Food, Agriculture and Natural Resources Policy Analysis Network (FANRPAN) Food, Agriculture and Natural Resources Policy Analysis Network (FANRPAN)
Site map|Contact us  


Determinants of crop income in Rural Mozambique, 2002-2005
Directorate of Economics: Research Paper Series
July 2009
David Mather

Acknowledgements: FANRPAN acknowledges USAID as the source of this document

Read document online

Executive summary

Crop income is the predominant source of income for most rural Mozambican households, accounting for 73% of rural household income on average in 2002, and greater than 80% of the total income of the poorest 40% of rural households. While the Government of Mozambique recognizes the need to improve agricultural productivity, there is little empirical evidence to date suggesting what mix of public and private investments would best foster improved agricultural productivity in rural Mozambique. This paper aims to better understand the determinants of household crop income in rural Mozambique, by using the TIA panel household survey of 2002-2005 to measure the impact of various private and public assets on crop income. We build upon Walker et al.’s (2004) analysis of TIA02 crop income by utilizing the econometric advantages of panel econometrics to obtain improved estimates of the impact of various private and public assets on crop income. Our principal focus is to measure the effect on total household net crop income of factors which are assumed to have a positive effect on crop productivity and profitability, including: private assets such as landholding; household use of improved inputs (fertilizer, animal traction) and diversification into tobacco or cotton; and access to public goods such as extension advice, market price information, and farm association membership.

While our analysis focuses on a few specific private and public assets, we control for many additional household and spatial factors which may affect household crop income, including drought, flood, and crop disease shocks. Although we do not find significant effects on crop income from village-level flood and crop disease shocks, we find that an additional day of drought during the principal growing season results in an average 5.5% loss in crop income in the south, and a 1.9% loss in crop income in the north. These results highlight the extreme sensitivity of crop income to weather shocks, and thus the potential value of: a) widespread promotion of smallholder access to low-cost methods of irrigation and/or conservation farming techniques to reduce the impact of drought – in contrast to the recent emphasis of heavy investment in formal perimeter irrigation schemes, which benefit only a very small proportion of the smallholder population; and b) investment in development and dissemination of drought-tolerant maize varieties as well as varietal improvement in traditionally drought-resistant crops such as cassava and sweet potato.

We find large significant effects of increased landholding on crop income, as a 5% increase in total landholding significantly increases crop income by 2.1% in the north, 2.8% in the center and 1.5% in the south. An important constraint to increased landholding could be the low use of animal traction in the center, and virtual non-existence of it in the north (attributed to trypanosomiasis spread by the tsetse fly). We investigate the determinants of landholding using regression analysis and find that adoption of animal traction use increases total landholding by 13.8% in the center and 18.5% in the south.

We also find that animal traction use significantly increases crop income by 33% in the center. Given that the crop income regression controls for total landholding separately, the return to animal traction use is likely due to improved soil productivity. The evidence of significant, large positive effects of animal traction use on both productivity (in the center) and total landholding (in the center and south) suggest that promotion of animal traction use could lead to increased crop income both through increased landholding and improved productivity. Public investment could potentially increase adoption of animal traction in the north by alleviating disease constraints to animal traction via medicinal subsidies and/or eradication of the tsetse fly. Because oxen ownership represents a high investment cost, support for rural financial services might help to address household financial constraints to financing traction rental. Given the lack of tradition of maintaining oxen in these areas, livestock extension could also play a valuable role in promoting oxen ownership or rental. While cost-benefit studies may be required to evaluate the expected ex ante rate of return to some of these investments, the high farm-level benefits which we find from animal traction use suggest that such investments could have large aggregate returns, and could help foster the emergence of more commercial farmers.

Diversification into cotton or tobacco resulted in very large and significant increases in total net household crop income. Central households which grew tobacco had 55% higher crop income than that of non-growers, while crop income among cotton growers was 194% higher than that of non-growers. Northern cotton growers had 33% higher crop income relative to non-growers; the effect of growing tobacco in the north was positive but not significant. While these crops have historically offered high returns, the option of growing such crops is only available for households which live within a reasonable distance from concession areas, and previous research has suggested that there appear to be landholding thresholds below which household participation is unlikely. Thus, tobacco and cotton are unfortunately not a panacea for widespread poverty reduction among smallholders due to these existing barriers to participation. It should also be noted that our evidence of relatively high returns to production of these crops should not detract attention from opportunities to improve the performance of Mozambique’s tobacco and cotton subsectors. Such opportunities for cotton include improving farmlevel profitability of cotton production via adoption of Bt cotton, combined with efforts to improve farmgate cotton prices through entry of new cotton firms, investments in ginning efficiency, and improved road infrastructure. Since the recent departure of a major tobacco company, tobacco growers have reported abuse of quality grading by the buying agents of the remaining tobacco company. Opportunities for tobacco may therefore involve efforts to promote the entry of other tobacco exporting companies, which may help ensure consistency in quality grading (by providing some measure of competition), and should expand the number of household with potential access to growing this highly renumerative crop.

