ITEM VIEW

Land rights in sub-Saharan Africa : measuring impact with satellite images, machine learning and citizen science

dc.contributor.advisorVon Fintel, Dieteren_ZA
dc.contributor.authorChingozha, Tawandaen_ZA
dc.contributor.otherStellenbosch University. Faculty of Economic and Management Sciences. Dept. of Economics.en_ZA
dc.date.accessioned2020-02-20T09:51:03Z
dc.date.accessioned2020-04-28T12:20:44Z
dc.date.available2020-02-20T09:51:03Z
dc.date.available2020-04-28T12:20:44Z
dc.date.issued2020-03
dc.identifier.urihttp://hdl.handle.net/10019.1/108126
dc.descriptionThesis (PhD)--Stellenbosch University, 2020.en_ZA
dc.description.abstractENGLISH ABSTRACT : The thesis employs satellite imagery to measure the impacts of land rights- in data-scarce sub-Saharan Africa (SSA). SSA governments are politically and financially constrained to provide objective and reliable research data at a reasonable spatial and temporal frequency. The thesis hence fills important data gaps. The research content highlights the importance of land tenure security enforcement and access to markets in rural SSA. It is widely acknowledged that colonial institutions, particularly private property rights, continue to affect modern development. Across SSA, the majority of people rely on agriculture as a source of livelihood. Hence, agriculture has an important role to play in alleviating poverty and inequality. A consequence of extractive colonial institutions is that they selectively introduced property rights, with the majority of indigenous farming in SSA remaining under customary tenure system. This system limits the extent to which there can be effective market participation. Low investments in public goods (in particular roads and railways) and the relatively poor quality of the land in these areas compounds the problem. Chapter 2 of the thesis investigates access to markets as an important pre-condition for land titles to affect agricultural growth. Using the case of Southern Rhodesia, we investigate whether land titles incentivised African large-scale holders in the Native Purchase Areas (NPAs) to put proportionally more of their available land under cultivation than their counterparts in the overcrowded Tribal Trust Areas (TTAs). We create a novel dataset by applying a Support Vector Machine (SVM) learning algorithm on Landsat imagery for the period 1972 to 1984 - the period during which the debate on the nexus between land rights and agricultural production intensified. Our results indicate that land titles are only beneficial when farmers are located closer to main cities, main roads and rail stations or sidings. In order to address past imbalances, SSA countries have attempted various reforms in the agricultural sector, including land redistribution and tenure reform. These reforms have not translated to agriculture growth; the main argument for stagnation post-reform is that governments do not follow through in enforcing property rights. Chapter 3 of the study focuses on Zimbabwe’s 2000 Fast Track Land Reform Program (FTLRP) that reallocated more than 80% of land previously held by Europeans to the African majority. We rely on a novel, countrywide dataset of the amount of land under cultivation and crop quality [Normalised Difference Vegetation Index (NDVI)] as the endogenous variables. No study has measured the national impact of the programme on agriculture. The wide scale of the FTLRP offers a unique opportunity to interrogate how incomplete property rights (enforced land titles) affect crop cultivation in SSA post land reform, within a natural experiment design. Our Difference-in- Difference (DID) and Spatial Regression Discontinuity (RDD) estimates suggest lack of follow-through. Land redistribution reduces crop cultivation and crop quality significantly. Measuring socio-economic change using remote-sensed data is also important within urban settings. The unplanned nature and unregistered status of commercial and residential informal establishments in urban SSA limit economic potential because of lack of land titles. Where settlements are unplanned and businesses are unregistered, trust lacks, land markets are imperfect and other opportunity costs arise. Owners or occupants cannot use informal