HBS 2000/01: March National Bureau of Statistics ANALYSIS OF THE HBS 2000/01 INCOME POVERTY
HBS 2000/01: March Overview of trends in the Tanzanian economy. GOT has achieved macroeconomic stability. Through financial liberalisation inflation is down to single digits, the budget deficit reduced to close to zero, modest depreciation of the exchange rate in line with inflation. GOT has made progress in reorienting its economy to a market based operation and begun to create incentives for private investors. But have the benefits of these efforts been realised by Tanzanians? INTRODUCTION
HBS 2000/01: March HOW HAVE THE ACTIVITIES OF TANZANIANS CHANGED OVER THE DECADE?
HBS 2000/01: March HOW HAVE THEIR MAIN SOURCES OF INCOME CHANGED?
HBS 2000/01: March Overview Data Issues Defining the consumption aggregate Defining a Fisher index Expenditure tables Defining the poverty line Estimating levels of poverty Conclusion WHAT HAS HAPPENED TO POVERTY LEVELS? – OUTLINE OF ANAYLSIS
HBS 2000/01: March House- hold Data Total Exp. Non- Cons. AE Poverty Line Not Poor OVERVIEW OF POVERTY ANALYSIS
HBS 2000/01: March THE CONSUMPTION DATA: QUALITY Over 1000 HHs in 2000/01 with out of range calorie consumption (> 8,000 or < 400, mostly too high). Causes included: Grams recorded as kilograms and other unit miscodes Purchased for trade and miscoded Home produced and stored/sold rather than consumed Under-reporting Large number of wrong units-codes in 1991/92 Corrections involved: correcting unit miscodes and imputation.
HBS 2000/01: March General rule followed in cleaning food data: 1. Calculate median consumption per adult equivalent for each of the 128 food commodities 2. Flag outliers as those with consumption over 20 times the median (confirm calories of household too high) 3. Impute consumption per adult equivalent for outliers using medians calculated from prices and quantities in the Fisher index CLEANING FOOD DATA
HBS 2000/01: March Used methodology described in Blaizeau: 1. To identify which items to take from diary or from annual records 2. To identify which items to include (due to user cost recovery, ratios on education, medical, water bills and telephone and postage suggested their exclusion) 3. Identified outliers based on 2 times the standard deviation with reference to the share of budget spent on the item. Mean values were imputed for outliers. CLEANING NON-FOOD DATA
HBS 2000/01: March Calculate household consumption levels: Based on 1 month diaries and 12 month recall Includes purchased and home produced items, payments in kind and gifts Standardise for household size and composition (adult equivalents) Includes food, some non-durables and some consumer durables (excludes health, education, telephone and postage and water bills) CALCULATING THE CONSUMPTION AGGREGATE
HBS 2000/01: March CALCULATING THE CONSUMPTION AGGREGATE Total Expenditure Consumption Expenditure Total Food Monetary Food Expenditure Non-Monetary Food Expenditure Non-Lumpy Consumer Durables Other Non- Durables Non- Consumption Expenditure
HBS 2000/01: March Have used Fisher Ideal Index Reasons for not using the CPI: does not reflect changes in consumption patterns (1994); weights some food items very heavily; not representative of poverty bundle; coverage – only urban areas, heavily weights Dar es Salaam COMPARING CONSUMPTION LEVELS OVER TIME AND SPACE
HBS 2000/01: March CPI & PRICE INDEXES FROM HBS DATA 1991/92 relative to 2000/01 Laspeyres index2.55 Paasache index 2.44 Fisher index 2.49 CPI (Dec 1991 relative to Jan 2000)4.36
HBS 2000/01: March /92 (per capita)2000/01 (per capita) Dar es Salaam73,531(13,268)91,012 (17,237) Other Urban 70,023(11,276)70,719 (12,719) Rural50,996(7,110)54,735 (8,305) Total54,598 (7,964)59,009 (9,423) All priced in 2000 TShs. MEAN HOUSEHOLD EXPENDITURE PER MONTH (real)
HBS 2000/01: March COMPOSITION OF HOUSEHOLD EXPENDITURE (real – 2000 TSh.)
HBS 2000/01: March HOUSEHOLD EXPENDITURE – items not in consumption aggregate (real – 2000 TShs.)
HBS 2000/01: March Setting the poverty line Price a New Food Consumption Basket for both 1991/92 and 2000/01 Advantage: large data set allows us to define a basket that reflects consumption patterns now POVERTY ANALYSIS
HBS 2000/01: March Define typical food consumption pattern for Tanzania using the consumption data (for the poorest 50% of the population) Define minimum quantities required to meet calorific requirements Price this basket with national median prices from HBS data – gives food poverty line Basic needs poverty line = FL*1/(food share) Adjust the lines (or for ease the consumption aggregate) to reflect different prices that regions and rural/urban households face, using a Fisher Index PRICING A NEW FOOD CONSUMPTION BASKET
HBS 2000/01: March /922000/01 Food poverty line 2,083 5,295 Basic needs poverty line (0.75) (0.73) 2,777 7, /92 lines adjusted using Fisher index (2.49) Food poverty line 5,189 Basic needs poverty line 6,915 Ratio of food baskets2.54 POVERTY LINES – TShs. For 28 days per adult equivalent
HBS 2000/01: March Poverty lines for 2000/01: Poverty lines for 1991/92: POVERTY LINES - cont
HBS 2000/01: March CALCULATING POVERTY ESTIMATES Compare consumption per adult equivalent to poverty line for each urban/rural region Household are poor if they fall below the line. Individuals are poor if they live in a poor household Sum the weighted number of poor individuals in each urban/rural region to obtain consistent estimates for larger populations
HBS 2000/01: March REVISITING THE ANALYSIS OF THE 1991/92 DATA To recap: 1991/92 now comparable with 2000/01 Data now systematically cleaned Consumption aggregate redefined Food basket redefined and priced Revised consumption aggregates, priced in 1991/92 prices, compared with revised poverty line priced in 1991/92 prices
HBS 2000/01: March POVERTY ESTIMATES
HBS 2000/01: March POVERTY ESTIMATES - cont
HBS 2000/01: March MEASURE OF INEQUALITY
HBS 2000/01: March FISHER INDEXES BY REGION
HBS 2000/01: March REGIONAL POVERTY ESTIMATES
HBS 2000/01: March FOOD POVERTY - HEADCOUNT
HBS 2000/01: March BASIC NEEDS POVERTY – HEADCOUNT
HBS 2000/01: March DSMUrbanRuralTotal 2000/ /92 (best estimate) /92 using Fisher index /92 with DSM adjustment SENSITIVITY OF FOOD POVERTY – P0
HBS 2000/01: March Differences due to: 1. The recalculation of the poverty lines. 2. The re-estimation of the Fisher index. POVERTY VERY SENSITIVE TO PRICES EXPLAINING DIFFERENCES WITH PREVIOUS ESTIMATES
HBS 2000/01: March CONCLUSIONS Rise in mean real expenditure Income poverty has fallen Inequality has increased Results are consistent – fall in food share improvement in housing construction macroeconomic reforms COMMENTS PLEASE