Poverty, Inequality, Terrorism The Wealth of Villages

1
Poverty, Inequality, TerrorismThe Wealth of
Villages

• -coauthor is John S. Felkner (post doc, NORC)
• Robert M. Townsend
• University of Chicago

2
TODAY, ONE PART, ONLY

• TO UNDERSTAND POVERTY, UNEVEN DEVELOPMENT AND
  THE POTENTIAL FOR TERRORISM LOCALLY
• NEED ECONOMIC MODELS TO UNDERSTAND UNDERLYING
  FORCES WITH FINE TUNED PREDICTIVE POWER
• ASSESS POLICY CHANGE

3
Data

• Socio-Economic Data Thai Community Development
  Department (CDD) biannual census data
• More than 3000 villages in four provinces,
  1986-1996
• Focus on four Thai provinces specifically chosen
  to represent a cross-section of Thai economic
  development fertile central plains versus
  poorer northeast- same as Townsend Thai project.
  Adding South/unrest
• Supplemental GIS spatial data collected from a
  variety of sources, including a number of Thai
  government agencies. Also utilized an archive of
  Landsat satellite imagery from 1979-2004

4
(No Transcript)
5
(No Transcript)
6
1986-1996 Thai high growth period

• Thai economy experienced some of the highest
  growth rates in the world, ranging from 7 to 12
  percent, often attributed to financial
  liberalization
• Average wealth doubled, rapid industrialization
• Extensive deforestation and urbanization

7
A Satellite View Of Industrialization
8
Wealth Index Spatial Distribution

• Chachoengsao, Lop Buri, Buriram and Sisaket
• 1986-1996

9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
(No Transcript)
14
GIS, Road Networks, and Accessibility

• Highly detailed geo-referenced data on road
  networks was used to calculate travel-time along
  road networks taking into account varying road
  speeds
• This allowed for the creation of variables as
  proxies for access to economic agglomerations,
  which could then be used in the testing and
  correction of simulation models

15
Sisaket Province, - Road Network withAverage
Road Speed
16
(No Transcript)
17
Dynamic Simulation of the Occupational Choice
Model

• villages as the data points
• Simulation begins with base year wealth
  distribution 1986 and produces results through
  1996
• Financial intermediation index imposed or not
  exogenously in each year of the simulation
  (binary from CDD)- occupation choice and end of
  period wealth a function of initial and talent
  (costs)
• The credit sector is weighted according to the
  exogenous intermediation fraction, and an
  equilibrium obtained giving a common market
  clearing wage and interest rate in credit mkt
• trace path of individual villages given the
  prices

18
(No Transcript)
19
Spatial and Temporal Testing of the Financial
Deepening Model The simulation did an
excellent job of capturing overall dynamic trends
20
(No Transcript)
21
(No Transcript)
22
Residuals structural models regressed onto
covariates

• Occupation choice onto
• wealth, education, an intermediation access and
  the agglomeration access proxies
• Results
• Wealth and education are never significant
• However, time-travel to nearest major
  intersections is positive and significant as
  model is over predicting with distance
• credit intermediation index is positive, as if in
  the model credit/saving access is too good

23
(No Transcript)
24
(No Transcript)
25
An Experiment

• Policy Simulation create new, hypothetical road
  networks and impose spatially varying estimated
  costs via m parameter
• does superior accessibility increase simulated
  entrepreneurial activity for villages close to
  new roads?
• Roads intersections were created using the GIS
  according to 2 criteria
• Located far from existing roads and major
  intersections
• Located in areas with low levels of
  entrepreneurial activity
• Model was re-simulated using the spatially
  modified model (with new estimated m parameter
  values with distance to new road intersections)
• Result dramatically higher levels of
  entrepreneurial activity near to the new major
  road intersections

26
(No Transcript)
27
Financial deepening model

• Model over predicts closer to spatial
  agglomerations
• Confirmed with Local Moran spatial statistical
  cluster detection
• Residuals also regressed onto agglomeration
  proxies, wealth and education, and significant
  and negative results for all 3 direct
  agglomeration proxy variables, and significant
  and positive results for wealth and education
• In sum, the simulation is over-predicting close
  to economic agglomerations- both wealth and credit

28
Spatial Modification

• Again, full sample stratified into bins 3 bins
  by equal number of villages along the axis of
  time-travel to major intersections
• Also, model simulated separately for commercial
  banks only, and then for BAAC only
• This allowed for the estimation across space of
  the variation in costs of using each major
  financial provider as captured by the q parameter

29

• Graph above displays relative costs by bin
  (results plotted in data wealth units)
• Note that for BAAC, costs are systematically
  lower than for commercial banks

30
.
31
.

32
Conclusions

• We begin with the assumption that spatial
  proximity acts to minimize transmission costs for
  ideas can we test whether spatial proximity to
  economic agglomerations facilitates the spread of
  entrepreneurial activity, wealth or access to
  credit?
• Consequently, we estimate transaction costs as a
  function of decreasing accessibility to economic
  agglomerations
• For the entrepreneurial choice model, the testing
  reveals that spatial proximity matters greatly in
  determining the cost of going into
  entrepreneurial activities the model performs
  much better after estimation of spatially varying
  entrance costs
• For the financial deepening model, the testing
  reveals an apparently policy distortion due to
  government support of the public credit provider,
  resulting in higher estimated costs closer to
  agglomerations

33
SES Predicted Income per capita