POLICY RESEARCH WORKING PAPER 2710 Breaking up the In the decollectivization of agriculture in Vietnam, local Collective Farm allocation of land use rights reduced overall inequality- Welfare Outcomes of Vietnam's thanks to initial conditions at Welfare Outcomes of Vietnam's the time of reform and Massive Land Privatization actions by the center to curtail the power of local elites. Martin Ravallion Dominique van de Walle The World Bank Development Research Group Poverty Team and Public Services for Human Development Team November 2001 | POLICY RESEARCH WORKING PAPER 2710 Summary findings The decollectivization of agriculture in Vietnam was a based privatization. The authors find that 95-99 percent crucial step in the country's transition to a market of maximum aggregate consumption (depending on the economy. But the assignment of land use rights had to be region) was realized by a land allocation that reduced decentralized, and local cadres ostensibly had the power overall inequality, with the poorest absolutely better off. to corrupt this process. They attribute this outcome to initial conditions at the Ravallion and van de Walle assess the realized land time of reform and actions by the center to curtail the allocation against explicit counterfactuals, including the power of local elites. simulated allocation implied by a competitive market- This paper-a product of the Poverty Team and the Public Services for Human Development Team, Development Research Group-is part of a larger effort in the group to undertake ex-post assessments of the efficiency and equity implications of policy reforms. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Catalina Cunanan, room MC3-542, telephone 202-473-2301, fax 202-522-1151, email address ccunanan@worldbank.org. Policy ResearchWorking Papers are alsopostedonthe Web at http://econ.worldbank.org. The authors may be contacted at mravallion@worldbank.org or dvandewalleCaworldbank.org. November 2001. (41 pages) The Policy Researcb Worleing Paper Seoes disseminates the findings of loric iy progress to encourage the excange of ideas about deiielopment issues. An objective of the series is to get the findings out qicickly, even if the presentations are less than fully polisbed. The papers carry the names of the authors and should be cited aceordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do, not necessarily represent the view, of the World Bank, its Executive Directors, or the couitrites tbey represent. Produced by the Policy Research Dissemination Center Breaking up the Collective Farm: Welfare Outcomes of Vietnam's Massive Land Privatization Martin Ravallion and Dominique van de Wallel World Bank, 1818 H Street NW, Washington DC, 20433, USA 12 November, 2001 Key words: Decentralization, land reform, privatization, equity-efficiency trade-off, Vietnam JEL codes: D60, P21, Q15 Correspondence: mravallion(@iworldbank.org. dvandewalledgworldbank.org. These are the views of the authors, and need not reflect those of the World Bank or any affiliated organization. Helpful comments and assistance were received from Dorothyjean Cratty, Eric Edmonds, Karla Hoff, Jean- Jacques Laffont, Michael Lipton, Dilip Mookherjee, Vijayendra Rao, Tomomi Tanaka, Carrie Turk, Chris Udry and participants at presentations at University of Michigan, Michigan State University, Yale University and the World Bank. The support of the World Bank's Research Committee is gratefully acknowledged. 1. Introduction Vietnam's land reform of 1988 abandoned the collective farming system that had been introduced in the 1960s. The 1988 Land Law and its key implementation directive-"Resolution 10"-gave individual households long-term use rights over the collectives' land and other resources.2 Four million hectares of land were thus scheduled for effective privatization. The economic significance of this new land law is obvious, given that (around 1990) three-quarters of the country's workforce depended directly on farming. Clearly, this land privatization was hugely important to living standards and their distribution in Vietnam. Implementation of Resolution 10 was decentralized to commune level; there was little choice, since the center could not control the local authorities, who were (naturally) much better informed about local conditions. So the center faced an accountability problem in this decentralized reform.3 Malarney (1997, p.900) describes the problem faced by the reformers: 5..given the institutional dominance of the Communist Party, local politicians with party backgrounds, which is to say all, are compelled by the party to be impartial and committed to official policies; yet, as politicians drawn from local kin and community, they are also pressured to nurture interpersonal relations, selectively avoid official dictates, and use their positions to bring advantages to kin and/or co-residents." This echoes concerns in recent literature and policy discussion about decentralized development programs (Bardhan and Mookherjee, 2000; Galasso and Ravallion, 2001). In developing country settings, the center often faces high costs of acquiring the information needed to control outcomes locally, and local agents may well have little sympathy with the center's aims. So there is a real risk that a decentralized program will be captured by a local elite that sees the intervention as an opportunity to enhance its own position. 2 What is widely known as the 1988 Land Law, was in actual fact enacted December 29 1987. We will follow the convention of calling it the 1988 Land Law. Resolution 10 followed in April 1988. 