LAND TENURE AND SOCIOECONOMIC INTERACTIONS Jame s Ssemam bo, James O. Bukenya and M arb el M ojca* Selected Paper to be presented at the Amercan Agrcultural Economcs Assocaton s Annual Meetng. Long Beach, CA July 23-26, 2006. Abstract The study analyzed the determnants of land tenure nsecurty n Uganda usng survey data collected by the Internatonal Food Polcy Research Insttute (IFPRI) durng the Polces for Improved Land Management Project n Uganda, 1999-2001. The survey ncluded a sample of 1322 farm households randomly selected and ntervewed usng a formal questonnare. The analyss revealed that tenure category, number of households n the vllage nvolved n dsputes outsde the vllage, and the number of households n the vllage who lost land as a result of a dspute over land were sgnfcant factors affectng tenure nsecurty. *Graduate Research Assstant, Assstant Professor and Research Assocate, respectvely n the Department of Agrbusness, Alabama A&M Unversty, Normal, Alabama 35762.
LAND TENURE AND SOCIOECONOMIC INTERACTIONS Introducton Uganda was known as a major food source n Afrca because of ts good sol, dependable ranfall and relatvely hgh agrcultural producton. Over tme however, the country s agrcultural sector has been characterzed as low-nput, sem-subsstence agrculture (Place and Otsuka, 2002). One of the concerns s the ssue of land tenure n the country. The country s confronted by land tenure conflcts of varous types. For example, cvl wars have resulted n ncreased threats to and conflcts over natural resources. The tradtonal rvalry between farmers and pastoralsts has been fuelled at tmes by based government polces. A common percepton s that the prevalng land tenure nsttutons are dscouragng farmers and nhbtng them from makng major agrcultural nvestments. The queston of the approprate tenure polcy has remaned unresolved, although the 1995 Land Consttuton and the 1999 Land Act emphaszed securty of tenure by Uganda s smallholders, protecton of women and other vulnerable groups from rresponsble land sales, and promoton of nvestment and smooth operaton of the market. Numerous land redstrbuton polces and soco-economc nteractons through the years have left many Ugandans landless and many n dsputes. Ths paper therefore ams to examne the factors affectng tenure nsecurty n Uganda. The paper relates tenure nsecurty to demographc characterstcs and socal nteractons. The paper n organzed as follows. Frst, a bref revew of the land tenure systems n Uganda followed by a defnton of tenure nsecurty and socal nteractons n secton three. Data and econometrc model are presented n sectons four and fve. The last two sectons present the result and conclusons. 1
Land Tenure Systems n Uganda The promulgaton of the 1995 Consttuton of Uganda brought wth t very sgnfcant changes. The radcal ttle to land was vested n the ctzens of Uganda, the Land Reform Decree was abolshed and the systems of land tenure that were n exstence before ndependence renstated. These were stated as customary tenure, malo tenure, freehold tenure and leasehold tenure: 1) Freehold tenure nvolves holdng of regstered land n whch the holder has full ownershp rghts. Land s held n perpetuty subject to statutory and common law qualfcatons. 2) Leasehold tenure nvolves holdng land for a gven perod of tme specfed on commencement (valdty) on such terms as may be agreed upon by lesser and lessee and may nclude sublease. 3) Malo land tenure nvolves holdng regstered land n perpetuty. Ths system has ts roots n the 1900 Buganda Agreement between Buganda and the Brtsh. In the central Ugandan regon of Buganda, the clan system was undermned wth ntroducton of the malo system, but malo land s stll subject to clan and lneage head approval. The Busuulu and Envujjo Law of 1928 made t dffcult for landlords to evct tenants and set a rent celng. The law ncreased nsecurty of tenure for malo owners by recognzng some of the secondary rghts of tenants as prmary rghts, whle extngushng many prmary rghts of the owners. The 1975 Reform Decree, then gave malo owners more powers to evct the tenant upon ssung a sx-month notce. Thus there was a shft n legal control over malo land, now n favor of the State and the owner. Although many malo owners stll perceved themselves as owners of the land, those wth tenants have rarely taken steps of evctng them. 2
