 Thank you and thanks everyone for coming it's exciting that there are a lot of folks out there interested in subjective well-being so since I'm presenting first I'm going to do a little bit of kind of laying the ground work and so my first question is you know why study subjective well-being and I would argue that one reason is that subjective well-being is an important lens through which we can understand the impacts of inequality on people's lives. When I'm using subjective well-being here I'm meaning how people perceive their lives and what I'm hoping to do in the next 20 minutes or so is answer two different puzzles. First whether or not relative deprivation is a universally experienced phenomenon. Relative deprivation is the discontent people experience when their reference group has more of something they believe that they're entitled to and this definition comes about comes from Runceeman back in 1966. Secondly I also look at which if any community attributes might matter for well-being. So the relative deprivation puzzle which Martin Ravellian and others have looked at is that there are consistent findings of relative deprivation from from high-income countries so there's an increase and what that means is that an increase in reference group income is associated with decreased respondents subjective well-being. Basically if people are doing better off than you it makes you feel worse off and the reason why this might matter is that increasing disparities between people and their social comparators could leave people feeling worse off even if their incomes increase. So this is sort of a relative income potential issue. There's been a lot of work on this starting with some sociologists in the 50s and you know even I would say before that as I'll talk about briefly soon. An interesting twist is that while it's fairly consistent that relative deprivation exists Firebound Schroeder in 2009 noted that in America relative deprivation appears to be spatially determined so that it's better to live in a rich neighborhood within a poor country or within a poor county or within a poor region and so what they're saying here is that like well maybe what's going on with these wealthy neighborhoods is that it might be schooling quality that we're picking up on or something that some other attribute that makes people feel better about being in this this particular reference group of a well-to-do neighborhood and so I want to come back to this community survey or the access of two community services shortly. Now the relative deprivation puzzle the piece that comes in is that there's mixed findings about relative deprivation from low and middle income countries. There's one study that absolutely finds relative deprivation in Nepal. There are a handful of other studies that don't find relative deprivation or in fact find a positive effect of having a wealthier comparator group and there's a lot of different explanations as to why this might be. One would be aspirations so if I see people around me doing better off than myself maybe I expect that I too if I wait long enough might also do well and therefore I'm not resentful of the people around me doing better off. Another might be mutual insurance if I break a leg and my friends and family are relatively better off than I am I can go to them for assistance so they offer almost like a social protection mechanism that might not be available elsewhere. There are other positive externalities that have also been hypothesized. You know a slightly darker interpretation might come from Durkheim who says well maybe people don't aspire to what they don't expect to be able to achieve so people maybe don't can't you know I don't expect I'm ever gonna do better so the fact that everyone else around me is doing well doesn't really matter because I just it's not gonna influence my life. Now one of the ways that relative the relative deprivation is most commonly measured using some form of arithmetic mean and when I was reading this literature something that kind of came to my mind was well what if the underlying distribution matters and what I mean by that is say I earn $11 an hour and the average in this room is $12 an hour but this half of the room earns $10 an hour and this half of the room earns you know $13 an hour that might be a different experience for me than if this half of the room earns $2 an hour and this half of the room earns $200 an hour and so what I started to think about was well what is the what what does dispersion matter for relative deprivation so that's that's the first piece I want to look at the second piece I'm looking at is this community attributes puzzle where compared to kind of relative deprivation studies of which there are many especially in in high income countries there's much less research on how community attributes and the inequality of those attributes might set a might shape subjective well-being and as we heard yesterday as well as elsewhere as well as you know kind of multiple times throughout the course of this conference you know communities are very different and a sort of I think Lipton talking about urban bias in 1977 kind of really highlights this and I think thinking about culture of poverty neighborhood effects you know even the financial of well-being improving someone's self reported health like there's all these different ways in which communities can influence a variety of different outcomes so I was quite curious as to try to identify which community characteristics and does dispersion of those community characteristics matter for subjective well-being in low income countries to highlight or to kind of preview my findings for 10 countries in Sub-Saharan Africa I don't find any evidence of mean-based relative deprivation and I'll talk a little bit as to why I think that is the case and this is quite similar to other studies of low income countries however I do find something that I'm terming inequality based relative deprivation and what I find is that when there is more inequality a higher genie coefficient a marginal increase in the consumption wealth index has a greater impact on subjective well-being so basically when you live some place with a lot of inequality a marginal increase