What influence do empowered women have? Land and the reality of women’s relative power in Peru

Abstract

Using primary data from rural Peru and a novel econometric framework, this paper evaluates the effects of gendered land inheritance on women’s relative power. We find limited evidence that increasing a woman’s landholdings increases her power; while an increase in landholdings of men in her household decreases her power. A coincident increase in land for both has significant empowering effects. Thus, gender policies need to go beyond interventions that exclusively target women—empowering women also requires empowering men in specific ways. We also provide fresh insight into using women’s empowerment for economic development by distinguishing power from the influence that women possess over specific household decisions because of their power. We characterize each influence (e.g., household expenditures) by two statistics—threshold (the level of empowerment required to awaken the influence) and sensitivity (the response of the influence to marginal increases in power). We find that different thresholds of power are required for women to influence different household decisions. Credit and land-rental require much greater power and are less sensitive than livestock, household-goods, or management of agricultural land. Distinguishing power from influence adds great value to the survey questions now used to create indices of women’s power.

Appendix 1

We consider two decision-makers, a man (m) and a woman (w), who have different preferences over two goods q^j, for j = 1, 2. For the sake of exposition, we assume q¹ is a good whose consumption is directly aligned with economic development. For instance, q¹ can be expenditures on children’s education which increases the human capital of the household. Alternatively, q¹ can be thought of directly as money secured through a loan or savings to buy a new technology for future production on the household farm (farm production and consumption decisions are assumed to be separable, for simplicity). In contrast, q² denotes a household durable (e.g., a vehicle) whose consumption is assumed to be less aligned with development outcomes. Decision-making in the household is assumed to be cooperative and results in pareto efficient outcomes. Thus, the household utility function can be written as a weighted sum of the individual utilities. The household’s optimization problem is:

$$\begin{array}{*{20}{c}} {} \\ {Max\,U} \\ {q^1,q^2} \end{array} = \lambda (z)\left[ {u_f(q^1,q^2)} \right] + (1 - \lambda (z))\left[ {u_m(q^1,q^2)} \right]{,}$$

$$s.t:q^1 + q^2 = I{,}$$

where I is the household budget. The function λ(.) ∈ 0, 1], assumed continuously differentiable in its arguments, represents the woman’s weight in the household utility function. In the literature it is customary to interpret λ as the woman’s bargaining power or empowerment (e.g., Reggio 2011; Doss 2013). λ(z) is identified by specifying the “distribution” factors, z, to include variables that affect household utility through bargaining power but not directly. Based on this framework, scholars have argued that women’s landholdings can increase women’s bargaining power and, thereby, their involvement in household economic decisions which, in turn, can lead to increases in outcomes such as q¹ consumption that help economic development.

Totally differentiating the FOCs for the household’s problem above, we obtain an expression for the effect of a land transfer on q¹:

$$\begin{array}{l}\left[ {\begin{array}{*{20}{c}} {\lambda (z)u_{11}^f + (1 - \lambda (z))u_{11}^m} & {\lambda (z)u_{12}^f + (1 - \lambda (z))u_{12}^m} & { - 1} \\ {\lambda (z)u_{21}^f + (1 - \lambda (z))u_{21}^m} & {\lambda (z)u_{22}^f + (1 - \lambda (z))u_{22}^m} & { - 1} \\ { - 1} & { - 1} & 0 \end{array}} \right]\left[ {\begin{array}{*{20}{c}} {\frac{{dq^1}}{{dz}}} \\ {\frac{{dq^2}}{{dz}}} \\ {\frac{{d\mu }}{{dz}}} \end{array}} \right]\\ = \left[ {\begin{array}{*{20}{c}} { - \left( {u_1^f - u_1^m} \right)\lambda \prime (z)} \\ { - \left( {u_2^f - u_2^m} \right)\lambda \prime (z)} \\ 0 \end{array}} \right]\end{array}{,}$$

where μ is the marginal utility of household income. Using Cramer’s rule, the derivative dq¹/dz is the ratio of two determinants $\frac{{|Ac|}}{{|A|}}$ where Ac is the bordered Hessian with the first column replaced by the vector on the RHS of (1). The numerator |Ac| is can be simplified as:

$$dq^1/dz = \frac{1}{{|A|}}\lambda \prime (z)\left[ {\Delta U} \right]{,}$$

where $\Delta U = (u_1^f - u_1^m) - (u_2^f - u_2^m)$ denotes the difference in the woman’s marginal utilities, relative to the man, corresponding to the consumption of q¹ and q², respectively. The sign of dq¹/dz depends on the sign of the numerator: dq¹/dz > 0 if λ′(z) > 0 and ΔU > 0. In words, dq¹/dz > 0 if z empowers women and if women, relative to men, have greater preferences for consuming q¹ and if their relative preference for q¹ is greater than that for q². Note that by the second order conditions for a maximum, the bordered Hessian |A| > 0.