 So, thank you all for attending the talk. I'm going to be presenting work unfortunately virtually and not in person, but this is joint work on refugee return and conflict with Austin Wright. And so as we just heard, there are more people forcibly displaced across borders now than at any point since record keeping began. And of course, one of the most prominent instances we just saw this past summer, when more than a million people attempted to flee Afghanistan as a result of the Taliban's rapid spread across the country. And at the same time that we're seeing increased forced displacement around the world, we've also seen a collapse of the so-called durable solutions for refugees, one repatriation from host countries to origin countries, which is the historically dominant solution accessible to refugees, two, naturalization in developing host countries, and three, resettlement in the West, which is hampered, of course, by discriminatory migrant backlash. Today I'm going to focus on refugee repatriation because it is the historically dominant solution for refugees. And understanding the dynamics of repatriation poses some major inferential challenges, namely because conflicts dynamics are directly endogenous to conflict trends in origin countries. So here we can look at a time series of refugee outflows from Afghanistan to Pakistan and then returns likewise of Afghans from Pakistan to Afghanistan. And what we see is that outflows, of course, spike when episodes of conflict begin or intensify, and then likewise repatriation spikes in brief moments of stability, for instance, around the collapse of the Taliban. And this pattern is not unique to Afghanistan. We see a kind of similar wave shape of refugee returns around these key moments of conflict stability in places like Iraq, Serbia, Myanmar, and Mozambique. So this is a regular phenomenon and it poses a real threat to inference, again, because repatriation and conflict are directly endogenous. We're going to get leverage on this question by studying a large cash grant program in Afghanistan that was rolled out in 2016 and saw the repatriation cash assistance offered by UNHCR to Afghan refugees interested in repatriating from Pakistan double. In response to this cash grant program, there was an unexpected and large-scale wave of Afghan returning from Pakistan, and they accounted for 66 percent of global documented refugee returns in 2016, so a huge number of people, upwards of half a million. The cash assistance that we study was economically meaningful in the Afghan context, amounting to about $400 per returnee, that's not per family, that's per individual. And it was rolled out at the start of the 2016 fighting season, so if you can imagine a situation where we would expect refugee returnees to exacerbate conflict, it's going to be at a period in time when the insurgent movement, the Taliban, is recruiting a large pool of kind of itinerant temporary fighters for summer offensives. And so this creates a ripe condition for finding the opposite of what I'll show you that we ultimately find. And we're going to leverage historical returnee settlement patterns in a Bartik style difference and differences design using some previously unreleased NATO and Afghan government combat records, which I obtained as part of dissertation research and provide a kind of follow-on to the earlier CIGACS dataset that has been released for the period from 2006 to 2014. And this data that I have covers 2015 to 2022, or 2020, rather. So the core findings here, we show that the encashment program and refugee returns under it reduce insurgent violence. And we provide evidence that this operates through an opportunity cost channel. The program raises reservation wages in destination communities and constrains insurgent recruitment. An unintended consequence of rising reservation wages is that it also constrains counterinsurgent tip buying. And so there is a negative effect on the efficacy of counterinsurgent bomb clearances. On the other hand, encash returns increase communal violence. And we show that this effect can be offset by social capital and strong local institutions. Okay, so there are three real perspectives on how refugee return at forced displacement more broadly might intersect with conflict. On the one hand, we can think about forced displacement worsening conflict, for instance, by providing a ripe source of recruits for insurgent groups, or by exacerbating institutional fragility. We might also think about how it could worsen social conflict, for instance, by exacerbating competition between non-migrant neighbors and migrants over jobs and other resources. And then a third perspective in the literature emphasizes how refugees might actually be associated with improving security because they come with aid and they've acquired human capital and skills abroad that might ultimately bolster livelihoods in destination communities. And we're going to adjudicate these perspectives in the context of the program I mentioned. So the context here, we're studying Afghan returns from Pakistan. Pakistan is a historically large host state of refugees virtually entirely from Afghanistan. In 2016, at the time that we study, there were more than one and a half million Afghans living in Pakistan. We're going to study returns induced by a cash assistance program, which in the graph at the bottom here, you can see induces a large spike in the time series of returns. So the program was precipitated by a terrorist attack against the school in Pakistan by the Pakistani Taliban in December 2014. No Afghan refugees were associated with the attack, but it kicks off a wave of anti-Afghan xenophobia in Pakistan owing to the perceived link between the Pakistani Taliban and the conflict in Afghanistan. And the government kind of stokes this link as well. In response to growing at anti-Afghan xenophobia in 2015 and 16, the Pakistani government threatens the expiration of refugee cards for registered refugees in Afghanistan