in my in your economic well-being kind of amplifies it's the or that there's an amplification of the of an increase in economic well-being on one's well on one's subjective well-being I also find that crime and the perception of government capacity both matter for subjective well-being although the self-reported access to although the reported access to community services doesn't matter and similarly the sort of the degree of inequality of crime within your community matters in the opposite direction as you might expect so as crime becomes more and more unequal you feel better off perhaps because you have a less of a chance of experiencing crime so what I do is I match data sets for ten different countries I use the Afrobarometer round four from 2008 and I match it with demographic and health surveys from round five I use also some UN you wider measure genie measures and the World Bank GDP average and those are the ten countries there so the subjective well-being question comes from the Afrobarometer survey and it asks people to describe their living conditions compared to others living in this country this is so my total sample is about 13,800 people and as you can see here there's sort of a normal distribution of responses you know most people are kind of clustering around you know relative to other citizens I'm doing a little bit better a little bit worse or about the same I think as some of the other speakers will talk about there are lots of factors associated with subjective well-being and what I'm going to do here today is really just try to focus on these community factors and on any quality measures I think there's a fair amount of research from psychologists saying genetics matter a lot getting good genetic data is extremely difficult so we have to put it aside now one of the limitations of the Afrobarometer data is that there is a lack of either a complete asset expenditure or income module and so Afrobarometer is quite interesting because it's a lot of breadth but there's not very much depth to the data set so as a result I use a principal components analysis to compute a consumption wealth measure and this might be controversial but what I do is I combine both consumption measures and asset measures and the reason why I do this is I'm quite concerned that using one or the other would leave out a tale of the kind of economic distribution so for example Afrobarometer includes asset information about you know how do you own a radio do you own a car do you want to be a TV etc and it's quite possible for someone who is poor to answer no to all of those questions it's also possible for someone who's ultra poor to answer no to those questions and I wouldn't be able to differentiate between those who are sort of right at the poverty line and those who are extremely poor so asset based measures leave out I think the kind of the left-hand tale a little bit too much and meanwhile the consumption measures from Afrobarometer kind of clump up the right-hand tale too much where it's quite possible for someone to say I never go without food I never go without water I never go without fuel which are the questions about consumption that get asked however it's you know it's quite possible that middle-class and high and you know and high income folks would answer yes you know this never happens to me all the time and I wouldn't be able to differentiate between them without including the asset measure so that's why I'm doing both it's a little unorthodox but I think it works actually quite well so what I have here is the mean consumption wealth score on the x-axis and the mean subjective well-being score on the on the y-axis for urban and rural groups within within my 10 countries so I've got 20 data points and you can kind of imagine a line running all the way up through here there are a couple of kind of exceptions and I think they're they're pretty understandable or at least I feel like there's a good reason why this might be the case the first exception is that urban and rural Kenya are kind of fairly well below where we went where in terms of well-being where you think they might be based on their consumption wealth scores this survey was fielded in 2008 after the 2007 2008 post-election violence in Kenya so I would suspect that people were really recovering from or quite concerned about what had recently happened in Kenya as they were answering their subjective well-being questions relatedly urban and rural Liberia are sort of much higher than one might expect given their their consumption wealth scores and it's quite possible that that their experience that they were sort of the memory of the Civil War maybe means people feel like you know what I'm doing pretty okay actually another interesting piece is that a lot of is that the vast majority of kind of the urban of urban respondents claim higher subjective well-being than rural respondents although this particular facet goes away when I control for or when I start including variables around crime and perception of governance and so on so what I do is I estimate a series of logit model ordinal logits the subjective well-being variables five categories so I use an ordinal to address that I have three different measures I'm going to show you or three different models I'm going to show you quickly today the first looks at community attributes and the community attributes I include as crime risk and experience with crime perceptions of governance and access to community services all three of these are also indices computed using principal components analysis I should note that the access to community services is not in fact self reported that earlier slides mistake I apologize it's a numerator report so an enumerator comes into a community and says you know yes this household has access to water this household has access to electricity and so on and I think there's a lot to be said that they might be wrong secondly the second set of findings I'm going to show are standard measures of relative deprivation this is mean based relative deprivation and I'm computing reference groups based on urban or rural residency status within a country this is an extremely broad definition of a reference group and I think that