Information Technology, Transparency and Corruption in Public - - PowerPoint PPT Presentation

▶
information technology transparency and corruption in
SMART_READER_LITE
LIVE PREVIEW

Information Technology, Transparency and Corruption in Public - - PowerPoint PPT Presentation

Feb 10, 2024 173 likes •572 views

Introduction Context, intervention and evaluation Results Conclusion Appendix Information Technology, Transparency and Corruption in Public Programs: Evidence from a Large Workfare Program in Bihar, India Cl ement Imbert with Abhijit

slide-1
SLIDE 1

Introduction Context, intervention and evaluation Results Conclusion Appendix

Information Technology, Transparency and Corruption in Public Programs: Evidence from a Large Workfare Program in Bihar, India

Cl´ ement Imbert with Abhijit Banerjee, Esther Duflo, Santhosh Mathew and Rohini Pande IGC Growth Week, London, September 2014

slide-2
SLIDE 2

Introduction Context, intervention and evaluation Results Conclusion Appendix

Motivation

  • Public Programs in developing countries are often not

implemented as they should

  • Corruption and leakage in funds is a part of the problem.
  • MNREGS: 40%-60% leakages in 2007-08 (Imbert Papp 2011).
slide-3
SLIDE 3

Introduction Context, intervention and evaluation Results Conclusion Appendix

Motivation

  • Public Programs in developing countries are often not

implemented as they should

  • Corruption and leakage in funds is a part of the problem.
  • MNREGS: 40%-60% leakages in 2007-08 (Imbert Papp 2011).
  • This is not for lack of trying to reduce corruption:
  • online documentation of expenditures (nrega.nic.in) + audits.
  • payments through post-offices and banks mandatory.
  • leakage down to 20% in 2011-12 (Imbert 2014).
slide-4
SLIDE 4

Introduction Context, intervention and evaluation Results Conclusion Appendix

Motivation

  • Public Programs in developing countries are often not

implemented as they should

  • Corruption and leakage in funds is a part of the problem.
  • MNREGS: 40%-60% leakages in 2007-08 (Imbert Papp 2011).
  • This is not for lack of trying to reduce corruption:
  • online documentation of expenditures (nrega.nic.in) + audits.
  • payments through post-offices and banks mandatory.
  • leakage down to 20% in 2011-12 (Imbert 2014).
  • But there are gaps in the system:
  • Entry delayed and incomplete (only 60% really required).
  • Accountability mostly through command and control.
slide-5
SLIDE 5

Introduction Context, intervention and evaluation Results Conclusion Appendix

Information technology and Corruption

  • Can real time data processing made possible by better

information technology reduce corruption?

  • Mularidharan et al. (2014) find that biometric smart cards to

handle payments to beneficiaries “delivered a faster, more predictable, and less corrupt payments process without adversely affecting program access”.

slide-6
SLIDE 6

Introduction Context, intervention and evaluation Results Conclusion Appendix

Information technology and Corruption

  • Can real time data processing made possible by better

information technology reduce corruption?

  • Mularidharan et al. (2014) find that biometric smart cards to

handle payments to beneficiaries “delivered a faster, more predictable, and less corrupt payments process without adversely affecting program access”.

  • Our intervention is complementary: we look at the process by

which money goes from the State to the Panchayat (before it reaches beneficiaries)

slide-7
SLIDE 7

Introduction Context, intervention and evaluation Results Conclusion Appendix

Intervention

  • Status quo: to obtain money the Panchayat sends request to

the Block, which sends a request to the district, which delivers the money to the Panchayat’s account

  • New system: the Panchayat enters worker payments details

and uploads them to the server, and that immediately generate a transfer to the Panchayat account.

  • A “full scale” experiment:
  • intervention rolled out by the Bihar Government in 12 districts

(covering 33 million rural people).

