<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Marshall Burke on Medium]]></title>
        <description><![CDATA[Stories by Marshall Burke on Medium]]></description>
        <link>https://medium.com/@marshall_182?source=rss-c25f44f76f48------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/2*GbX-0u37ok5vcw9RyVGF3A.jpeg</url>
            <title>Stories by Marshall Burke on Medium</title>
            <link>https://medium.com/@marshall_182?source=rss-c25f44f76f48------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 01:37:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@marshall_182/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Who’s growing in Africa?]]></title>
            <link>https://medium.com/atlasai/whos-growing-in-africa-744cc7e1bf9?source=rss-c25f44f76f48------2</link>
            <guid isPermaLink="false">https://medium.com/p/744cc7e1bf9</guid>
            <category><![CDATA[africa]]></category>
            <category><![CDATA[economic-development]]></category>
            <category><![CDATA[satellite-imagery]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[remote-sensing]]></category>
            <dc:creator><![CDATA[Marshall Burke]]></dc:creator>
            <pubDate>Thu, 22 Aug 2019 14:51:43 GMT</pubDate>
            <atom:updated>2019-08-22T22:39:25.474Z</atom:updated>
            <content:encoded><![CDATA[<p>Marshall Burke and Apoorva Lal, AtlasAI</p><p>Our <a href="https://medium.com/atlasai/the-tyranny-of-national-statistics-d3a79af526a9">last post</a> looked at how national averages can be deceptive when trying to get an accurate picture of economic well-being. Using new satellite- and deep-learning-based estimates of asset wealth across the continent, we showed that about half of the variation in economic well-being across the continent in a given year is within countries rather than between them. This tells us that just making comparisons at the country level — which is what is currently done when <a href="https://sdg-tracker.org/">things like the Sustainable Development Goals</a> get tracked — could give a very incomplete picture of who is wealthy and who is poor.</p><p>What about when we look at how things are changing over time? Maybe there are a lot of differences in average wealth levels within countries, but over time everyone is being lifted up in some countries but not in others. If so, this might tell us that we can get away with country-level tracking when looking at growth rates.</p><p>Unfortunately, the data don’t really support this notion either. Using the same satellite imagery/ deep learning pipeline we’ve developed here at <a href="https://www.atlasai.co/">AtlasAI</a> and described <a href="https://medium.com/atlasai/the-tyranny-of-national-statistics-d3a79af526a9">in our last post</a>, we can look at changes in asset wealth over the last few decades at a very local level. As before, we are measuring asset wealth as an index, where higher values indicate households that own more assets. In our training data, which comes from household surveys done across the continent, households in the top quintile of this index almost all have electricity and own a radio, tv, fridge, and phone. In the bottom quintile, almost all do not have electricity and do not own any of these items.</p><p>The animation below shows what happens when we track asset wealth over time in Nigeria, from the country level down to the district level. At the country level, asset wealth was overall growing across the country, rising about 0.25 standard deviations between 2005 and 2017. (Consistent with <a href="https://www.economist.com/middle-east-and-africa/2019/05/30/nigerians-got-poorer-in-muhammadu-buharis-first-term">the story from Nigerian GDP data</a>, this growth was strong prior to 2015 but then basically slowed or stopped after that.) But again this country level average hides enormous variation at both the state and the district level.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*upsFSooW8Z8E4TcrpvC0KQ.gif" /></figure><p>We estimate that some districts were growing quite rapidly, for instance by more than 0.5 standard deviations over the 15 year period. Other districts actually showed declines in asset wealth over the period, by as much as 0.2 standard deviations. The map below shows this clearly. Conflict-stricken areas of northeast Nigeria, we estimate, showed meaningful declines in asset wealth over the period, while parts of the center and south of the country showed robust growth.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Kwh1y1Hknd33798LrYjxFw.png" /><figcaption>Map shows AtlasAI estimates of growth in asset wealth between 2003 and 2018 in Nigeria. Units are in terms of an asset index as described above, and are computed as 4-yr growth rates in this index.</figcaption></figure><p>Interestingly, for Nigeria we find a modestly positive relationship between how wealthy a district was at the start of our dataset, and subsequent growth (left plot below). This is the opposite of what would be predicted by standard models of <a href="https://en.wikipedia.org/wiki/Solow%E2%80%93Swan_model#Conditional_convergence">convergence</a> — i.e. that poor places should catch up over time as resources and technology get transferred to these areas. For other countries in Africa — e.g. Ghana, right plot — things look more like what standard economic thought would predict: poor places have grown faster over the last few decades.