4) Customary tenure s a system of land tenure regulated by customary rules often admnstered by clan leaders. The customary system predomnates n Uganda. Wthn ths system there can be both ndvdual and communal land ownershp, but the land s not generally ttled or regstered. Snce 1966, tenure securty n ths land area has been low and unpredctable (Place and Otsuka, 2002). Tenure Insecurty and Socal Interactons Tenure nsecurty s defned as the perceved probablty or lkelhood of losng ownershp of a part or the whole of one s land wthout hs/her consent (Sjaastad and Bromley 1997, Alemu 1999). The strength of ths percepton may have a bearng on how farmers manage ther land and ths n turn has an effect on agrcultural producton and sustenance of the people who drectly depend on t. A lot of authors have stated that tenure nsecurty dscourages farm operatons and land nvestments. Ksamba-mugerwa (1989), found that there was a consderable uncertanty as to future land rghts wthn malo land. They also found out that on land occuped by tenants, t was the owner who especally felt nsecure about long-term land rghts, because of possble government land tenure reform. Some studes addressed the effect of dfferences n tenure systems and tenure securty on agrcultural nvestment whle studes to determne the causes of tenure nsecurty whch may dentfy the root cause of the problem s lackng. Socal nteractons n the communtes such as dsputes over land have been the root cause of cvl wars and revolutons, much so because land has been the prmary means of generatng lvelhood for the overwhelmng majorty of rural populaton n these countres (Dennger and Feder 1998).Furthermore, the way land tenure s nsttuted and the consequent perceptons and expectatons of the land holders may drectly affect the way 3
farms are managed (Frmn-Sellers and Sellers 1999; Maxwell and Webe 1999; Sjaastad and Bromley 1997; Gavan and Ehu 1999; Alemu 1999) and ths may have effcency as well as sustanablty consequences. Data To analyze the determnants of land tenure nsecurty n Uganda, we draw on data from the Internatonal Food Polcy Research Insttute (IFPRI). The survey data were collected durng the Polces for Improved Land Management Project n Uganda, 1999-2001. In all, one hundred vllages were selected across the country and the questonnares were admnstered to 107 communtes, the lowest admnstratve unts n Uganda called Local Councl 1 or LC1. The study regon covered most of Uganda, ncludng more densely populated n the southwest, central, eastern and parts of the north, representng seven of the nne major farmng systems of the country. Wthn the study regon, communtes were selected usng a stratfed random sample, wth the stratfcaton based on populaton densty and development domans defned by the dfferent agro-ecologcal and market access zones. Topcs n the vllage survey ncluded lvelhood strateges, land use, land tenure and land markets, labor, wage rates and credt, crop producton, commercalzaton and management, lvestock management and commercalzaton, tree product and commercalzaton. The survey also collected nformaton on the respondents demographc characterstcs. Table 1 presents the varables ncluded n the analyss and ther descrptve statstcs. The model used a dummy ndependent varable, tenure nsecurty (SECURITY), coded as 0 when the household feels tenure secure and 1, otherwse. The majorty (76 percent) of the respondents stated that 4
they were tenure nsecure. Tenure category (TENCAT) was represented as: 1 = freehold, 2 = leasehold, 3 = malo, and 4 = customary. Involvement n land dsputes n the vllage (DISPVILL) was coded 1 when the respondent experenced or s nvolved n a land dspute n the vllage and 0, otherwse. The number of households nvolved n dsputes n the vllage (HHDSINV) was treated as a contnuous varable. Respondents nvolvement n dsputes outsde the vllage (DISPOUTV) was coded 1 when the household had an nvolvement n dsputes outsde the vllage and 0, otherwse. The number of households n the vllage nvolved n dsputes outsde the vllage (HHDINVOV) was used as a contnuous varable. If the respondent dd lose land as a result of dsputes, LOSEACC was represented by 1 and 0, otherwse. The number of households who lost access to land (HHDLACC) was used as a contnuous varable. Gender was coded as 1 for males and 0 for females. Respondents age was a contnuous varable whle HHDHEAD was coded as 1 f the respondent s a household head and 0 otherwse. ---------- Table 1 about here ------------- Model A logt model was used to examne the relatonshp between tenure nsecurty and respondents demographc characterstcs as well as socal nteractons. Specfcally, the model was used to estmate the lkelhood of a household beng tenure nsecure. The percepton of tenure nsecurty and the factors affectng t are modeled as, N = X+ e, (1) where N = 1 f the household feels tenure nsecure, and 0 otherwse, X s an array of factors that may cause or are assocated wth tenure nsecurty, s a vector of parameters, and e, a vector of error 5
