 So what I'm going to talk about today is household decision-making and technological adoption. So I'm going to take a narrow focus and really think about this decision at the household level to adopt a new technology. When Alain kindly invited me to this session, he mentioned a previous project that we did in part with a wider funding on technology adoption, I mean related to technology adoption, where the question we asked was what are the effects of technological adoption or land pressure on household structure, household composition, and the organization of agriculture within the household. And the conclusion or one of the punchline of this project was that when new technology become available, the efficiency losses related to these collective farming in very large West African households become especially important, leading to individualization of production, both within the household where you see a rising importance of what we call individual plots where subunits of the household would cultivate their own plots, or a growing splitting of large households into more nuclear units. What I will talk about today is something quite different where it still has to do with household but we will take the household composition as given. And what we ask is really how within the household individual members interact to decide about technology adoption. And this is a joint project with Jean-Marie Ballant, Ludovique Bequet and Clarice Ronas at the University of Namur. And I have to say it's very recent, so we were still collecting data last month, so this is a preliminary work, but I think I still think it's there are some thought-provoking ideas that this project brings. So there's a large literature on what are the households, what are the determinants of technology adoption, and risk preference play a large role in this matter. The idea is that as a new technology brings some new uncertainty, there are risks associated with adoption, and it's typically the more risk-tolerant household who are the first to adopt. But when we talk about risk preference in this literature, it's typically the preferences of the household head. So in the empirical literature, it's always what has been measured, nearly always. And this is a bit surprising if you compare this with the level of sophistication of the literature on inter-household relationship for technology adoption. So for example, there's a lot of work using network analysis to identify who should extension, agricultural extension, effort targets within the village, who should be the household that you first convince to adopt so that the adoption spreads. So the focus of the literature has been on who to target in the village to boost adoption rather than who to target in the household or how to think about these household decision-making processes. So what we do more specifically, we investigate with the help of some experiments, how couples take agricultural decisions when it involves risk return trade-off. And we do this in the Philippines, in Mindanao in particular, which is one of the main corn producing regions. GMOs were introduced in the early 2000s that have been relatively widely adopted. Now, a lot of farmers produce what they call sige-sige, which is glyphosate tolerant corn, but without the terminator gene. So it's kind of a pirate GMO, which is apparently the result of a crossing between a local white corn and some stolen seed from Monsanto. Corn production is extremely risky, so there are droughts, floods, storms, droughts, landslides, typhoons. So it's really an area where farmers have extremely risky production. It's a rain-fed agriculture. Women are typically in charge of the management of the household budget. So they decide about expenditure allocation, but men are in charge of the agricultural decision. So we surveyed household and we, so we collected detailed information about adoption, this adoption, and we played live in the field games individually with spouses and in couple. And I will focus here on the results of two games that were framed as choices over corn varieties that involve the risk return trade-off. So specifically, we played a kind of classic game of eliciting to elicit risk aversion, and what we did, we tried to elicit risk aversion of individuals and of the couples. So we had people choose over these various corn varieties. So you see that there's, as you go down the list, there's an increasing expected return from adopting these corn, but an increasing risk. Also, this helps us elicit individual risk aversion, and then we asked people to in couple choose a variety, and all this was incentivized. So we elicit a couple risk aversion. Then we tried to get at this measure of the trust for taking appropriate investment decisions. So what was the idea here is people would have the possibility to invest in a risky, in a technology, in a risky technology. So basically that the choice between two corn varieties with a risk return trade-off. But they did not take the decision themselves. Basically, they had to send money to the spouse, and the spouse would take the decision individually. And they could choose how much money to send. Regardless of the amount sent, everything was then put in the common pot and divided by two. So every spouse got half of the total amount. And so the choice were really, were quite limited. So people had an endowment of $2,000, and they could choose to send everything $1,000 or $0,000. If an individual would take this decision on its own, it would always, even if it's very risk averse, invest everything because you see there's a risk less option that has a positive return. But what we see is that when they have to trust the spouse to invest, they actually send, I mean, the majority do not send everything. So the first thing I'm going to show you is that this couple risk aversion is related to real life behavior. So this is, this is very preliminary and it's, it's a simple correlation. But what we see is that past investment behavior in GMO, in particular, is correlated with this level of couple risk aversion that really sits more than with the husband's level of risk aversion, which actually doesn't seem to explain much in this context. So this is what we show here. So either we look at whether the household has adopted GMO or whether the household has ever used this type of loans that is a full package to adopt GMO. And we see that indeed, it's relative, it's, it correlates with the couple's level of risk aversion, not really much with the husband's risk aversion. And it's also correlated with the wife's risk aversion. What we