Somchai Jitsuchon Thailand Development Research Institute 5 April 2006 Vientiane, LAO PDR

Preview:

DESCRIPTION

Poverty Policy Formulation, Monitoring and Reporting Results: Thailand ’ s Experience. Somchai Jitsuchon Thailand Development Research Institute 5 April 2006 Vientiane, LAO PDR. Outline. 1. Overview of Thailand’s Poverty 2. Poverty Policy Formulation - PowerPoint PPT Presentation

Citation preview

1

Somchai Jitsuchon Thailand Development Research Institute

5 April 2 0 0 6 Vientiane, LAO PDR

Poverty Policy F ormulation, Mo nitoring and Reporting Results:

Thailand’s Experience

Outline

1. Overview of Thailand’s Poverty

2. Poverty Policy Formulation Fundamental Changes of Policy Architects National vs. Area-based Policy

3. Monitoring and Reporting Poverty Data Small Area Estimation Poverty Map

3

Thailand’s Poverty Overview

Poverty TrendThailand’s Poverty Declined Rapidly over the Past 40-50 Years

If using old definition (before 2004), head-count ratio would be only around 5%

4542

34

28

1915

1820 21

1915

11

51

45

38

33

25

1719

21 2119

16

11

0

10

20

30

40

50

60

1986 1988 1990 1992 1994 1996 1998 1999 2000 2001 2002 2004

Consumption Poverty Income Poverty

0

0.1

0.2

0.3

0.4

0.5

0.6

Gini C

oeffic

ient

But Income Inequality Remains High..One of the World’ Highest

Income Share by IncomeQuintile

1986 1988 1990 1992 1994 1996 1998 2000 2002 2004Poorest 20% 55.87 54.37 56.97 58.98 57.23 56.53 56.13 57.45 55.91 54.86Quintile 2 20.02 20.62 19.50 18.90 19.68 19.91 19.82 19.83 20.07 20.16Quintile 2 12.09 12.38 11.70 11.11 11.67 11.83 12.00 11.50 12.07 12.41Quintile 2 7.67 8.05 7.54 7.06 7.35 7.55 7.75 7.27 7.72 8.04Richest 20% 4.36 4.58 4.29 3.96 4.07 4.18 4.30 3.95 4.23 4.54Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00Richest/Poorest

12.81 11.88 13.28 14.90 14.07 13.52 13.06 14.55 13.23 12.10

These ratios are frequently quoted in public debates on poverty/inequality

Consequences on Tar get Groups

Destitute poor (absolute poverty) has been dwindling in nu

mber, but some pockets of chronic poverty might exist.

Relative poverty increasingly important stubborn to economic growth, if inequality persists.

began to dominate public debates/policies. More ‘poverty measur

es’ are devised for the relative poor, not the poorest.

Problems of vulnerability also increasingly important, but

still largely neglected.

Rural and urban poverty more linked than in the past, due t

o convergence of economic activities.

8

Poverty Po licy Formulation

Changes in Poverty Policy Architects

In the past national poverty policy either did not ex

ist, or was an unsubstantial part of ‘Nati onal Plan’. Poverty declined mainly thr ough growth process.

Technocrats were thus key (and sole) ar chitects of poverty policy at national lev

el. Politicians mostly influenced sectoral po

- licies, or minor area specific policies.

Changes in Poverty Policy Architects

Present Poverty policy was nationalized by the TR

- T party around the year 2 0 0 0 1 , alo ngwithgl obal i nt er est i n pover t y r educt i on . Politicalsuccess of TRT par t y was par t l y due t hisshi f t .

‘ National Plan’ now plays very little role, a long with its technocrat architects. Pover

ty policy was basically transferred to politi cians’ hands.

Consequently, mostpovertypoliciesar e nowmor e t ar get i ng, mor e sect or a l. Oneexcept i on i s t he uni ver sal heal t h ca rescheme.

-National vs. Area based

Mostofthetime(pastor pr esent ), al l maj or pover t y pol i c iesarecent r al l y concept ual i zed and i mpl em entedbycent r al gover nment ’s bur eaucr at i c arms.

However, therehasbeenat t empt t o decent r al i zed i mpl ement ations to ‘local governments’.

