Soil Quality in Uganda: Do Transfer Rights Really Matter?

Although land tenure security is generally defined as “the certainty that a person’s rights to land will be recognized by others and protected in cases of challenges” (FAO 2002), one of its dimensions is whether the plot manager believes that he or she has complete transfer rights (Bambio and Bouayad 2018; Hayes et al. 1997; Place and Hazell 1993). In this paper, we focus on this dimension of tenure security.

Due to data limitations, we only look at the use of manure and chemical fertilizer.

Chemical fertilizers are short-term investments—they last in the soil for only one cropping season. Manure is a long-term investment—it lasts in the soil for more than one cropping season; thus, has long-lasting benefits for soil.

Others (Besley 1995; Place and Otsuka 2002; Deininger and Jin 2006; Deininger et al. 2008; Abdulai et al. 2011; Bambio and Bouayad 2018) examine land rights impact on tree planting. However, tree planting may not necessarily increase the soil quality; in fact, it may reduce it.

However, the effect of inorganic and organic fertilizers on soil quality can be complicated. Although chemical fertilizers are high in nutrient content and are used rapidly by crops and the use of chemical fertilizers increase the fertility of soil and crop productivity especially in short-term (Chen et al. 2010), long-term, excessive and inappropriate use of chemical fertilizers may pose risk to soil quality (Dong et al. 2012; Guo et al. 2010). Conversely, although organic fertilizers release nutrients in a more sustainable way, they are comparatively low in nutrient content, and the nutrient release rate is too slow to meet crop requirements in a short time. Nevertheless, long-term application of organic manure improve the quality of soil organic matters (SOM) (Chen et al. 2010). Due to these complexities, studies suggest an appropriate combination (Chen et al. 2010; Liu et al. 2017).

In Uganda, communal land includes gardens, pastures, grazing areas, burial grounds, and hunting areas commonly known as common property regimes. Communal land is mainly found in Northern Uganda. It is mainly utilized by the pastoralist communities in Northern Uganda and parts of the cattle belt in the West. User rights are guaranteed in form of farming, seasonal grazing, pasture, and other community activities. No specific ownership rights of control are conferred on users. Control and ownership are through the family, clan or the community. Similarly, under individual/family or clan customary tenure, emphasis is also in use rather than ownership. Male elders are the custodians of customary land in most communities and determine distribution of the land. However, the family has more control in the land utilization rather than the community (Busingye 2002).

Responses from recall back question on soil parcel identification asked in 2012 survey suggest that soil samples were collected from 763 households in 2003 and 841 households in 2012. The fact that we observe few households with soil samples in the data is evidence that many samples got spoiled during the process.

Thus, the first attrition related to soil sample households is about 27% (409 out 559). When we treat households with missing values of soil samples as attrite cases, the attrition is also not random (Table 8, column 3). As a robustness check, we also weight the estimations by attrition weights estimated from this type of attrition. The results (not reported to economize space) remain unaffected.

The 2012 survey asked a recall back question to identify the soil parcel for the 2003 survey. Moreover, the 2012 survey identifies the soil sample parcels. This allows us to match the parcels.

For example, although rainfall is important for moisture availability, which is important for soil health; excessive rainfall may lead to substantial leaching and depletion of soil nutrients through soil erosion. Conversely, drought may negatively affect nitrogen-fixation.

In the analysis, we also examine how each individual soil variable correlates with land rights.

For the remaining five soil variables, generally more is better.

Soil indicators from different year samples were obtained by calibration; the calibration was done year by year separately.

Respondents to this question are household heads who are also the decision-makers on land uses, investments and agricultural production activities.

One may be concerned that the change of crops may also affect soil quality results. However, since soil samples were taken from maize plots or non-maize cereal (millet, sorghum or wheat) plots if the household did not cultivate maize, we do not think that the change of crops in this case may affect soil quality because all these cereal crops have similar characteristics.

A further exploration of the data revealed that 21.1% of the 313 households (parcels) changed their perception on transfer rights from perceiving to have full transfer rights to perceiving having partial or no transfer rights, 13.4% changed from perceiving to have partial or no transfer rights to perceiving to have full transfer rights and 65.5% did not change.

Because we have one parcel per household, it could also be considered as parcel fixed effects and control any parcel level heterogeneity that are invariant overtime.

We also examine those observations that changed their transfer rights more closely (Tables 12, 13). We isolate those households to exploit this variation to see how soil quality is related to farmers who changed their perception on transfer rights. Put differently, is the move from not perceiving to have full transfer rights to perceiving to have full transfer rights associated with an improvement in soil quality? We find that moving from not perceiving to have full transfer rights to perceiving to have full transfer rights is associated with an improvement in soil quality (Tables 5, 6). Conversely, moving from perceiving to have full transfer rights to not perceiving to have full transfer rights is not associated with improvement in soil quality (Tables 14, 15). This provides further supportive evidence of our argument.

In Eq. (2), we use the more accurate attrition weights (313 out 940) because the first attrition level (779 out of 940) may be misleading here.

Transferability of land rights is likely to evolve with commercialization. Similarly, fertilizer is likely to rise with commercialization. Failure to control for commercialization may lead to endogeneity problem arising from omitted variable. In \(V_{jt}\), we also control for distance to the nearest district town (a proxy for commercialization).

The share of the households that changed perceptions on transfer rights is 34.5% (see Table 2 where we use all the 313 households and Table 4 where we use a subsample of households that did not experience changes in perceptions on transfer rights).

Here we use the full sample i.e., regardless of whether the soil sample was collected from the parcel or not. We also run the regressions with partial samples (soil samples) to see how the results match up. The results not reported to economize space are largely similar.