3 Local accountability problems were apparently also common in the 1960s' collectivization program (Fforde, 1989) 2 This paper assesses the assignment of land-use rights achieved by Vietnam's de- collectivization. We use two counter-factuals for assessing impact. One is an equal allocation of (quality-adjusted) land and the other is the allocation that would have maximized aggregate consumption, as would have been achieved by a competitive market-based privatization under ideal conditions. Comparing this with the actual allocation allows us to estimate the implicit value that was placed on efficiency versus distributional goals in the allocation of land. We also characterize the specific distributional outcomes of the realized land allocation; possibly efficiency was sacrificed, but the poor would have been better off if it had not been. The following section describes Vietnam's de-collectivization, and the factors leading up to it. Section 3 outlines our approach in theory, while section 4 describes our empirical specification. Section 5 discusses our survey data, collected 2-3 years after the reform was completed. Section 6 presents the regressions and section 7 discusses their welfare implications. Section 8 concludes. 2. Privatizing the collective's land With the aim of raising agricultural productivity, Vietnam's 1988 Land Law and Resolution 10 abandoned collective farming and granted households long-term use rights over land and the freedom to cultivate it as they wished.4 Land remained the property of the State, reverting back to the authorities when a household moved or stopped farming.5 The de- collectivization was virtually complete by 1990 (Ngo 1993). 4 Use rights for crop land were granted for 10 to 15 years; longer periods applied to tree crops. Some flexibility was allowed in that 10-15% of the cooperative's land could be kept aside for new households and demobilized soldiers, and available for hire by households in the meantime (Tran 1997). 5 Although Resolution 10 affirms the right to transfer land use and legate it to one's offspring, such rights were not fully guaranteed legally (Bloch and Oesterberg 1989). It did not recognize the right to exchange, lease, or mortgage land. These rights were only extended in the 1993 Land Law. Land policies have evolved since, but it is the impact of the 1988 Land Laws that is our main interest in this paper. 3 The new Land Law made recommendations on how the land was to be allocated across households. It recommended that allocations take account of the availability of land resources, on households' labor force and the land that households had been cultivating. It also placed certain limits on how much land could go to any one household.6 Resolution 10 acknowledges claims to the land farmed prior to collectivization. It also entreats the cooperatives to provide appropriate jobs and good arable land to the families of war heroes and martyrs, to those who significantly contributed to the revolution, to the injured and unable-bodied and to others facing considerable difficulties. But it then dilutes this request by adding that the well-being of these groups is really the responsibility of the local Peoples' Committees and that the Ministry of Labor, War Invalids and Social Affairs and the Ministry of Finance will devise policies on social assistance to them (Vietnam Communist Party 1988). While the new land law extended some guidelines, it left local cadres with considerable power over land allocation and the conditions of contracts. The center's directives were disseminated by Provincial Peoples' Committees, who in turn relied on the local authorities, apparently allowing them wide berth in adapting the guidelines to local conditions, priorities and customs. One can expect foot dragging on their part, and the pursuance of quite different objectives in implementing the central directives. Those who were making the decisions locally were essentially the same cadres who had positions of relative privilege as the managers of the cooperatives, and relatively high living standards under the collective mode of agricultural production (Selden, 1993). The reform threatened to undermine their power and privilege. 6 Article 27 of the 1988 Land Law stipulates that the land allocation to each household should not exceed ten percent of the total farm land area of each concemed village. It further decrees regional per capita land ceilings for those contracting land for long term use from state operated farms. It has been claimed that ceilings were officially set at two hectares in the fertile but densely populated Red River Delta and three hectares in the South (ANZDEC 2000), though we find no mention of this in Resolution 10 or the 1988 Land Law. 4 So there was a real risk here that the benefits of reform would be captured by self- interested local cadres, potentially undermining the center's aims. There is anecdotal evidence of abuse of local power, against the center's interests. Kolko (1997, p. 92) argues that: "From its inception, the land redistribution was marred by conflict, ambiguity and corruption. Cadres in many villages immediately began to distribute the best land to their families and relatives, and abuse was rife." There were numerous public disputes at the time, stemming from (amongst other things) conflicting historical claims over land, disputes over village and commune boundaries and complaints about corrupt party cadres (Nguyen 1992; Pingali and Xuan 1992; Kolko 1997). It has also been argued that those with the weakest prior claims on plots did poorly in the land allocation. For example, Ngo (1993) argues that war veterans, demobilized soldiers and their families were short-changed in the land allocations