establishments as collateral to access credit, for example. Informality also has direct costs if urban services and amenities buckle under pressure. High informality results from a migration rate that exceeds job creation in urban areas. Authorities can choose between destroying informal establishments and efficient urban planning and enforcing tenure security to manage urban densification. The latter requires the development of cadastral databases and land use maps – an exercise which may be costly and resource intensive. Chapter 4 investigates the use of citizen science to classify the informal sector and different land use types from Very High Resolution (VHR) satellite images. It explores the conditions or factors affecting the precision of generating land cover maps/cadastral databases through citizen science. Cost minimization should not significantly sacrifice quality. The chapter presents a pilot study with a group of 41 Stellenbosch University students, who volunteered to classify different land use types. We use a sample of satellite images before and after Operation Restore Order (ORO) (a 2005 clean-up operation that destroyed informal structures in Zimbabwe’s main urban areas). Estimates show that the higher the number of classifications, the better the precision axiom in accordance to Linus Law does not hold; possibly due to the irregular, small and sparsely distributed nature of informal structures. It is also shown that learning effect (experience and training), as well as the demographic variables of race, sex, nationality and student volunteer hands-on experience with the land use type in question are important factors affecting classification accuracy.en_ZA
dc.description.abstractAFRIKAANSE OPSOMMING : Die tesis gebruik afstandwaarnemingsbeelde om sosio-ekonomiese verandering in data-skaars Afrika suid van die Sahara (sub-Saharan Africa = SSA) te meet. Regerings van SSA word polities en finansieel beperk om objektiewe en betroubare navorsingsdata teen ’n billike ruimtelike en tydfrekwensie te verskaf. Hierdie tesis vul gevolglik belangrike datagapings. Die navorsingsinhoud beklemtoon die belang van die handhawing van grondbesitsekerheid en toegang tot markte in landelike SSA. Daar word wyd erken dat koloniale instellings, en veral privaat eiendomsreg, steeds moderne ontwikkeling beïnvloed. Oor die hele SSA heen vertrou die meerderheid mense op die landbou as ’n bron van hul lewensbestaan. Die landbousektor speel dus ’n belangrike rol by die verligting van armoede en ongelykheid. ’n Gevolg van die onttrekking van koloniale instellings is dat hulle eiendomsreg selektief ingestel het, terwyl die meerderheid van die inheemse boerdery in SSA onder die gewone besitstelsel bly funksioneer. Hierdie stelsel beperk die mate waarin daar effektiewe markdeelname kan wees. Lae beleggings in openbare goedere (veral paaie en spoorweë) en die betreklik swak gehalte van die grond in hierdie gebiede vererger net die probleem. Hoofstuk 2 van die tesis ondersoek toegang tot markte as ’n belangrike voorwaarde waarop grondtitels die groei in die landbousektor kan beïnvloed. Met behulp van die geval van Suid-Rhodesië ondersoek ons of grondtitels die aansporing insentief was wat Afrika se grootskaalse titelhouers in die inheemse aankopegebiede (Native Purchase Areas = NPA’s) aangespoor het om proporsioneel meer van hul beskikbare grond te bewerk as hul eweknieë in die oorvol stamtrustgebiede (Tribal Trust Areas = TTA’s). Ons skep ’n nuwe datastel deur ’n steunvektormasjien-leeralgoritme (Support Vector Machine = SVM) op Landsat-beeldmateriaal vir die tydperk 1972 tot 1984 toe te pas – die tydperk toe die debat oor die neksus tussen grondregte en landbouproduksie verskerp het. Ons resultate dui daarop dat grondtitels slegs voordeel inhou as boere nader aan die belangrikste stede, hoofweë en spoorwegstasies of sylyne geleë is. Ten einde die wanbalanse van die verlede aan te pak, het die SSA-lande gepoog om verskillende hervormings in die landbousektor teweeg te bring, onder meer grondherverdeling en besithervorming. Hierdie hervormings het egter nie tot groei in die landbou gelei nie; die belangrikste argument vir stagnasie ná hervorming is dat regerings nie voortgaan en eiendomsreg afdwing nie. Hoofstuk 