  • we randomly selected one third of the blocks in each of these

districts to receive the new system.

slide-8
SLIDE 8

Introduction Context, intervention and evaluation Results Conclusion Appendix

Fund Flow in Control

St Status tus ¡ ¡Qu Quo

State ¡Pool ¡in ¡Central ¡Bank ¡of ¡India ¡ ¡

CPSMS ¡ ¡ Access ¡ Panchayat ¡ Savings ¡Account ¡

Fund ¡Request ¡

Panchayat ¡

Fund ¡Transfer ¡

Block ¡ District ¡

1 2 3 4 5

slide-9
SLIDE 9

Introduction Context, intervention and evaluation Results Conclusion Appendix

Fund Flow in Treatment (Labor Payments only)

In Inter erven en'on

  • n

State ¡Pool ¡in ¡Central ¡Bank ¡of ¡India ¡ ¡ ¡

CPSMS ¡ ¡ Access ¡

Panchayat ¡ Savings ¡ Account ¡

Fund ¡Request ¡

Panchayat ¡

Fund ¡Transfer ¡

1 2

slide-10
SLIDE 10

Introduction Context, intervention and evaluation Results Conclusion Appendix

Parsing the intervention

  • 1. The intervention makes it more difficult/dangerous to cheat
  • In the regular system, entry into nrega.inc is delayed and

incomplete: only 60% of expenditures are accounted for.

  • In the new system, entry must correspond to a name/days

worked: it becomes immediately possible to audit.

  • Audits are few and far between but they do carry large

penalties (prison).

slide-11
SLIDE 11

Introduction Context, intervention and evaluation Results Conclusion Appendix

Parsing the intervention

  • 1. The intervention makes it more difficult/dangerous to cheat
  • In the regular system, entry into nrega.inc is delayed and

incomplete: only 60% of expenditures are accounted for.

  • In the new system, entry must correspond to a name/days

worked: it becomes immediately possible to audit.

  • Audits are few and far between but they do carry large

penalties (prison).

  • 2. The intervention also changes who can takes a cut:
  • In the regular system, cheating is done by the lower level. The

block level and district levels are in a position to collect a tax.

  • In the new system, the district level is entirely cut out. The

block level remains critical because it has the infrastructure to access the system.

slide-12
SLIDE 12

Introduction Context, intervention and evaluation Results Conclusion Appendix

Sample

12 Districts 69T 126C Blocks 1002T 2029C Panchayat

slide-13
SLIDE 13

Introduction Context, intervention and evaluation Results Conclusion Appendix

Data sources

  • On MGNREGS expenditures:
  • CPSMS: financial transactions from all Panchayat savings

account.

  • MIS: monitoring system of the Ministry of Rural Development.
slide-14
SLIDE 14

Introduction Context, intervention and evaluation Results Conclusion Appendix

Data sources

  • On MGNREGS expenditures:
  • CPSMS: financial transactions from all Panchayat savings

account.

  • MIS: monitoring system of the Ministry of Rural Development.
  • On MGNREGS employment and works:
  • Household survey: 9500 households in 195 blocks, May – July

2013.

  • Survey of 4165 MGNREGS asset randomly sampled from

nrega.nic.in.

slide-15
SLIDE 15

Introduction Context, intervention and evaluation Results Conclusion Appendix

Data sources

  • On MGNREGS expenditures:
  • CPSMS: financial transactions from all Panchayat savings

account.

  • MIS: monitoring system of the Ministry of Rural Development.
  • On MGNREGS employment and works:
  • Household survey: 9500 households in 195 blocks, May – July

2013.

  • Survey of 4165 MGNREGS asset randomly sampled from

nrega.nic.in.

  • On politicians and bureaucrats:
  • Mukhiyas interview in each surveyed village.
  • Asset declaration of all MNREGS employees.
slide-16
SLIDE 16

Introduction Context, intervention and evaluation Results Conclusion Appendix

Timeline

  • July 2012: Preparation (prepare infrastructure)
slide-17
SLIDE 17

Introduction Context, intervention and evaluation Results Conclusion Appendix

Timeline

  • July 2012: Preparation (prepare infrastructure)
  • Sept 1st 2012: Launch
  • Sept 18th: State Pool runs dry.
  • Dec 11th: State Pool replenished.
  • Dec 15th: Strike of Panchayat Personnel
slide-18
SLIDE 18

Introduction Context, intervention and evaluation Results Conclusion Appendix

Timeline

  • July 2012: Preparation (prepare infrastructure)
  • Sept 1st 2012: Launch
  • Sept 18th: State Pool runs dry.
  • Dec 11th: State Pool replenished.
  • Dec 15th: Strike of Panchayat Personnel
  • April 1st 2013: Intervention is rolled back.
  • May 15th - July 15th 2013: Endline survey
slide-19
SLIDE 19

Introduction Context, intervention and evaluation Results Conclusion Appendix

Expenditures (CPSMS data)

slide-20
SLIDE 20

Introduction Context, intervention and evaluation Results Conclusion Appendix

Weeks worked (Households)

Source: Household survey (May-July 2013)

slide-21
SLIDE 21

Introduction Context, intervention and evaluation Results Conclusion Appendix

Results Summary

  • Drop in Expenditure (about 25%) =0.25 million per

Panchayat.