and were over-represented as protagonists in disputes. It is unimaginable that such an enormous land reform was free from corruption. However, the interpretation of the existing qualitative evidence on this issue is unclear. Cases of extreme abuse of power by local elites were visible when they boiled up in local protests-Vietnam's "hot spots" (Kolko 1997, cites many examples). However, the fact that local protests were possible can also be interpreted as evidence that there were constraints on local abuse of power. The possibility for bias in the historical-qualitative account is clear; the cases of abuse may well have been uncommon but far more visible. Objective village-level assessments were rare; in the only village study we know of to address this issue, Tanaka (2001) describes the elaborate efforts of the "land allocation committee" in a North Vietnamese village to equalize land allocation. Such efforts are unlikely to have attracted much publicity at the time. While one would not want to generalize from one village study, it is no less hazardous to infer from the available evidence that capture by local elites was the norm. 5 There clearly were constraints on the power of the cadres, in part due to actions by the center. Formally, Article 54 of the Land Law extends the threat of punishment for officials found to have abused their power in the allocation process. Enforcement is, of course, another matter. However, it is important to note that Vietnam's peasants had sacrificed heavily through long periods of war, in the hope of a better life after reunification. The outcomes for their living standards had been disappointing. Peasant resistance to the collective system was common in the 1980s, and has been identified as a factor motivating the center's de-collectivization reforms (Beresford 1985, 1993; Selden, 1993; Kerkvliet 1995). With the center's support, the Vietnam Peasant Union (VPU) was created in 1988 with the explicit aim of giving peasants a stronger voice in reform policies and-implicitly at least-promoting the center's reforms locally. As with past peasant unions, it seems that the VPU was eventually captured by local elites; Wurfel (1993, p.32) argues that by 1990 the VPU had been "tamed by local party cadre, who had interests to protect." But for a critical period the VPU appears to have acted as a counter-weight to the cadres (Wurfel, 1993). During the reform period, the center also gave greater freedom to the press; the press subsequently carried much criticism of the bureaucracy, again helping the reform process (Wurfel 1993). The reform movement was clearly driven by more than the center's concerns about the welfare of peasants. The same inefficiencies of the collective farming system constrained the resources available to the center for its industrialization plans, and created food shortages in urban areas (Beresford 1993; Kerkvliet 1995). Arguably then, the reforms were only possible through an implicit coalition between the peasants and reformers at the center-a coalition that clearly aimed to constrain the power of local cadres to capture the process. Recent history provided a reference point in deciding how the land should be allocated. 6 Collectivization came soon after the completion of land reform programs that had gone a long way toward redressing the high inequality of land ownership under French Colonial rule (Beresford 1985; Pingali and Xuan 1992). The pre-collectivization allocation may have influenced land allocation at de-collectivization. There are reports that some households simply went back to farming the land they had originally handed over to the cooperative or collective, or land they had some historical claim to.7 While there was no legal commitment to restore the pre- collectivization land allocation, that was an option for the local authorities. There were important differences between the North and the South at the time of the reforn. In the South's Mekong Delta, farmers had resisted collectivization, and by the time Resolution 10 was introduced less than 10 percent of all the region's farmers had been organized into collectives. In contrast, virtually all of the crop land in the North and in the South's Central Coastal provinces-where joining the collectives was seen as a means of rebuilding after the war-was collectivized by the time of the reform (Pingali and Xuan 1992; Ngo 1993). Southern Vietnamese farm households who participated did so for a much shorter period, while many never participated in the collectives, notably in the Mekong Delta. However, the land allocation in the South was still administratively determined and periodically re-allocated (Pingali and Xuan 1992); the difference with the more collectivized North is that in the South (especially the Mekong Delta) farmers continued to farm individually rather than collectively. Prior to reunification, agricultural land in the South had also undergone a series of land reforms.8 Resolution 10 allowed farmers in the South to recover land owned prior to 1975, though former 7 Smith and Binh (1994) quote a number of Son La households in the North as professing, in 1994, to be farning the same land they had at the time of the departure of the French. Tran (1997) claims that land was redistributed according to household original contributions to the cooperatives in some areas. 8 The South's land reform programs prior to reunification had initially consisted of lease price control and ownership ceilings, but were followed in 1970 by substantial land redistribution and titling under a "land-to-the tiller" program (Callison 1983; Pingali and Xuan 1992). 