Thread by @busbyj2

Great and challenging conversations on my book on climate and security at UMass Boston, MIT and Harvard. Thanks @MIT_SSP for bringing me up. Thanks to @StacyDVanDeveer @fravel & @dustintingley for being terrific hosts at different institutions. 1/

https://twitter.com/mit_ssp/status/1651243283820040192

I had a number of trenchant questions the last two days on the project and want to use this thread to more fully address and puzzle through answers as the questions stayed with me. 2/

By way of background, the book uses paired cases to explain how climate hazards contribute to negative security outcomes in some situations and not others 3/

It’s anchored in the climate security literature that emerged in the mid 2000s to examine whether climate change will contribute to conflict 4/

I sought to broaden the security aperture to include other negative security outcomes, namely large scale humanitarian emergencies 5/

Using the same explanatory framework of state capacity, political inclusion, and international assistance, I sought to explain differential outcomes in three sets of paired cases from different regions of the world - Africa, the Middle East, and Asia 6/

That included two cases of humanitarian emergencies (Somalia, Myanmar) and one case of civil war onset (Syria). 7/

Each was paired with regional counterparts who also experienced exposure to climate hazards but didn’t experience the famine (Somalia yes, Ethiopia no), civil war (Syria yes, Lebanon no), or large scale loss of life (Myanmar yes, Bangladesh and India no). 8/

The bet was that in the wake of hazard exposure, similar processes of state service delivery and government representation, coupled with flows of international assistance, could help explain why a range of catastrophic outcomes occurred in some situations not others. 9/

The most straight forward cases in the book are the humanitarian ones, mostly because the connection between the climate hazards (droughts and cyclones) and the security outcomes (large scale loss of life) are more proximate in time and steps along the causal chain 10/

Cases of civil conflict potentially go through other lengthier steps between hazard exposure and conflict onset such as migration that makes it harder to establish that the climate hazard had anything to do with conflict emergence (which can happen months or even years later) 11/

The most challenging question is whether the same explanatory framework is fit for purpose to explain both humanitarian emergencies and civil war onset. I thank Barry Posen for his provocation on that score. 12/

There is a rich literature on civil war onset, with now classic debates between different perspectives on greed and grievances. 13/

In the climate security space, arguments have focused on a variety of direct & indirect pathways potentially contributing to violence, climate shocks/reduced livelihoods both making rebellion more attractive and state repression less feasible 14/ annualreviews.org/doi/10.1146/an…

Studies have surfaced the importance of conflict incidence in countries that have high political exclusion, suggesting inadequate attention to the needs of underrepresented groups in government, can fuel their resentment in the wake of hazard exposure. 15/ pnas.org/doi/10.1073/pn…

The role of political exclusion in wider studies of conflict outside of the climate security space has also emerged. 16/ cambridge.org/core/books/abs…

Plane taking off now so will resume in a bit. 17/

Other studies of conflict emphasize other parameters of governance, namely state capacity, as potent risk factors for armed conflict. 18/ nature.com/articles/s4158…

Both state capacity and political exclusion I thought highly relevant not only for understanding armed conflict (emergence & incidence) but also humanitarian emergencies in the wake of hazard exposure. 19/

State capacity gets at the ability of states to respond to climate hazards while political exclusion gets at their willingness to respond to all at-risk groups. 20/

So, if anything, I’m taking an emergent understanding of the drivers of armed conflict to the natural disasters/humanitarian emergency literature rather than the other way around. 21/

My understanding of state stability underpins my approach to both kinds of cases. I see state stability largely a function of elite compacts between groups who decide to resolve issues through politics rather than guns. 22/

The breakdown or the lack of inclusivity of those compacts can lead to both humanitarian emergencies and conflicts. 23/

If elite compacts are exclusive so some groups aren’t represented in government, those excluded groups will fare poorly in the wake of hazard exposure because governments will ignore them or actively discriminate against them. 24/

If those excluded groups are able, that unequal service delivery will serve as a potent source of grievance that can manifest in conflict alongside other slights. 25/

On some level, this question of biased service delivery to some groups and not others presumes a level of state capacity in the first place. Some governments like Somalia have struggled to deliver any meaningful services to anyone, though inclusion may matter on the margins. 26/

While I distinguish between infrastructural power to deliver services and coercive power to repress violence, I recognize the former is in a sense conditional on the latter. It’s hard to do much state service delivery if the state lacks a monopoly of violence. 27/

This is a belabored way to say I think it’s reasonable to think that governance factors relevant for conflict are also relevant for humanitarian emergencies. 28/

