Poor farmers have been blamed for environmental degradation in developing countries. Struggling to survive, they have been claimed to overexploit their natural resource base; but empirical testing has provided confusing evidence. To clarify the discourse, Thomas Reardon and Steven Vosti proposed the concept of investment poverty. It identifies the potentially large group of households that are not poor by traditional welfare poverty measures, but that are too poor in that their surplus above the welfare poverty line is too small to allow them to make investments for the conservation of their natural resource base. They also emphasize the role of assets in generating welfare for rural households, thus affecting production and investment decisions. This research pursues two questions: How can investment poverty be conceptualised and measured? And, which are the welfare-enhancing effects of household assets?

Empirical data is collected in the province of Herrera of the Republic of Panama. 402 farmer households are interviewed with a structured questionnaire, using a non-probabilistic quota sampling frame. Household welfare is measured with household consumption and income aggregates, and traditional welfare poverty lines are calculated. This allows to divide the studied households by their per capita consumption aggregates into extremely poor (15,4% of households), moderately poor (33,6%) and non-poor (51,0 %). Comparisons reveal that welfare poverty groups differ significantly not only in asset endowments, income and investment strategies, but also in household structure and in their access to infrastructure and extension services.

Although clearly defined in theory, the empirical measurement of investment poverty is challenging. In Herrera, pure conservation investments are not found and welfare is not associated with the sustainability of land use. Hence, the investment poverty line is studied as a level of welfare below which household welfare ceases to be linearly associated with investments that have both productive and conservation aims. Such an investment poverty line, located roughly at twice the annual price of a minimum nutrition, is surpassed by 42,3% of studied households. Investment non-poor households are characterized by both spouses working, high level of social involvement, high farm productivity, growing input-intensive crops, and naturally by having accumulated more land, cattle, vehicles, education, and better housing than the investment poor households.

The contribution of assets to household welfare is explored by hypothesizing mechanisms through which land, livestock, education and labour force could enhance welfare, and under which conditions. Results of multinomial logistic regressions using the elaboration method reveal that among extremely and moderately poor households, education is the only studied asset contributing to higher welfare levels. Among the moderately and non-poor households, on the other hand, education, land and labour force contribute to higher welfare through several mechanisms, but depending on conditioning variables, such as input use. There is no automation by which assets generate welfare flows towards households. And, asset accumulation strategies change as welfare increases; the mechanisms are different among the poorest from among the richest households.

The relationship between poverty and the environment remains problematic. Overcoming the investment poverty line can be seen as the necessary price of overcoming poverty and accumulating assets in the long run, but under current circumstances it is way out of reach for the poorest Herreran households. Even beyond the investment poverty line environmental sustainability is not necessarily achieved, if cattle herds increase and pastures spread.