periodic common access) are also forest ecosystems.ReferencesAlkire, S (2002): Valuing Freedoms: Sen’s Capability Approach and Poverty Reduction, Oxford University Press, New Delhi.Amacher, G S, W F Hyde and K R Kanel (1996): ‘House-hold Fuelwood Demand and Supply in Nepal’s Tarai and Mid-Hills: Choice between Cash Outlays and Labour Opportunity’,World Development 24 (11), 1725-36.Angelson, A and S Wunder (2003): ‘Exploring the Forest-poverty Link: Key Concepts, Issues and Research Implications’, CIFOR Occasional Paper No 40, Centre for International Forestry Research, Bogor, Indonesia.Beck, T, Ghosh and G Madan (2000): ‘Common Property Resources and the Poor: Findings from West Bengal’,Economic & Political Weekly, 35(3): 147-53.Bromley, D W (1989): Economic Interests and Institu-tions: The Conceptual Foundations of Public Policy, Basil Blackwell, Oxford and New York.Byron, N and M Arnold (1999): ‘What Futures for the People of the Tropical Forests?’World Develop-ment, Vol 27 (5), pp 789-805.Cavendish, W (2000): ‘Empirical Irregularities in the Poverty – Environment Relationship of Rural Households: Evidence from Zimbabwe’,World Dev 28 (2000), pp 1979-2003.Chopra, K and P Dasgupta (2002): ‘Common Pool Resources in India: Evidence, Significance and New Management Initiatives’, Final Report of DFID sponsored project on Policy Implications of Knowledge with respect to Common Pool Resources undertaken jointly with University of Cambridge, UK. Chopra, K and S C Gulati (2001): Migration, Common Property Resources and Environmental Degrada-tion: Interlinkages in India’s Arid and Semi-arid Regions, Sage Publications, New Delhi.Chopra, K, and G K Kadekodi (1991): ‘Participatory Institutions: The Context of Common and Private Property Resources’,Environmental and Resource Economics, 1, 353-72.Chopra, K, G K Kadekodi and M N Murty (1990): Participatory Development: People and Common Property Resources, Sage Publications, New Delhi.CWS (2001): ‘Common Pool Resources in Semi-arid India: Dynamics, Management and Livelihood Contributions’, Centre for World Solidarity, A P Regional Report.Dasgupta, P (2006): ‘Common Pool Resources as Development Drivers: A Case Study of NTFPs in Himachal Pradesh, India’, SANDEE Working Paper (refereed) No 15.Fisher, M (2004): ‘Household Welfare and Forest Dependence in Southern Malawi’,Environment and Development Economics, Vol 9, 135-54.Godoy, R (1992): ‘Some Organising Principles in the Valuation of Tropical Forests’,Forest Ecology and Management, 50:174–175. Godoy, R, H Overman, J Demmer, L Apaza, E Byron, THunca, E Perez, V Reyes-Garcia, V Vadez, A Cubas, K McSweeney and N Brokaw (2002): ‘Local Financial Benefits of Rain Forests: Compar-ative Evidence from Amerindian Societies in Bolivia and Honduras’,Ecological Economics, 40, 397-409.Godoy, R and K Bawa (1993): ‘The Economic Value and Sustainable Harvest of Plants and Animals from the Tropical Forest’,Economic Botany, 47 (3), 215-19.Gunatileke, H and U Chakravorty (2003): ‘Protecting Forests through Farming’, Environmental and Resource Economics, Vol 24 (1), 1-26.Hecht, S, A B Anderson and P May (1988): ‘The Subsidy from Nature: Shifting Cultivation, Successional Palm Forests and Rural Develop-ment’,Human Organisation, Vol 47, pp 25-35.Hotchkiss, D and S Kumar (1988): ‘Consequences of Deforestation for Women’s Time Allocation, Agricultural Production and Nutrition in Hill Areas of Nepal’, International Food Policy Research Institute, Washington DC.Iyengar, S and N Shukla (1999): ‘Regeneration and Management of Common Property Land Resources (CPLRs) in India: A Review’, Working Paper No 110, Gujarat Institute of Development Research, Gota, Ahmedabad. Jodha, N S (1997): ‘Management of Common Property Resources in Selected Dry Areas of India’ in J MKerr, D K Marothia, K Singh, C Ramasamy and W R Bentley (eds),Natural