in private meetings and negotiations with the UNHCR. And in response, UNHCR as a way of mitigating the hosting burden borne by Pakistan offers to double repatriation cash assistance payments with the hope of getting more people to return home. And so the payment doubles from $200 to $400 in the period in gray, you can see. This assistance is economically meaningful in the Afghan context, the median family under the program receives $3,600, which is equivalent to an entire fighting season's worth of income if the men in the group were to mobilize on behalf of the Taliban. It's also equivalent to at least several months of listed wages. And we have some survey-based validation throughout that I'm happy to talk about in Q&A. I'm going to gloss briefly over the encashment procedure, but essentially once you registered for return in Pakistan from a camp, you had a week or so to get to the red stars denoted in Afghanistan, which are encashment centers where the grant was paid out to you and from which you could return to any district of your choice. So we draw on these novel records of insurgent violence gathered by NATO in the Afghan government. We are also using a U.S. government-sponsored conflict tracker from Afghanistan to chart communal violence. Happy to talk more about this in Q&A. And to show you the kind of distribution of violence here, it comports with a lot of what we know already about conflict dynamics in Afghanistan, particularly severe levels of conflict along the border with Pakistan. All right. We're drawing on repatriation data on documented refugee returns from the UNHCR Afghanistan office. In the map here, you can see the distribution of returns under the program that we study. And again, we have survey-based validation using a survey of attorneys fielded by the Asia Foundation. The empirical strategy is a Bartik-style differences and differences design where we have this time series shock coming from the program and then we're leveraging cross-sectional variation coming from earlier historical return settlement patterns, specifically in 2013, the first year that we have return data. And we're going to focus on this reduced-form least squares equation, but I can also show you instrumental variables results, for instance. And to build some initial confidence in the design here, you can see that historical return patterns strongly predict contemporary destinations. And so some nice face validity. Of course, the key assumption here is parallel pre-trends. And again, I'm happy to go more into this in Q&A, but we think we have a pretty good case here for parallel trends in the pre-treatment period. All right. So briefly, the core results, the first and main effect here is that we see the encashment program associated with a large and substantively meaningful decline in insurgent violence during the program period in destination communities. And yet this is counterveiled by an equivalently large and substantively meaningful increase in incidences of communal violence, so for instance, feuds over land. Now I'm going to go into some results where we try and probe the mechanisms. So first, we might expect that rising reservation wages owing to the encashment program constrain insurgent recruitment. And if this is the case, we should see insurgents substituting from more labor-intensive tactics to more capital-intensive tactics. And so the most labor-intensive tactic is going to be direct fires, something like a gunfight with government forces. Capital-intensive attacks are going to be things like explosive violence and IEDs. And what we see is that the negative effect of the program on overall insurgent violence masks heterogeneity to this effect, declines in direct and indirect fires and an increase in explosive violence. We can also push this a bit further by looking at the lethality of these different attacks. And again, we see declining lethality of labor-intensive violence and increasing lethality of capital-intensive attacks, which suggests to us in a pretty clear way that the opportunity cost channel is what's at work here. And we're also responding to declining levels of recruitment by shifting out of labor-intensive attacks. We might also expect rising reservation wages to increase the price of tips that counter-insurgents are trying to buy from the civilian populace. We don't have data on tips for the period covering 2015 to 2020, but we can exploit the rate at which IEDs that were in place were found and cleared versus successfully detonated by the insurgents. This measure is strongly sensitive to civilian informing. And we provide evidence that the rate of IED clearances by the government is declining in the program, again, suggesting that rising reservation wages have this negative unintended consequence of constraining counter-insurgent informing. You are running out of time. If you could please accelerate a little bit. Thank you. Yes. Almost done here. Okay. So communal violence can be offset by social capital as proxied by the share of refugees returning to an origin province or district. And then finally, we show that strong local institutions, things like village-level sheriffs and access to elders also help offset the increase in communal violence. All right. So in conclusion, we provide the first causal evidence here on refugee return and conflict. We show that return affects different types of violence in different ways and think that this is a ripe avenue for future work. We find that the cash transfer induces repatriation and spills over to non-migrant neighbors still illuminating some of the dynamics of refugee aid. Again, provide some evidence on mechanisms. The program specifically constrains insurgent recruitment, but also counter-insurgent tip-buying. And then finally, we find that building social capital and local institutions, especially informal institutions, is likely to be a key antecedent for safe refugee return because these help blunt any positive effect on communal violence. Thank you.