States are anchored in wider international relationships, which affect their access to resources in times of crisis and in between. Some can or are willing to tap into those international networks. Others can’t or won’t. 29/

This book focuses on triggering events, emanating from relatively short-lived climate shocks, really short-lived hazards like cyclones that unfold over a period of days or slow onset hazards like droughts that unfold over a series of months or years. 30/

Long run climate change is likely to contribute to permanent long run changes in rainfall and temperature. As others have observed, the field has yet to really adequately address such long term trends. 31/ journals.sagepub.com/doi/full/10.11…

As I argue in a forthcoming piece in @ps_polisci, our field's methods are not not well suited to study the long-run effects of a problem that is largely prospective. 32/

It's for this reason, I largely followed convention by studying the effects of relatively short-run climate shocks. @jcaverley asked what makes climate shocks special or are they similar to other kinds of economic shocks. 33/

My facile answer is I'm interested in climate shocks and other colleagues like Alan Kuperman have treated climate shocks as just another kind of shock. 34/ amazon.com/Constitutions-…

I could seek even greater generalizability for my argument by claiming that same factors that help explain whether climate shocks lead to negative security consequences in some places & not others also help us explain whether shocks more broadly lead to catastrophic outcomes. 35/

Capacity, inclusion, assistance are likely relevant to other circumstances but the specific pathways associated with climate are likely especially relevant to agriculturally dependent societies, as @EMeierding emphasizes in a forthcoming review of my book in @HDiplo 36/

I wanted to say a bit more on what is state capacity. @kathryn_sikkink pressed me on what it is and observable indicators for it. I draw on Michael Mann, @FukuyamaFrancis & others who have tried to unpack the concept and the challenges of measuring it. 37/ cgdev.org/sites/default/…

State capacity is the ability of states to delivery services, but we have only rough and poor proxies for observing state capacity as @cullenhendrix rightly notes in his review of my book coming in @HDiplo and in his excellent scholarship on the topic 38/ jstor.org/stable/20752162

We have only rough proxies for state capacity and in the book I start with investor survey driven accounts of government effectiveness, control of corruption, and regulatory quality, perceptual measures that may have their own biases. 39/

In each of the case chapters, I try to nail down what state capacity means with respect to particular hazards, namely early warning, water management & food security policies and practices with respect to drought & forecasting/early warning and emergency response to cyclones. 40/

I lay out a series of questions that can help one assess what constitutes hazard specific state capacity in a given area, not all of which are easy to answer in every case. 41/

Taking off now for real so tweeting may be delayed. Would like to say more about the cases and complementary work to mine. 42/

So speaking of complementary research, I was asked about whether I should have focused on a research strategy based on how hazards can differentially affect rebel groups and governments. Tobias Ide has a terrific project that does just that using QCA. 43/ amazon.com/Catastrophes-C…

Let me turn back to the cases, why I selected them and what the evidence suggested. The idea of using paired cases of countries that faced similar environmental harms but different security outcomes was a strategy
@marc_a_levy proposed years ago. 44/ jstor.org/stable/2539228

Isolating countries that experienced similar exposure to climate hazards and were like enough to be comparable was a challenge. For example, Ethiopia and Somalia both have faced severe droughts, but not always at the same time. I ended up using droughts from different years. 45/

The Levant faced a drought in the 2006-2010 period. I knew I wanted to write about Syria but paired with? Iraq already faced armed conflict (Turkey too). Jordan's share of agricultural employment was too small (3.9%) compared to Syria (17.5%) but Lebanon's was closer (13.5%). 46/

Syria and Lebanon had shared origins story of being created out of the same decolonization process from France and both being multi-ethnic polities with similar agro-ecological conditions. But many differences, population size among them. 47/

Despite these differences, I still thought the puzzle posed by @cullenhendrix in Political Geography was worth exploring. 48/ sciencedirect.com/science/articl…

@cullenhendrix Pairing these cases was challenged further by the fact that they are not truly independent cases, given Syria's history of occupation of Lebanon. 49/

Nonetheless, pairing Syria and Lebanon was more attractive than trying to compare countries in other regions with different ecological and social conditions. 50/

All of this is to say that pairing cases is really hard going. There are no perfect pairs. There may be more sophisticated quantitative strategies at the state or subnational level. Two examples come to mind. 51/

Kyosuke Kikuta uses the synthetic control method to show how DRC's Great War contributed to deforestation, more than what would have been expected had the war not happened. 52/ journals.uchicago.edu/doi/abs/10.108…

