In this section, a summary of the four papers highlighting on the rationale, objectives, methods and major findings is presented. The reader is encouraged to see the full results in context. The underlying assumption in all the papers is that individual farm households allocate time and other resources in order to achieve utility or income maximization, given their resource endowments and constraints. A theoretical model that assumes farm households who operate under imperfect market conditions and asset poverty is used as a general decision framework. Primary data collected through a cross-sectional farm household survey from a densely populated village in southern Ethiopia is used for all the papers. The interpretation of analytical results was enriched through field observations and qualitative information gathered through discussions with farmers, researchers, policy makers and development agencies operating in the area.
Paper I: Between Malthus and Boserup: farm size-productivity relationship under population pressure
The aim of this paper is to analyze the relationship between farm size and land productivity under intense population pressure. We ask if higher levels of land use intensification may lead to an inverse farm size-productivity relationship. The paper extends past work using plot specific land quality attributes. We build a theoretical framework that reviews previous assumptions about the relationship between population pressure, farm size and intensification, and the roles of market imperfections, or heterogeneous land quality for the inverse farm size-productivity relationship.
Farm household and plot level data were used for the empirical analysis. First, we tested whether population pressure and market imperfections explain the inverse farm size- productivity relationship at the farm household level. Next, we applied a household random effects (RE) model on the plot level data and tested for the effects of observed plot quality attributes.
The results from farm household level analysis of the farm size-productivity relationship exhibited persistence of the inverse relationship for all the regressions that tested for different hypotheses such as labour market imperfections, population density, and asset poverty. The household level analysis did not fully support the labour market imperfection explanations for the inverse relationship; family workforce endowments were insignificant in all the
regressions and their inclusion did not change the inverse relationship. This may indicate that labour is abundant while marginal returns to labour are very low. When we included a range of observable soil quality indicators in the regression, using plot level data, the inverse relationship became positive and statistically insignificant. These results imply that it is important to control for land quality when this type of analysis is carried out and analysis at plot level is preferable to be able to do this well. Matched plot panel data would be required to control for unobservable time-invariant plot characteristics.
An inverse relationship after controlling for land quality may reflect the increased intensification efforts undertaken by some households under conditions of land scarcity when markets are imperfect. In line with this, we included a dummy for presence of a perennial food crop and the amount of fertilizer applied on a plot in the regression. Both variables were positively and significantly related to productivity but the farm size variable was still insignificant although its positive sign increased in magnitude. The elasticity of productivity in response to population pressure was low, less than 0.5 and 0.3 at household and plot levels of analyses respectively. The intensification efforts of the households, most probably being constrained by credit market imperfections, lack of livestock, and asset poverty, were not sufficient to get them out of the Malthusian trap. Policies that enhance the availability and use of credit for productive purposes could help to boost the smallholders’ intensification efforts through increased use of productivity-enhancing purchased inputs, e.g., fertilizer and improved seeds. This is important for increasing the propensity of households to cope with declining land availability under population pressure. However, also this strategy has clear limitation on the small farms in the study area given that the production is rain-fed and the population growth continues at a very high rate while off-farm employment opportunities are very limited.
Paper II: Too poor to invest? Poverty and farm intensification decisions in Southern Ethiopian highlands
This paper analyzes farm intensification decisions of smallholders by investigating how land scarcity, imperfect markets, available technologies, and livestock and land endowments shape farm intensification decisions. The rationale of the paper is to test whether asset poverty, market imperfections and subsistence constraints under land scarcity force farm households to intensify production in starchy perennials. Using a two-period intensification model with crop-livestock interactions, we investigated whether there are significant synergy effects
between livestock and a perennial food crop (enset) and if poverty in livestock assets can lead to a poverty trap. Parametric and non-parametric methods such as fractional response, Tobit and OLS models and kernel density distributions were used in the empirical analyses of farm household data and to further test the links between asset poverty and farm intensification. In this paper, farm intensification is measured in terms of enset area share of cropped area and enset stock per unit of cropped area (see Figure 2 for the structure of enset).
