A couple of weeks ago, I blogged about a new approach to measuring within household decision making. Continuing in that vein, I was recently reading a paper (ungated version here) by Almas, Armand, Attanasio, and Carneiro which offers a really neat way to look at bargaining power within the household.
But before I get into the weeds on measurement, let’s take a look at the headline result. Many cash transfer programs choose to give the transfer to women. The logic goes she will spend more on the kids (there is some evidence for this, but it’s a topic for another post). And sometime folks make the argument that the cash transfer will empower women to boot. Almas and co. offer some nice insight into this question using results from an experiment in North Macedonia where conditional cash transfers (for kids staying in school) were randomly allocated to either the head of household (usually a man) or a woman.
Almas and co. look at impacts on female empowerment in three dimensions. First, they examine the more standard measures of household decision making (e.g. who makes decisions regarding household finances), where they put a bunch of these questions into a household decision making (HDM) index. Second, they construct a domestic violence (DV) index, which is based on questions about the acceptability (not actual perpetration) of domestic violence. Finally, they bring their new measure to the table, a lab in the field experiment done with women which elicits their willingness to pay to avoid having a (smaller than the CCT) transfer go to their husband. This, they argue, is picking up bargaining power within the household – women who are willing to pay less than others have more say in the household. There is more on how this is constructed and what it means below, but first let’s turn to the results.
Almas and co. start off by looking at the impact on the more standard measures. Here, what they find is that there is some suggestion that a woman getting the CCT versus a man is associated with more empowerment as measured through the decision making or violence indices, but nothing is statistically significant. However, when they use the lab in the field experiment, they find stronger, statistically significant results. A woman CCT recipient is willing to pay much less to keep the transfer away from the husband. Indeed, Almas and co. go a step further and use the randomly assigned CCT as an instrument for the woman’s share of household income and then look at how this lines up with the lab in the field results. And with this they find that shifting all of the income in a household from the male to female partner would lead to a decrease in the willingness to pay of 28 percent. So, bottom line: using this new measure, there is some evidence that the CCT has increased women’s bargaining power and not by tiny amounts.
So the nifty thing about this paper is this new method of measuring power within the household. Let’s go on and dive in. As indicated above, they do this lab experiment with women only – both direct CCT recipients and those in which the head of the household got the CCT. The starting point for the experiment is the unitary model. The unitary model is, as the name implies, when the household acts as a unit. This can be because they are seamlessly run by a dictator or because all preferences perfectly align. In this world, a woman should not be willing to forsake any amount of cash to keep the transfer from going to her husband versus her. In non-unitary households, bargaining power (e.g. stemming from income shares or outside options) matters for allocation.
So the lab experiment is designed to tease this out. To start with, the participant is asked to choose between an amount A she gets or the amount B that will go to her husband. And for round 1, A is less than B. If she chooses to keep it (A) then the amount of A in the next round is reduced by 75%. If she chooses A again, it keeps on going until it reaches 20 Macedonian Denars (MKD), which Almas and co. chose as a stopping value. Now, what if in round 1 she chooses B, i.e. to give it to the husband. Then the amount A that she could keep goes up by 50 MKD. If she persists, this goes up by 50 MKD (1 MKD =$0.02 at the time of the project) for 3 rounds maximum and then stops.
So far so good, but now, what if a woman switches from choosing herself vs. the husband over the course of the rounds. Things get more complicated, and now the amount offered to her next is the average of the present round and the amount from the previous round where the participant made a different decision than the current one. So, basically, the amount can go up or down and is designed to tease out a very much more continuous model of people’s willingness to pay to keep their spouse from getting control. This is a distinct contrast from the more discrete cases of who makes decision (her alone, him alone, together) and even some other take-it-or-leave-it lab in the field games of this family.
Finally, it is important to keep in mind that folks are playing with real money here – with cash given to either the woman or her husband -- albeit with random round choices so he can’t totally figure out what she did – but there is full public information about earnings.
So what do people do? The average woman is willing to give up 19 percent of the maximum total amount in order to keep the cash from her husband. And, there are 7 percent of the women who are willing to give up 90 percent or more. On the flip side, there are 35 percent of women who have a zero or negative willingness to pay. Almas and co. are well steeped in the models of the household and they note that “the interpretation of a zero or negative WTP is not straightforward.” Indeed. So its consistent with a household where she rules everything. But it could also be a case where her preferences have zero weight. Or it could be that the experiment amounts are just not enough to shift her bargaining power in a meaningful way. Or maybe she just doesn’t want to decide. Understanding this in more depth – which may harken back to using some of the tools I talked about last time, plus some other things, would be an interesting area to probe in further research. And thus, we have another new tool which opens up a richer set of questions in understanding the black box of the household. And, lest one despair that this is economics at its navel gazing best, remember this is also a paper showing how this set of household dynamics not only matter for policy effectiveness, but are also impacted by policy choices.
