Jump to... Introduction Food security Project areas Market orientation Project intervention Women's work burden Conclusions Tables References

Dependent variable

Three dependent variables have been used, corresponding to three different indicators of food security: (i) the current level of food consumption within the household, (ii) the change in the level of consumption and (iii) the change in the ability to cope during lean seasons.

The respondents were asked about the average annual number of months during which they have enough cereals to feed their families reasonably well. Their answers have been taken as the measure of the current level of food security. They were also asked about the situation five years previously, and the reported change in the number of months during which they had enough cereals has been taken as the measure of the change in the level of food security. Finally, their responses to the question of whether their ability to cope during lean seasons had gone up, gone down, or had remained unchanged over the previous five years have been used as the indicator of the change in their coping ability.

Independent variables

Among the possible determinants of food security, current household income is the only important variable on which the survey has not collected any data, although it has collected information on how household income changed, in the opinion of the respondents, in the five years previous to the survey. In the absence of concrete information on the level of income, proxies have had to be used. The current size of a household’s operating land has been one, but, since this would not serve as a good proxy among those households that derive their income mainly from the non-farm sector, the level of food consumption over the previous five years has been used as an additional proxy.

Another determinant is the strength of the intervention programmes supported by IFAD. This has been measured by the number of years during which the women respondents have been members of the groups that have been formed as part of the programmes.

Market orientation has been measured in two ways: first, by capturing the potentially beneficial effect of diversified livelihood structures and, second, by capturing the potentially harmful effect of a reduced cushion of subsistence production. For the former, a diversification index has been constructed on the basis of the information collected on the number of months of food obtained by households from each source of household income. The index varies from 0 to 1, a higher value signifying a higher degree of diversification. Information on both the number of income sources and the relative contribution of each source has been used in constructing the index; the greater the number of sources and the more equal the relative contributions, the higher a household’s degree of diversification. Algebraically, the index is given by the formula D = Ö (n.e^-c), where n is the number of income sources from which a household derives its food, and c is the coefficient of the variation in the amounts of food (as measured by the number of months of food) obtained from the different sources.

In order to capture the potentially harmful effect of a reduced cushion of subsistence production, two different variables have been used. A dummy variable has been used to distinguish those households for which subsistence income has declined more than any other category of income during the five years previous to the survey. Another dummy variable has been used to distinguish those households for which subsistence production has declined as a result of push factors from those households for which subsistence production has declined as a result of pull factors. The rationale behind the use of the second variable is that a reduction in subsistence production may not be so harmful if better prospects in other sectors pull households away from subsistence production; the really serious cases are those in which households have been pushed out of subsistence production against the will of the household members, for example because of a loss of land, poor rainfall, reduced family labour, and so on.

In addition to these variables, other variables have been used to capture the possible effect of household size and any area-specific effect in the three survey regions.

Regression methods

The regressions on current levels and changes in the levels of food consumption have been carried out by ordinary least squares, and the regression on changes in coping ability have been carried out with the logit method.

Explanation of symbols

Dependent variables

MFFWN – The number of months during which a household could eat well at the time of the survey.

MFFWC – The change in the number of months during which a household could eat well at the time of the survey compared with five years previously

DUMCOPE – A dummy variable for the change in coping ability compared with the situation five years previous to the survey: 1 = ability up, 0 = ability down or unchanged

Independent variables

GTJYD – The number of years during which the woman (respondent) has been a member of a group formed through the IFAD project

DIVINDX – The index of the diversification of a household’s livelihood structure. A higher value of the index signifies greater diversification

MFFW5 – The number of months a household could eat well five years previous to the survey

OPRLE – The extent of the land operated by the household

DEPT – The number of dependents in the household

DUMINCC – A dummy variable for the change in the total income of the household over the five years preceding the survey: 1 = income up, 0 = income down or unchanged

DINSINM – A dummy variable to distinguish those households for which the subsistence income increased more than the income from any other source during the five years preceding the survey: 1 = these households, 0 = all other households

DINSDCM – A dummy variable to distinguish those households for which the subsistence income decreased more than the income from any other source during the five years preceding the survey: 1 = these households, 0 = all other households

DUMPUSH – A dummy variable to distinguish those households for which the subsistence income has decreased due to push factors as opposed to pull factors: 1 = push factors, 0 = pull factors

DUMLQC – A dummy variable for the change in the quality of the land: 1 = quality up, 0 = quality down or unchanged.

DUMTN – A dummy variable for Tamil Nadu

DUMNP – A dummy variable for Nepal

Regression results

Regression on Food Security

* Significance at less than the 10% level.
** Significance at less than the 5% level.

Major findings

The major findings of the regression analysis can be summarized as follows:

(a) The greater the degree of diversification of the livelihood structure, the higher the current level of food consumption, other things remaining the same. Longer involvement in IFAD-supported programmes and a larger amount of operating land also contribute to higher levels of food consumption.

(b) A change in the level of food consumption is more favourable among those households whose land quality has improved over time, for those who were involved with IFAD-supported programmes for longer periods, and for those whose livelihood structure was more diversified. But households which were pushed out of subsistence production against their will tended to suffer a reduction in food consumption.

(c) Diversification of livelihood structure and involvement in IFAD-supported programmes have also helped improve the ability of households to cope during lean seasons. However, those households among which the income from subsistence crops has declined more than the income from any other source have tended to experience a reduced ability to cope.

The implications of the findings

Two major conclusions follow from this analysis. First, the evidence confirms the point made in the text that increased market orientation can have two opposing effects on household food security. By allowing increased diversification in livelihood structures, increased market orientation improves both the level of food consumption in normal times and the ability to cope during bad times. However, if market orientation is accompanied by a big fall in subsistence production, especially if this has been caused by push factors, it can have a deleterious effect on food security.

The lesson for IFAD and other national and international agencies is that their interventions should include a component of which the objective is to safeguard a minimum level of subsistence production. There is nothing wrong with encouraging poor households to produce for the market; there is also no harm if, in the process of producing for the market, households reduce the amount of time they spend on subsistence production. But, in that case, measures must be taken to improve the access of households to better food production technologies and to more complementary inputs (for example, fertilizer and irrigation) so that the amount of food produced by the households does not drop to a precariously low level. The implication of this point is that single-component projects, such as a pure credit delivery programme or a programme for providing inputs only for cash crop production, may not be the best kind of intervention from the point of view of food security.

Second, there is clear evidence that IFAD-supported programmes have made significant contributions to HFS in the survey areas. The longer a household has been involved in a project, the better it has done in all dimensions of food security. This is a pleasing confirmation that IFAD’s interventions have been effective in improving HFS, but understanding the precise mechanisms through which this effect operates requires further investigation.

There are several possibilities. For example, these programmes have helped many households diversify their livelihood structures by providing credit for self-employment activities in the non-farm sector and by providing inputs for cash crop production. By providing irrigation facilities and other aids for land improvement, they have also helped many other households improve the quality of their land. All these measures might have contributed to food security. But since the diversification in livelihood structures and the improvement in land quality have been included in the analysis as separate variables, they already capture any effect that project interventions may have exerted through them. So, the variable representing the duration of project participation must refer to some other effect. It has been argued in this report that this variable captures the indirect effect of project participation that operates through the empowerment of women. The argument proceeds as follows: by mobilizing women into groups, the projects enable them to gain greater access to resources and to acquire greater control over decision-making within the household. These advantages in turn contribute to household food security. Evidence in support of this argument has been presented in the text.