Resource Econom-ics: Theory and Application in India, Oxford and IBH Publishing Company, NewDelhi.Jodha, N S (1986): ‘Common Property Resources and Rural Poor in Dry Regions of India’, Economic & Political Weekly, 31 (27), pp 1169-81. Kadekodi, G K and A Perwaiz (1998): ‘Dimensions of Wastelands and Common Property Resources in India’, Working Paper Series No E/190/98, Insti-tute of Economic Growth, Delhi.NSSO (1999):Common Property Resources in India, NSS 54th Round (January 1998-June 1998), Government of India.Pasha, S A (1992): ‘CPRs and Rural Poor: A Micro Level Analysis’,Economic & Political Weekly, November 14, 2499-2503.Pattanayak, S K and E O Sills (2001): ‘Do Tropical For-ests Provide Natural Insurance: The Micro-economics of Non-timber Forest Product Collection in the Brazilian Amazon’,Land Eco-nomics, Vol 77, 595-612. Ravindranath, N H, P Sudha and K M Indu (2000): ‘Participatory Forestry; Indian Experience in Community Forestry and Joint Forest Management’, Centre for Ecological Sciences, Indian Institute of Science, Bangalore.Rao, K K (2000): ‘Vegetation and Non-timber Forest Products Assessment under JFM in Eastern Ghats of Andhra Pradesh, India’, Department of Environmental Sciences, Andhra University, Visakhapatnam, mimeo.Sen, A (1992): Inequality Re-examined, Harvard University Press, Cambridge, Massachussets. – (1999):Development as Freedom, Knopf Press, New York.Singh, K, N Singh and R P Singh (1996): ‘Utilisation and Development of Common Property Resources – A Field Study in Punjab’, Indian Journal of Agricul-tural Economics, Vol 51, Nos 1 and 2, 249-59.forests. These households represent the typical case of NTFP collection as a subsistence activity.To recapitulate, the results for Karnataka in particular, and, Bihar and Maharashtra in a less dramatic fashion, indicate that non-poor households collectNTFPs for sale as well, provided access and property rights conditions are favourable. This is significant and provides pointers towards the development of NTFP related economic activity as an income diversification route for relatively affluent rural households. Our study indicates that in certain pockets of the country,CPRs are providing the basis of income generation for households with multiple options, quite distinct from their role as providers of subsistence incomes. This is particularly true for collection of NTFPs. These results point towards the possibility of a new role for NTFP collection from CPRs in the context of market oriented development, a role that has significant implications for the paradigm of development with and through conservation.Institute for Studies in Industrial Development (ISID)ICSSR Institutional Doctoral Fellowships 2007-08ISID is a national-level policy research institution affiliated to the Indian Council of Social Science Research (ICSSR). Applications, in the prescribed format, are invited from candidates for award of Doctoral Fellowship at the ISID. The thrust areas of research are: Industrial Development; Corporate Sector; Trade, Investment and Technology; Employment and Labour. Applicants should have postgraduate degree in Economics (those having Masters degree in allied subjects like Statistics, Business Management, Commerce can also apply) with at least 55% marks and registered for doctoral programme of any Indian university. The university rules should permit one of the faculty members of the ISID Faculty as co-supervisors. The fellowship amount is Rs. 6,000/-per month for two years, extendable by one year in exceptional cases only, plus contingency grant of Rs. 12,000/-per annum. Application Form and other details canbe downloaded from Institute’s Website:http://isid.org.in. Completed applications should reach the Director, ISID, 4, Institutional Area, P.B. No. 7513, Vasant Kunj, New Delhi – 110 070 by March 18, 2008.