Political Economic Consequences of Pakistan's Linguistically Fractured Educat...
Structural Constraints on Public Goods Provision: Evidence from Pakistan by Dr. Adeel Malik and Mr. Rinchan Ali Mirza, Oxford University, UK
1. Structural Constraints on Public Goods Provision
Interim Report
Adeel Malik
Globe Fellow in the Economies of Muslim Societies
Oxford University
Rinchan Ali Mirza
Doctoral Candidate in Economic History,
Oxford University
2. Ongoing work streams
Index of religious diversity
Mapping of current Tehsils on colonial boundaries
Collection of historical/archival data
Measures of political concentration
Snippets of work in progress
Data
Puzzles
Correlation
Research Direction
3. Measuring religious diversity
Herfindhal Index of religious fragmentation as shown below:
= 1 - Σj Sj^2
where Sj is the share of an ethnic group (j) in the total population of
a Tehsil (i) at time (t)
Religious polarization as shown below:
= 1 - Σj ( (½ - Sj) / (½) )^2 X Sj
where Sj is the share of an ethnic group (j) in the total population of a
Tehsil (i) at time (t)
The polarization measure, different from the fragmentation
measure, says that the largest level of ethnic tension in a Tehsil
will be felt when there are two ethnic groups with an almost
equal share of the population
4. Mapping Tehsils
Tehsil boundaries changed overtime due to several rounds of
administrative restructuring
Main Contribution: first ever empirical exercise to map current
development outcomes at the colonial tehsil level
120 current tehsils were collapsed to 69 historical tehsils
Step 1: Chose base year of 1931 census upon whose tehsils
statistics on current development outcomes are computed
Step 2: Identify administrative restructurings in each tehsil from
1931 to the most recent year (i.e. 2008)
Type 1: A tehsil in 1931 or any later census year was split into two or more
smaller tehsils during the period up until the recent data year
Type 2: Portions of two or more tehsils in 1931 or any later census year
combined to form a new tehsil during period up until the recent data year
5. Mapping Tehsils: Type 1
Identified every instance of restructuring where a district was
split into two or more tehsils between 1931 and the recent year
As number of tehsils increased they were collapsed to their
previous census round
Example:
Tehsil Z split into tehsils X and Y between two rounds: later and earlier
Used the following formula to make tehsils comparable between the two rounds
[(data for tehsil X) + (data for tehsil Y)] in later round = [data for tehsil Z] in
earlier round
Starting from 1931: repeat the above exercise for every census
round till the most recent year
Outcome: current tehsil variables at colonial tehsil boundaries
6. Mapping Tehsils: Type 2
Identified instances when portions of two or more tehsils in 1931
or any later census year were combined to form a new tehsil in
period leading up to the recent year
Example:
Tehsil X formed by carving out portions of N pre-existing tehsils between two
rounds: later and earlier
Used the following formula to make tehsils comparable between the two rounds
[(area given to X from pre-existing tehsil 1/total area of X)*(data for X) + (data for
pre-existing tehsil 1)] in later round = [data for tehsil 1] in earlier round
[(area given to X from pre-existing tehsil N/total area of X)*(data for X) + (data for
pre-existing tehsil N)] in later round = [data for tehsil N] in earlier round
Starting from 1931: repeat the above exercise for every census
round till the most recent year
Outcome: current tehsil variables at colonial tehsil boundaries
7. Mapping Tehsils: Limitations
Assumption: Development indicators uniformly distributed
across the area of the Tehsil
Implication: Allows us to break up development indicators in the
new Tehsil into portions which are then added back to the parent
Tehsil in the previous round
Development performance proportional to areas
Example:
A new tehsil Kotli Sattian was carved out from areas of Murree and Rawalpindi
20 % of Murree tehsil from earlier round given to the new Kotli Sattian tehsil in
later round. Furthermore, 30% of Rawalpindi tehsil from earlier round given to
the new Kotli Sattian tehsil in later round.
(20/50=40%) of development indicator of Kotli Sattian tehsil was then added
back to Murree tehsil to make Murree tehsil comparable across the two rounds.
(30/50=60%) of development indicator of Kotli Sattian tehsil added back to
Rawalpindi tehsil to make Rawalpindi tehsil comparable across the two rounds
8.
9.