terms. Assumng that the dsturbances are normally dstrbuted, the relatonshp between tenure nsecurty and the varous factors used as ndependent varables was specfed as: N= 0 + 1 TENCAT + 2 DISPVILL + 3 HHDSINV + 4 DISPOUTV + 5 HHDINVOC (2) + 6 LOSEACC + 7 HHDLACC + 8 SEX + 9 AGE + 10 HHDHEAD + where, N = Percepton of tenure nsecurty of a household (dependent varable). If a household feels tenure nsecure, the varable takes the value of 1, and 0 otherwse. 0 through 10 are unknown parameters to be estmated, and s the error term. The probablty of a household beng tenure nsecure was examned usng the equaton: P = E (Y = 1 X ) = β 0 + β X (3) Where, P s the probablty that Y equals 1 for a gven value of X. X ( = 1, 2, 3 n) represents the explanatory varables. β 0 represents ntercept and β represents coeffcents to be estmated. Equaton 3 can be represented as: P = E( Y = 1X ) 1+ e 1 _( βo+ βx) Where, e s the base of the natural logarthm; approxmately 2.718. For smplfcaton, equaton 4 can be wrtten as: (4) P 1 = 1 + e Z (5) Where, Z = β 0 + β X. Z s a lnear combnaton of (β 0 + β 1 X 1 + β 2 X 2 + β 3 X 3 + β n X n ) and ranges from to + ; P ranges between 0 and 1. If P n Equaton 5 represents the probablty of a household 6
beng tenure nsecure, then (1 - P) represents the probablty of a household beng tenure secured, hence, 1 1 P = Z (6) 1+ e Combnng equatons 5 and 6, we derve: P 1+ e 1+ e Z Z = = e Z 1 P (7) where P 1 P s the odds rato of beng tenure nsecure, whch s the rato of the probablty that a household s tenure nsecure. Equaton 7 s the condtonal probablty that Y s equal to 1 gven X. Ths was denoted as P (Y = 1 X ). Conversely, the quantty Y = 1 - P gves the condtonal probablty that Y s equal to zero gven X, P (Y = 0 X ). By takng the natural log of equaton 6.0, the result wll be: P L = β ln = Z = β 0 + X (8) 1 P where L s the natural log of the odds beng tenure nsecure. The model s that natural log of the odds equal to the constant (β 0 ) plus the product of the estmated coeffcents β and X. The change n probablty was calculated as: P = β P 1 P) (9) ( where P = s the estmated probablty of a household beng tenure nsecure n each observaton; and β s the estmated coeffcent. The change n probablty ( P ), therefore, s a functon of the 7
probablty of a household beng tenure nsecure (P ) multpled by the probablty of beng tenure secured (1- P ) and the estmated coeffcent (β ) consderng other varables constant. Results and Dscussons Table 2 presents the logt estmaton of the tenure nsecurty model. The estmated results were nterpreted usng the change n probablty (Equaton 9). The results show that tenant category s a sgnfcant factor affectng tenure nsecurty n Uganda. The coeffcent exhbts a postve sgn, as expected, whch shows that the movement of tenure category from freehold land to customary land s n the same drecton as tenure nsecurty. Ths means that tenure nsecurty ncreases as tenure category moves from freehold to customary. The change n probablty suggests that as a household moves from one tenure category to the next category, the household s 10 percent more lkely to feel tenure nsecure. Another sgnfcant varable affectng tenure nsecurty determned by the model s the number of households n the vllage nvolved n dsputes outsde the vllage (HHDINVOV). As hypotheszed, the parameter estmated showed a negatve relatonshp between tenure nsecurty and havng a land dspute outsde the vllage. Ths can be explaned by the reason that the level of tenure nsecurty are ndependent between vllages, that an ndvdual s relatvely secure when nvolved n a dspute n another vllage as the dsputes don t drectly affect hs status n the vllage where he s. Land dsputes and tenure nsecurty n ths scenaro are dealt wth separately from one vllage to another. The change n probablty ndcates that as the number of households n the vllage nvolved n dsputes outsde the vllage ncreases, the lower the lkelhood (4.6 %) that the households wll be tenure nsecure. Ths analyss s n relaton to testng the level of tenure nsecurty wth households nvolved n dsputes nsde the vllage. The number of households who lost access to any land as a result of a dspute over land rghts was determned to be a sgnfcant factor determnng tenure securty. The postve sgn of HHDLACC ndcates 8
