Technology and Low-Income States

Development Economics

September 24, 2024

Building State Capacity: Biometric Smartcards (2016)

• Many developing money invest in poverty-reducing measures, but the money is lost to inefficiencies and corruption along the way (leakage)
• Biometric payment schemes (theoretically) give an effective way to authenticate and transfer payments in environments with limited literacy
• Issues:
  • It's hard to implement
  • Rent-seekers may limit effectiveness
  • Technical problems
  • Displaced corruption to other places
• Rolled out smartcard in 157 subdistricts with 19 million rural population (an intervention at scale)

Technology and Low-Income States

Development Economics

September 24, 2024

Building State Capacity: Biometric Smartcards (2016)

• Faster collection of payments
  • 20 percent less time collecting and up to 10 days sooner
  • Fewer delays
• Reduced leakage
  • 24 percent increase in household earnings
  • 41 percent reduction in work-fare program leakage
  • 47 percent reduction in social security pension leakage
• No loss on the extensive margin: 7.1 percentage points increase in households participating
• No evidence of vulnerable populations made worse off
• "Ghost" workers as one mechanism

Technology and Low-Income States

Development Economics

September 24, 2024

E-governance (2020)

• If a program is funded nationally but implemented locally, how do we transfer the funds?
  • Notoriously "leaky" to send funds in advance
  • "Just-in-time" funding emerging as an alternative
• Admin data shows local expenditures (the GP) reduced by 24%
  • Some delays in funding

Technology and Low-Income States

Development Economics

September 24, 2024

E-governance (2020)

Technology and Low-Income States

Development Economics

September 24, 2024

E-governance (2020)

• If a program is funded nationally but implemented locally, how do we transfer the funds?
  • Notoriously "leaky" to send funds in advance
  • "Just-in-time" funding emerging as an alternative
• Admin data shows local expenditures (the GP) reduced by 24%
  
  • Some delays in funding
    
• GP officials successfully lobbied to end the program but program was eventually rolled out nationally after full results
  
• Full results:
  
  • Funds distributed to beneficiaries unchanged
    
  • Some improvement at the extensive margin
    
  • 5% reduction in the number of households unmatched (potential ghosts)
    

Technology and Low-Income States

Development Economics

September 24, 2024

Band-aid on a Corpse (2008)

• Multi-tiered, local public health system in India should be ideal way to deliver in a developing context, but falls short. Why?
• One reason might be high absenteeism of nurse providers
  • Health centers closed 56% of the time
  • 43% absenteeism for nurses
• Experimentally implement incentives (fines and punishments) for nurses that are extremely effective after 6 months

Technology and Low-Income States

Development Economics

September 24, 2024

Band-aid on a Corpse (2008)

Why does the intervention fall over time?

Technology and Low-Income States

Development Economics

September 24, 2024

Band-aid on a Corpse (2008)

Poorly aligned system incentives:

• Machine breaks => no need to check in => don't show up to prevent machine repair => continue to be absent
• Officials allowing more exemptions

Technology and Low-Income States

Development Economics

September 24, 2024

No Bulls (2023)

• What does fixing information asymmetry do to service provision?
• Looks at AI (not that kind of AI) in rural Pakistan
• Farmers cannot usually directly observe quality of vet care (and results signal is noisy
• If we provide this information directly (via a Yelp-like service), find that farmers get 25% higher AI rates, but pay no higher prices
  • Does not involve switching vets
  • Instead, driven by vets exerting more effort

Technology and Low-Income States

Development Economics

September 24, 2024

No Bulls (2023)

Technology and Low-Income States

Development Economics

September 24, 2024

No Bulls (2023)