Why should we care about GIS in Economics?

It is not new concept in Economic thought!

[...] if economics profession has notably neglected it [geography], this is not because economists have been uninterested in the subject, but because they have regarded it as intractable.

Source: Fujita, M; Krugman, P and Venables, A., 1999.

So, what's new?

[1] New amazing GIS data sets available out there

provides a original source of variation

Remotely sensed information

(Donaldson and Storeygard, 2016).

access to information difficult to obtain by other means

unusually high spatial resolution

wide geographic coverage

Other examples: Henderson et al (2012), Pinkovskiy & Sala-i-Martin (2016), Hodler & Raschky (2014), Michalopoulos & Papaioannou (2013, 2014), Alesina et al (2016), among others.

Historical Maps

recovery old (persistent) characteristics

Digitized by Nunn (2008). He found a negative relationship between the number of slaves exported from a country and current economic performance.

Then use by Nunn & Wantchekon (2011), Michalopoulos & Papaioannou (2013, 2014, 2015), Alsan (2015), Alesina et al. (2016), etc.

Geo-Coded individual-level data sets

exploring geography and location as an important dimension

(Duranton, D. and Overman, H 2005) → location patterns of industries in UK

Others examples of geo-coded data sets: .LSMS-ISA (World Bank) -> geo-coded household level 6 African countries

[2] GIS may improve empirical strategy stronger

reducing the omitted-variable bias (I)

gathering more information from neighborhood

Gibbons, S. and Machin, S (2005) ↑ transport → ↑ house prices

Other nice examples are: (i) Conley & Udry (2010) social learning in the diffusion Ghana ; (ii) Acemoglu, D; Garcia-Jimeno, C. and Robinson (2014) direct and spillover effects of local state capacity in Colombia

reducing the omitted-variable bias (II)

building instruments

Nunn (2008) Current African Countries & Slave Trades

Instrument Slave Export → distances from each African country to the locations where slaves were demanded

New identification strategies

Spatial Regression Discontinuity

Dell (2010) Current African Countries & Slave Trades

Colonial Institutions (Mita) → current outcomes

Exogenous unit of analysis

Montalvo and Reynal-Querol (2016)

[3] Provides a new dimensions to gathering micro-level data

Improve your data collection

Most collection data software allows to gather GPS location (e.g. ODK Survey, Survey CTO, CAPI)

Cracking Geo-Spatial Datasets

"Geo-spatial" data

The terms "geo-spatial" refers to information that is located on the earth's surface using coordinates.

"Geo-spatial" data

• Location (e.g. coordinates)
• Type (e.g. line,area)
• Attributes (e.g. temperature,...)

How can I represent those points, lines, and areas on the surface of the earth?

Geodetic Location

Latitude → north-south direction, relative to the equator

Longitude → east-west direction, relative to an arbitrary point (typically the location of the Royal Observatory in Greenwich - England)

... yet earth is not a perfect sphere, it is more like an oblate spheroid.

The mathematician Carl Gausse proved that it is mathematically impossible to project $R^3$ shape (e.g. sphere), onto a flat plane without introducing some sort of distortion.

What should we do?

... we need to choose a distortion that fits your needs

• cylindrical
• conical
• azimuthal

Cylindrical Projections

It include the Mercator Projection, the Equal-Area Cylindrical Projection, and the Universal Transverse Mercator Projection.

Conical Projections

It includes the Albers Equal-Area, the Lambert Conformal Conic Projection, and the Equidistant Projection.

Azimuthal Projections

It includes the Gnomonic Projection, the Lambert Equal-Area Azimuthal Projection, and the Orthographic Projection.

Coordinate System

• Unprojected Coordinate System
• (point directly on the earth's surface)


• Not Projections $\rightarrow$ $(x_0,y_0)$


• Projected Coordinate System
• (point on a two-dimensional map that represents the surface of the earth)


• Use Projections $\rightarrow$ $(\hat{x_0},\hat{y_0})$

.. still something missing

yet, we said that the earth is not a perfect sphere

Datum is a mathematical model of the earth used to describe locations on the earth's surface. A datum consists of a set of reference points, often combined with a model of the shape of the earth.

Most common datum: NAD 27, NAD 83 and WGS 84

"Geo-spatial" data

• Location (e.g. coordinates)
• Type (e.g. line,area)
• Attributes (e.g. temperature,...)

How can I represent those points, lines, and areas on the surface of the earth?

Vector Data

Point $(lat,lon)$	Linestring	Polygons	Polygons

Raster data (images as data)

Geo-located map set

Pixel characteristics as main attribute

Classified (Remote sensitive data)

Raster Data

Raster Data

Polygons

Polygons

Linestring

Point

"Geo-spatial" data

GIS data format

Getting geo-spatial dataset into a file (s)

Many formats available

• Raster Data


• DEM,geo-tiff,band based files
• Vector Data


• Shapelife, Simple features, TIGER/Line,...

micro-format (WKT,GeoJSON,...)

Shapefile developed by ESRI

Other files such as: .prj(projections)...

Same file name, yet different extensions

GIS tools

C, C++ and Python based solutions (objected-oriented syntaxis)

Source: Lawhead, 2013

Point

ArcGIS

• Commercial Software


• Pre-defined python package - ArcPy (+ Model Builder)


• Memory demanding; Windows only

QGIS

• Open source software; easy to create map images;
• Pre-defined python package - PyQGIS
• compatible with any OS (even android)

More differences

Standalone solutions

• Complete control of the underlying algorithms, data, and execution (no black box)


• Integration wit others python packages


• Solution no-software depending (replication and collaborations)


• GIS analysis beyond the bottoms


• Integration with new high-performance environment (i.e. cluster,cloud)

Ready for starting?