As one of our Development Impact bloggers got married this past weekend, I took some time to look at the recent literature on marriage and what happens afterwards.
A nice paper by Carolina Castilla in the May AER gives us some insight into trust and reciprocity between spouses in India (gated version here, longer, more detailed working paper here). The premise is nice: where better to find trust than a sample of married couples? Castilla works with 150 married couples in Uttarakhand, India to check this out. She plays a Berg, Dickhaut and McCabe trust game with them.
The game goes like this. The two spouses are separated. One (randomly chosen) gets 75 rupees and chooses how much to transfer to her/his spouse (this is the sender). Whatever goes to her/his partner gets tripled. The receiver then chooses how much to give back. So, of course, to maximize household revenue, the sender should send all of her/his money.
What happens? Not so much maximizing – only 3 percent of households send the entire endowment to their spouse. On average, the sending spouse is remitting 57 percent of her/his endowment, and the receivers are given 53.7 percent back. So, these married couples are leaving significant money on the table.
So what’s going on? Well these numbers look a lot like 50 percent, so maybe spouses are applying a 50/50 split. Nope, these shares are actually statistically different from 50. So maybe they don’t like inequality. Nope again – the sending and receiving earnings are statistically different from each other. Castilla pursues things further – she runs the receiving spouses through a dictator game (they get an additional 75 rupees and have to decide how much to give to their spouse, who gets no say in the matter). Spouses give significantly less in the dictator game (49.5 percent of their endowment). As Castilla concludes: “this suggests receivers’ motivation to return money is partially driven by altruism, and partially by reciprocity.” So apparently that’s the secret of marriage.
How do married people compare to others? Castilla compares her results to other experiments done with general population type folks in developing country contexts. Her married folks are giving more (both on the sender and receiver sides) than folks in Paraguay and Zimbabwe. Her conclusion: “spouses are both more trusting and more reciprocal on average.”
The working paper version gives us some insight into gender differences. Remember that either the husband or the wife is randomly assigned to be the sender/receiver – this lets Castilla tease apart who gives more in which role. And the answer? Women give 6 percentage points less on average (significant at 10 percent) when they are the sender. When they are the receiver, they return 10 percentage points less on average.
Castilla looks at expenditure patterns prior to the game and the ratio of female to household expenditure to see if these can help shed some light on why this might be the case. On the sender side, it turns out that higher spending on tobacco (which is almost entirely consumer by men) is associated with lower amounts given by women, while men give more as senders when their wives are paying for school expenses (the sample is fairly evenly divided on whether males/females are paying for school fees). On the receiver side, the potential factors are less clear – here again tobacco seems to play a role (now on the extensive not intensive margin) as it is associated with women sending back less. Men in households that don’t spend that much on tobacco (as a share of total expenditure) send more back to their wives, but in households where the share is larger, they actually send back less.
So, when we think about interventions like a cash transfer program, these kind of results should give us pause – and send us looking more into the household to understand what are the dynamics, what is driving them, and how that will affect the outcomes the cash transfer is intended to change. And, in the meantime, here’s to the married couples and their higher levels of trust and reciprocity.
A nice paper by Carolina Castilla in the May AER gives us some insight into trust and reciprocity between spouses in India (gated version here, longer, more detailed working paper here). The premise is nice: where better to find trust than a sample of married couples? Castilla works with 150 married couples in Uttarakhand, India to check this out. She plays a Berg, Dickhaut and McCabe trust game with them.
The game goes like this. The two spouses are separated. One (randomly chosen) gets 75 rupees and chooses how much to transfer to her/his spouse (this is the sender). Whatever goes to her/his partner gets tripled. The receiver then chooses how much to give back. So, of course, to maximize household revenue, the sender should send all of her/his money.
What happens? Not so much maximizing – only 3 percent of households send the entire endowment to their spouse. On average, the sending spouse is remitting 57 percent of her/his endowment, and the receivers are given 53.7 percent back. So, these married couples are leaving significant money on the table.
So what’s going on? Well these numbers look a lot like 50 percent, so maybe spouses are applying a 50/50 split. Nope, these shares are actually statistically different from 50. So maybe they don’t like inequality. Nope again – the sending and receiving earnings are statistically different from each other. Castilla pursues things further – she runs the receiving spouses through a dictator game (they get an additional 75 rupees and have to decide how much to give to their spouse, who gets no say in the matter). Spouses give significantly less in the dictator game (49.5 percent of their endowment). As Castilla concludes: “this suggests receivers’ motivation to return money is partially driven by altruism, and partially by reciprocity.” So apparently that’s the secret of marriage.
How do married people compare to others? Castilla compares her results to other experiments done with general population type folks in developing country contexts. Her married folks are giving more (both on the sender and receiver sides) than folks in Paraguay and Zimbabwe. Her conclusion: “spouses are both more trusting and more reciprocal on average.”
The working paper version gives us some insight into gender differences. Remember that either the husband or the wife is randomly assigned to be the sender/receiver – this lets Castilla tease apart who gives more in which role. And the answer? Women give 6 percentage points less on average (significant at 10 percent) when they are the sender. When they are the receiver, they return 10 percentage points less on average.
Castilla looks at expenditure patterns prior to the game and the ratio of female to household expenditure to see if these can help shed some light on why this might be the case. On the sender side, it turns out that higher spending on tobacco (which is almost entirely consumer by men) is associated with lower amounts given by women, while men give more as senders when their wives are paying for school expenses (the sample is fairly evenly divided on whether males/females are paying for school fees). On the receiver side, the potential factors are less clear – here again tobacco seems to play a role (now on the extensive not intensive margin) as it is associated with women sending back less. Men in households that don’t spend that much on tobacco (as a share of total expenditure) send more back to their wives, but in households where the share is larger, they actually send back less.
So, when we think about interventions like a cash transfer program, these kind of results should give us pause – and send us looking more into the household to understand what are the dynamics, what is driving them, and how that will affect the outcomes the cash transfer is intended to change. And, in the meantime, here’s to the married couples and their higher levels of trust and reciprocity.
Join the Conversation