 Okay, so let me start by thanking Mohammed and Victor for inviting me to talk at this conference. I must say it's been a really excellent two days, so congrats on a great programme. It would of course have been nice to be there with you in Toulouse, but I won't hold you to account for a global pandemic. But you have put me in the kind of unenviable position of standing in between the participants and their Friday evenings or afternoons, as the case may be. So I'll attempt to at least be relatively brief and snappy. I'll put my timer on, but you'll let me know if I start to run out of time, if I don't police myself well. So this is a paper which is about transportation technology, by which I mean transportation infrastructure, which is why I thought it would be a nice fit for this particular session. And I'm going to look at the impact that transportation infrastructure has through its impact on individual mobility, that is the mobility of individuals in society, on their ability to mobilise socially or politically mobilise, if you will. To be a bit more concrete about what I do, is I take an episode from Swedish history at the end of the 19th or the beginning, to the beginning of the 20th century, when the railway network was becoming truly national. That's the, you know, expansion of transportation technology, if you will. This gave individuals a greater opportunity to travel. It effectively shrunk distances between places that were formerly quite far apart in space, as well as, you know, in popular imagination, if you will. And I'm going to show how this shrinking of distance between places, which was enabled by the expansion of the railway network, allowed people to travel to, and with them bring their ideas. This resulted in a mobilisation of society within a range of social movements that became very important for the Swedish democratisation process. So I'm going to be much more clear about what I mean by all of these individual parts as I go, but that's to give you a very brief summary of what it is I do in this paper. I should say that, you know, this is kind of a non-obvious intersection of disciplines. And I think that speaks to, you know, a number of different things that we've touched on over the last couple of days, even though I won't, in this presentation, do a kind of meta commentary on some of these techniques in the way that many speakers have done, and that's been very interesting. I hope it can show that we can use tools and data from, you know, quite disparate parts of the discipline which weren't intentionally intended to be used together, but which we can combine in interesting ways to say something interesting, I think. As a result of this being a kind of non-obvious intersection of things, I always like to begin when I present this paper by motivating in terms of thinking about, you know, what's grassroots social movements can do and why and how. And so we know from just taking three examples offhand here that social movements can be powerful forces for change socially, politically, economically. In this century during the Arab Spring, you know, it's been shown that mobile, you know, social mobilization through the use of social media was very important for the overthrow of elites. In the last century, in the 1900s, the US Civil Rights Movement depended to a large degree on local chapters of the NAACP organizing members locally to, you know, build national momentum for an expansion of the legal rights for formerly heavily discriminated against parts of the population. And by going back even further to the 19th century, which is going to be most relevant for me today, we can think about suffrage movements which build popular support in order to together mount sufficient pressure on the elite to extend franchises. We can think of this in the case of the Chartists in the UK, for example, we can think of it throughout Europe. And in my case, I'm going to be looking at Sweden where these movements for very much an important force behind expanding the narrative, relatively narrow franchise. So we know that these movements can be important, but we know less about what determines how they spread and what makes them successful. There's a, you know, a large tradition, a long tradition of work in, in economic sociology going back to people like Tilly and Granavetta, who have shown that knowing what other people in my network, if you will, are doing is going to be key for me to determine myself if investing costs the effort into mobilizing is going to be worthwhile. But there's a, you know, a literature that's emergent and parallel to that also posits that, well, knowing what my peers are doing might not necessarily facilitate aggregate mobilization. Mansur Olsen made the case quite strongly that if I know that someone else is going to put in cost the effort, why should I, the free rider problem. We know about this stuff and there's ways to put this stuff in these kinds of ideas in much more elaborate network models, which I won't go into but we can put bells and whistles on this. One of the key takeaways that it's not clear, theoretically, if I become more able to interact with people in this potential network of mobilization, that that should on aggregate improve things or make things easier to mobile. The empirical work on this topic has, you know, there's some very recent very good papers that I've looked at this, but then the focus seems to be more on short run bursts on relatively extreme forms of collective actions, collective action. These papers focus on protest events that take place over a matter of weeks, months, or maximum number of years. I think that in order to understand this kind of social mobilization and the kind of latent support for it that needs to build up for these types of rapid burst of collective actions actually have bite, we need to take a longer time frame. And that's what I do in this paper. So my main research question is how this technology that facilitates social interaction interactions between individuals in society. What is the fact that they are actually able to travel and meet. How does that then shape the spread of these movements that can arise from from these interactions. And what's the role of a kind of process of spatial social diffusion underlying this. Like I alluded to, I'm going to use Swedish history in the late 19th, early 20th century as a natural testing ground for these types of questions. So during this period we saw not the beginnings of but certainly the, you know, a large expansion which resulted in a truly international transportation network through the railways, and