  • Drop in number of days worked and payments according to
  • fficial data.
  • No drop in number of days worked or payments according to

households.

  • No decline in asset built according to official reports and

found in the field.

slide-22
SLIDE 22

Introduction Context, intervention and evaluation Results Conclusion Appendix

Results Summary

  • Drop in Expenditure (about 25%) =0.25 million per

Panchayat.

  • Drop in number of days worked and payments according to
  • fficial data.
  • No drop in number of days worked or payments according to

households.

  • No decline in asset built according to official reports and

found in the field.

  • Where is the missing missing money?
slide-23
SLIDE 23

Introduction Context, intervention and evaluation Results Conclusion Appendix

Liquid Assets of NREGA functionaries [partial data]

Log liquid assets During * Treatment Group

  • 0.109
  • 0.114

(0.185) (0.203) After * Treatment Group

  • 0.253**
  • 0.322**

(0.126) (0.132) During * Block Employee 0.650*** (0.116) After * Block Employee 0.362** (0.156) During * Treatment * Block Employee 0.00467 (0.205) After * Treatment * Block Employee 0.364* (0.217) Observations 2,746 2,746

slide-24
SLIDE 24

Introduction Context, intervention and evaluation Results Conclusion Appendix

Decline in Assets is consistent with the indirect evidence of reduction in Leakage

  • Overall decline in expenditures (average by block): 3.75-5

million

  • Overall decline asset one year after from MGNREGS

functionaries (average by block): 2.4 million.

  • The order of magnitude is right: the difference may be funds

lost by the district, non-money asset, and increased consumption.

  • Block officials not hurt: surprising.
slide-25
SLIDE 25

Introduction Context, intervention and evaluation Results Conclusion Appendix

Did households pay part of the costs?

  • No decline in payments reported by households.
  • Increased delays +35 days (/72 days in control!) but only in

the first period of intervention.

  • No change in consumption (May-July 2013 Survey)
slide-26
SLIDE 26

Introduction Context, intervention and evaluation Results Conclusion Appendix

Conclusion and epilogue

  • The new CPSMS led to a reduction in expenditure without

change in the level of transfers provided: less leakage.

  • This was done on a very large scale, in difficult circumstances
slide-27
SLIDE 27

Introduction Context, intervention and evaluation Results Conclusion Appendix

Conclusion and epilogue

  • The new CPSMS led to a reduction in expenditure without

change in the level of transfers provided: less leakage.

  • This was done on a very large scale, in difficult circumstances
  • However... this was shut down
  • The district officials hated the system and anecdotes have it

that it is their reluctance that ultimately convinced the state bureaucracy to pull back.

slide-28
SLIDE 28

Introduction Context, intervention and evaluation Results Conclusion Appendix

Infrastructures available in the Block

July ’12 Jan ’13 Apr ’13 Req. T C T T C Computer 3 1.32 1.06 2.49 2.06 1.61 DEO 3 1.22 0.86 2.20 1.75 1.27 Generator 1 0.67 0.56 0.97 0.90 0.85 Internet 1 0.38 0.33 0.85 0.71 0.60 Scanner 1 0.57 0.37 0.73 0.81 0.65 Printer 1 0.59 0.43 0.71 0.83 0.76 Sampled Blocks 69 126 66 69 123 Source: Phone surveys of Program Officers.

slide-29
SLIDE 29

Introduction Context, intervention and evaluation Results Conclusion Appendix

Treatment effect on daily spending (CPSMS)

Before Sept-Dec 12 Jan-Mar 13 After Treatment

  • 204.1
  • 842.5***
  • 1,433***
  • 113.1

(305.7) (262.7) (314.3) (291.0) Observations 2,918 2,919 2,920 2,920 District FE Yes Yes Yes Yes Mean in Control 5323 4393 4574 5207 Effect as % of Mean