10. Narowal Sheikhupura
Lyallpur
.6 Jaranwala
Lahore
Kasur
Shakargrah
Gujranwala
Chunian
Sialkot Pakpattan
Bahawalnagar Daska
Pasrur
Minchinabad Toba Tek Singh
.5
Shahdara
Montgomery
Okara
Sargodha
Samundri
Khanewal Nankana Sahib
Rawalpindi
Wazirabad
.4
Dipalpur
Rajanpur
Multan
Mailsi
Phalia Jhang
Shorkot
Hafizabad
Khairpur
.3
Bhakkar Lodhran
Bahawalpur
Alipur Ahmadpur Sharqi
Kabirwala
Shujabad Shahpur
Dera Ghazi Khan
Leiah Ahmadpur Lamma Gujrat Bhalwal
Muzaffargarh Rahim Yar Khan
Chakwal Chiniot
Taunsa Mianwali
Kahuta Addu
Kot Khanpur Kharian
.2
Jampur Gujar Khan Jhelum Khushab
Pindigheb Pind Dadan Khan
Fateh-jang
Isa Tallagang Allahabad
Khel
Attock
.1
Murree
.05 .1 .15 .2 .25
Proportion depressed caste (district)
Herfindahl Index of Religious Diversity Fitted values
11. Kharian
.2
Phalia
e( prop_remit | X ) Gujrat
.1
Jhelum Wazirabad
Sialkot Kahuta
Gujar Khan Dera Ghazi Khan
Chakwal
Khanpur
Rahim Yar Khan Daska
Jampur
Ahmadpur Lamma Toba Tek Singh
Allahabad Bahawalpur Rajanpur Shahdara
Leiah
0
Alipur
Tallagang Lodhran
Khushab Mailsi
Bhalwal Isa Khel
Khairpur Shorkot
Kot Taunsa
Pind Dadan Khan AhmadpurAddu
Pindigheb Sharqi
Muzaffargarh Khanewal
Bhakkar Gujranwala
Murree Fateh-jang
Attock Shujabad Mianwali
Kabirwala SamundriSargodha
Bahawalnagar
Montgomery
Rawalpindi
Chiniot Hafizabad
Shahpur Multan
Pasrur Pakpattan Jaranwala
Lyallpur
Okara Minchinabad
Shakargrah
Jhang
Dipalpur Nankana Sahib Chunian Sheikhupura
Narowal
Kasur
Lahore
-.1
-.3 -.2 -.1 0 .1 .2
e( index | X )
coef = -.2266273, (robust) se = .06128053, t = -3.7
X=density, latitude and longitude
Is religious diversity only acting as a proxy for selection into occupational
groups?
15. Plots and Puzzles I:
Rainfall and development in Punjab
1
.8
JHE RAW MUR
GJK
PAS SIA KHT
CHK
LHR WAZ GUJ
GUR KRN
DAS
LYP PDK SKG
TAL
ATK FTJ NAR
.6
TTS SAR PHA
SHK PIG BWL
MUL MIA
OKASDH KHU
BWPKHA SHP
JRWHAF
KAB MON
SAM
KOT JHA
LYA
RYK SHO
BHN BHA
MAI KHP
KNP TAU ISKCHI KSR
PAKNKSCHU
LOD
.4
AHLMUZ DIP
DGK
ALBALI
SHU
JAMMIN
RJPAHM
.2
0 20 40 60
Average rainfall in tehsil in 1923
Literacy Rate for over 15s Fitted values
16. Plots and Puzzles II:
Public goods provision falls with distance from Lahore!
Rajanpur
.2
.15
Kot Addu
Jampur
.1
Dera Ghazi Khan
Kasur
Isa Khel
Muzaffargarh Ahmadpur Sharqi
.05
BahawalpurLeiahTaunsa Khanpur
Chunian Shujabad
Bhakkar
Lodhran Alipur
Mianwali
Kharian Chiniot Kabirwala
Jhelum Kahuta Jhang Rahim Yar Khan
Pakpattan
Bahawalnagar
Khushab Multan
Nankana Sahib Shorkot Rawalpindi
Khairpur Attock Ahmadpur Lamma
Shahdara
Lahore SheikhupuraOkaraSamundriToba Minchinabad
Sialkot
Pasrur Shakargrah
Lyallpur
Jaranwala
Wazirabad
Gujrat Murree Pindigheb
GujranwalaMontgomery Chakwal Khan Fateh-jang
Gujar
Narowal Hafizabad Bhalwal TallagangMailsi
DaskaDipalpur PindSargodha Tek Singh
Phalia Shahpur
Dadan Khan Khanewal Allahabad
0
0 200 400 600
Distance of tehsil from Lahore
PBGsch5kmore Fitted values
17. Measuring political concentration
Political concentration: Extent of the replacement of traditional
elites by new families in a district at a particular time
Number of new families in power in district (it) /
Total number of political appointments in district (it)
How do we differentiate new from old?
Pre-partition nobility, mentioned in Punjab Chiefs and District
Gazetteers, land grants from the British, other titles, awards and
influence in the British era, elected in at least two previous
elections, etc.
18.
19.
20.
Number of
Electoral New
Constituencies Arrivals New_prop
Bahawalpur 35 6 0.171429
D. G. Khan 23 2 0.086957
Muzaffargarh 41 3 0.073171
Khanewal 20 4 0.20
Jhang 47 8 0.170213
Attock 26 3 0.115385
Jhelum 23 7 0.304348
Rawalpindi 52 28 0.538462