Making and Un-Making Poverty Anirudh Krishna Copenhagen, May 5, 2008 [Ak: insert picture]

Research Questions 1.ORIGINS: How do people come to be poor? 2.REMOVAL: How – for what reasons – do people escape? Basis Vectors…

Few answers available Stocks measured (rarely flows) Handful of panel studies in developing countries Rarely probe WHY questions Methods gaps…critical

Grassroots Investigations in 400 diverse communities of INDIA KENYA PERU UGANDA NORTH CAROLINA, USA Poverty Dynamics tracked for > 35,000 households

STAGES OF PROGRESS METHODOLOGY Seven Steps Step 1. Getting together a representative community group Step 2. Discussing the objectives of the exercise Step 3. Defining “ poverty ” collectively in terms of Stages of Progress

Stages of Progress (Rajasthan) 1. Food for the family 2. Send children to school 3. Some clothes to wear outside the house 4. Start repaying debts 5. Repair the existing shelter Poverty Cutoff 6. Dig a well 7. Purchase cows and buffaloes 8. Construct a pakka (brick) shelter 9. Purchase ornamentsProsperity Cutoff 10. Radio, tape recorder, refrigerator 11. Motorcycle 12. Tractor, car

Initial stages before the poverty cutoff: STAGERajasthanGujaratW. Kenya 1.Food 2.Primary education Some Clothing (to wear outside) 3.Some Clothing (to wear outside) Primary educationHouse repair (roof renovation) 4.Retiring accumulated debt Primary education 5.House repair (roof renovation) Small animals (chicken, sheep, goat) 6.Hiring in a small tract of land

STAGES OF PROGRESS METHOD Step 4. Define “ X years ago ” in terms of a well known signifying event Step 5. List all households. Ascertain stage for each household at present time and in earlier period Step 6. Categorize all present-day households: A. Poor 25 years ago and poor now (Remained poor) B. Poor then and not poor now (Escaped poverty) C. Not poor then but poor now (Became poor) D. Not poor then and now (Remained non-poor)

STAGES OF PROGRESS METHOD Step 7. For a random sample (25-40%) of households in each category, investigate reasons for change or stability I. Community inquiry: Comparative perspective Probing II. Household inquiry: To verify what the community has said and to go deeper into reasons Training is critical

Stages-of-Progress Methodology Robust local understandings: similar within contexts Relatively long time-horizons Triangulating recall data Close correspondence with “objective” data Identification of reasons

The Good News Escaped Poverty Rajasthan (35 villages) 11% Gujarat (36 villages) 9% Andhra (36 villages) 14% W. Kenya (20 villages) 18% Uganda (36 villages) 24% Peru (20 communities) 17% North Carolina (13 communities) 23%

Escaped Poverty Became PoorChange in Poverty Rajasthan (35 villages) 11%8% 3% Gujarat (36 villages) 9%6% 3% Andhra (36 villages) 14%12% 2% W. Kenya (20 villages) 18%19% -1% Uganda (36 villages) 24%15% 9% Peru (20 communities 17%15% 2% North Carolina (13 communities) 23%12% 11% The Entire News

Not a “rising tide” but a bathtub of poverty Falling into poverty: NOT isolated, marginal or temporary Large numbers – one-third – of poor people not born poor

Bathtub result: not methodology-dependent CountryPeriodStudyEscaped Poverty % Fell into Poverty % Uganda (1300) Deininger- Okidi (2003) 2912 India (3139 rural) Bhide - Mehta (2004) 2313 S. Africa (1171) Carter- May (2001) 1025 B’desh (379) Sen (2003) 2618

Signs of Rising Vulnerability Uganda: Central and Western Region % of all households First Period Second Period Escaped Poverty 13%12.2% Fell into Poverty 5.6%10.9%

Increased Vulnerability in North Carolina (13 rural communities) % of all households First Period Second Period Escaped Poverty 16%13% Fell into Poverty 6%12%

Increased Vulnerability in Kenya (Countrywide:71 rural and urban communities) % fell into poverty Zone 1Zone 2Zone 3Zone 4Urban Zone National Average First period ( ) 7%10%12%15%7%10% Second period ( ) 11%14%13%22%15%14%

Lessons learned (1) Poverty is constantly being created, even as some of it is removed Descents could be becoming more frequent generally But hardly anything is being done to PREVENT descents

GENERAL REASONS FOR DESCENT Falling into poverty is slow and cumulative. Multiple contributing reasons. BAD HEALTH AND HIGH HEALTH CARE EXPENSES: primary reason (59% in Rajasthan; 73% in W. Kenya; 88% in Gujarat; 77% in Uganda; 75% in Andhra; 67% in Peru; 41% in North Carolina) Other location-specific reasons :  Social and customary expenses; high-interest debt; crop disease, land exhaustion, drought; job loss Not significant in any region: Laziness, Alcoholism

