Many anti-poverty programs are ‘bundled’, in that they have multiple
active program components. We shed light on the interplay between two
important types of programs by randomly assigning rural households in
Mozambique to subsidies for modern agricultural inputs, formal savings
facilitation programs (either a ‘basic’ or a ‘matched’ savings program), or
both subsidy and savings programs. We take household consumption per
capita as the summary welfare measure of interest. Impacts on consumption are very similar, from an economic and statistical standpoint, for all
treatment combinations (subsidy alone, either savings program alone, or
combinations of subsidies and savings), amounting to 7.7 percent on average. Theoretically, such a pattern can emerge if access to input subsidies
affects the uses to which formal savings are put: savings may be used
for mainly self-insurance (buffer stocks) when subsidies are available, but
otherwise may serve both investment and insurance purposes. Auxiliary
predictions of the theory are borne out in analyses of treatment effects on
consumption variance and on the responsiveness of consumption to shocks.
A key gain from combining subsidy and savings interventions appears to
be an improved ability to cope with risks that arise from exploitation of
high-return economic opportunities.