The question of wealth ranking is a vexing one. It seems to be assumed that if we are doing participatory research into sustainable development for the poor, we have to have a wealth ranking exercise. It has become part of the standard PRA menu, and yet one often encounters complaints that it is difficult to do because people are resistant, or apologies for not being able to do wealth ranking.

Local differences in wealth and poverty are certainly very much within the concern of this project. Why then is wealth ranking seen as difficult to implement? One reason must be that, as generally practiced, it is seen as intrusive. And no wonder. Imagine a team from a university in Iraq trying to survey the assets of the families of a small rural community in the USA. How would they be received? What sort of motives would people impute? To try to identify and rank every household in a community according to its assets or wealth by whatever method - whether by public pair ranking, or by key informants in semi-public or secret (as recommended by some), or by questionnaire survey - is surely most obviously something that could easily lead to suspicion or resentment. It is only natural for people to fear to reveal their wealth, or their poverty, for fear that such information may be used to their disadvantage.

Wealth ranking is usually very difficult in pastoral communities (less so in agricultural ones) and if done may simply yield spurious results. In any case, there is no good reason for trying to make it a public exercise. It is not clear why so many participatory researchers seem to feel that it is necessary to try to undertake wealth ranking by household, especially at an early stage in field work; and then to have to apologize for the difficulties they encountered in trying to do so.

We should ask, therefore, what it is we are trying to find out. Wealth and poverty analysis, which is broader than ranking, may be carried out for many possible reasons, and with a variety of tools, the most important of which include:

• Poverty programming and targeting: a poverty alleviation program wishes to ensure that its relief or development benefits are not hijacked by the local elite;
• Establishing the level and trend of socio-economic equity: to find out the distribution of wealth in the society, for example as in a Lorenz curve, from which we can derive statements like "10 percent of the population own 90 percent of the cattle, and 90 percent of the population own 10 percent of the cattle , and (by comparison) the trend is getting worse".
• To understand the dynamics of wealth and poverty in a given society: what characteristics and social relations enable or hinder prosperity, and what characteristics intensify poverty. For example, one might discover that both the wealthiest and the poorest in a given society are more recently settled.

Of these three main purposes above, only one (the second) requires some kind of household ranking, and even in that case a sample survey would be sufficient. For the others, an analysis by category would be enough. And it is very non-contentious in the local situation.

Instead of ranking, categorizing a population by wealth or poverty might be sufficient, particularly for the purposes of this study. We recognize that wealth is a continuum, that the boundaries of the categories are fuzzy, that there are marginal individuals and households, and that there is mobility between categories over time. Wealth categories are nevertheless a useful heuristic device. We expect that those people in certain categories have sustainable livelihoods, other people are at risk, while yet others have no prospect of sustainable livelihoods. We expect that people in one category would have certain options or adaptive strategies open to them under stress, while those in other categories would have different options, or no choice. We expect that certain policy changes might benefit people in one category while ignoring or even prejudicing those in another category. We may need the categories for the purpose of constructing a stratified sample.

The first step therefore is to obtain through discussion with community members some of the key local criteria or characteristics of poverty and wealth, and then, by applying these characteristics in abstract, to determine a set of useful categories that make sense in local discourse. It is important to try to avoid derogatory classifications, choosing instead terms that have broad social acceptability. For example, one community in Uganda agreed on a four-fold classification of:

1. "Those who can manage", a euphemism for the relatively wealthy who could look after their own interests without help from anybody;
2. "Those who have something", i.e. had some small assets, such as a few animals, some equipment, and an opportunity of developing;
3. "Those who earn slowly", i.e. they had no real capital assets, but struggled on a daily basis to make ends meet; and
4. "Those who cannot manage" the obverse of the first group who were destitute and therefore required external support to survive (Rennie, 1994).

The general characteristics of each social group could be described. Broad approximation can be made of the relative proportions of these, either by observation or social mapping. Further insights can be obtained by identifying households with serious problems, e.g. food shortage, lost all their animals. Yet further insights can be obtained by individual household analysis, biographies, key informants etc.

What was particularly interesting about this four-fold classification of wealth and poverty in Uganda was that it corresponded quite well with a recent World Bank analysis of poverty in that country, thus again allowing linkages between local and contemporary to be made.