The results indicated that increasing land scarcity contributed to intensification in the production of the perennial, and crop-livestock interactions facilitated the process of farm intensification. Livestock ownership appears to be a key to more successful intensification in response to population pressure even if land poverty also appears to stimulate investment in perennials. Livestock- and land-poor households appeared to be less able to intensify and got lower returns to their investment than more livestock-rich but land-poor households. The results also indicated that the most asset-poor household group had more limited access to off-farm income, has poorer access to formal credit markets and stood out with less crop production per consumer unit. However, the livestock change analysis revealed that this poorest household group also was able to rebuild their livestock endowments to the same level as other households after a few years, indicating an ability of breaking out of the household group-specific poverty trap.
Paper III: Are the poor forced to cut the branch they are sitting on? Perennial crop harvesting decisions of food insecure smallholders
This paper provides a comprehensive quantitative study on farm households’ harvesting decisions of a perennial food crop, enset. The perennial is a staple starchy crop that takes 4-5 years to reach maturity, but evidence shows that many households harvest this crop before its physiological maturity, sometimes after less than two years. By doing so, they forgo some future consumption as they lose higher yield obtained from a mature plant. The paper assesses whether poor households are caught in a poverty trap and are compelled to cut down the seeds in order to meet immediate consumption needs. First, we calibrated a simple simulation model in order to assess the relevance of a Faustmann rotation model in our case study area. We used experimental production data to estimate a simple growth function for the starchy perennial crop, enset. The simulation results indicate, as expected, that higher discount rates shorten the rotation period.
With imperfect labour and land markets and borrowing constraints, individual households may have idiosyncratic discount rates, depending on their poverty status. This may imply different harvesting strategies in terms of rotation times. We therefore expanded the Faustmann rotation model to account for local situations and integrated this into our empirical analysis. Based on the expanded Faustmann optimal rotation model and using cross-sectional data, the empirical analysis investigated factors related to average rotation period, proportion of young plants harvested, and the number of young plants harvested. The econometric methods applied in the analysis include OLS, fractional response and Cragg’s models.
The empirical study finds that food insecurity is the key driving factor for early harvesting of the starchy perennial crop. The results reveal that food insecure households could be in a vicious circle of food insecurity: early stage harvesting may help them to alleviate the immediate subsistence constraints, but it reduces future availability as fewer plants are saved to reach full maturity. The paper provides quantitative explanations for the underlying reasons for the early harvesting decisions. It also contributes to the existing literature on the application of the Faustmann rotation model in developing country conditions, using a perennial crop which has not been studied elsewhere.
Paper IV: Soil degradation, poverty, and farmers’ willingness to invest in soil conservation
This paper analyzes farm households’ perceptions of land degradation and their willingness to invest in soil conservation activities. Households consider soil erosion and nutrient depletion to be the most important land degradation problems on private farms in the study area, with gully formation being an additional problem on communal land. About 45% of the sample households ranked soil erosion as the most serious problem on their farms followed by nutrient depletion (41%).
Based on assumptions of land and labour market imperfections and thus non-separability of production and consumption decisions, we used a two-period investment model (Bellman equation) to identify the determinants of farm households’ willingness to invest in soil conservation on their private farms. The contingent valuation method (CVM) was used to elicit farmers’ willingness to invest in soil conservation in terms of in-kind labour contribution and cash payments, the two possible payment vehicles.
Both OLS and Tobit regression results for the willingness to invest in soil conservation on private farms in terms of labour contribution and cash payments indicate that livestock wealth in tropical livestock units (TLU) has positive correlation with the willingness to pay (WTP) both in labour days and in cash. The econometric results revealed that a unit increase in TLU would lead to an increase in WTP by 7.84 birr and 27 person days. The results also indicated that the shadow wage rates are very low which could be attributed to labour abundance, limited employment opportunities and financial constraints in the area. Methodologically, the WTP studies that employ both labour and cash as payment vehicles could provide more sensible results than those using only cash. Farm households are generally willing to invest in beneficial conservation practices that are less cash-intensive and able to use locally available labour resources.
3. Conclusions and some policy implications