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3lThK1RuEioB23zkkpsz1Q.png" /><figcaption>Each dot represents a district. Horizontal axis is the average value of the asset wealth index in that district in 2005, and vertical axis is the subsequent change in the index between 2005 and 2017 in that district.</figcaption></figure><p>Looking across Africa, we again see huge variation in where asset wealth is growing and where it’s not. Known success stories (Ghana, Zambia, parts of East Africa) stick out, as do known problem areas that have struggled with conflict (southern Cote d’Ivoire, DRC, Somalia).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-oOl71twZkTHWVv7bGwdzw.png" /><figcaption>Map shows AtlasAI estimates of growth in asset wealth between 2003 and 2018 at the district level. Units are in terms of an asset index as described above, and are computed as 4-yr growth rates in this index.</figcaption></figure><p>Overall, we calculate that 69% of variation in growth rates is within countries rather than between them. Put another way, country level factors (e.g. country level institutions) can explain less than a third of the overall variation in asset growth across Africa, as measured in our data. This number is even higher — 81% — if we focus on just sub-Saharan Africa and omit North African countries. Our data and results further emphasize the importance of subnational data for understanding which areas are on track to meet sustainable development goals, and which areas are not. Look for a public data release on <a href="https://www.atlasai.co/">our website</a> in the coming months!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=744cc7e1bf9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/atlasai/whos-growing-in-africa-744cc7e1bf9">Who’s growing in Africa?</a> was originally published in <a href="https://medium.com/atlasai">Atlas AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The tyranny of national averages]]></title>
            <link>https://medium.com/atlasai/the-tyranny-of-national-statistics-d3a79af526a9?source=rss-c25f44f76f48------2</link>
            <guid isPermaLink="false">https://medium.com/p/d3a79af526a9</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[poverty]]></category>
            <category><![CDATA[sustainable-development]]></category>
            <category><![CDATA[remote-sensing]]></category>
            <category><![CDATA[deep-learning]]></category>
            <dc:creator><![CDATA[Marshall Burke]]></dc:creator>
            <pubDate>Mon, 22 Jul 2019 19:45:58 GMT</pubDate>
            <atom:updated>2019-07-23T16:41:44.055Z</atom:updated>
            <content:encoded><![CDATA[<p>Marshall Burke and Apoorva Lal, AtlasAI</p><p>Signed four years ago, the <a href="https://sustainabledevelopment.un.org/?menu=1300">Sustainable Development Goals</a> represent an ambitious effort to achieve improvements in human livelihoods around the world. The SDGs set explicit targets to be reached by 2030, and designate a large set of indicators that are to be tracked to understand whether appropriate progress is being made.</p><p>These indicators — when they are able to be tracked — are tracked at the national level (e.g. see this very nice <a href="https://sdg-tracker.org/">SDG tracker</a>). There are at least two reasons for this country-level measurement. First, the national level is the level at which data are historically available. Countries periodically conduct nationally-representative household surveys, or utilize systems of national accounts, that are designed to generate reliable country-level statistics. However, most do not collect (or do not publicly release) representative data below the country level.</p><p>A second argument for country-level measurement is a <a href="https://en.wikipedia.org/wiki/Why_Nations_Fail">large academic literature</a> that suggests that country-level institutions, such as respect for the rule of law, protection of private property, and inclusive decision-making, are paramount in explaining why some parts of the world are much wealthier than others. Under this argument, it makes sense to measure things at the country level, because between-country differences are the main source of overall differences in prosperity.</p><p>But is national-level tracking what we should be doing? What if there were other credible ways to track indicators below the country level? And what if the variation in outcomes of interest was actually larger <em>within</em> countries than between them? And without reliable subnational data, what do you do if you want to target a program to poor people, or figure out whether a particular anti-poverty program is actually working?