1/ Excited about new paper on Economics of Maps w/ @sstern_mit in the JEP Winter 2020 issue!
pubs.aeaweb.org/doi/pdfplus/10…
Inspired by "the map is not the terrain", we argue that maps matter & examine the economic antecedents & consequences of mapping distortions
tweetstorm below 👇🏾
pubs.aeaweb.org/doi/pdfplus/10…
Inspired by "the map is not the terrain", we argue that maps matter & examine the economic antecedents & consequences of mapping distortions
tweetstorm below 👇🏾
@sstern_mit 2/ Our starting point is the observation that distortions in mapping info have shaped the course of history. We highlight the example of the Martellus Map that was used by Columbus to argue for a westward route to India 
3/ we then review and unify recent papers strewn across the econ literature that "maps matter" for central economic outcomes. my faves? London subway map distortions and route choice, Italian satellite data & property tax evasion and HOLC redlining maps and urban segregation.
4/ We also provide an overview of the business of mapping and the geospatial industry informing the casual reader about this historically important and fast growing $400 billion dollar industry
5/ Our second point is that even though maps matter, the "map is not the terrain". Maps must necessarily abstract away from some features of the terrain and highlight others.
6/ We argue that maps distort our view of the terrain on two dimensions -- data & design. They might simply not possess certain pieces of data about the terrain, or even conditional on the same data maps might look vastly different based on design choices.
7/ A prominent design choice is level of aggregation. For example: Does Oklahama ever vote in favor of democratic candidates? Your view of the geography of Oklahoma politics might differ substantially based on which map of the 2016 election you saw:
8/ Since (a) maps matter but that (b) maps are rarely "mirror images" of nature, but rather chosen carefully chosen representations we state our key proposition: data and design distortions are not random, but are systematic deviations shaped by economic forces
9/ We first discuss the central question: will a free market supply an optimal level of maps. Basic conclusion: likely see underinvestment in both mapping data & design but with design you have the additional problem of overuse -- i.e. overproduction due to imitative copycats.
10/ Beyond the question of over/undersupply, our core argument is that *who* is mapping and *why* they are mapping matters a great deal for what type of map is produced. Consider the map for Manhattan by Google Maps and Apple Maps. Both used by millions of people.
11/ Even though these maps are of the same place, they differ dramatically! Google labels relatively more roadways and transit, while Apple Maps favors landmarks and shops. Label overlap is only 10 percent!
12/ We argue that these differences are not random. They likely reflect Google's priorities on its API partners such as Uber and Apple its relatively affluent end users in NY. HT to @justinobeirne for this cool comparison.
@justinobeirne 13/ Our core framework highlights 5 key forces that shape mapping distortions: (1) costs of mapmaking (2) nature of demand (3) intellectual property & competitive environment (4) innovation in mapmaking technology and (5) incentives of mapmaking organizations or individuals.
14/ For each one of these forces, we provide comparative statics that clarify how they are likely to shape mapping distortions.
15/ Perhaps my fav. example is the one about mapmaker incentives & features Disney's map of Disney World. Looking at it, you'd believe that Disney World was surrounded by pristine wilderness & only Disney hotels! Alternate maps reveal many more hotel choices & regular urban dev!
16/ What are the implications for researchers? We have spent a lot of time using them, but maps are powerful forces that can shape your fav. economic outcome in fields like dev, public finance, urban econ, innovation etc. Need more studies documenting their market effects.
17/ We also need more theory on equilibrium properties of maps. Why do maps that ostensibly have similar goals look different from one another in different settings and contexts (say, subway maps versus automobile maps)?
18/ Finally, how can policy-makers use mapping info to shape economic outcomes? 2 egs here: @sstern_mit, @infowetrust and @JorgeGuzmanCBS are developing "startup maps" to help investors and policy-makers figure out the location of the next big idea. startupcartography.com
19/ Similarly Raj Chetty and team are developing @OppInsights maps that help policymakers figure out census blocks with the most potential for urban mobility. opportunityatlas.org
20/ If you take our article seriously, know that both of these maps, like all maps, are necessarily biased -- how can you use this bias as a force for good and which biases are most efficient from a social welfare perspective?
21/ Finally, while our discussion has largely been limited to geographic maps, a lot of these points apply to other kinds of data, including genomic maps in pharma decision-making & a lot of the current discussions on bias in training data & AI. Excited about these directions!
21/ If you enjoyed this article, we'd love to hear your reaction and thoughts! Personally, @sstern_mit was the original impetus for the conceptualization of this project and a tour-de-force on its execution. Many of my fav. examples above come from @sstern_mit's creative genius!
22/ And we must also thank a number of people who made this paper much better than when it started. JEP editors Moretti, @TimothyTTaylor, Williams and Hanson as well as @mslaurabliss @infowetrust @JorgeGuzmanCBS & the excellent research assistance of Melissa Staha.
23/ So, in sum, maps are cool, they matter a LOT, and there is a lot of low hanging fruit in studying how their distortions and how they shape markets and economic outcomes near-and-dear to your heart. March forth and keep mapping! FIN.