this is partially why I don't find any evidence of relative deprivation given sort of data limitations this is kind of the finest grain cut I think I can get at the moment but but what I do find is that I exchange these mean relative deprivation measures for measures of inequality within a reference group computed as a genie coefficient and then I interact that genie with the respondents own kind of index for in order to get a sense of where they stand relative to to others so there are a lot of variables that are suppressed here but what I want to point out is first is that this consumption wealth score is highly statistically significant at the 1% level and it's also positively related to subjective well-being as we would hope the perception of government index which which asks people how well do you think the government is dealing with food insecurity how well do you think they're dealing with crime with health etc basic human development sort of factors and that's also strongly correlated and so I think this is a potentially quite interesting finding where people feel positively about what their government is doing they're also more likely to feel positively about their lives I don't know what direction that is happening but I think it's a it's an interesting finding I also the experience and fear of crime indexes is also quite statistically significant and it's negative so as your as your fear of crime increases your subjective well-being declines I also looked at a variety of different health measures here is looking at wasting rates within a community it turns out none of these are particularly significant or interesting but I had this vision that there are going to be an important aspect of what people's lives what might be contributing to people's lives was sort of a general sense of how healthy their community was but doesn't look like that thank you so in the second set of findings I look at I'm again I'm suppressing I'm suppressing a fair number of things and just including a handful of things I'm looking at relative deprivation the you know the access to community services measure is not significant so I've kind of put that aside but what I find here is neither leave out mean consumption wealth index or my leave out mean crime and fear of crime index are statistically significant so I don't find any evidence of relative deprivation as constructed with my extremely broad reference groups so what I do instead is start looking at inequality and I compute genie coefficients and interact those genies with each individual's own consumption wealth index measure as well as their own experience in fear of crime index measure and what I find is I do find that they are statistically significant I'm going to go straight to the marginal effects because it's hard to interpret them without you know when we have an interaction term we can't interpret them unless we're looking at the marginal effects pieces so what we find what I find here is that the marginal effect for a consumption wealth index evaluated at different values of the consumption wealth genie matters a lot so here's my reference group here so these are all statistically significant I didn't put the stars in because it made my tables ugly so here's the the genie coefficient for urban been in which is relatively low and the genie coefficient for rural been in which is relatively high in my you know across my 20 reference groups and what I'm finding is that places with higher inequality a marginal increase in the consumption wealth index has a higher proportional odds of changing someone into the next higher level category of subjective well being so what I'm finding is that inequality amplifies the impact of an increase in the consumption wealth index oh and so I guess I should remark back here interestingly the odds versus not the case there is no statistical significance associated with an increase in in the genie coefficient for different levels of the consumption wealth value so for the marginal effects piece for crime risk what here I find a slightly higher increase in the crime risk genie dramatically increases someone's proportional odds of reporting a higher value of subjective well-being and so the idea of a crime genie to kind of reiterate is that as crime becomes more unequal as it becomes less likely for you yourself to experience crime you feel better off and in fact this this effect is even larger than the consumption wealth index effect which I think is also quite interesting so to summarize you know Durkheim wrote in suicide in 1897 that desires depend upon resources to some extent and I find that desires also depend on how both how equitably these resources are distributed as well as you know the particular resources we're looking at matter quite a bit so I find that crime and crime inequality matter as does the perception of governance relative deprivation in this sample is an inequality based phenomenon and what I'm finding is that inequality amplifies the impact of economic factors on people's perceptions of their lives I also find that relying on mean measures of relative deprivation rather than distributions might explain some of the lack of findings of relative deprivation in other places and might miss some of the important ways in which inequality shapes objective well-being I should also comment that I have another paper in process that looks at different sorts of relative reference groups and there I do find that the type of reference group matters quite a bit so looking at a village level reference group I find relative deprivation in Ghana quite strongly when I look at social networks within those villages especially reciprocity based social networks I find it actually a positive impact so it's much better to have your social network do well whereas you might feel quite quite badly if your if your community is doing much better than you and so this is I think consistent with rebellion and lock sheen's work as well as with my belief kingdom and night's work and so I think the fact that I don't find relative deprivation here to me indicates that the reference groups are probably overly broad however I do think that the inequality pieces is still quite interesting so thank you