7 "landlords" were explicitly barred from doing so (Pingali and Xuan 1992). There are reports that in the Mekong Delta the implementation of Resolution 10 often entailed restoring the land allocation that prevailed prior to reunification (Hayami, 1993; ANZDEC, 2000). The collectives had also owned and controlled the farm capital stock (tools, machinery, draft animals) that also had to be allocated among farm households. It is sometimes claimed that this process more easily allowed cooperative officials to favor themselves, their families and friends than the more visible land allocation process. Because of data limitations we can only focus on land allocation, though we will look for signs of differing returns to land associated with political connections. 3. Theoretical model of the actual and counter-factual land allocations Motivated by the above observations we shall test whether the local implementation of decollectivization served distributional goals-possibly reflecting capture by local elites-at some loss to aggregate consumption. We construct a model that allows us to estimate that loss, and to compare the observed allocation against explicit counter-factuals. One of those will be an equal allocation of land per capita. This is easily calculated. The other is the allocation that maximizes aggregate consumption of the commune; this requires a model of consumption. The actual decision-making process might be anything from administrative fiat (according to the cadre's personal preferences) to a complex bargaining game. We only assume that the outcome (however it is reached) is Pareto optimal, in that no commune member's utility can be increased without someone else being worse off. To characterize all possible solutions, we represent the problem as maximizing a weighted sum of welfare levels across all farm- households. The Pareto weight attached to the utility of household i is wi = w(Xi) where Xis a vector of exogenous household characteristics. Naturally, different weighting functions imply 8 different distributions of land and utility. If the weights tend to be negatively (positively) correlated with welfare (to be defined) then one can say that the outcome will tend to be "pro- poor" ("pro-rich"). The utility of the i'th farm-household is assumed to depend solely on its consumption of a composite commodity. The household receives Li of land, which yields an output of F(Li, Xi ). (For now we treat land as homogeneous; in the empirical work we allow for observable heterogeneity, and we consider the consequences of latent heterogeneity in the next section.) The household also has (positive or negative) non-farm income, Y(X,) .9 (At the time of the reform, and since, agricultural labor markets were virtually non-existent in Vietnam, so to simplify the exposition we close off this market in our model.) The household's consumption is then: Ci = C(Li, Xi) = F(Li, XI) + Y(X,) (1) We assume that the function F is increasing and strictly concave in L. Utility is in turn an increasing concave function of consumption, Ui = U(C,). The commune selects an allocation of the total available land nL across n households, with mean L. The observed land allocation is: (L1,..,LL,) = arg max[I w(Xi)U[F(Li, Xi) + Y(Xi)] ELi = nL] (2) i=l i=1 which solves: w(X,)U'(Ci)FL (L, Xi) = ,u for i = 1,.., n (3) where FL (Li, Xi) is the marginal product of land and ,u is the shadow price of land in the 9 The functions F and Y may vary between households in different communes. We will deal with this possibility in the empirical work by including a complete set of commune effects in the regressions. 9 commune (the Lagrange multiplier on the aggregate land constraint in equation 2). It is readily verified that the land allocation is strictly increasing (decreasing) in X. (thej'th element of the vector Xi) if the sum of the elasticities of w(X;) and U'(Ci)FL (Li,Xi) to X, is positive (negative). Compare this to the allocation that maximizes the commune's aggregate consumption: (4L*,..,) = arg max[ C(Li, Xi) n Li = nL] (4) i=l i=l We call this the "consumption-efficient allocation." This equates CL (La, Xi) = FL (L', Xi) with the multiplier 2 on aggregate land in (4), giving L' =L(Xi, A) for i=l, ..,n (5) Mean consumption is then: n G = X C(L , Xi)/n (6) i=1 The consumption loss from the actual allocation is then C - C where C is the actual mean. The consumption-efficient allocation is also the competitive equilibrium given (X, ,.., XJ). In a market-based land allocation, each household's consumption will be F(Li, Xi) + Y(Xi) - AL; where A is the market price of land. Demands equate FL (Li, Xi) =A over all i, which is the allocation that maximizes aggregate consumption under the non-market allocation. Naturally the market solution will also vary with the joint distribution of the Xs. There are some caveats to this interpretation of the consumption efficient allocation. A competitive market is unlikely to have been a feasible option at the time in Vietnam. For one thing, agricultural land markets were virtually non-existent. And other markets (notably for credit) and institutions (for property rights enforcement) were probably not functioning well 10 enough to assure an efficient market-based privatization of land. However, against these observations, it should also be noted that under communism very little mobility had been allowed between communes. People may well have been sufficiently well inforrned within each village to know if one family attached an appreciably higher value to extra land than another, even though a market did not exist. It should also be noted that if holding land gives utility independently of consumption then the competitive market allocation of land will differ from the consumption-maximizing one. For example, if land provides insurance against risk then it will have value independently of current consumption. Then our interpretation of the consumption-maximizing allocation as the market solution would also require that risk markets worked perfectly.'0 Since we have no basis for assigning a value to land independently of the current consumption it generates we cannot calculate a "conditional" market solution (conditional on other market failures). Nonetheless, the consumption-maximizing allocation remains a natural benchmark for assessing the realized allocation. It will be of interest to see how close the non-market allocation is to this benchmark. 