We also find evidence that some non-farm income sources may reinforce and multiply growth in the agricultural sector. Fishing/charcoal resource extraction activities have a significant positive association with crop income in the north and center (24% and 40%), while high-return medium-small enterprise activities such as trading and construction are associated with 20% higher crop income in the north. It is important to note that it is not possible to claim that these activities have a causal effect on crop income, as these activities may be undertaken in response to (or simultaneously with) the household’s crop production that season. On the other hand, skilled nonfarm income in the center appears to compete for family labor, as it has a rather large and significant negative association (-30%) with crop income. What is surprising about these results is while non-farm income opportunities are most viable in southern Mozambique – where crop potential is limited and demand for labor from South Africa is a reality – there are no significant associations between non-farm income sources and crop income in that region. This suggests that non-farm income is not reinvested into agriculture in the south. Although we find a significant association between some non-farm income sources and crop income in the center and north, policymakers should bear in mind that the types of non-farm activities available in the center/north are largely dependent upon growth in agriculture.

We find that household receipt of market price information significantly increased crop income by 23% in the center and 31% in the south. There are several potential policy implications from these results. First, considering the size of these farm-level benefits of market information, and the widespread receipt of market price information (MPI) by rural households (40% of rural households in 2005), it is likely that the rate of return to Agricultural Market Information System of Mozambique (SIMA) investments to date is quite large, and would justify the restoration of SIMA funding to previous (higher) levels. Second, these results suggest that there would also be large returns to investments which increase household access to market price information. Given that radio is the predominant and lowest-cost (per household) method of dissemination price information (74% of households receiving price information received it via radio), there appear to be at least three ways to increase household-level access to MPI via radio broadcasts.

One means to do this would be to expand radio coverage to areas previously not served (19% of households live in villages which do not report receiving SIMA radio broadcasts). However, the number of households which live in villages which receive MPI, yet did not personally receive MPI in 2005 (32% of households) is actually larger than the total number of households living outside these areas (20% of the rural population). Thus, it would appear that the principal constraint to increased household access to MPI is not necessarily a lack of radio coverage in rural Mozambique. Low frequency of broadcasts may explain why we observe that 45% of households which own a radio (and live in a village which receives price information via radio) said that they did not receive MPI. Thus, within areas already receiving radio coverage, another option to increase the number of household which receive MPI would be to increase the number of SIMA broadcasts in a given area of the country, using existing radio stations.

Another potential constraint to increased household use of MPI may be related to the appropriateness of the content of the SIMA price information relative to the marketing needs of farmers in specific areas of rural Mozambique. For example, SIMA could possibly deliver price information in local languages, and report on all the major crops on a provincial basis. In addition, SIMA could perhaps add some analytical content to their messages, such as price forecasts and trends, or potential markets and transport costs. Addressing these types of constraints would require additional investments to set up provincial SIMA units (SIMAPS) that collect, analyze and disseminate more province-specific market information on a broader range of products than the national SIMA – which are specifically targeted to the needs of the radio audience in that province.

Because some extension advice may only result in improved productivity over time, we created extension variables to measure the possibility that an extension visit has an immediate impact on crop income (that season) or an impact which is realized over time. We also constructed separate extension variables for tobacco/cotton growers because of the advantage those growers have in terms of net returns as well as access to credit, fertilizer, etc. With respect to non-tobacco/cotton growers, we find no significant effects of the number of cumulative extension visits on crop income in any region, but we find that households in the south which received an extension visit in 2002 had 81% higher crop income in 2005 than other households (although this result in the south is not robust to the inclusion of outlying observations). The results from these two extension variables suggest that extension messages do not improve a household’s crop income in the year in which the visit is made, but that, in the south, this advice led to an increase crop income over time. One policy implication of these results is that caution may be warranted prior to substantial increases in extension funding without a better understanding of what kinds of extension are working well and which are not, and why impacts are only found over time (and only in the south).

With respect to tobacco/cotton growers, we find that an increase in the number of cumulative extension visits improves crop income by 41% in the north, while the effect of an extension visit in 2002 over time is significant and leads to an 82% increase in crop income over time in the center. A policy implication from these tobacco/cotton extension results is that it is possible for extension advice to result in higher crop income in both the year of receipt and over time. This extension effect for tobacco/cotton farmers might derive from higher returns to their cash crops, or it could be that these farmers are better able to implement extension recommendations due to better input access. Further investigation is therefore needed to discover to what extent the tobacco/cotton extension result is driven by increases in tobacco/cotton income versus increases in income from other crops.

We find that farm association membership is associated with a nearly-significant 22% increase in crop income in the north, yet had no significant effects elsewhere. A policy implication of this result is that caution may be warranted prior to substantial increases in funding to assist farm associations, without a better understanding of why associations in the north have had a nearly-significant effect on crop incomes, while those in the center and south have not. In order to measure the effect of market access on crop income, we use road density and village remoteness as proxies of market access. Because these variables are only observed in 2002, they drop out of our panel regression model, thus we run a separate regression (pooled ordinary least squares (OLS)) to measure the effect of these variables on crop income. Given the methodological shortcomings of these road variables as well as the conflicting results from them, we consider the results on these market access variables to be inconclusive.

Finally, this study suggests ways to improve the effectiveness of the TIA survey to serve as an instrument to monitor the contribution of the agricultural sector to poverty reduction. These include collecting additional information on: agricultural technology use; the specific activities of farm associations in each village; the nature of extension advice received by households and the organizational affiliation of the extensionist; and market access, preferably based on the satellite coordinates of interviewed communities, overlaid with a grid of roads in a geographic information system (GIS) format.

Top of page   -   Home   -   Contact us   -   Disclaimer
Food, Agriculture and Natural Resources Policy Analysis Network
FANRPAN Remote Access FANRPAN Webmail
Octoplus Information Solutions