But before I get into the weeds on measurement, let’s take a look at the headline result. Many cash transfer programs choose to give the transfer to women. The logic goes she will spend more on the kids (there is some evidence for this, but it’s a topic for another post). And sometime folks make the argument that the cash transfer will empower women to boot. Almas and co. offer some nice insight into this question using results from an experiment in North Macedonia where conditional cash transfers (for kids staying in school) were randomly allocated to either the head of household (usually a man) or a woman.
Almas and co. look at impacts on female empowerment in three dimensions. First, they examine the more standard measures of household decision making (e.g. who makes decisions regarding household finances), where they put a bunch of these questions into a household decision making (HDM) index. Second, they construct a domestic violence (DV) index, which is based on questions about the acceptability (not actual perpetration) of domestic violence. Finally, they bring their new measure to the table, a lab in the field experiment done with women which elicits their willingness to pay to avoid having a (smaller than the CCT) transfer go to their husband. This, they argue, is picking up bargaining power within the household – women who are willing to pay less than others have more say in the household. There is more on how this is constructed and what it means below, but first let’s turn to the results.
Almas and co. start off by looking at the impact on the more standard measures. Here, what they find is that there is some suggestion that a woman getting the CCT versus a man is associated with more empowerment as measured through the decision making or violence indices, but nothing is statistically significant. However, when they use the lab in the field experiment, they find stronger, statistically significant results. A woman CCT recipient is willing to pay much less to keep the transfer away from the husband. Indeed, Almas and co. go a step further and use the randomly assigned CCT as an instrument for the woman’s share of household income and then look at how this lines up with the lab in the field results. And with this they find that shifting all of the income in a household from the male to female partner would lead to a decrease in the willingness to pay of 28 percent. So, bottom line: using this new measure, there is some evidence that the CCT has increased women’s bargaining power and not by tiny amounts.
So the nifty thing about this paper is this new method of measuring power within the household. Let’s go on and dive in. As indicated above, they do this lab experiment with women only – both direct CCT recipients and those in which the head of the household got the CCT. The starting point for the experiment is the unitary model. The unitary model is, as the name implies, when the household acts as a unit. This can be because they are seamlessly run by a dictator or because all preferences perfectly align. In this world, a woman should not be willing to forsake any amount of cash to keep the transfer from going to her husband versus her. In non-unitary households, bargaining power (e.g. stemming from income shares or outside options) matters for allocation.
So the lab experiment is designed to tease this out. To start with, the participant is asked to choose between an amount A she gets or the amount B that will go to her husband. And for round 1, A is less than B. If she chooses to keep it (A) then the amount of A in the next round is reduced by 75%. If she chooses A again, it keeps on going until it reaches 20 Macedonian Denars (MKD), which Almas and co. chose as a stopping value. Now, what if in round 1 she chooses B, i.e. to give it to the husband. Then the amount A that she could keep goes up by 50 MKD. If she persists, this goes up by 50 MKD (1 MKD =$0.02 at the time of the project) for 3 rounds maximum and then stops.
So far so good, but now, what if a woman switches from choosing herself vs. the husband over the course of the rounds. Things get more complicated, and now the amount offered to her next is the average of the present round and the amount from the previous round where the participant made a different decision than the current one. So, basically, the amount can go up or down and is designed to tease out a very much more continuous model of people’s willingness to pay to keep their spouse from getting control. This is a distinct contrast from the more discrete cases of who makes decision (her alone, him alone, together) and even some other take-it-or-leave-it lab in the field games of this family.
Finally, it is important to keep in mind that folks are playing with real money here – with cash given to either the woman or her husband -- albeit with random round choices so he can’t totally figure out what she did – but there is full public information about earnings.
So what do people do? The average woman is willing to give up 19 percent of the maximum total amount in order to keep the cash from her husband. And, there are 7 percent of the women who are willing to give up 90 percent or more. On the flip side, there are 35 percent of women who have a zero or negative willingness to pay. Almas and co. are well steeped in the models of the household and they note that “the interpretation of a zero or negative WTP is not straightforward.” Indeed. So its consistent with a household where she rules everything. But it could also be a case where her preferences have zero weight. Or it could be that the experiment amounts are just not enough to shift her bargaining power in a meaningful way. Or maybe she just doesn’t want to decide. Understanding this in more depth – which may harken back to using some of the tools I talked about last time, plus some other things, would be an interesting area to probe in further research. And thus, we have another new tool which opens up a richer set of questions in understanding the black box of the household. And, lest one despair that this is economics at its navel gazing best, remember this is also a paper showing how this set of household dynamics not only matter for policy effectiveness, but are also impacted by policy choices.
Join the Conversation