these were reduced the cost of travel through various parts of the country. That's going to be, you know, the shock on the right hand side if you will, which facilitates the movement of individuals and therefore their interaction with each other. And at the same time we see the nascent of these social movements that I that I mentioned and I'll describe in a bit more detail, which became dominant socially economically and politically well into the 20th century. These included things like temperance organizations who of course had as their main objective. They were a push to prohibition essentially and they were almost successful in in sort of introducing us style prohibition in Sweden in the early 1900s. They include also various free churches that were religious organizations separate from the state Lutheran church. On the, you know, more on the labor side, we have trade unions and various leftist parties that organize the labor movement. So these are quite different organizations but what unified them was this push for extensions to what at this time was a very relatively narrow franchise in Sweden and that's, that was surprising to me to understand just how narrow it wasn't restricted to the wealthiest to meet essentially. So unifying these organizations was this push for democratization. What I'm going to do is to assemble a data set from various sources that I that I'll describe, which allows me to construct a yearly panel which covers over 2000 parishes over a 30 year period. I'm going to be able to say in every year and every parish, what organizations are present present there and what are their memberships. I'm able to measure by digitizing older railway maps, the proximity of these parishes to the constructed railway network, and I'm also going to be able to calculate in various ways, least cost paths between pairs of parishes as a result of this, which will be time varying as a result of this expansion of the railway network. So in the subset of these parishes I'll be able to say, you know, using very detailed station level data on passenger flows, and the flow of freight or the flow of goods. And that will be helpful for me to show that, as we'll see if I have time, the effects I'm going to document on the social movement side are driven by the flow of passengers into a place, not by a flow of goods. And so that's going to allow me to distinguish between you know this mobility of individuals versus the kind of mobility of goods of all sorts, more generally, which can be indicative of economic growth taking place as a result of really expansion. Empirically I'm going to use a variety of different strategies. I'll be able to show you some very simple spatial correlations which are I think very interesting just in and of themselves. You know, taken with a grain of salt of course because of the various, you know, endogeneity problems you might care to throw at me with those, but I think they're nevertheless quite informative. So in a no less specification, throw some, you know, parish and near fixed effects at this control quite richly for things like geography, baseline demographic characteristics occupational structure and so on which I take from the four count population synthesis. Then I'm going to also explain how I can construct an instrument using proximity to initial railway plans that were drawn, and which informed subsequent construction of the railway network. And this is of a straight line instruments so you can think of two important places being targeted. Say you want to connect with a trunk line stock on to Gothenburg. All the places that are in between that straight line of those important places are, you know, speaking slightly sloppily here going to be connected more or less by accident and I'm going to show you some checks that that various characteristics in the baseline, such as the initial presence of these movements as well as various things like occupational structure are going to be unrelated with distance to this instrument in the absence of actual railway construction. And then finally, I won't probably won't have time. I'm glancing at my time in here, probably won't have time to show the event study but in case, you know, you know, you can take issue with parts of my instrumental variable strategy. I will also present an event study where I use essentially the time periods before and after large reductions in the distance to the railway of a parish and see what happens before and after showing that things developed very smoothly in a parallel fashion before these large reductions and then seem to take off after these large improvements in accessibility of the railway network. So the findings are that as parishes become better connected. They are, they become more likely to host any of these organizations. So that's an extensive margin result if you will. They subsequently see larger membership numbers that is more of the individuals locally are going to become engaged and become members of these organizations. And he holds also when I control for population numbers of course as an important confound that these places might grow as a result of becoming connected, but the result here I'm going to document membership holds even controlling these developments. And then I'm going to show that these places host a larger number of different organizations so it's not just that the labor movements come in and take over and run the shop but rather, you know, there's going to be a variety of different types of organizations. Free churches and so on. I'm going to show, I wrote a great lengths in the paper to show that these results are driven by what I call greater influence from membership elsewhere or what I call social market access. The thing by this is you can think of the membership or any of these outcomes in a given parish in a given year to be a weighted average of how much influence is exerted by existing membership in other parishes with that weighted average. The weights in those in that average is going to be essentially how far apart those parishes are taking the least cost path along this transportation network. Well, if I have time go into much more detail on that, because I think that's a novel way of using something that's been used for many years, thinking about market access but using it in a new way to address a different question. And then the result I alluded to already about flows of passengers and freight indicates that it's about the movement of people not the movement of goods. Of course I wouldn't be so bold as to claim that this is the only thing that's driving the rise of