Forexample, provincesaregrantedmorepower (fi nanci al and bur eaucr at i c). More room for local initiatives. But most of l ocal eff or t s i s st i l l devot ed t o car rynat i onal pover t y pol i ci es desi gned b ynat i onal pol i t i ci ans.

Consequences The current policy quickly favors the r

elative poor, rather than the absolute poor (except the universal health care

).

There is urgent need for reliable pove rty data at disaggregated areas level (

at least at provincial level).

Also urgent need for high frequency p overty data (at least annually), to sup

port the ‘Poverty Eradication within 3 years’ agenda by TRT party leader.

13

Poverty Data

Poverty Data Household Surveys on Consumption/Income

Census (pop census, agri cultural census, industrial

census) Administration Records Participatory Reports Hybrids

Thailand’s Pov erty Data

A. Use household surveys (SESs) alone. OK at national/regional level - but inadequate for true area based po licy implementations (e.g. SESs produc

e zero poverty in many provinces).

B. Rural Village Data: Nrd2C and BMN(basic minimum need)Ad hoc ‘poverty line’ composite index (monetary & non-monetary), with ad hoc formula

C. Poverty Registration (TRT part y initiative)completely self-report

Nrd2C/BMN

Census-type Rural Survey. Nrd2C every 2 years, BMN every year.

Data collection/reporting by village committees.

There are doubts that some villages do not actually collect data, but report anyway.

Strengths: - Allow non monetary dimensions of poverty Frequent, yet Low cost

Weaknesses: cover only rural areas data quality may suffer from non-standard

data collection method and (more importantly)impartial evaluation.

2Nrd C VariablesVariable NRd2C Development Variables Variable NRd2C Development

Variables1) Infrastructure 4) Water Sources

I1 Land Title and Types i20 Drinking WaterI2 Electricity Accessibility i21 Non-Drinking WaterI3 Transport and Communication i22 Water for AgricultureI4 Right to Use Land 5) Knowledge, Education, and

Culture2) Outputs, Income, and Employment i23 Villagers’ Education Level

I5 Village Business i24 Continuing Education Ratei6 Earning and Employment i25 Knowledge Provided by the

Governmenti7 Wage Rate i26 Places for Educationi8 Outputs from Rice Farming i27 News and Information Service

Placesi9 Outputs from Other Plant-Farming i28 Activities in Religion, Cultural,

and Sporti10 Other Occupation 6) Natural Resources and

Developmenti11 Migration to Work i29 Foresti12 Farmer Grouping and Cooperation i30 Soil Resourcesi13 Off-Season Agriculture Activities i31 Water Resource

3) Healthi14 Protection in Drug Usagei15 Protection from Contagious

Diseasesi16 Mental Healthi17 Environment SanitaryI18 Work SafetyI19 Participation in Health/ Sanitary

Activities

5Rural Villages that ‘fail’ more than criteria are ‘targeted villages’

Individual level data (not househo -ld), nation wide.

- Completely self reporting method potential use as ‘poverty map’ Strengths

Allow ‘poor’ people to report their specif ic problems. Finding solutions to povert

y problem is thus straightforward Weaknesses

- Over report of problem potential severe targeting problem (regi

- -stration by non poor, and non registrati on by poor)

Used more as political propaganda

Poverty Registration

- Mis targeting Proble m of

Poverty RegistrationNon-Poor Poor Total

Non-registered 82.0% 71.6% Registered 18.0% 28.4%Total 100.0% 100.0%Within Registered 89.9% 10.1% 100.0%

If not complimented by other datasources,

7 1 .6 % of poor people will be neeeeeeeee

20

An AdditionalTool: Small Area Est imation Poverty Map

Potentials and Performance

SAE Poverty Map

Simple Idea: Get estimates of household income/consumption on large dataset (usually Census) based on models built on household surveys (SESs).

SESs have both (Y,X) but Census has only X.The models also allow for ‘location effects’

Advantages: •Combine Census’s Large Coverage with SESs’ Reliability.•Esitmated Y’s enable many applications (poverty, inequality, social security).

