Intra-household bargaining over household technology adoption

Abstract

We examine the barriers to adoption of improved cook stoves (ICSs) in rural India, using a large, nationally representative dataset. We develop a collective household model to derive testable hypotheses about whether women’s intra-household influence, together with their relatively strong marginal preference for ICSs, affects adoption. Using a joint adoption-influence econometric model, we find compelling evidence that women’s influence over intra-household decisions significantly increases adoption. We further distinguish between alternative sources of women’s influence, and argue that our distinction has potential implications for ICS dissemination policies. We find that while there is significant variation in women’s influence across rural India due to cultural and other sociological factors, the effect of intra-household influence on adoption has a significant bargaining power component. Our results suggest that ICS programs may be able to increase adoption by marketing stoves in ways that empower women.

Appendix

The likelhood function for our model is can be derived by rewriting the adoption equation (17) as $Prob(v_{i}=1)=\Phi ( {{{\varvec{\alpha }}}}^{\prime }{{\mathbf{X}}}_{{{\mathbf{i}}}})$ and the influence equation (19) as $Prob(\lambda _{i}=j)=\Phi (\mu _{j}- {{\mathbf{b}}}^{\prime }{{\mathbf{Z}}}_{i})-\Phi (\mu _{j-1}-{{\mathbf{b}}}^{\prime }{{\mathbf{Z}}}_{i}).$ The explanatory variable matrices X and Z contain common elements but are not identical to allow identification. The likelihood function is built up from the joint density of the random variables $v_{i}$ and $\lambda _{i}:$

$$\begin{aligned} L&= {} \underset{v_{i}=0,\lambda _{i}=0}{\Pi }\Omega \left( -{{{\varvec{\alpha }}}}^{ \prime }{{\mathbf{X}}}_{i},-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}} _{i};-\rho \right) \\&\quad\times \underset{v_{i}=0,\lambda _{i}=1}{\Pi }\Omega \left( -{{\varvec{\alpha }}}^{{{\mathbf{\prime }}}}{{\mathbf{X}}}_{i},\mu _{1}-{{\varvec{\beta }}}^{\prime}{{\mathbf{Z}}}_{i};-\rho \right) -\Omega \left( -{{{\varvec{\alpha }}}}^{\prime} {{\mathbf{X}}}_{i},-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}}_{i};-\rho \right) \\&\quad\times \underset{v_{i}=0,\lambda _{i}=2}{\Pi }\Omega \left( -{{{\varvec{\alpha }}}}^{ \prime }{{\mathbf{X}}}_{i},\mu _{2}-{{\varvec{\beta }}}^{\prime}{{\mathbf{Z}}}_{i};-\rho \right) -\Omega \left( -{{{\varvec{\alpha }}}}^{\prime }{{\mathbf{X}}} _{i},\mu _{1}-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}}_{i};-\rho \right) \\&\quad\times \underset{v_{i}=0,\lambda _{i}=3}{\Pi }\Omega \left( -{{{\varvec{\alpha }}}}^{ \prime }{{\mathbf{X}}}_{i},\infty -{{\varvec{\beta }}}^{\prime } {{\mathbf{Z}}}_{i};-\rho \right) -\Omega \left( -{{{\varvec{\alpha }}}}^{\prime }{{\mathbf{X}}} _{i},\mu _{2}-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}}_{i};-\rho \right) \underset{v_{i}=1,\lambda _{i}=0}{\Pi }\Omega \left( {{{\varvec{\alpha }}}}^{ \prime }{{\mathbf{X}}}_{i},-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}} _{i};\rho \right) \\&\quad\times \underset{v_{i}=1,\lambda _{i}=1}{\Pi }\Omega \left( {{\varvec{\alpha }}}^{\prime }{{\mathbf{X}}}_{i},\mu _{1}-{{\varvec{\beta }}}^{ \prime }{{\mathbf{Z}}}_{i};\rho \right) -\Omega \left( {{{\varvec{\alpha }}}}^{\prime } {{\mathbf{X}}}_{i},-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}}_{i};\rho \right) \\&\quad\times \underset{v_{i}=1,\lambda _{i}=3}{\Pi }\Omega \left( {{{\varvec{\alpha }}}}^{ \prime }{{\mathbf{X}}}_{i},\infty -{{\varvec{\beta }}}^{\prime } {{\mathbf{Z}}}_{i};\rho \right) -\Omega \left( {{{\varvec{\alpha }}}}^{\prime }{{\mathbf{X}}} _{i},\mu _{2}-{{\varvec{\beta }}}^{\prime }{{\mathbf{Z}}}_{i};-\rho \right) \end{aligned}$$

where $\Omega$ is the bivariate normal distribution.