The critical significance of Health  Increasing out-of-pocket costs and “catastrophic” health expenses ( Xu et al. 2003; Sen, Iyer and George 2002).  “Medical poverty trap” (Whitehead, Dahlgren and Evans 2001 )  Average long-term income loss of 17 percent (Yao 2005)  More than half of all personal bankruptcies in America (Himmelstein et al. 2005)  Poor people pay more (Fabricant et al. 1999, Farmer 1999)  Macro Evidence (EQUITAP 2005; Milly 1999; Scruggs and Allan 2006)

GENERAL REASONS FOR ESCAPE Escape also usually occurs over a period of time. People work upon strategies that take them upward. Diversification of Income Sources: Agriculture and Informal Sector (70% Rajasthan, 73% W. Kenya, 79% Uganda, 71% Andhra, 69% Peru, 70% Gujarat) OTHER REASONS (much less frequent) Jobs – in Government and Private Sector Not very significant: Government assistance programs

AGRICULTURE: A pathway to prosperity? Average land held by poor households: Uganda acres Gujarat acres Peru acres Kenya acres Prospects are equally limited in the informal sector

Lessons learned (2) Escape and Descent in parallel Asymmetric Reasons Consequences for Economic Policy: Separate policies : (1) promote escapes, (2) prevent descents Consequence for Political Analysis: Subgroups better than “The Poor”

Distinct subgroups (separate needs and experiences): (range) –Persistent Poor % –Newly Poor % –Recently Escaped % –Never Poor National statistics do not collect these data

HYPOTHESIS Because different reasons affect their lives and influence their strategies, different subgroups of poor people have substantially different demands from the state

Empirical Test 1,032 households in 36 communities (Andhra Pradesh, India) Poverty status assessed: 1997 and 2004 Classified within 4 subgroups Asked to rank their major demands from the state

Major demands by subgroups (% within each subgroup) Health Service HousingIrrigationHigh School Wage Labor Persistent Poor Newly Poor Escaped Poverty

Lessons learned (3) “The Poor”: merely a figure of speech Not a valid category for political analysis Because Ins and Outs regularly refresh the composition of poverty

Stage – and assets (recorded 7 years ago) STATUS7Land (bighas) Large Animals Small Animals Kaccha house Very Poor (Stage 1-3) % Poor (Stage 4-5) % Middle (Stage 6-8) % Better Off (Stage 9+) %

Also worrisome: Very limited upward mobility 71 Rajasthan villages (last 10 years) Accountant (2)Lineman (7) Advocate (4)Panchayat Secy. (4) Computer Operator (4)Patwari (11) Constable (8)Messenger (6) Clerk Typist (10)Sub-Inspector (4) Doctor (1)Schoolteacher (50) Driver (4)Soldier (Jawan) (32) Civil Engineer (2)Software Engineer (1)

And in Karnataka Karnataka villages (last 10 years) Accountant (3)Engineer (3) Advocate (4)Lineman (2) Computer Operator (4)Nursing Assistant (1) Constable (8)Patwari (3) Clerk Typist (6)Peon (6) Doctor (1)Schoolteacher (20) Driver (2)Soldier (Jawan) (8)

Similarly Low Glass Ceiling in other Countries… UGANDA (40 communities) – 1 Doctor, 1 Bank Manager, 14 “businessmen” PERU (20 communities) – 1 Civil Engineer, I Agricultural Engineer, 2 Nurse Practitioners, several “merchants” Available pathways (agriculture and the informal sector)  limited upward mobility

Example of NC “Stages of Progress” 1. Food 2. Shelter 3. Transportation 4.Clothing (extreme poverty) 5. Phone 6. Television 7. Car 8.Entertainment (poverty) 9.Savings 10. Buy Home 11. Saving for Kids’ Education

NC Findings (2) Primary Factors in Escaping Poverty: Employment Family Budgeting Primary Factors of Falling Into Poverty: Health Care Job Loss Family

Five years or more school education (% in 20 Rajasthan villages) years years years 61+ years Upper caste OBC SC AVERAGE (n=26,124)

Five years or more school education (% in 20 Karnataka villages) years years years 61+ years OBC SC ST Muslims AVERAGE (n=23,067)

STUDY: SOFTWARE ENGINEERS IN BANGALORE 150 newly recruited software engineers selected at random from three different firms Invited by to take part in a Web-based survey Remarkable response rate (73 percent)

Results: THREE SIGNIFICANT GAPS Rural Gap: 7-11% from rural areas Wealth Gap: 15% lower middle class; none “poor” Generational Education Gap – most significant – College graduate fathers and high school graduate mothers

Limits to Upward Mobility Parents’ education sets apart new entrants Small sliver of Indians years old have equally educated parents (no more than 4-7%)  Why should parents’ education matter so much?

Career aspirations of youth (14-22 yrs) attending school n=1,456 RajasthanKarnataka School teacher43%39% Police constable11%12% Army recruit13%5% Other low-paid government15%22% Other low-paid private5%11% TOTAL LOW-PAID87%89%