1. Building State Capability in the Context of Statistical Systems: Evidence, Analysis, Action
Michael Woolcock
Development Research group, World Bank
Kennedy School of Government, Harvard University
Workshop on Statistical capacity Development
OECD, Paris, france
11-12 December 2017
Free download available at:

2. Overview Rationale: Why this matters
State capability as a key development cooperation issue
Situating Problem-Driven Iterative Adaptation (PDIA)
Evidence, Analysis: Why an alternative is needed
The (unhappy) state of state capability
Explaining the limits of past, current approaches
Action: Putting PDIA to work in Statistical Capacity
Using different types of data to navigate complex spaces
Statistical capacity development as itself a complex challenge

3. 1. Rationale
Why building implementation capability is a key development cooperation issue
1. The development challenges only get harder
SDG 3: Healthy lives, ‘for all at all ages’
SDG 4: Education: from enrollment to ‘lifelong learning… for all’
SDG 16: Peace, inclusion, ‘justice for all’, accountability ‘at all levels’
Such laudable goals raise expectations on states…
… but current capability is low, and population growth alone (esp. in Africa) will double in next 30 years…
…while citizens and firms will require ever-better education, health, etc
Especially important in fragile states…
…but also rising middle-income countries
With rapidly rising demands on data quality, quantity and interpretation

4. ‘Like running Firefox on DOS…’ The virtues and limits of adaptive approaches
2. Any policy is only as good as its implementation
SDG 17: “Strengthen the means of implementation…”
But technical and adaptive problems are really different
IFI systems were (primarily) designed to engage with the former
…so, when asked to engage with adaptive problems, their operating systems routinely “crash”
Erecting hospitals, providing bandages isn’t curative care
Problem-Drive Iterative Adaptation (PDIA) as one option among several in the broader ‘Doing Development Differently’ (DDD) movement
Complement to, not substitute for, orthodox approaches
Not seeking ‘paradigm’ status; best suited to addressing ‘complex’ projects (or aspects thereof)
Quick overview of PDIA

5. How does PDIA differ? “Big D” (e.g. WB, agencies) “small d”
(e.g. NGOs)
PDIA
What drives action?
Pre-determined solutions (“institutional mono-cropping”, “best practice”), more inputs
Niche, parallel solutions (via variety of antidotes – e.g. “participation” “community driven”)
Planning for action?
Lots of advance planning (implementation of secondary importance)
Boutique; starting very small with no plans for scale
Feedback loops?
Monitoring (short, on financing, compliance, inputs) and Evaluation (long feedback loop on outputs, maybe outcomes)
Casual; geared to advocacy, not systemic learning
Scale?
Top-down: the head learns, implementation is just muscle (“political will”)
Small is beautiful…
Or, just not logistically possible

6. How does PDIA differ? “Big D” (e.g. WB, agencies) “small d”
(e.g. NGOs)
PDIA
What drives action?
Pre-determined solutions (“institutional mono-cropping”, “best practice”), more inputs
Niche, parallel solutions (via variety of antidotes – e.g. “participation” “community driven”)
Problem-Driven: looking to solve particular problems, locally nominated and prioritized
Planning for action?
Lots of advance planning (implementation of secondary importance)
Boutique; starting very small with no plans for scale
Assuring authorizing environment promoting positive deviation, purposive crawl of the design space
Feedback loops?
Monitoring (short, on financing, compliance, inputs) and Evaluation (long feedback loop on outputs, maybe outcomes)
Casual; geared to advocacy, not systemic learning
MeE: integration of rigorous “experiential” learning into tight feedback loops
Scale?
Top-down: the head learns, implementation is just muscle (“political will”)
Small is beautiful…
Or, just not logistically possible
Diffusion of feasible practice across organizations and communities of practitioners

7. 2. Evidence, Analysis
Only the 13 ‘historically developing countries’ (in green) are on a plausible path to strong capability by the end of the 21st C
Rapid negative
(g<-.05)
Slow
Rapid positive
(g>.05)
Negative
(-.05<g<0)
Positive
(0<g<05)
Strong (SC>6.5)
BHR, BHS, BRN
CHL(0), SGP(0), KOR(0), QAT(0)
ARE(0) 8 3 4 1
Middle (4<SC<6.5)
MDA, GUY, IRN, PHL, LKA, MNG, ZAF, MAR, THA, NAM, TTO, ARG, CRI
PER, EGY, CHN, MEX, LBN, VNM, BRA, IND, JAM, SUR, PAN, CUB, TUN, JOR, OMN, MYS, KWT, ISR
KAZ(10820), GHA(4632), UKR(1216), ARM(1062), RUS(231), BWA(102),
IDN(68), COL(56), TUR(55), DZA(55), ALB(42), SAU(28), URY(10), HRV(1) 45 13 18 14
Weak
(2.5<SC<4)
GIN, VEN, MDG, LBY, PNG, KEN, NIC, GTM, SYR, DOM, PRY, SEN, GMB, BLR
MLI, CMR, MOZ, BFA, HND, ECU, BOL, PAK, MWI, GAB, AZE, SLV
UGA(6001), AGO(2738), TZA(371), BGD(244), ETH(103), ZMB(96) 32 12 6
Very weak (SC<2.5)
YEM, ZWE, CIV
SOM, HTI, PRK, NGA, COG, TGO, MMR
SDN(7270), SLE(333), ZAR(230), IRQ(92)
NER(66), GNB(61), LBR(33) 17 7
102 30 40 28
Source: Authors’ calculations of state capability from Quality of Government, Failed State Index, and World Governance indicators
Number in brackets is years to the level of the lowest OECD country (‘Portugal’)

8. Education in India Successive cohorts doing worse…
Parents smuggling answers to students doing exams (Bihar, India)

9. Poor service, even for the wealthy…
Source: Kaffenberger and Pritchett 2017

10. …but especially in ‘complex’ tasks (i. e
…but especially in ‘complex’ tasks (i.e., high discretion, transaction-intensive, contentious)
Anthropology Sociology Popular Culture
On implementing a national
social protection program
for women in rural India
On implementing responses
to the AIDS crisis in Malawi
On implementing the welfare
state under ‘austerity’ in Britain

11. 3. Action How can we ‘do development differently’ (DDD)?
Building state capability by expanding local successes
Different kinds of implementation problems require different kinds of solutions
Thus need different kinds of evidence and strategies
Organizations, like individuals, acquire capability through practice
Cf. learning languages, playing musical instruments, sports
“You can’t juggle without the struggle”
A wide array of initiatives already underway around the world
Disseminated through an evolving global community of practice

12. Applications to statistical capacity
Using data to navigate complex spaces
Documenting, learning from wide variation in outcomes (ex post)
Using real-time feedback to build implementation capability (in situ)
Both require integrating different types, sources of evidence
Including forms amenable to non-literate / non-numerate citizens
Afghanistan (theatre, photography), India (Social Observatory)

13. Applications to statistical capacity
Using data to navigate complex spaces
Documenting, learning from wide variation in outcomes (ex post)
Using real-time feedback to build implementation capability (in situ)
Both require integrating different types, sources of evidence
Including forms amenable to non-literate / non-numerate citizens
Afghanistan (theatre, photography), India (Social Observatory)
Collecting, curating and interpreting data as themselves complex development challenges
Even data collection practices with long, venerable history (e.g., SNA) are far from stellar (Jervens)
‘Open Data Initiative’ as an important step…
Accommodating critics, rivals using this data also requires high capability