Designing Shock Protection for Vulnerable Households: A Howto Guide

1
Designing Shock Protection for Vulnerable
HouseholdsA How-to Guide
Jonathan Zinman Dartmouth College
BASIS CRSP Policy Conference Impact Evaluation of
Innovations in Rural Finance June 13, 2006
2
Overview of Talk

• What are shocks?
• Why should we focus on interventions to help
  smooth shocks?
• What works? Dont know much, but on the right
  track
• Scientific standard the right one given stakes
• Research adds value what works and why
• What do we need to know? (The substance)
• Are there underlying frictions in micromarkets?
  Failures?
• How do our targeted consumers/entrepreneurs make
  decisions?
• What works?
• (Traditional evaluation approach answers only a
  small part of 4 at best)
• How do we get there? The process
• Experimentation and the learning organization
• Real innovations guided by science and
  practitioner experience

3
Why Worry About Shocks?

• Temporary setbacks should be smoothed
• Long-run prospects unaffected
• Smoothing getting households, businesses, back
  on their feet
• Failure/inability to smooth is inefficient
• Utility is concave in consumption a little
  consumption means a lot (of marginal utility)
• Starving is very costly
• Dont want viable businesses to fail
• Irreversible investment problem high
  continuation value
• Discourages risk-taking
• Stock options example
• Safety net for permanent shocks

4
How Do People/FirmsDeal With Shocks?

• Saving (for a rainy day)
• Borrowing (on the rainy day)
• Insurance (its raining cash)

5
Why Should WeWorry About Shocks?

• Some reason (arguably ample reason) to believe
  that microfinance and microinsurance markets
  dont function well
• Information asymmetries
• Consumer biases
• Undersaving. Underinsuring?
• Overborrowing on sunny days
• Inadequate safety net
• Especially in LDCs
• Incomplete formal markets perverse
  consequences? (crowd-out informal)
• Theory also shows that when markets are missing,
  shocks can have disastrous consequences
• Poverty traps
• Debt traps

6
Why Should WeWorry About Shocks?

• Much more theory, practice than conclusive
  evidence here
• Evidence is somewhat indirect when shocks
  happen, how well can households smooth/recover?
• Shocks are a problem Gertler and Gruber AER 2002
  Gertler et al 2004
• Shocks not a big problem (informal
  markets/networks do the job) Townsend 1995,
  etc.
• Morduch 2002 (World Bank Observer) reviews

7
Can Micromarket Interventions Help?

• The little available evidence suggests yes
• Gertler et al microcredit and health shocks
• Ongoing BASIS-funded research, including.
• Karlan-Zinman 2006
• more consumer credit access
• avoid shocks, smoother consumption

8
What works?

• More research needed.
• Scientific standard appropriate given level of
  resources committed to development projects
• Randomized-control trials gold standard
• Research adds value, saves donor money in
  medium-haul
• What works.
• Work relative to what? (opportunity cost)
• Replicate effective interventions
• Why (doesnt) work?
• Guide next round of innovations

9
So What Do We Need to Know?The Three Questions

• Are there market frictions?
• How do targeted folks make decisions?
• What works?
• s 1-2 about whether and how to design
  interventions that works
• the R in RD
• 3 about what works
• Evaluation as traditionally defined

10
So What Do We Need to Know?

• Are there market frictions?
• Important because tells use whether and how to
  intervene
• Example asymmetric information
• Biggest motivation for microfin/insurance
• But little empirical evidence
• Moral hazard and adverse selection have different
  practical and policy implications
• E.g., invest in enforcement? Screening?

11
Asymmetric Information Progress

• Karlan-Zinman 2005 randomize offered/actual/and
  future rate in a consumer credit market (Manfred)
• Gine-Karlan-Zinman 2006 randomize premiums
  offered on private hospitalization insurance
  (Omar)
• Karlan-Mullainathan-Zinman randomize enforcement
  (Diego)

12
So What Do We Need to Know?

• 2. How do targeted folks make decisions?
• If we build it, will they come?
• What should we build in the first place?
• Example are people rational or
  psychological?
• Impacts how market, price, design products

13
Client Decision Making Progress

• Example Say you want to help people save (or
  access credit or insurance) to help protect
  against shocks.
• Only psych agents want commitment devices
  (Omar SEED also anti-smoking)
• Only psych agents may respond more to
  marketing/presentation than to price (BKMSZ, with
  Manfred)
• Rational will benefit from expanded access to
  consumer credit, psych may not (Manfred)

14
Smoothing and Targeting

• Thinking about specific motivations for
  intervention (Questions 1-2) raises another key
  question
• Whom should we target?
• Why has consumer credit been the bastard
  stepchild of microcredit?

15
Why not Target Consumers?

• Pat answer consumer credit not productive
• But theory and (a bit of) evidence suggest that
  benefits from smoothing are large, and that
  micromarkets could help
• Evidence shows that entrepreneurial credit used
  for consn smoothing (Menon) and vice versa
  (Karlan-Zinman credit cards in the U.S.)

16
Why Not Target Consumers?

• Pat answer consumers dont know whats good for
  them. Biases lead to overborrowing (BKMSZ
  Stango-Zinman)
• But consumer the business in closely-held
  firms
• Do people wear different hats?
• Are entrepreneurs more rational than their
  wage-earning counterparts?

17
Question 3. What Works?

• What is what?....
• Product design, market access has many
  elements
• Pricing
• Other contract terms (e.g., loan maturity)
• Risk assessment
• Enforcement
• Product development
• Product presentation (marketing and sales)
• Targeting

18
What Works?Beyond Evaluation Experimentation

• What helps consumers deal with shocks?
• All of our related projects answer this question
• We test specific interventions/innovations. More
  examples
• Does access to health insurance improve
  well-being? (ongoing)
• Expanding consumer credit supply DOES improve
  well-being (Manfred)
• Does expanding access to microcredit via credit
  scoring improve household well-being? (Reggie)
• But approach is more ambitious than traditional
  evaluation

19
What Works?Beyond Evaluation Experimentation

• Traditional evaluation
• After-the-fact
• Often ad-hoc
• Experimentation
• Ground-floor and continuous
• architecture vs. inspection approach
• Build in tests of the why questions as well
• Feed back into design and operations
• Experiment again

20
Experimentation the Learning OrganizationA
Virtuous Cycle
21
Experimentation asProcess Innovation

• Experimentation does more than answer the 3
  questions
• Can transform organizations
• Mircomarkets democratize finance
• Experiments democratize scientific approach to
  business, policy

22
Experiments are Process InnovationThe Learning
Organization

• Experiments make organizations more
• Strategic
• Systematic
• Scientific
• Capable
• Examples Green Bank ICICI Other credit card
  companies HR Block Amazon)
• Experimental approach also makes knowledge
  transfer easier methodology and assumptions are
  transparent

23
Summing Up

• Research adds value
• What works and why
• Helps with evaluation and design
• What we need to do
• Answer the 3 questions
• Frictions/failures?
• Decision Making by our targets?
• What works?

24
Summing Up

• How we do it process innovation
• Learn by doing
• Learn from doing
• The learning organization.

25
Experimentation the Learning OrganizationA
Virtuous Cycle