that f the household dd lose land n the past, they are more tenure nsecure for fear of losng more land, as the cause/s of losng land n the past can be possble causes of losng another land n the future. The change n probablty shows that as the number of households who lost land ncreases, the lkelhood of beng tenure nsecure ncreases by 4.7 percent. --------- Table 2 about here ------ Other varables ncludng land dspute nvolvement n the vllage, number of households nvolved n dsputes n the vllage, dspute nvolvement from outsde the vllage, losng land as a result of dsputes, respondents age, gender, and beng a household head were tested but were found to be nsgnfcant. Although the R 2 (Nagelkerke) s low, 0.101, a low R 2 s acceptable n logt regresson (Hosmer and Lemeshow, 2000). The model predcted a 77 percent of the responses. Conclusons The paper analyzed the effects of tenure nsecurty and soco-economc nteractons on farm households n Uganda. The results provde nsghts on the factors determnng tenure nsecurty. It can be concluded that there s tenure nsecurty n the study area as 76 percent of the respondents stated that they feel nsecure about ther land tenure. In the model, tenant category was found to be a sgnfcant varable affectng tenure nsecurty. Increases n the number of households n the vllage nvolved n dsputes outsde the vllage were also found to ncrease the lkelhood of beng tenure nsecure. Smlarly, the number of households who lost access to any land as a result of a dspute over land rghts was found to be a sgnfcant factor determnng tenure securty wth a postve relatonshp. 9
References Alemu, Teke. 1999. Land Tenure and Sol Conservaton: Evdence from Ethopa. Ekonomska Studer (92). Goteborgs Unverstet (Kompendet-Goteborg). Dennger, Kalus and Gershon Feder. 1998. Land Insttutons and Land Markets. A paper prepared as a background for the comng Handbook on Agrcultural Economcs. World bank, Washngton, D.C. Frmn-Sellers, Kathryn and Patrck Sellers. 1999. Expected Falures and Unexpected Successes of Land Ttlng n Afrca. World Development 27(7), 1115-1128. Gavan, Sarah and Smeon Ehu. 1999. Measurng the Producton Effcency of Alternatve Land Tenure Contracts n a Mxed Crop-Lvestock System n Ethopa. Agrcultural Economcs, 20, 37-49. Holden, S. and H. Yohannes (2001). Land Redstrbuton, Tenure Insecurty, and Intensty of Producton: A Study of Farm Households n Southern Ethopa. Land Economcs. 78(4):573-90. Hosmer, Davd and Stanley Lemeshow. 2000. Appled Logstc Regresson, 2 nd edton. Wley- Interscence Publcaton. John Wley & Sons, Inc. pp 375. IFPRI. Uganda: Polces for Improved Land Management Dataset, 1999-2001. Accessed February 2005, www.fpr.org/data/uganda01.htm. Ksamba-Mugerwa. 1989. Land Tenure and Agrcultural Development n Uganda. Land Tenure Center, Unversty of Wsconsn. Place, Frank and Kejro Otsuka. 2002. Land Tenure Systems and ther Impacts on Agrcultural Investments and Productvty n Uganda. The Journal of Ddevelopment Studes 38 (6):105-128. Sjaastad, Espen and Danel Bromley. 1997. Indgenous Land Rghts n Sub-Saharan Afrca: Appropraton, Securty and Investment Demand. World Development 25(4), 549-562. 10
Table 1. Defnton and descrptve statstcs of varables Varables Descrpton Mean Standard Devaton SECURITY TENCAT DISPVILL HHDSINV DISPOUTV HHDINVOV LOSEACC HHDLACC Does the respondent feel secure? 0-Secure; 1-Insecure Tenant category 1-Freehold; 2-Leasehold; 3-Malo; 4-Customary Respondents nvolved n land dsputes n the vllage 1-Yes; 0-Otherwse Number of households nvolved n the dspute n the vllage Contnuous People nvolved n dsputes from outsde the vllage 1-Yes; 0-Otherwse Number of households n the vllage nvolved n dsputes outsde of the vllage Contnuous Dd anyone lose land followng the dspute? 1-Yes; 0-Otherwse How many households lost access? Contnuous 0.7602 1.9750 0.5590 4.9811 0.3472 6.2640 0.1929 0.7451 0.4271 1.1491 0.4967 10.0612 0.4763 30.2117 0.3947 1.8712 SEX Sex 1-Male; 0-Female 0.6188 0.4859 AGE Respondents age Contnuous 37.7678 12.2469 HHDHEAD Is the respondent a household head? 1-Yes; 0-Otherwse 0.6528 0.4763 11
Table 2. Probablty Estmates for Tenure Insecurty n Uganda Varable Coeffcent Std. Error Sg. Level Probablty INTERCEPT TENCAT * DISPVILL HHDSINV DISPOUTV HHDINVOV * LOSEACC HHDLACC * SEX AGE HHDHEAD Model predcton success = 77% Nagelkerke R-square = 0.151 Log-lkelhood = 1364.22 Sample Sze = 1322 *denote sgnfcance at 1 percent level 0.0860 0.4480*** -0.1995-0.0091-0.2479-0.0186*** 0.2541 0.1924*** 0.3314 0.0080-0.2314 0.2633 0.0683 0.1723 0.0070 0.1820 0.0038 0.3088 0.0769 0.2081 0.0058 0.2182 0.7439 0.0000 0.2468 0.1897 0.1732 0.0000 0.4105 0.0123 0.1112 0.1695 0.2888-0.1044-0.0497 0.-0.0023 0.0603-0.0046-0.0631 0.0472 0.0793 0.0019-0.0575 12