The Complementarity Between Property Rights and Market Access for Crop Cultivation in Southern Rhodesia: Evidence from Historical Satellite Data

ABSTRACT

Agriculture plays a central role in the efforts to fight poverty and achieve economic growth. This is especially relevant in sub-Saharan Africa (SSA) where the majority of the population lives in rural areas. A key issue that is generally believed to unlock agricultural potential is the recognition of property rights through land titling, yet there is no overwhelming empirical evidence to support this in the case of SSA. This paper investigates access to markets as an important pre-condition for land titles to result in 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 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.

Keywords:

• land titling
• crop cultivation
• access to markets
• machine learning
• Landsat images

JEL Codes:

• C81
• N37
• Q13
• Q15

Disclosure statement

No potential conflict of interest was reported by the authors.

SUPPLEMENTARY DATA

Data is available from https://doi.org/10.1080/20780389.2019.1584526.

ORCiD

Dieter von Fintel http://orcid.org/0000-0001-6771-1315

Notes

2 With land titles, it is much easier for land to change from one person to another.

3 Jayne and Jones (1997:1506) point out that state intervention sought to “prevent African farmers from eroding the viability of the less efficient European producers”. It is acknowledged that Africans had traditionally been successful agriculturalists (Phimister 1974). For example, Phimister (1988) (cited in Andersson and Green 2016) observes that the native commissioner of Chilimanzi (now Chirumanzi) wrote that by 1904 the African produced 90% of the country’s crop production available for market.

4 Clarke (1975) puts 60–70% as the figure for Africans living and depending on rural land.

5 The crop coverage for TTAs, NPAs and EAs are the preferred variables used in the analysis. The highest correlation is between TTA crop hectarage and communal area output at 0.9994.

6 For the sake of robustness, we also estimate the models with a Tobit specification, acknowledging that the dependent variable is censored from the bottom at 0 and from the top at 1.

7 As shown in Online Appendix 3, Landsat images are available as frames or tiles. To obtain an image mosaic for the whole country, 22 tiles (that represent areas with different geographies and potentially differential data quality) are pieced together. Frame Fixed Effects (FEs) are introduced into the specification to absorb any systematic characteristics related to these images.

8 The period 1972–9 is rather eventful and the war of independence pitting the African nationalist fighters and the Rhodesian army is the major highlight. We introduce year FEs in an attempt to account for these fluctuations.

9 Southern Rhodesia (Zimbabwe)’s rain fed agricultural season starts around late October and ends around April. Ideally, the machine-learning algorithm should be applied on images for this period. However, the MSS 1–5 represents first generation sensors and the image time frequency is low. For some frames and years, images are selected even if they fall outside the agricultural season in order to ensure full geographic coverage of the country. Even where images are available for the same month (within the farming season), some are unusable due to cloud cover and the only option is to select another image from another month, preferably within the farming season. The month and season FEs attempt to correct these differences in image acquisition dates.

10 It is also for this reason that our inference does not account for serial autocorrelation in addition to spatial dependence.

11 Without fixed effects, we detect no differences in crop coverage between the various land classes. This, however, results because usable images were non-randomly distributed across various seasons in the three areas.

12 Main effects represent households that live 1 km from the various infrastructures, since ln(distance)=ln(1)=0 and this eliminates the interaction term.

13 The first and last year naturally only comprise a two-year window. We pool regressions to smooth over idiosyncratic fluctuations in specific years, and to gain statistical power, as the year-specific analyses contain few NPA observations relative to TTAs.

Additional information

Funding

Von Fintel gratefully acknowledges funding from the National Research Foundation [grant number IFR170208222264] and the Elite Fund of the Faculty of Economic and Management Sciences. Errors remain those of the authors, and opinions expressed do not necessarily reflect those of the funders.