4. Empirical model In our empirical implementation of the above model we make the following assumptions: Assumption 1: Utility is given by log consumption: U(Ci ) = In C(Li, Xi) (7) Assumption 2: Log consumption is given by: lnCi = a +f,lnLi + Xir + si (8) where 0 < 8 < 1 and ci is a zero-mean i.i.d. error term uncorrelated with ln Li and Xi . 10 To give another example, Burgess (2000) argues that food market failures exist in rural China such that holding more land reduces the shadow price of food. 11 Assumption 3: The welfare weights take the form: In wi =Xb + vi (9) where vi is a zero-mean error term uncorrelated with Xi. Assumptions 1-3 imply that the land allocation satisfying equation (4) can be written in explicit form as the regression model:" ln Li = ln(,8 /,u) + Xib + vi (10) This identifies directly the parameters of the implicit welfare weights of the local land-allocation authority. Substituting (10) into (8) generates the reduced form equation for consumption: ln Ci = a + , ln(i / pt) + Xi (,ib + Y) + 8i +/3v1 (11) The consumption-maximizing allocation by contrast is given by:12 InL 1 =a + ,lS + i (12) Comparing (10) and (12), it can be seen that if y /(1 - ,B) = b then the actual allocation responds to changes in Xthe same way as the consumption-efficient allocation. So if the two allocations are essentially the same then we should be able to accept the restriction that b = fib + y when imposed on the reduced form equations, (1 0) and (11). If we cannot accept this restriction then it is of interest to calculate the consumption-efficient land allocation, (L, , ,..., Ln ), from which we can then measure the distribution of consumption losses implied by the actual allocation, using the fact that the proportionate consumption loss for household i is (L* / Lj ) - 1. To verify the following equation, take logs through (3) and note that note that (7) and (8) imply that U'(.)FL = 8 / Li. Equation (10) then follows using (9). 12 Given (8), the consumption-efficient allocation to household i solves ln L, = ln(,B / A) + In C where lnC, = a + Iln Li + Xiy +. 12 While allocated land is endogenous in this model, it is taken to be exogenous to consumption (i.e., Cov( v, s )=O). This is a standard assumption in past empirical work for Vietnam and in other settings in which land allocation is done administratively rather than through markets.'3 The assumption can also be defended on the grounds that the land allocation preceded the survey-based consumption measure by 3-4 years. Our estimates of the parameters of equation (8) will be biased if there are omitted variables that jointly influence the welfare weights and consumption levels. The most serious concern in this respect is heterogeneity in land quality. Higher land quality will probably result in higher consumption at given land quantity. Assuming that the quality differences are public knowledge within the commune, the administrative land allocation will take them into account, with more land being used to compensate for lower quality. We will include available controls for differences in the average quality of land holdings. However, latent heterogeneity will create a negative correlation between the error terms in the estimated consumption equation and the land allocation equation (Cov( v, e ) 6 yrs old). 0.459 0.25 0.422 0.28 0.454 0.28 0.469 0.25 0.485 0.24 0.452 0.26 Labor age adult member is handicapped 0.008 0.09 0.007 0.09 0.006 0.08 0.018 0.13 0 0 0.007 0.08 SOE: h'hold member has primary or secondary occupation in State owned 0.006 0.08 0.032 0.19 0.012 0.11 0.007 0.08 0.011 0.11 0.019 0.14 enterprise and had it 5 years ago Gov't job: member has worked for gov't in primary/secondary occupation for 5+ yrs, or 0.068 0.25 0.040 0.21 0.069 0.28 0.047 0.23 0.084 0.30 0.058 0.25 did so 5 yrs ago or retired from gov't Social subsidy: h'hold is recipient of gov't tranfers to war heroes, martyrs, disabled etc 0.103 0.30 0.118 0.32 0.134 0.34 0.091 0.29 0.050 0.22 0.101 0.30 Household head's years of education 6.252 3.71 7.226 3.70 7.051 3.80 4.562 3.79 4.312 3.13 6.162 3.83 Other h'hold adults' years of education 9.808 9.25 10.681 8.56 11.174 9.54 10.203 9.93 9.765 9.55 10.441 9.24 Log allocated irrigated land equivalent (m2) 7.197 0.73 7.447 0.62 7.400 0.79 7.603 0.73 8.416 1.29 7.587 0.93 Allocated irrigated land equivalent (m2) 1679.569 1117.37 2007.701 997.03 2084.141 1312.36 2621.580 2403.59 7296.937 6514.12 3003.256 3646.40 H'hold's private irrigated land (in2) 159.616 238.56 157.051 167.05 86.213 157.35 136.424 545.33 279.165 1505.35 155.887 648.13 H'hold's private non-irrigated land (M2) 242.92 401.20 113.382 521.38 250.951 389.62 310.033 598.75 209.016 1561.83 218.544 921.38 H'hold's private perennial land (m2) 278.719 507.38 120.698 353.67 90.713 204.60 188.533 463.52 903.740 1672.80 343.747 1453.46 H'hold's private water surface land (m2) 58.320 163.23 60.732 176.88 30.012 116.36 0 0 116.259 1102.29 