social movements in this period. But I do consider a range of various alternative explanations and when I throw these explanations into my specification they don't at least rule out my my proposed mechanism. On the man side, you can think that, you know, religious or occupational structure might be changing as a result of railway developments. Indeed they are, but but not in a way that completely diminishes the effect that I find. There's been work that shows that the possibility of emigrating of immigration that is exiting from the political system and moving to say to America was important for these movements to develop. Essentially measure measure for how easy just to get to the main immigration ports as a using the railway network and show that my results still hold. And then on the supply side, I digitized information on the telegraph system as well as on the postal service to show that it's not flows of other types of information. That's explaining the full extent of the effect, but it's something about the movement of people which seems to be important in at least in the initial stages of setting up these organizations in the first place. Of course, communicating through the telegraph or through post is going to be important once you want to coordinate these various organizations that are active in different parts of the country. But for the initial setup, it seems that that personal interaction is quite important and these things can't explain away at that part of the result. And then finally, and I'll just allude to this here very briefly because I certainly wanted time to discuss it at the rate that I'm going. But this is the kind of so what question that comes at the end, which is, did this matter. Okay, you show us that railways allow these movements to develop. But you know, what's the, why do we care. And what I what I do is to show that in the first election with universal male suffered here. There is another election, 10 years later that which also introduces female suffered which which I am, you know, currently looking at because that's interesting as well. There is an election with universal male suffered for the first time. I show that places, correlationally places with more social movements do exhibit higher turnout and do show a larger vote share for the social democrats. So these things did matter in terms of shaping political engagement as well as outcomes of the political process. So, let me begin with some motivating observations. And the first slide here, I'm showing, you know, the intersection, not in terms of data but in terms of previous historical scholarship that has been done on on these two topics on railways and on social movements. So the first quotation here is from Lucas, who was writing in the 70s, you know, well, many decades later and is one of the key social historians of these movements in Sweden. So what was that the channels of communication for these new ideas were primarily personal visits by culprits, basically people who would travel and set up local reading groups and so on. Preaches for these free churches agitators and labor unions and so on. And a key example of this is August palm, who was an early. He was a key social democrat in the early stages of Swedish social democracy. He went on what he called agitation travels that is he would travel extensively on on on the railway and go from essentially from parish to parish and speak to to various coming at it from the other angle is Eli Heksher. He was writing in the early 20th century. And in an essay on the Swedish railway system, and he mentions that travel has been extended to infinitely wider strata of the population. In contrast with the past transport has an apparent democratizing effect in the present. And he doesn't really expand on this so I'm not sure exactly what he had in mind. But what I find here seems to be seems to be indicative of one part of this democratizing effect that he may have been alluding to, just as an aside. And I always tell this story, and I'm not sure if it's true but it's interesting. This is the same Hector that later became known for the Hector only in trade model, which you may remember from your grad student days of learning trade models. And they won the Nobel Prize for that, of course, but Hector always look more fondly on his work on Swedish economic history and particularly on the railways, but of course that's not at least that side of Sweden what is known for. And that just goes to show that you can't fully pick what you become remembered and famous for but that's just an aside. So how does this look then, you know very descriptively in the data. So here I'm going to show you two maps of the Swedish railway system. The first one, which is as it stood in the beginning of my sample period in 1881 already fairly well developed railway construction began in the 1850s. And quite a substantial portion of the 1860s. And then kind of ground a bit to a halt. And this was the state of the world essentially in 1881. You can see that you know the main population centers Stockholm, Gothenburg, Malmö, even going up into the center of the country to Östersund. These places have been already relatively well connected, but lots, lots, large parts of the country are not connected. And even in the places that were the areas that were connected density was still relatively low. But by 1910 you see, you know, a very different situation. The population centers in the north have become connected, even in the south, the network is much more dense. So if you imagine on the on the left side here, if you were to go from Gothenburg on the west coast, I don't know if you can see my cursor, but it's there on the coast. If you wanted to go on the railway down to Malmö, which is in the southwest, you'd have to go far inland and then back down. Whereas in the in 1910 you can go essentially on a straight line down the coast so much more direct connection. At the same time, we see the rise of these various movements which I described so let me take the temperance movement graph as an example and explain how you should read these graphs. So in the solid line I'm showing you on the left axis the proportion of these. Sorry, I bring the municipalities they should be parishes, but in the Swedish sense they are essentially the same. So I'm showing you here on the left axis, the proportion of these parishes, which had at least one of these movements present. And you can see that this goes from a basically a zero at the beginning up to something like 70% by the end of by the early 1900s. In terms of total membership in thousands it goes