 My name is Ben Wood. I'm from the International Initiative for Impact Evaluation. This work is co-authored with my former dissertation advisor, Carl Nelson at the University of Illinois, and Todd Killick and Shabon Murray from the World Bank. It's nice to piggyback on a Malawi presentation, so I don't need to kind of introduce Malawi in the agricultural system because we already talked about it. Smallholders in the agricultural is very important in Malawi. This research stems from the idea of value chains in trying to increase the efficiency of land. There's been kind of recycling of an idea from 30 years ago in agricultural economics of agricultural commercialization and trying to move up the value chain. So one of the questions I was going to ask Blessings and didn't have a chance to was if the 40% of GDP includes consumption. So one of the issues with agriculture in Malawi is how do you capture the fact that a lot of the farmers are subsistence farmers and they're eating what they're growing. So one of the suggestions has been to move towards a commercialized crop which in Malawi is typically tobacco. Malawi is very well situated climatically for tobacco, but there's many places in the world that can produce tobacco. And the reason this is important is because when people move small holders that have very limited land move towards tobacco production were implicitly saying they should rely on the market to buy the food they're no longer producing on their farm. So if you're moving towards this model you're no longer a subsistence farmer. You're selling your crop and you're buying your maize from the market. And recently there has been some fluctuations as I'm sure many of you know about food prices around the world. And Malawi experienced this large drought, back to back droughts in 2001 and 2002, which I'm trying to show here. You know, kind of condensed in the poster. And the question is how does crop adoption decisions how do these decisions affect the health of children who are very sensitive to food price shocks or food price spikes. And this research has been looked at previously kind of in a correlational context with the World Bank. In 1999 they were able to show that there was an issue with or a correlation between your decision to produce tobacco and lower nutritional health outcomes for children. But all they're able to do is kind of look at it from a descriptive statistics perspective. And so we're trying to move beyond descriptives into a causal model context. And we find that adopting tobacco during or before a price spike has a negative effect on children's health. I think it's kind of a logical conclusion. And the reason it's important is because if there's domestic drought that happened in 2000 and 2001 and 2002, the maize prices are going to go up and tobacco prices aren't going to move. Because tobacco, the prices are dictated internationally. Malawi is not a price maker. They're a price taker. And so when the production goes down in maize and the price in maize goes way up the relative losses for tobacco producers are very high because their prices aren't changing and the market that we've all been telling them to rely on to buy their food at suddenly is going really crazy and the prices are a lot higher. And so we're moving into this nutrition and how do we assess nutrition of children in Malawi. We're using a height for age Z-score. So it's stunting, which is nutritionally a internationally recognized measure of children's health. And I feel comfortable with it after going into the field and seeing how they're collecting the measures in the sense that I think that the height measurements are fairly accurate and easy to transport into the field. So ideally we'd like to look at this nutritional health measure, stunting, after controlling for a number of child household and community specific effects or variables and then simply throwing tobacco adoption into an OLS and see whether or not this adoption decision has a negative effect on children's health during a food price spike. Unfortunately there's this correlation between adoption and health so it makes the estimation a little bit harder. We're following Woolridge's suggestion when using a binary, which one trying to instrument for a binary endogenous variable by predicting the probability of adoption with two instruments, which I'm going to get into in a minute, in the first stage probit and then using a general method of moment estimation. So the two instruments that we use are the number of tobacco producers by district in 1998. This instrument I argue is valid in the sense that networks, people see whether or not you're producing tobacco in the area but these number of producers were from before the survey that we're actually using. So none of the children that we're measuring were alive when we use this instrument. And then the other instrument is the relative land suitability between tobacco and maize. So this would have an effect on your decision to adopt but would not necessarily have an effect on children's health outcomes the relativity of it. So our first stage instruments are fairly significant in the first stage if high F statistics and inability to reject the validity of the instrument. I know this table is kind of long and there's a lot more to it but I'm only presenting the significant results here. But I'm really looking to focus on the variable of interest which is tobacco producers. So in a tobacco producing household if we follow kind of what the World Bank did to the logical conclusion it would be to run an OLS and kind of this ideal model here and we see what we found before in 1998-1999 is still happening in 2004-2005 in the sense that tobacco production is still negatively associated with children's health. When I input the instruments we see a stronger negative results still highly significant. And then what we try to do here and I wasn't able in my seven minutes to get into it a lot is to look at it in a more nutritional from a nutritional standpoint which is that there's this argument that the first thousand days of child's from inception to kind of two years are the most important from a nutritional health standpoint. And so this price spike occurred in late 2001 through the middle of 2002 and so you would have been in utero we argue up until this point. When you were born between here then you were arguably exposed to the price spike in your early development window and if you were born right before or after then you were not exposed to it in your early development window. So when we split the household up between the children that were exposed in the early development window and those who were not exposed then we find this negative result is still there and it's still highly significant but for the non-exposed group it's still negative but the significance has gone away. And I would like to emphasize in my last 30 seconds that the research is not meant to argue against cash crop adoption or the idea of moving up the value chain. It's simply to think that there's context to all these decisions and farmers, small-holder farmers have a lot of things going on. So there's been research and questions about why aren't people doing this more often, why don't we see more adoption and this could be one reason that there's a lot of risk involved in adoption and if you're especially choosing to adopt a crop that you can't eat in times of trouble could be a problem.