</p><p>Unfortunately, even national-level tracking of key indicators such as poverty rates is difficult in many parts of the world. In recent decades, <a href="http://documents.worldbank.org/curated/en/700611468172787967/pdf/WPS7252.pdf">published estimates indicate</a> that over 50 countries in the world have either zero or one poverty estimate available, which is below the necessary two observations that would allow even a basic understanding of how poverty is changing. And, again, these are national-level estimates that reveal nothing about how livelihoods are changing within countries. Given this data landscape, saying that we are going to achieve the Sustainable Development Goals is a little like signing up for a weight loss program but not having a scale on which to weigh yourself.</p><p>Other credible options for measuring SDG-relevant outcomes are now emerging, however, and these could change the game in terms of how we understand and spur progress towards achievement of the SDGs. In particular, there have been a host of efforts in recent years to use non-traditional data sources to study patterns of economic development below the county level. Data from <a href="http://www.jblumenstock.com/files/papers/jblumenstock_2015_science.pdf">mobile phones</a>, <a href="https://www.unglobalpulse.org/projects/nowcasting-food-prices">social media activity</a>, <a href="https://www.pnas.org/content/114/50/13108">street-level photos</a>, and <a href="https://science.sciencemag.org/content/353/6301/790/tab-e-letters">satellite imagery</a> have all shown promise in accurately measuring local-level economic activity. These efforts suggest that the first reason for country-level tracking — that the country level is the only level at which relevant data are available — is beginning to make less sense.</p><p>Analysis of the granular economic estimates produced from these new data sources also suggests the second reason for measuring things at the country level — that most of the variation is between countries rather than within them — is also not very compelling.</p><p>Here’s some evidence that our team at <a href="https://www.atlasai.co/">AtlasAI</a> has generated on this point. At Atlas, we’re combining satellite imagery with deep learning models to generate accurate, local-level estimates of a range of outcomes relevant to the SDGs. We’ve trained a deep learning model to predict village-level asset wealth across Africa from satellite imagery. Here we are measuring asset wealth as an index, where higher values correspond to households owning more assets. In our training data, households in the top quintile of this index almost all have electricity and own a radio, tv, fridge, and phone. In the bottom quintile, almost none have electricity nor own any of these items.</p><p>In held-out villages — i.e. villages in countries that the model was not trained on — the model is able to explain an impressive three-quarters of the variation in asset wealth measured in household surveys, using only satellite images as inputs. This means that a model can be trained in one country where data are available (e.g. Uganda) and accurately applied to a nearby country were imagery is available but household survey data aren’t (e.g. South Sudan).</p><p>For the five most populous African countries, the figure below shows how the distribution of asset wealth changes as we move from country level averages, to state level averages, down to district level averages. It’s immediately clear that country-level averages hide enormous sub-national differences in asset wealth. For instance, Ethiopia is the poorest country on average among these five countries, but we estimate that some of the richest districts in Ethiopia are wealthier than the average in Egypt, the richest country in the sample. Conversely, some of the poorest Egyptian districts are poorer than the Ethiopian country average.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fCdnKWQgoxnZ_2Jq7pjShQ.gif" /><figcaption>Distribution of asset wealth under increasing disaggregation, from country down to district.</figcaption></figure><p>Using these data across all African countries, we calculate that 43% of the overall variation in asset wealth is <em>within</em> countries, and nearly a quarter is <em>within</em> states within countries. This number is even higher if we focus on just sub-Saharan Africa — 65%. Similar numbers have been found for other outcomes, including <a href="http://web.stanford.edu/~mburke/papers/BurkeHeftNealBendavid_2016.pdf">child health</a>. These within-country differences are starkly apparent in the corresponding map, which shows how wealth estimates change when country-level averages are disaggregated down to the state and then district level.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ow5WNrPptdpqB8XHhFoGTQ.png" /></figure><p>These data illustrate the growing ability — and clear importance — of measuring outcomes below the national level. We believe that data such as these can have a large range of applications, from better tracking progress towards the SDGs, to better targeting of anti-poverty programs, to better evaluation of whether these programs are working. Look for a public data release on o<a href="https://www.atlasai.co/">ur website</a> in the coming months!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d3a79af526a9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/atlasai/the-tyranny-of-national-statistics-d3a79af526a9">The tyranny of national averages</a> was originally published in <a href="https://medium.com/atlasai">Atlas AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>