do next is also look at, use this a measure of bargaining power that we also elicit experimentally to see in which type of couple this seems to play a role. And so the first thing we see is that it's in couple where women appear to have relatively more bargaining power that the couple's level of risk aversion matters the most. So because I told you that husbands are the primary decision maker for agriculture, you, I mean, the first idea would be to say, well, this makes sense. So when the woman has Monse, maybe indeed her, the couple's preference is going to matter more. But it's not that simple because we have also a result that's hard to interpret, which is that it's in the same couple that the husband level of risk aversion also seem to have more, more greater importance in this decision. So this is, I don't want to insist, we have some ideas about some delegation within the household that may explain this, but it's very preliminary still. Okay. So the next step once we see that indeed this couple risk aversion seems to be correlated with real life behavior is to look at the determinants of the couple risk aversion. And in a recent review of the literature on this topic, Monroe mentioned that there's no evidence today that individual risk preferences are aligned with what couple would choose. And no particular pattern, and there's no particular pattern in the way preferences are resolved. And you also write in the same review that the couple is not a simple is not simply a weighted average of two individuals. It has its own decision making procedure. And in a sense, the result we get experimentally really is in line with this conclusion. So first we find that couples are more risk averse than individuals. The couple level of risk aversion seems closer to the risk aversion of the risk, the more risk averse spouse, and the husband seems to have more weight in this decision. So we just show you very descriptively evidence in that direction. So first, in 44% of the couple, which I think is a lot given the menu of choices, husband and wife make the same variety choice in the risk aversion game. When they do not, the more often choose a variety closer to the husband choice. And when again, when choices differ, 67 of couples jointly choose a variety closer to the choice of the most risk averse of both spouse. So here's just descriptive statistics again. I mean, if we look at the LCT coefficient of relative risk aversion, you see that for couple, it's actually larger than for husbands or for spouse on average. And it's actually statistically different. So if we try to look at the determinants of this couple risk aversion, it seems that the husband's coefficient of relative risk aversion is a stronger determinant. So if you probably don't see the coefficient, coefficient on the husband level of risk aversion is constantly twice as large as the coefficient on the woman risk aversion. Yes, and I'm going to move on. Okay, so what do we learn from this second type of game? There's large levels of underinvestment when a couple do not communicate and have to trust the other spouse to make the decision. So as I told you, we, an individual would always invest the 2000. Yet when they delegate the choice to the spouse, many, many only invest a thousand, despite a large alignment of individual choices. So if you see how husbands and wives play individually, I'm going to show it to you, they do the same. Yet when you ask them to send money for the other one to do it, they send actually less money than they should. And despite the fact that everything is divided by two at the end. So what's also very interesting, individuals, and this is also verified in the other game, appear to systematically overestimate the risk taking behavior of the spouse. So they play very much in line, husband and wife. Yet when you ask them, what would the other spouse do, the much more often than the other spouse actually do, they think the spouse will take the the risk, risky option. So it's as if there's some, they don't know exactly what the other one would do. There's and there's some ambiguity regarding this house variety choice that discourages investment. They don't really know and they prefer not to take this risk. So this is, those are just the raw results. So you see, if you look at the last panel, you see that, no, no, middle panel, sorry, most individuals, three quarter individual go for the safe choice, only less than 25% choose the variety B. Yet when you ask them and husband and wife play remarkably similarly, when you ask them, what do you think the other one would do, more than 40% think the other one would actually go for the risky choice. And if you see how much they send, more than 65% actually sent only a thousand. And yeah. Okay. So what can I tell you, I need to wrap up. So if we look at this amount sent, it negatively correlates with risk aversion of the standard, which you would expect. Yet, and this is a little bit, I mean, we need to understand that better. We would also expect that it correlates with the belief you have regarding the others, the way the other one would play. The only way you can really rationalize this is if you think the, your spouse would take a risky decision, then you don't want to send everything. Yet when we look at these interaction where we don't find, we don't find a strong effect of this belief regarding the other's choice. Okay. So what, what do we take from, from this? First, the couple's risk aversion seems to better correlate with real life behavior than the household heads risk aversion, which is, as I told you, still the, at least the clear to be the primary decision maker. Joint decision of a risky project prospect and not simple averages of individual decision. We, we do find that, yes. So this, this suggests that reducing the technological adoption decision to the choice of one individual, which would be the household head is probably not appropriate. And seconds, this is more the result of the last thing I talked about. It seems that individual do not trust the spouse to take appropriate agricultural decision, even if preference seems to be relatively well aligned. And thus, thus forgoing quite profitable investment. So this suggests that targeting only the household head and trying to convince him to adopt and then hoping that the wife would somehow contribute either financially or through work, maybe a little bit problematic. And this raises the question whether extension efforts should not target couple or inform the people involved in the decision simultaneously. This is an open question. Thank you.