3 van die studie fokus op Zimbabwe se 2000 Fast Track Land Reform-program (FTLRP) wat meer as 80% van die grond wat voorheen deur Europeërs besit is, aan die meerderheid Afrikane herverdeel het. Ons gebruik ’n nuwe, landwye datastel wat die hoeveelheid bewerkte grond en gewasgehalte [Normalised Difference Vegetation Index = NDVI] as endogene veranderlikes bevat. Geen studie het nog die nasionale impak van die program op die landbou gemeet nie. Die wye skaal van die FTLRP bied ’n unieke geleentheid om uit te vind hoe onvolledige eiendomsreg (afgedwonge grondtitels) gewasverbouing in die SSA ná grondhervorming binne ’n natuurlike eksperimentontwerp beïnvloed. Ons beramings ten opsigte van verskil-in-verskil (Difference-in-Difference = DID) en ruimtelike regressiediskontinuïteit (Spatial Regression Discontinuity = RDD) dui op ’n gebrek aan deurvoer. Grondherverdeling verlaag die gewasverbouing en gewasgehalte aansienlik. Die meting van sosio-ekonomiese verandering met behulp van afstandwaarnemingsdata is ook belangrik in stedelike omgewings. Die onbeplande aard en ongeregistreerde status van kommersiële en residensiële informele instellings in stedelike SSA beperk ekonomiese potensiaal weens ’n gebrek aan grondtitels. Waar nedersettings onbeplan is en ondernemings nie geregistreer is nie, is daar ’n gebrek aan vertroue, is die grondmark problematies en ontstaan daar ander geleentheidskostes. Eienaars of bewoners kan byvoorbeeld nie informele instellings as aanvullende sekuriteit gebruik om toegang tot krediet te verkry nie. Informaliteit het ook direkte koste wanneer stedelike dienste en geriewe onder druk meegee. Hoë informaliteit is die resultaat van ’n migrasiekoers wat werkskepping in stedelike gebiede oorskry. Owerhede het die keuse tussen die vernietiging van informele instellings en doeltreffende stedelike beplanning, en die toepassing van verblyfsekerheid wat die bestuur van stedelike verdigting betref. Laasgenoemde vereis die ontwikkeling van kadastrale databasisse en grondgebruikkaarte – ’n oefening wat duur en hulpbronintensief kan wees. Hoofstuk 4 ondersoek die gebruik van burgerlike wetenskap om die informele sektor en verskillende soorte grondgebruik aan die hand van satellietbeelde met ’n baie hoë resolusie (Very High Resolution = VHR) te klassifiseer. Dit ondersoek die toestande of faktore wat die akkuraatheid van die generering van grondbedekkingskaarte of kadastrale databasisse deur middel van burgerlike wetenskap beïnvloed. Kosteminimalisering moenie veroorsaak dat gehalte beduidend ingeboet word nie. Die hoofstuk bied ’n loodsondersoek aan met ’n groep van 41 studente aan die Universiteit Stellenbosch, wat vrywillig aangebied het om verskillende soorte grondgebruik te klassifiseer. Ons gebruik ’n steekproef van satellietbeelde voor en ná Operation Restore Order (ORO) (’n 2005-opruimingsoperasie waarin informele strukture in Zimbabwe se belangrikste stedelike gebiede vernietig is). Ramings toon dat hoe hoër die aantal klassifiserings is, hoe meer hou die presisie-aksioom volgens die Linus-wet nie steek nie; waarskynlik as gevolg van die onreëlmatige, klein en yl verspreide aard van informele strukture. Daar word verder aangetoon dat die leereffek (ervaring en opleiding), sowel as die demografiese veranderlikes van ras, geslag, nasionaliteit en die studentevrywilligers se praktiese ervaring betreffende die betrokke grondgebruiktipe, belangrike faktore is wat die klassifikasie-akkuraatheid beïnvloed.af_ZA
dc.format.extentxviii, 172 pages ; illustrations, includes annexures
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : Stellenbosch Universityen_ZA
dc.subjectLand tenure -- Political aspects -- Africa, Sub-Saharanen_ZA
dc.subjectLand reform -- Africa, Sub-Saharanen_ZA
dc.subjectMarket participation -- Africa, Sub-Saharanen_ZA
dc.subjectCitizen scienceen_ZA
dc.subjectInformal sector (Economics)en_ZA
dc.subjectRemote sensingen_ZA
dc.subjectUCTD
dc.titleLand rights in sub-Saharan Africa : measuring impact with satellite images, machine learning and citizen scienceen_ZA
dc.typeThesisen_ZA
dc.description.versionDoctoralen_ZA
dc.rights.holderStellenbosch Universityen_ZA


Files in this item

Thumbnail
Thumbnail

This item appears in the following Collection(s)

ITEM VIEW