  • 3.834
  • 19.18
  • 31.32
  • 2.173

Source: CPSMS Credit Debit Data. S.e. are clustered at the block level Total Estimated effect =-0.24million per Panchayat

slide-30
SLIDE 30

Introduction Context, intervention and evaluation Results Conclusion Appendix

Treatment effect on annual spending (MIS)

Apr 12-Mar 13 Apr 13-Mar 14 Expenditure items Labour Material Labour Material

  • 2.270***
  • 1.077**
  • 0.277

0.313 (0.760) (0.526) (0.725) (0.534) Observations 2,947 2,947 2,954 (0.760) District FE Yes Yes Yes Yes Mean in Control 13.83 7.717 13.61 8.373 Effect as % of Mean

  • 16.42
  • 13.96
  • 2.032

3.735 Source: nrega.nic.in. Annual Expenditures from MIS data. S.e. are clustered at the block level Estimated effect in CPSMS=-1114*212days=-0.24million Estimated effect in MIS is -0.33 millions

slide-31
SLIDE 31

Introduction Context, intervention and evaluation Results Conclusion Appendix

Participation

Since July 12 During Intervention Treatment

  • 0.0006
  • 0.0015

0.0030 0.0029 (0.0086) (0.0089) (0.0059) (0.0062) Observations 10,036 9,989 10,036 9,989 District FE Yes Yes Yes Yes HH Controls No Yes No Yes Mean in Control 0.0936 0.0936 0.0482 0.0482 Effect as % of Mean

  • 0.623
  • 1.626

6.293 6.071 Source: Household survey (May-July 2013) Standard errors are clustered at the block level

slide-32
SLIDE 32

Introduction Context, intervention and evaluation Results Conclusion Appendix

Days Worked

Work Spell ended: Before Sep-Dec Jan-Mar After Treatment

  • 0.166***

0.0347 0.128 0.0516 (0.0505) (0.132) (0.169) (0.357) Observations 9,966 9,966 9,966 9,966 District FE Yes Yes Yes Yes HH Controls Yes Yes Yes Yes Mean in Control 0.231 0.499 0.693 1.825 Effect as % of Mean

  • 71.77

6.959 18.48 2.830 Source: Household survey (May-July 2013) Standard errors are clustered at the block level

slide-33
SLIDE 33

Introduction Context, intervention and evaluation Results Conclusion Appendix

NREGA projects built

Registered Completed % Found (MIS) (MIS) in Survey Treatment 0.0494 0.372 0.0176 (0.263) (0.342) (0.0176) Observations 390 390 4,165 District FE Yes Yes Yes Mean in Control 13.82 2.126 0.855 Effect as % of Mean in C 0.357 17.48 2.057 Source: MIS and NREGA Asset survey (May-July 2013) Standard errors are clustered at the block level