Sam Barrett matches villages in Malawi that faced similar climate hazard exposure but some received formal adaptation finance and others were left to fare for themselves through informal action. 53/ sciencedirect.com/science/articl…

In my book, in addition to paired cases, I also take advantage of within-case analysis to show how countries changed over time. Those of a certain age will remember how Ethiopia experienced large-scale famine in the 1980s but hardly anyone died after drought in the 2010s. 54/

Similarly, Bangladesh before it seceded from Pakistan experienced a cyclone that killed 300,000 but by the 2010's, hardly anyone died after severe cyclones. How do we explain that? 55/

Naturally, I think my argument of capacity, inclusion, and assistance can give us some traction, though salient factors vary between and within country cases. 56/

Now, as I turn to the methods you might say, hang on a minute, how are these humanitarian emergencies "security" problems? Isn't that conceptual stretching? 57/

As I argue in the book, what distinguish security problems from other kinds of problems isn't the agency of external or internal actor trying to do the state or its citizens harm, it's the severity of the threat. 58/

It's for this reason that problems like pandemics and climate change have been elevated to security problems because the potential for large-scale loss of disruption is severe enough to regard as security threats. 59/

I'm not the first to make this argument. @MBazilian @cullenhendrix made the case for the need for security studies to take into account the rising importance of these "actorless" threats. 60/ justsecurity.org/72939/an-age-o…

These kinds of problems are also ones that hard security institutions are increasingly having to deal with, whether it be mobilization of the National Guard to deal with COVID 19 or for military deployments for climate-related humanitarian operations at home and abroad. 61/

Militaries might not think it is their job to deal with these kinds or problems, but in many countries around the world, including advanced industrialized countries, sometimes the military has unique capabilities which are required to deal with these sorts of problems. 62/

If you now (somewhat) accept my premise that both civil wars and humanitarian emergencies are security issues and potentially explainable using a common theoretical framework, how do I go about it in the book? 63/

I try to get at it through process tracing with efforts to trace cause and effect along the causal pathway. Here, I'm indebted to one of my graduate advisors @IRgetsreal for his pathbreaking work in this space. 64/ mitpress.mit.edu/9780262572224/…

In chapter 3, I diagram a series of simplified possible causal pathways potentially connecting climate hazard exposure to security outcomes including but not limited to conflict. 65/

I also develop a typology of pathways to these outcomes based on the configuration of capacity, inclusion, and assistance. 66/

In each chapter, I need to do the following. 67/

I start by documenting comparable hazard exposure through a variety of measures. For drought, its national level growing season rainfall and temperature deviations or subnational level rainfall or agricultural stress indices. 68/

For cyclones, its severe cyclones that had the potential to kill large numbers of people. 69/

I then try to show that there were potentially devastating impacts that could lead to catastrophe absent adequate government policy. For drought, this shows up first in agricultural declines and impacts on livestock. 70/

In some cases, I can go beyond these initial impacts to talk about numbers of people at risk of food insecurity or economic losses that fall short of catastrophic loss of life or conflict, for which there is still time for state intervention. 71/

The end outcome is known in all these cases, so I then trace each aspect of my argument - capacity, inclusion, and assistance and discuss how each separately relates to the cases before bringing them all back together. 72/

Before each case, I provide an advance coding of the cases and theoretical expectations going in based on the typology in chapter 3. 73/

In cases where there is within-case variation, that advance coding also reflects my understanding and expectations for how the cases changed over time. See Ethiopia's change from type 1 in 1984 to type 6 or 8 in 2015. In cases like Ethiopia, multiple pieces change over time. 74/

Those on some level reflect my ex ante codings of the cases. There is a bit more complexity as I go through the cases. 75/

I start by coding state capacity both broadly using investor surveys of government effectiveness and then specifically in the issue area at hand. It's easier in some cases than others. 76/

It's easier where there are stark differences - no forecasting, early warning, or cyclone shelters in Myanmar compared to 72 hour forecasting, tens of thousands of early warning volunteers, and thousands of cyclone shelters in Bangladesh. 77/

It's harder in Syria & Lebanon where metrics related to H₂O management have to do with training & quality of staff or where differences aren't as stark. Or, where you have differences in the costs of policies like subsidies which are more downstream reflections of capacity. 78/

So, having established where state capacity differs between countries or within countries over time, I then turn towards measuring political inclusion. I start with off the shelf metrics from the Ethnic Power Relations and V-DEM. 79/