55.738 459.87 H'hold cultivates swidden land=1 0.289 0.45 0.037 0.19 0.043 0.20 0.225 0.42 0.020 0.14 0.104 0.31 Share of good irrigated land 0.281 0.391 0.563 0.371 0.439 0.344 0.319 0.405 0.246 0.413 0.428 0.400 Share of good non-irrigated land 0.376 0.422 0.729 0.404 0.815 0.327 0.369 0.414 0.942 0.213 0.663 0.429 H'holds in regression sample n-484 956 506 276 443 2810 Source: 1992/93 Viet Nam Living Standards Survey. Note: * We identify goverrunent work through professional codes 20 and 21. Table 2: Determinants of consumption Northem Red River North Coast Central Coast Mekong Delta Full sample Uplands religion -0.086 -0.007 -0.041 0.124 -0.059 -0.022 (2.07) (0.14) (0.54) (1.18) (1.00) (0.82) ethnic -0.062 -0.193 -0.117 -0.649 0.141 -0.070 (0.86) (2.34) (1.23) (2.90) (1.90) (1.65) local bom -0.077 0.027 0.101 -0.138 -0.062 -0.035 (1.57) (0.78) (1.53) (3.65) (0.86) (1.29) age of head -0.0002 0.016 -0.003 0.002 0.005 0.007 (0.02) (2.32) (0.32) (0.21) (0.47) (1.83) age2 of head x IO 0.038 -0.158 0.025 0.012 -0.046 -0.059 (0.42) (2.19) (0.33) (0-15) (0.45) (1.46) Log household size 0.451 0.462 0.534 0.532 0.452 0.482 (6.90) (7.62) (10.24) (6.24) (6.92) (15.73) dependency ratio -0.066 -0.026 -0.120 -0.186 -0.110 -0.071 (0.65) (0.41) (1.71) (1.73) (1.19) (2.00) gender of head 0.074 0.030 0.014 0.025 -0.078 0.008 (1.65) (0.75) (0.37) (0.61) (1.34) (0.34) disabled adult -0.348 0.003 -0.432 -0.067 -- -0.162 (3.81) (0.01) (1.37) (0.61) (1.68) govemment j ob 0.103 0.149 0.103 0.296 0.181 0.140 (2.13) (3.10) (1.70) (4.15) (3.72) (4.83) SOE job 0.540 0.109 -0.044 0.498 0.046 0.130 (4.16) (2.26) (0.58) (1.45) (0.40) (2.74) education of head 0.021 0.027 0.024 0.033 0.009 0.025 (3.87) (5.45) (4.48) (4.93) (1.46) (9.48) educationofother 0.010 0.011 0.013 0.005 0.010 0.011 adults (4.72) (7.74) (4.89) (1.89) (4.21) (11.32) social subsidy recipient 0.007 0.044 0.041 -0.034 -0.025 0.031 (0.17) (1.10) (0.56) (0.52) (0.30) (1.15) Log allocated irrigated 0.097 0.084 0.052 0.214 0.188 0.131 land equivalent (2.82) (2.30) (2.39) (3.81) (6.89) (7.45) private irrigated x 103 0.137 0.239 0.236 0.049 0.017 0.028 (3.34) (2.32) (3.01) (1.04) (1.56) (2.54) private non-irrigated 0.015 -0.002 0.089 0.047 0.022 0.012 x IO, (0.31) (0.05) (2.50) (0.77) (1.24) (0.98) private perennial x 103 0.064 0.109 0.038 0.033 0.042 0.019 (3.47) (1.73) (0.40) (0.51) (3.59) (1.76) private waterx103 0.189 0.175 0.313 -- 0.016 0.040 (2.15) (3.40) (4.16) (0.72) (1.50) cultivates swidden land 0.070 -0.082 -0.092 -0.018 0.112 -0.009 (1.15) (0.86) (0.70) (0.26) (3.83) (0.24) share good irrigated 0.017 0.032 0.084 -0.055 0.111 0.042 land (0.25) (0.57) (1.21) (0.63) (1.55) (1.47) share good non- -0.004 0.004 -0.008 -0.037 0.016 0.020 irrigated land (0.06) (0.10) (0.27) (0.54) (0.20) (0.81) Constant 13.320 13.415 13.377 12.712 13.300 13.474 (41.53) (49.55) (50.75) (28.17) (37.69) (68.80) R2 0.679 0.671 0.703 0.666 0.570 0.673 RMSE 0.305 0.318 0.300 0.383 0.367 0.340 F stat 53.10 971.45 456.46 71.89 438.67 92.43 Prob>F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 n 484 956 506 276 443 2810 Note: The dependent variable is log household consumption expenditures. Commune fixed effects included. T-ratios in parentheses are based on standard errors corrected for heteroskedasticity and clustering. 34 Table 3: Actual land allocations compared to consumption-efficient allocations Northern Uplands Red River North Coast Central Coast Mekong Delta Full Sample actual efficient actual efficient actual efficient actual efficient actual efficient actual efficient religion -0.123 -0.095 0.047 -0.007 0.130 -0.043 0.035 0.157 0.162 -0.073 0.078 -0.025 (1.48) (2.03) (0.66) (0.14) (0.93) (0.54) (0.47) (1.16) (1.86) (1.00) (1.24) (0.81) ethnic 0.023 -0.068 0.309 -0.210 0.462 -0.124 -0.116 -0.826 0.364 0.174 0.013 -0.080 (0.39) (0.85) (2.56) (2.29) (2.78) (1.22) (0.47) (2.73) (1.51) (1.91) (0.11) (1.65) local born 0.032 -0.086 -0.029 0.030 -0.086 0.107 0.024 -0.176 0.146 -0.077 0.045 -0.040 (0.72) (1.59) (0.54) (0.78) (0.51) (1.51) (0.38) (3.69) (1.73) (0.87) (1.07) (1.29) age of head -0.008 -0.0002 -0.0003 0.017 -0.012 -0.003 0.045 0.002 0.028 0.006 0.003 0.008 (0.49) (0.00) (0.81) (2.29) (0.88) (0.32) (2.23) (0.22) (1.75) (0.47) (0.46) (1.82) age2 of headx 103 0.081 0.042 -0.065 -0.173 0.041 0.026 -0.486 0.016 -0.213 -0.057 -0.056 -0.068 (0.44) (0.42) (0.81) (2.17) (0.27) (0.33) (2.32) (0.14) (1.39) (0.45) (0.75) (1.45) log h'hold size 0.724 0.499 0.794 0.504 0.696 0.563 0.661 0.676 0.243 0.557 0.695 0.555 (6.63) (8.01) (14.38) (9.41) (5.18) (10.79) (4.39) (7.50) (2.21) (8.08) (11.93) (18.70) dependency ratio -0.498 -0.073 -0.478 -0.029 -0.386 -0.127 -0.292 -0.237 0.092 -0.135 -0.420 -0.082 (2.59) (0.66) (6.91) (0.41) (2.52) (1.71) (1.34) (1.68) (0.50) (1.21) (6.07) (2.02) gender ofhead 0.070 0.082 0.070 0.032 0.147 0.015 0.103 0.032 0.155 -0.096 0.094 0.009 (0.77) (1.69) (1.90) (0.75) (2.43) (0.37) (1.19) (1.68) (1.21) (1.34) (2.82) (0.35) disabled adult -0.125 -0.385 -0.086 0.003 -0.094 -0.456 0.118 -0.085 -0.053 -0.186 (1.19) (3.81) (0.70) (0.00) (0.45) (1.36) (0.57) (0.61) (0.64) (1.68) gov'tjob -0.221 0.114 -0.122 0.162 -0.200 0.109 -0.049 0.377 0.095 0.223 -0.160 0.161 (1.28) (2.17) (1.90) (2.93) (1.63) (1.71) (0.29) (3.48) (0.92) (3.74) (2.75) (4.77) SOE -0.767 0.598 -0.232 0.119 0.134 -0.046 -0.049 0.634 0.342 0.056 -0.174 0.150 (2.26) (3.90) (4.09) (2.22) (0.43) (0.57) (0.13) (1.35) (0.88) (0.40) (2.32) (2.69) education of head -0.012 0.024 -0.006 0.029 -0.009 0.026 -0.018 0.042 0.018 0.011 -0.001 0.028 (1.06) (3.81) (1.10) (5.26) (1.10) (4.38) (2.53) (4.35) (1.46) (1.45) (0.30) (9.12) education of other -0.005 0.011 0.002 0.012 0.005 0.014 0.004 