from zero again up to something like two and a half, not 2,500,000 members. And you should read the same the other graph in the same way where you can see that temperance movements and free churches took off a bit earlier unions and leftist parties took off a bit later. But essentially, the sum of all this is that by 1910 say that combined membership across all of these movements was around 700,000 people and you have to compare this to a population in Sweden at this time of around 6 million. So that that really is quite a substantial popular mobilization in these movements. How do these things look if we put them together. Here I'm just showing you two plots, one in 1881 and one in 1890 to show you spatially how the railways relate to where these movements are and how strongly appear to be. And you can see, and of course here I'm not claiming any kind of causal relationship yet but you can see already the spatial relationship between these two. Two things where there seem to be railways and in particular where always seem to form these important meeting points where we're essentially they connect parts of the network. We see that these movements tend to spring up and places that are far from the network seem much less likely to develop these movements. This is going to hold you also if we throw this in a very straightforward correlation where I throw all the observations in. This is very parsimonious in a way I'm just throwing some course geographical fixed effects for the 24 counties, residualizing based on that, as well as controls for latitude and longitude. But that's it. Apart from that is unconditional and what we see is this quite striking relationship across all four types of movements where the further you are away from the railway and this is on the log scale here. The less likely you it is that one of these movements is present in the parish. And this holds also for membership numbers and for the number of organizations, just to show you and not hide them behind the link to show you that it also holds for membership, a bit less tightly. It also holds but still that negative strong relationship is there. Okay, I see I have about seven or 10 minutes left so I should certainly speed up, but I should spend at least a minute or two on the data because you know that's part of why we're here. We're here to see where, you know, various different bits of data from the universe of data that's available on Sweden and which is very rich can be put together to look at these questions which are not exactly obvious to look at. On the social movements side, I am very much in debt to Andrea and Chris who in the 90s went through pinkstaking work to go to these local parish archives to collect information about when these social movements were set up, what their membership numbers were and so on. And as far as I'm aware there's also, at least in some of these parish archives individualized membership lists, which these people did not digitize or at least that's not available as far as I can tell. But this is perhaps an avenue for future research to go back and see what we can actually do with this individual level data. So it's very interesting to connect to things like the population censuses, also to look at, you know, multiple membership and different types of organizations, looking perhaps at social networks that arise as a result of membership in the same organization so there's potentially lots to be done here but I'm not sure exactly how well preserved these individual lists of members are everywhere. So right here is where I can claim to make at least a marginal data contribution because here I've gone to these historical railway maps that were published by Statistics Sweden, in which I've hand digitized for the for the 30 years from 1881 to 1910. For my instrumental variable strategy I also have these initial railway plants which I hand digitized from a separate source. And then I have a bunch of other data from various sources. The population census has of course come from the international, the international version of IPUMS or the North Atlantic population project which is made available, you know the full count population census from 1880, 1890, 1900 and 1910, and which is very useful for a range of different applications. And I digitized the data and other communications, such as the telegraphs and the postal service also from Statistics Sweden, and then a bunch of, you know, more standard geographical data from NASA. Right, so in the last five minutes or so I'm going to describe my empirical strategy and show you basically what I do in the paper to flesh out the, you know, very correlational patterns I already show but it's only basically to strengthen that main point that I've already made, verbally and So essentially what I do is to regress some social movement outcome, YIT, so say the presence of these movements in a given parish in a given year, and that's going to be regressed on the distance of that parish to the railway network in that particular year. Sorry, in the year before. And the reason for that is that the maps I'm using actually document the state of the network as of the 31st of December in that year. So it, in principle, if things were constructed quite near the end of the year, it can only realistically affect things in the following year so that explains this one one year like structure essentially on the distance of railway. So I'm going to throw out this parish fixed effects alpha year fixed effects gamma, and as well as some important geographical and baseline demographic controls in the vector x. You know, a whole account for, you know, a range of concerns you might have with the specifications such as the fact that places that were close to the capital say we're more likely to see railway expansion, and they're also more likely, presumably to develop these movements because essentially under at least on the labor organization side, a lot of this was actually directed from from stock. In the presence of a lot of controls thrown at this specification, you may still have some concerns about an endogenous relationship between how well connected a parishes and what their social movement outcomes might be. You can have in mind, for example, the fact that in Sweden at this point in time, the development of the railway network was very much a top down thing. Directed by economic and political elites, as well as military elites, a lot of the reasoning behind expanding the railway network was for military concerns to connect the major population centers in a way that didn't rely on coastal travel, for example, the