Scholars advocate that participating in the market chain increases household food security, but the empirical evidence is scarce. We will shed light on this issue by applying a hybrid empirical approach - which combines machine learning algorithms and causal inference methods - on a unique cross-country household sample, the IFAD10 database. Specifically, we will test for the causal linkages between market positioning, resilience, and vulnerability. We expect that the results will confirm that households positioned downstream to the market chain are more resilient to shocks and, in turn, experience better food security levels (i.e., they are less vulnerable to food insecurity). A mediation analysis will be carried out to investigate whether participation in the market chain accounts for a significant share of the positive effects of household resilience on food security. These findings would suggest that policymakers should prioritize interventions to improve market positioning other than participation as key means to boost household resilience and, in turn, decrease welfare fluctuations and vulnerability to food insecurity.
One of the leading forces behind today¿s economic development in low-income countries is crop commercialization. Agricultural commercialization enhances efficiency and the gains from trade, leading to economic growth and welfare improvement. Furthermore, the integration of smallholder farmers into traditional markets is supposed to have strongly pro-poor outcomes, thanks to a virtuous cycle of efficiency, which increases household income, consumption, food security, and nutritional outcomes (Montalbano et al., 2018). Nevertheless, on the other hand, participation in the market chain may be less beneficial to the food security levels of the poorest and most vulnerable groups, who are often unable to catch the grain from increased market orientation (Bouis & Haddad, 1990; von Braun et al., 1991; Abbi et al., 1991; Kennedy & Cogill, 1987; Popkin, 1980).
Indeed, when facing global markets, farmers may become more vulnerable to external shocks for reasons like risk-aversion to price changes and bargaining power (Bellemare et al., 2013). Yet, factors like market power, marketing costs, government intervention, and asymmetric information limit efficient spatial and vertical price transmission (Meyer & Cramon-Taubadel, 2004). Hence, crop markets tend to be oligopolistic in developing areas, limiting, in turn, small farmers welfare (Sexton, 2013; Muratori, 2016; Swinnen & Vandeplas, 2012; Swinnen & Vandeplas, 2014; Kikuchi et al., 2016; Fa¿kowski, 2010; Osborne, 2005). Furthermore, in developing contexts market participation is limited to lower value activities (African Development Bank et al., 2014), constraining farmers positioning to backward stages in the market value chain.
In this framework, the causal relationship between market participation and food security is not straightforward. In the presence of incomplete or missing markets, as is generally the case in developing contexts, farming households perceive food self-sufficiency also as a source of protection against price risks in food markets (Fafchamps, 1992; de Janvry & Sadoulet, 2006). In this respect, food production acquires an insurance value which is added to its regular contribution to income, and the supposed benefits of agriculture commercialization on food security may be offset by transaction costs, risk aversion and/or weak resilience (Montalbano et al., 2018). On the other hand, in the market option, prices for small producers depend on their positionality within the farmer-producing class. As such, small farmers tend to rush post-harvest production to sell their crop to the market when the price volatility on the market goes down the value chain, pushing small producers into a vicious cycle of low productivity, low quality, and prices (Purcell, 2018).
Empirical evidence on this issue is weak. Most scholars have carried out quantitative assessments based on single-country case studies and context-specific frameworks about market participation rather than market positioning. Broader empirical assessments are hampered with difficulties both in terms of data and methodology: the market choice sets its bases on several factors influencing both households decision-making process and their food security (Key et al., 2000) whereas alternative commercialization options have mixed impacts on food security (Swinnen & Vandeplas, 2014; Wohlgenant, 2001; Weldegebriel, 2004; McCorriston et al., 2001; Wang et al., 2006).
The novelty of this work stands on three main facts: first, unlike previous literature, we assess food security by looking at the volatility of its inner stochastic components rather than at its overall variation in levels; secondly, we use a hybrid empirical approach combining machine learning algorithms and an instrumental variable approach to adapt a standard vulnerability measure (Chaudhuri in 2001 and 2003) to the analysis of food security; thirdly, by exploiting a unique dataset of household surveys available for eight countries in the world, we provide a replicable setting to test the complex interactions among market positioning, resilience, and volatility of food security.