0.007 0.010 0.012 0.003 0.013 adults (1.31) (4.81) (0.74) (6.95) (0.86) (5.04) (0.84) (1.87) (1.45) (4.09) (1.29) (11.04) social subsidy 0.005 0.008 -0.079 0.048 0.035 0.044 -0.192 -0.044 -0.371 -0.030 -0.088 0.036 recipient (0.07) (0.17) (1.61) (1.09) (0.37) (0.57) (1.58) (0.52) (3.50) (0.30) (2.26) (1.15) private irrigated 0.471 0.152 0.399 0.261 0.084 0.249 0.144 0.063 0.028 0.021 0.151 0.033 X 103 (2.79) (3.41) (3.28) (2.28) (0.71) (2.95) (5.79) (1.07) (2.05) (1.55) (3.14) (2.54) private non irrigated -0.033 0.017 -0.013 -0.003 0.174 0.093 -0.086 0.059 -0.004 0.027 -0.012 0.014 x lo0 (0.66) (0.30) (0.21) (0.00) (1.62) (2.49) (1.85) (0.78) (0.16) (1.24) (0.69) (0.97) privateperennial 0.015 0.071 0.028 0.119 0.054 0.040 -0.084 0.042 0.022 0.052 0.005 0.022 X 103 (0.27) (3.59) (0.62) (1.68) (0.38) (0.40) (2.32) (0.50) (0.62) (3.63) (0.57) (1.76) private water x 103 -0.017 0.209 0.041 0.192 0.346 0.330 -- -- 0.058 0.020 0.063 0.046 (0.11) (2.17) (0 .7) () (2. (2.62) (4.42) (6.50) (0.71) (5.00) (1.50) 35 cultivates swidden 0.121 0.077 0.230 -0.089 0.050 -0.097 0.046 -0.023 0.465 0.138 0.078 -0.010 land (1.10) (1.26) (2.32) (0.87) (0.40) (0.69) (0.40) (0.26) (7.90) (4.08) (0.93) (0.24) share of good 0.302 -0.005 -0.032 0.035 -0.100 0.089 0.050 -0.070 0.051 0.136 0.013 0.048 irrigated land (1.20) (0.00) (0.74) (0.57) (0.78) (1.20) (0.52) (0.62) (0.48) (1.57) (0.18) (1.48) share of good non- -0.434 0.019 -0.201 0.004 0.029 -0.008 0.221 -0.047 -0.019 0.020 -0.035 0.023 irrigated land (2.89) (0.24) (3.46) (0.10) (0.44) (0.26) (3.30) (0.53) (0.38) (0.04) (0.89) (0.81) Constant 5.729 6.881 -- 4.778 -- 6.614 -- 7.003 -- 5.876 -- (18.39) (38.40) (12.97) (15.73) (17.44) (13.74) R2 0.543 0.630 0.627 0.610 0.771 0.675 RMSE 0.512 0.389 0.503 0.482 0.648 0.545 F stat (14, 15)= (20,31)- (16,17)= (10,11)= (18,22)= (21,109) 135.92 2020.27 2120.20 230.57 1066.59 =874.10 Prob>F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 obs 484 484 956 956 506 506 276 276 443 443 2810 2810 Note: Cormmune fixed effects included. T-ratios in parentheses are based on standard errors corrected for heteroskedasticity and clustering. 36 Table 4: Mean consumption, inequality and poverty under alternative land allocations Northern Red River North Coast Central Coast Mekong Delta Full Sample Uplands Actual allocation Mean consumption ('000 dongs)/ h'hold 4725.083 4594.556 4183.381 5725.078 7300.921 5258.276 Inequality in per capita expenditures 0.101 0.085 0.079 0.124 0.130 0.115 Headcount index of poverty (%) 81.322 67.523 85.143 61.975 49.919 68.455 Squared poverty gap index (xO00) 13.014 7.386 13.464 9.719 5.639 9.271 Consumption-efficient counterfactual Maximum consumption 4821.796 4656.408 4227.616 6000.305 7688.655 5448.437 ('000 dongs)/ h'hold (%) loss (1-actual/efficient) 2.006 1.328 1.046 4.587 5.043 3.490 Inequality of consumption under the 0.120 0.101 0.087 0.185 0.176 0.150 efficient land allocation Headcount index of poverty under the 78.393 66.691 83.959 59.664 50.526 66.331 efficient land allocation (%) Squared poverty gap index under the 13.564 8.083 13.712 11.976 6.724 10.330 efficient land allocation (xlO0) Equal land counterfactual Mean consumption at equal land per 4773.223 4620.384 4205.749 5829.239 7546.890 5345.507 household (%) loss 1.009 0.559 0.532 1.787 3.259 1.632 Inequality of consumption at equal land 0.101 0.087 0.080 0.122 0.117 0.116 allocation Headcount index of poverty at equal 79.620 66.985 84.653 61.134 46.440 66.505 land allocation (%) Squared poverty gap index at equal 12.700 7.411 13.331 9.167 4.548 8.928 land allocation (xl 00) Note: Inequality is given by the difference between log mean consumption per capita and the mean of log consumption per capita. 37 Table 5: Mean consumption, inequality and poverty with mobility between communes Northern Red River North Coast Central Coast Mekong Delta Full sample Uplands Consumption-efficient counter factual Maximum consumption ('000 dongs)/ h'hold 4836.772 4674.562 4245.877 6111.004 8386.244 5580.237 (%) loss 2.309 1.712 1.472 6.315 12.942 5.770 Inequality under the efficient allocation 0.117 0.101 0.088 0.186 0.223 0.146 Headcount index of poverty 78.195 66.422 82.898 59.174 49.555 64.562 under the efficient allocation (%) Squaredpovertygapindexunder 13.271 8.029 13.605 11.551 6.526 9.457 the efficient allocation (xlOO) Equal land counter factual Mean consumption at equal land per household 4792.570 4639.757 4226.545 5938.658 8105.723 5488.358 (%) loss 1.408 0.974 1.021 3.596 9.929 4.192 Inequality at equal land 0. 098 0.087 0.080 0.122 0.149 0.113 Headcount index of poverty at 79.185 67.107 83.755 60.294 46.804 65.004 equal land allocation (%) Squared poverty gap index at equal land allocation (xlOO) 12.419 7.345 13.218 8.724 4.282 8.076 38 Appendix: Reduced form regressions for consumption Northern Red River North Coast Central Coast Mekong Delta Full Samrle Uplands religion -0.098 -0.003 -0.034 0.131 -0.02 -0.005 (1.96) (0.06) (0.56) (1.60) (0.62) (0.25) ethnic -0.059 -0.167 -0.093 -0.674 0.210 -0.086 (1.14) (1.84) (0.71) (2.40) (2.15) (2.39) local born -0.074 0.025 0.097 -0.133 -0.035 -0.024 (1.55) (0.49) (1.47) (1.64) (0.62) (1.02) age of head -0.001 0.016 -0.003 0.012 0.010 0.008 (0.11) (2.56) (0.40) (1.03) (1.02) (2.15) age2 of head x 0.046 -0.163 0.027 -0.092 -0.086 -0.065 103 (0.45) (2.46) (0.34) (0.82) (0.88) (1.77) log hhold 0.521 0.529 0.570 0.673 0.498 0.573 size (8.58) (12.56) (10.92) (7.93) (8.02) (23.64) dependency -0.114 -0.067 -0.140 -0.249 -0.092 -0.128 ratio (1.31) (1.22) (1.94) (1.89) (0.91) (3.61 h gender of 0.081 0.036 0.022 0.047 -0.049 0.018 head (1.86) (1.19) (0.54) (0.67) (0.90) (0.94 F disabled adult -0.360 -0.005 -0.437 -0.042 -- -0.17 (2.20) (0.04) (2.43) (0.22) (2.08) gov't job 0.081 0.138 0.093 0.286 0.199 0.118 (1.34) (2.66) (1.83) (2.58) (3.03) (4.25j SOEjob 0.466 0.090 -0.037 0.488 0.110 0.107 (2.43) (1.60) (0.28) (1.66) (0.61) (2.23) education of 0.020 0.026 0.024 0.029 0.013 0.025 head (3.77) (6.50) (5.04) (3.28) (1.69) (10.25) education of 0.010 0.011 0.013 0.006 0.012 0.011 other adults (4.32) (6.35) (6.21) (1.64) (4.16) (10.87) social subsidy 0.008 0.037 0.043 -0.075 -0.095 0.018 recipient (0.15) (0.98) (0.90) (0.84) (1.00) (0.71) private 0.183 0.272 0.241 0.080 0.022 0.067 irrigated x 103 (2.46) (3.00) (2.22) (1.48) (1.33) (5.83) private non 0.012 -0.004 0.098 0.028 0.021 0.011 irrigated x Io, (0.27) (0.13) (2.02) (0.56) (1.32) (1.25) private 0.066 0.112 0.041 0.015 0.046 0.020 perennial x 103 (1.75) (2.59) (0.53) (0.24) (3.77) (3.66) private water x 0.187 0.179 0.331 -- 0.027 0.048 *103 (2.01) (2.79) (2.54) (1.53) (3.26) cultivates 0.082 -0.063 -0.089 -0.008 0.199 -0.010 swidden land (1.77) (0.88) (0.83) (0.09) (1.23) (0.31) Share of good 0.025 0.015 0.079 -0.044 0.120 0.054 irrigated land (0.38) (0.32) (1.47) (0.59) (1.38) (2.02) Share of good -0.025 0.001 -0.006 0.011 0.012 0.016 non-irrigated (0.47) (0.04) (0.17) (0.14) (0.20) (0.78) land constant 13.874 -- -- 14.128 14.619 13.879 (71.90) (46.60) (47.99) (137.88) R 2 0.670 0.668 0.700 0.641 0.522 0.657 RMSE 0.309 0.320 0.301 0.397 0.387 0.349 F stat 25.220 40785.07 32665.58 14.003 10.360 39.568 Prob>F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 N 484 956 506 276 443 2810 test of F(36,894) F(53,1804) F(39,932) F(31,486) F(42,796) F(129,5340) y/( - f)= b =8.68 = 179.15 = 151.65 =6.66 =31.45 =28.37 Prob:>F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Note: The dependent variable is log household consumption expenditure. Commune fixed effects were also included. 39 Figure 1: Distribution of consumption losses relative to the efficient allocation (ii) National % loss from actual allocation 50 O0 0 09- 0 0 00 0 0 0 °00's,%j>D 6 0 .1 o 0o 00 00 0 00 O 40 0 I 0 6q 00 0 00 0600 0 0 0 0 -40 12 14 16 log consumption per capita (ii) Northern Uplands % loss from actual allocation 50 0 0 0 0~~~~~~~~~~~~~~~~~~~~~ -40~~~~~~~~~~~~~~~~~~ 0 ~ ~~~ ~ ~ ~ ~~~ ~~ 000 0e 00 0 O 0 9, ~~0 e o Cb 0 50 -b -40- 12 14 16 log consumption per capita 40 (iii) Red River % loss from actual allocation 50 -40~~~~~~~~~~~~~~~~~~ * ,_ i;°0t 0 0 0 12 14 16 log consumption per capita (iv) North Coast % loss from actual allocation 50 °~~~~~~~~~~~ _ 0~~~~~~~~~~~~~~~~~~~~ 0~~~~~~~~~~~~~~~~~~~~~ -40 - 12 14 16 log consumption per capita 41 (v) Central Coast % loss from actual allocation 50 - 50~~~~~~~~~~~~~ 0 0 0 0 0 0 00 00 ~ ~ 0 O-~~~~~~A o0 00 0 00O e 0D 00 e o ° %%&o 0o -0 O~~~~~~~~~~~~~~~ 12 14 16 log consumption per capita (vi) Mekong Delta 0 loss from actual allocation 50~~~~~~~~~~~~~ 0 O o 000O o 0o o Oo00 00 000 0 o0 a0 00 0 0 o lo consupton pe capita o0 o ° o0 00 o oO ~ 0 50 - 0 0 00 o 0 00 0~~~0 C0 o o 0 O 0 0 0 q0 0 09 .0 0 oo 000 0. .0 0 0 0 -0 0 0 12 14 16 log consumption per capita 42 Policy Research Working Paper Series Contact Title Author Date for paper WPS2693 Helping People Help Themselves: David Ellerman October 2001 B. Mekuria Toward a Theory of Autonomy- 82756 Compatible Help WPS2694 Financial Development and Financing Inessa Love October 2001 K_ Labrie Constraints: International Evidence 31001 from the Structural Investment Model WPS2695 Trade, Credit, Financial Intermediary Raymond Fisman October 2001 K. Labrie Development, and Industry Growth Inessa Love 31001 WPS2696 Firms as Financial Intermediaries: Asli Demirguq-Kunt October 2001 K. Labrie Evidence from Trade Credit Data Vojislav Maksimovic 31001 WPS2697 Regional Integration and Industrial Dorsati H. Madani October 2001 L. Tabada Growth among Developing Countries: 36896 The Case of Three ASEAN Members WPS2698 Foreign Bank Entry: Experience, George Clarke October 2001 P. Sintim-Aboagye Implications for Developing Countries, Robert Cull 38526 and Agenda for Further Research Maria Soledad Martinez Peria Susana M. Sanchez WPS2699 Benefits and Costs of International Pierre-Richard Agenor October 2001 M. Gosiengfiao Financial Integration: Theory and Facts 33363 WPS2700 Business Cycles, Economic Crises, Pierre-Richard Agenor October 2001 M. Gosiengfiao and the Poor: Testing for Asymmetric 33363 Effects WPS2701 Trade and Production, 1976-99 Alessandro Nicita November 2001 L. Tabada Marcelo Olarreaga 36896 WPS2702 Productivity versus Endowments: Hiau Looi Kee November 2001 L. Tabada A Study of Singapore's Sectoral 36896 Growth, 1974-92 WPS2703 Integrating Independent Power Fiona Woolf November 2001 Energy Help Desk Producers into Emerging Wholesale Jonathan Halpern 30652 Power Markets WPS2704 Regulatory Governance and Chile's Ronald Fischer November 2001 G. Chenet-Smith 1998-99 Electricity Shortage Alexander Galetovic 36370 WPS2705 Concession Contract Renegotiations: Antonio Estache November 2001 G. Chenet-Smith Some Efficiency versus Equity Lucia Quesada 36370 Dilemmas Policy Research Working Paper Series Contact Title Author Date for paper WPS2706 Household Income Dynamics Jyotsna Jalan November 2001 C. Cunanan in Rural China Martin Ravallion 32301 WPS2707 Financial Intermediary Development Thorsten Beck November 2001 A. Yaptenco and Growth Volatility: Do Mattias Lundberg 38526 Intermediaries Dampen or Magnify Giovanni Majnoni Shocks? WI/PS2708 Accountability and Corruption: Daniel Lederman November 2001 P. Soto Political Institutions Matter Norman Loayza 37892 Rodrigo Reis Soares \\PS2709 Explaining Leakage of Public Funds Ritva Reinikka November 2001 H. Sladovich Jakob Svensson 37698