Baltic at this point in time was still seems a very dangerous place. I guess you can imagine who was the main perceived threat. And for this reason, the railway avoided the coast and that, you know, can be correlated with many other things. Of course, I will control for that in the vector x but nevertheless that these endogeneities might remain. And so in order to get some exogenous variation in this distance the railway measure. I'm going to instrument it in the first stage which uses these initial plans for the network that I mentioned. And I'm going to write down a specification like this because these initial plans are time invariant, the cross sectional at a point in time, and in order to use them in this, in this a panel setting and going to interact them with decade dummies essentially and then explain what this, what the logic behind this is in a few slides. But why is this relevant instrument interesting. So it seeks to leverage the history of railway construction in Sweden, and this follows the paper by Torbaria and Chastinium for who introduced this instrument for the Swedish case in 2017 in a journal of urban economics paper. And it's similar to other types of straight lines from in such as bonds known for for a US highway and in state highway network. The logic behind this instrument is that the pre railway transportation in Sweden was unpredictable and fragmented, you can think of roads that were very primitive. Whatever waterways they were were very seasonal because Sweden is a cold country and they were frozen over for large parts of the year, especially in the north. And so the state essentially viewed itself as having a paramount role in rolling out a national railway network. And two master plan proposals where put forward with the aim to connect the main cities stock on Gothenburg Gothenburg and so on. Avoid the natural barriers, particularly the large lakes in the center of the country, and avoid the coast for the reasons that I discussed for essentially military strategic concerns. And although these plans were ultimately rejected by parliament they greatly informed eventual construction nevertheless. The side effect of this is that parishes between places that were targeted by the plan became connected, and not for anything necessarily related to their own potential for developing economically or socially but rather because they happen to be in between two targeted parts of the network. And here I'm just showing you the two various plans that were proposed. And you can see how they quite closely resemble the eventually constructed network here in the beginning of my sample period. In addition to this I'm going to use these nodal lines which by Iranian flu have gone back to the historical documents and documentation at the time, which seek to kind of get at the spirit of these plans and to connect these five nodal towns that were identified. And so I'm going to take a parish and see how close it is to any of these various straight lines. And that's going to be my instrument in the cross section. And where it becomes interesting using this in a panel setting, because in each individual year we see a very strong elasticity between how far you are to the planned network, and how far you are to the actual network. So throughout this is a positive elasticity, but importantly this elasticity flattens over time. So as the network expands, how far away you were from these initial plans becomes less and less important. You can see that these places that were from the get go very well connected. They don't really move on this graph it's these places that were very far from the plan network that really I was driving this flattening of the elasticity of the time. And that's exactly exactly what we pick up. If we estimate this more rigorously in a, you know, first stage specification, where the emitted decade will be the first one between 1881 and 1890. And we can see this flattening of the elasticity, even as we throw very rich controls at this relationship. And I see I'm really out of time so I should speed up and wrap up. I'm showing the paper that this, if we take the subset of parishes that have not yet been connected and correlate various variables for these parishes to how far away they are from the plan. This is prior to actually becoming connected to the network. We see no correlation between the likelihood that they have a movement present their membership or organization numbers, as well as other demographic and occupational structure outcomes such as a sharing agriculture or an industry and so on. This is true in 1881 in 1890, I can also do this in growth rates between the two. So results very briefly in the final couple of minutes. Basically it confirms what I showed you correlation. So, both in the OLS with these various controls and fixed effects, and in the IV specification, we find that as the distance to the nearest railway shrinks. It becomes more likely that that movements are present or rather, you know, this is a distance. So of course this is negative here but if you think about this as a decrease in the distance that is a negative change in distance would interact with this with this negative coefficient to produce a positive effect on the presence of the movement. The IV picks up exactly the type of or solves the endogeneity problem in exactly the kind of way that we would have expected. If it's the case that the railway railways were strategically directed in a way that would seek to avoid the mobilization of these social movements because it appears that or less if anything is biased towards zero. The same is true for membership where being closer to the railway increases membership numbers. The closer you are to the railway, the more organizations you see. I have a range of further results. I described a lot of these already verbally you can check the check out the paper for much more. But at this point I think I should wrap up and I'm very happy to return to any number of these in the Q&A if the opportunity arises. But for now let me conclude by saying that I've used this formative period of Swedish history to show that infrastructure facilitates the spread of social movements in a way that is not necessarily obvious. And I've used data in a non-obvious way to show this. And individual mobility seems to be a key mechanism. And I didn't get a chance to show this but at the end of the day the social mobilization matters because these movements then shaped participation and as well as vote shares in Sweden's first election with universal suffrage. So at that I'll conclude and thanks very much.