slide-34
SLIDE 34

Introduction Context, intervention and evaluation Results Conclusion Appendix

Durable goods of mukhyas

Table 14: Treatment effect on Mukhiya asset holding

Assets%owned%by%the%Mukhiya% % TV% Two%wheeler% CD/DVD%player% Gas%stove% Bicycle% % (1)% (2)% (3)% (4)% (5)% (6)% (7)% (8)% (9)% (10)% Treatment% 0.0339% 0.0329% L0.0173% L0.0549% L0.0767% L0.0976% L0.0546% L0.0584% 0.109% 0.0966% %% (0.0695)% (0.0696)% (0.0953)% (0.0909)% (0.0677)% (0.0693)% (0.0809)% (0.0803)% (0.0949)% (0.0945)% Observations% 354% 354% 354% 354% 354% 354% 354% 354% 354% 354% District%FE% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes% HH%Controls% No% Yes% No% Yes% No% Yes% No% Yes% No% Yes% ControlMean% 0.648% 0.648% 1.009% 1.009% 0.455% 0.455% 0.725% 0.725% 0.721% 0.721% Effect%as%%% 4.763% 4.549% L2.723% L6.193% L21.66% L26.05% L9.103% L9.920% 13.22% 10.99% % Chair/Bed% Mobile%Phone% Four%Wheeler% Fridge% Asset index (11)% (12)% (13)% (14)% (15)% (16)% (17)% (18)% (19) Treatment% 0.404% 0.122% 0.0126% L0.0189% L0.0510% L0.0242% L0.0465% L0.0431% !.0347' %% (0.951)% (0.952)% (0.517)% (0.539)% (0.169)% (0.165)% (0.0488)% (0.0489)% (.0567)' Observations% 354% 354% 354% 354% 354% 354% 354% 354% 354' District%FE% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes' Controls% No% Yes% No% Yes% No% Yes% No% Yes% Yes ControlMean% 10.49% 10.49% 3.163% 3.163% 0.730% 0.730% 0.262% 0.262% % Effect%as%%% 2.654% L0.682% L0.784% L2.574% L6.781% L4.371% L19.86% L18.38% % Note:%The%unit%of%observation%is%a%Panchayat.%The%dependent%variable%is%the%number%of%assets%owned%by%the% Mukhiya%household.%The%data%was%collected%by%a%representative%survey%of%350%Mukhiya%from%12%sample%districts% households%in%MayLJuly%2013.%Treatment%is%a%dummy%which%is%equal%to%one%for%the%blocks%selected%for%the% intervention.%%

slide-35
SLIDE 35

Introduction Context, intervention and evaluation Results Conclusion Appendix

Cattles of Mukhyas

Table 15: Treatment effect on Mukhiya cattle holding

Assets%owned%by%the%Mukhiya% % Cows% Buffaloes% Goats% Chickens% Cattle&Index& % (1)% (2)% (3)% (4)% (5)% (6)% (7)% (8)% (9)& Treatment% E0.0891% E0.105% E0.0117% 0.00803% E0.224**% E0.199**% E0.0937*% E0.0867*% ..0946*& %% (0.218)% (0.215)% (0.113)% (0.114)% (0.0976)% (0.0930)% (0.0557)% (0.0520)% (.0519)& Observations% 354% 354% 354% 354% 354% 354% 354% 354% 354& District%FE% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes% Yes& Controls% No% Yes% No% Yes% No% Yes% No% Yes% Yes& ControlMean% 1.515% 1.515% 0.438% 0.438% 0.279% 0.279% 0.129% 0.129% 0% Effect%as%%% E8.613% E10.50% E3.379% 0.905% E78.50% E73.55% E71.18% E68.77% % Note:%The%unit%of%observation%is%a%Panchayat.%The%dependent%variable%is%the%number%of%animals%owned%by%the% Mukhiya%household.%The%data%was%collected%by%a%representative%survey%of%350%Mukhiya%from%12%sample%districts% households%in%MayEJuly%2013.%Treatment%is%a%dummy%which%is%equal%to%one%for%the%blocks%selected%for%the% intervention.%

slide-36
SLIDE 36

Introduction Context, intervention and evaluation Results Conclusion Appendix

Some increased delays in payments

Work Spell ended: Before Sep-Dec Jan-Mar After Treatment

  • 18.06

35.03** 15.84

  • 2.467

(24.03) (17.47) (9.770) (5.630) Observations 123 170 225 381 District FE Yes Yes Yes Yes HH Controls Yes Yes Yes Yes Mean in Control 73.44 72.80 49.95 38.41 Effect as % of Mean

  • 24.60

48.12 31.71

  • 6.423

Source: Household survey (May-July 2013) Standard errors are clustered at the block level

slide-37
SLIDE 37

Introduction Context, intervention and evaluation Results Conclusion Appendix

Confirmed by the Mukhiya

CPSMS creates delays Spontaneously When Prompted Treatment 0.120*** 0.106** (0.0429) (0.0425) Observations 354 354 District FE Yes Yes Mean in Control 0.119 0.144 Effect as % of Mean 101.4 73.44 Source: Mukhiya Survey S.e. clustered at the block level

slide-38
SLIDE 38

Introduction Context, intervention and evaluation Results Conclusion Appendix

No change in Consumption

Log Monthly Expenditures All Frequent Recurrent Rare Treatment

  • 0.00367
  • 0.00796
  • 0.0282

0.00284 (0.0215) (0.0170) (0.0268) (0.0393) Observations 9,041 9,040 9,026 9,024 District FE Yes Yes Yes Yes HH Controls Yes Yes Yes Yes