The Ethnic Power Relations dataset works better for multi-ethnic countries like Syria and Lebanon. Not at all for Syria where all are ethnic Somali but where clan differences are more salient. Here, I have to rely on qualitative discussions of political inclusion. 80/

Even in countries where differences in inclusion by these metrics are stark like Syria and Lebanon, there is a more nuanced political discussion of how elites have stayed in power or delivered services, in some cases reinforcing the metrics and in others not. 81/

Now you might call for a time out and say, why are going down the road of political inclusion at all, why don't you use the conventional metric of regime type and reference Amartya Sen's classic argument that famines don't happen in democracies? 82/ amazon.com/Development-as…

The reason for this is that there are countries like Ethiopia that are not fully or even all that close to being democratic that are (or were) what @LoviseAalen described as an inclusive autocracy, reasonably representative of all social groups. 83/ washingtonpost.com/news/monkey-ca…

It's this quality of inclusion, where the government derives legitimacy from its social compact with the populace to provide services despite not being democratic, which I think is more salient than regime type for countries like Ethiopia and Bangladesh. 84/

Having documented capacity and inclusion, I then turn to assistance, looking for patterns of international assistance, including development assistance and humanitarian aid in the lead up and aftermath of the hazard events. 85/

It's this flow of resources which can compensate for weak state capacity in the moment and build state capacity over time. I use patterns of ODA and humanitarian assistance, in the aggregate, as a share of GDP, in per capita terms, even remittances, whatever is available. 86/

Where relevant, these tend to show differences between Ethiopia and Somalia, Bangladesh and Myanmar in their access to resources as some countries are aid darlings while others are relative pariahs. 87/

My simplified causal pathways don't reflect this iterative quality as they are mostly focused on the snapshot of the hazard event and its aftermath but clearly investments by international donors over time have buttressed state capacity in Ethiopia and Bangladesh. 88/

I also anchor this general discussion in the specific politics of the moment to show how aid appeals or offers related to the hazards unfolded. Somalia for example experienced large decline in humanitarian assistance right before the famine because ... 89/

Al Shabaab wouldn't allow aid actors in and the US threatened to prosecute groups for violations of the Patriot Act, for fear aid would be diverted to al Shabaab. 90/

In Syria's case, an international aid appeal largely fell on deaf ears as the Assad regime was already something of a pariah state even before the civil war. The US didn't contribute to that drought relief appeal. 91/

For countries where aid was received, I suggest how that was put to good effect for different episodes. 92/

For all three cases, I try to pull these pieces together to show how capacity, inclusion, and assistance explain the outcomes in question. 93/

I try to draw these connections for both positive and negative security outcomes. 94/

Pausing as we are landing. Will come back to Syria case and conclude in a bit. 95/

Alright, the finale section. There is a robust debate on whether the drought in Syria had anything to do with the subsequent conflict. In the final sections of my Syria-Lebanon chapter, I try to bring the pieces together and then reflect on alternative explanations. 96/

It's mostly a story of political inclusion. There are some differences on the margins in state capacity and assistance, but it's mostly an inclusion story. 97/

There is a robust debate between some who think there are strong connections between the drought (Femia/Werrell, Kelley, Gleick) and the civil war and those who don't (Selby, de Chatel, Frohlich, Daoudy). 98/

Tobias Ide has a helpful piece assessing the differences. My summary of his piece. 99/ link.springer.com/article/10.100…

So, how do we get to civil war from the drought? What do we need to show? All accounts lead through migration from the northeast of the country that bore the brunt of the drought with people moving to peri-urban areas in other parts of the country in large numbers. 100/

Part of the disagreement is about how large those numbers were and part of the disagreement is about whether it was drought or bad policy that resulted in such large food insecurity problems. But first, we have to reckon with what prompted the exodus in the first place. 101/

In terms of whether drought or policy drove the migration, I'm not hung up on differentiating those as drivers. You have a drought that causes major declines in agriculture, with policy amplifying that problem with the removal of fuel and fertilizer subsidies. 102/

As for the size of the migration, there are estimates on the high side of as many as 1.5-2 million displaced. Here Ide says its truly hard to know with precision. I agree with his take. 103/

So, if we accept that there are several hundred thousand people who migrated to the cities, what is the relationship with the civil war and does this have any relationship with my argument? 104/

I quote a number of Syria scholars who think the migration was important. Take Hinnebusch 105/

See also Schmidt. 106/

And Lawson. 107/

Goldsmith echoes these claims of exclusion that swelled the mass of young men available to be recruited for the uprising. 108/