Limitations: •Only monetary definition of poverty.•Census is every 10 years (may use other dataset---BMN). •Huge data work, complicated econometric procedures.

chchch uy βxln

chcchu

First Map in 2000 (Joint projectNESDB/NSO/WB/TDRI) Use household survey 2000, Census 2000, and

village survey 1999 (provides location variables for rural map)

Second Map in 2002 (Join projectNSO/NESDB/WB/TDRI) household survey 2002, Census 2000, and village

survey 2002

- Third Map in 2004 (on going effort)

SAE Poverty Maps in Thailand

2000Comparing SAE Ma 1 9 9 9 2 ‘

p’Nrd2C Classification

SAE Poverty Map Non-Target ornon-matched

Target Total

Non-Poor 39,781 9,511 49,292

Poor (30% up) 12,296 4,707 17,003

Total 52,077 14,218 66,295

The two maps are significantly differen t. Either (or both) may have the proble m of including the wrong villages as we

ll as excluding the right villages. Whic h one?.

Why Validation?•Survey Sampling Errors • Model Error• Omitted Variable problem• Inconsistency between SAE and Nrd2C

First Field Validation• To verify SAE 2000 Map, and 1999 Nrd2C•ee eeeeee eee eeeeeeeee eeeeeeee eeee

th) Second Field Validation

• 2002To verify SAE Map• 3In provinces: Pitsanulok (north), Nonb

eeeeeee( ), (l)

Field Validation

FFFFF FFFFFFFFFF FF FF (2000

Findings:• 2Both SAE and Nrd C maps did well in som FFF FF FFFFFFF/, .• F FFFF FFFFFFF F FFFFF FFFF2 ,

• outdated data• ‘ target villages’ were targeted from ‘developmen t dimensions’ not relevant to poverty level.• Doubtful reporting (less often than commonly tho ught), possibly to manipulate government budge t allocation.

• For SAE map, most problems are wrong preFFFFFFFF FF FFFF FFFFFFF (arise possibly from fail ure to define appropriately ‘location variables’ impo rtant to general income level, e.g. how widespread t - he rubber tree growing was)

• FFFF FF FF FFFF FF FFFFFFFFFF FFFF-FFFFF FFFF‘’ . Both can be used together, with improved data quality/models.

Focus on accuracy of ‘poverty ranking’ at sub-district level (not poverty rate)

Method1. Compare with related official

documents (e.g. tax record, BMN)

2. Key Informant Interview (District chiefs, local development agencies, provincial statisticians )

3. Area Observation (geographic, soil quality, water sources, business community, house conditions)

4. Interview with People (farmers, shop owners, street walkers)

Second Field Validati on (2002 Map)

Interviews with people gave the most reliable information

- Head count Rati o Municipal (urban) - Non municipal (rur

al)

Poverty Map for Pitsanulok (disa ggregated at district level)

Poverty Map for Pitsanulok (Gini coefficient at district level)

Verification P oint:

Wang Tong district had high inequalit

y WangTong

Poverty rates va ri edconsi derabl

y

- Some sub districts were clearly bette

- r off(Pai Khodon, Baan Grang).

-Central District Pitsanulok (head - count at sub district level)

Central District: Valid ation Points

BMN Database SAE Poverty RankSub-districts

Income Rank Poverty Rank Income ConsumptionBaan Krong (A) 1 12 19 15Baan Grang (B) 15 9 1 2Pai Khodon (C) 13 17 2 1Ta Poh (D) 19 2 4 3

- With BMN ranking: (B) and (C) sub districts will get morebudget

With SAE map rannking: Both should get less budget tha n (A)

(): Interview with a farmer

Baan Grang (B): Group In terview (farmers)

- -(A) wascl ear l y a bet t er off sub di st r ict, suppor t i ng SAE r esul t s

One of the poorest province

According to SAE maps, all districts were poor, except for Noan Sung district

However, one key informant insisted Noan Sung was relative poorer than other district.

Verification Point: Which was true?

Poverty Map for Nongbualumpoo (head-count at district level)

Noan Sung

Noan Sung: A Farm Field with Double

Cropping

All other districts could not do double cropping.

Also found additional occupation (fishery)

Preliminary Evaluation of SAE

method SAE Poverty Map is fairly accurate in

predicting poverty ranking by area.

Poorer (Reality)

Better-Off (Reality)

Poorer (SAE) Predict

Predict few (X)

Better-Off

(SAE)Not

PredictPredict

SAE can benefit from improvement in the accuracy of surveyed income/consumption.

Need to simplify the method (underway), and overcome the theoretical and empirical issues of poverty map updating.