Dukhan provides some nuance on how rural economies led to the collapse of tribal communities, making those dispersed populations less easy to lead by traditional authorities. 109/

One more from Kilcullen and Rosenblatt. 110/

These are impressionistic, sweeping claims, many by leading scholars. However, there are those who take a different view. Ababsa for one says yes lots of displaced people but they did not participate in the protests. 110/

Daoudy goes a step further - no protests in the drought affected areas and no evidence the displaced participated in the protests (which she supports based with some interview evidence). 111/

So what are we to make of these discrepant claims? Ash & Obradovich use nights at light to show places that gained population in the lead up to the civil war were more likely to have protests but.../ 112

But drought didn't really feature into the protester's concerns. 113/ journals.sagepub.com/doi/10.1177/00…

I don't think that's necessarily a problem for the argument. As my former student @tgsmitty argued in his work on food protests, food doesn't necessarily have to feature as a central claim in protester activity to have been a motivation for protests. 114/ newsecuritybeat.org/2014/12/feedin…

So, I close my discussion of Syria with more extended observations on how challenging it is to get at the identities of those who ultimately participated in the protests and later civil war. 115/

We have to go beyond Tom Friedman's interviews with a few former farmers who became rebels to make claims that farmers played a significant role in the conflict, if not at the outset later on. 116/

At this point, it may be difficult to try to reconstruct this with available evidence. 117/

It may come down to whether we think the migration of some unknown number of people from the northeast, most or many of then Sunnis potentially recruitable to violence as their co-ethnics were repressed, made a difference in the emergence of the civil war. 118/

I linger on the Syria case because it's the most difficult case to establish the causal chain between drought and the security outcome, with perhaps the most ambiguity. But I'm also left with some dissatisfaction with the Lebanon case. 119/

The Lebanon drought story ends mostly with the government subsidizing both farmers and urban consumers to shield them from production losses and food price spikes. 120/

There isn't much of a record in Lebanon of larger agricultural losses or food insecurity in 2010 beyond a temporary decline in wheat production, which makes me wonder if Lebanon was simply not sufficiently parallel to the Syria case in terms of the intensity of the drought. 121/

Or, is Lebanon just not as dependent on agriculture? Or, did I not do enough digging into the geography and economic impact of the 2010 drought on the Bekaa Valley? 122/

Lebanon also looks more and more like a close call, that it has avoided a resurgence of civil war, not necessarily because of better quality governance than Syria or even more political inclusion but perhaps due to luck or historical memory of the past civil war. 123/

Its predatory elites have managed to service themselves rather than the public and not covered themselves in glory whether it be the garbage crises, management of wildfires, the fertilizer explosion, the wider economic crisis post-COVID. 124/

It wouldn't surprise me if they did descend into more widescale violence at some stage, which with recent attacks from Israel, might be international in nature. 125/ bbc.com/news/world-mid…

Lebanon's confessional power-sharing system may have insulated them somewhat, for the time being, from internal conflict but just barely. Inclusion only gets you so far if nobody is picking up the trash. 126/ aljazeera.com/economy/2021/1…

I don't think this is the last word on either the Syria or Lebanon cases or climate and security for that matter. 127/

I've got a few irons in the fire to try to nail down the Syria case even more, looking at death data in the first two years of the conflict, comparing governorate of birth to death, to get at whether someone had likely migrated and died away from home. 128/

Right now, it looks like people mostly died where they were born. 129/

But, setting all that aside, I think we're left with a story that is consistent with my argument with a few puzzles of unfinished business to try to address. I think that's true for the other cases as well. 130/

The takeaway for policy is actually somewhat hopeful in that climate vulnerable countries at relatively low levels of income can make focused investments in state capacity with the help of international donors to reduce the risks of large-scale loss of life and disruption. 131/

Poor countries don't have to accept their fate that climate exposure is destiny because it isn't. Donors can help them, but it's harder for donors to shape political inclusion. Well-meaning outsiders need to have a theory of state building & the limits of what is possible. 132/

The last thing I'll note and then stop is the backsliding is possible and progress can be upended, particularly with respect to political inclusion, as Ethiopia has shown in recent years. 133/ theconversation.com/droughts-dont-…

So, in any case, I'll stop now with this thread. I come out of talks now and again and you think I really nailed that answer or I wish I had said X, Y, or Z. Here, I've gone through the alphabet and then some so thanks for reading if you've made it this far. 134/

And should you like to read a copy of the book, it is available on Amazon for a reasonable price with full color maps I'll have you know. Thanks everybody. 135/ END amazon.com/States-Nature-…