Household Expenditure Survey HES Data for PovertyPPPs Weights for Aggregation

1
Household Expenditure Survey (HES) Data for
Poverty-PPPsWeights for Aggregation

• D.S. Prasada Rao
• School of Economics
• University of Queensland
• Brisbane, Australia

2
Basic Data Needs

• Establish a reference population
• Data on prices of the items in the consumption
  basket of the poor
  
• Expenditure weights
• For the reference population
• Coverage
• National Urban and rural

3
HES Basic uses

• Regularly conducted and results disseminated in
  summary form and also in detailed unit-record
  format on CDROM.
  
• Major and reliable source of data on expenditure
  patterns.
  
• HES are widely used on determining the poverty
  lines converting caloric needs into food
  poverty line
  
• Used in the estimation of poverty incidence and
  poverty gap
  
• Analysis of household expenditure behaviour

4
Countries in the most recent ADB study Poverty
in Asia
East Asia China, P.R. 1998 Mangolia 1998 So
uth East Asia Cambodia 1999 Indonesia 2002
Laos PDR 1997 Malaysia 1999 Myanmar 199
7 Philippines 2000 Thailand 2002 Vietna
m 2002 South Asia Bangladesh 2000 India 1
999 Nepal 1996 Pakistan 1999 Sri Lanka
1995 Maldives 1998
Central Asia Azerbaijan 2001 Kazakhstan 2002
Kyrgyz Rep. 2000 Tajikistan 2003 Turkmenistan
1998 Uzbekisthan 2000 Pacific Fiji 1990
FS Micronesia 1998 Kiribati Marsha
l Islands P.N.G. 1996 Samoa 2002 Tonga 20
01 Tuvalu 1994 Vanuatu
Bhutan, Hong Kong, Singapore, Taiwan
5
HES as a Source of Prices

• In most countries HES provide data on expenditure
  and quantities consumed
  
• Possible to define unit values
  expenditure/quantity
  
• Data mainly available for food items
• Recent studies make use of unit value data for
  price comparisons Aten and Menezes (2002), Rao
  (2003), Coondoo et al (2004) and Deaton (2004)
  
• Unit values and quality differences

6
PPPs for Income Groups
Ethiopia 1.000 1.015 1.029
Uganda 1.000 1.279
1.672
7
Aggregation scheme
8
Expenditure shares by commodity groups
9
Aggregation scheme - continued
10
Aggregation scheme - continued
11
Expenditure shares by commodity groups
12
Exchange Rate and PPPs for FoodNo. of Schillings
1 Birr
13
Indian Study
Source Coondoo et al. (2003)
14
HES and the Reference Population

• In the calculation of Poverty-specific PPPs we
  need to identify the poor
  
• Items of relevance
• Consumption pattern of the poor
• There is inherent circularity in this process
• How do we break this?
• Use country-specific poverty estimates to
  identify the poor
  
• Use an arbitrary cut-off like the poorest 30 or
  40
  
• Use an iterative process technically sound
  method
  
• In all these three approaches, we need HES data

15
HES and the Expenditure Weights

• We need weights to combine basic heading PPPs
• Depends upon the particular formula used.
• Fisher and Tornqvist binary index numbers
• EKS, Geary-Khamis and weighted CPD method
• Use ICP basic headings from consumption
• Allows the use of ICP-price data to be combined
  with HES expenditure weights
  
• Establish a mapping between ICP basic headings
  and HES expenditure data
  
• HES classification and ICP classification do not
  match
  
• The extent of mismatch varies across countries
• There is need to split some categories and in
  some cases aggregate HES categories to match ICP
  categories
  

16
HES and the Expenditure Weights

• Mapping between ICP basic headings and HES
  expenditure data
  
• Process depends upon the aggregation method
• If a binary comparison between two countries, A
  and B, is needed, then only HES of the two
  countries are used. This leads to one mapping
  between ICP and HES headings.
• If another pair C and D is used, this may lead to
  another mapping.
  
• If we use the EKS method which makes use of
  binary Fisher index numbers, the mapping process
  could be tedious.
  
• If CPD or GK methods are used then there is a
  need to have a single mapping
  
• ICP Global office is currently working on this
  problem.
  

17
HES and the Expenditure Weights

• Determination of expenditure weights
• Weights available for each poor household in each
  country
  
• Weights for all the households below the poverty
  line
  
• Weights for all the households at the poverty
  line
  
• Choice depends upon the purpose
• If PPPs are for converting an international
  poverty line, then households on the poverty line
  may be considered.
  
• If PPPs are for converting expenditures of poor
  in different countries, then we need weights for
  all households below poverty line are important.
  

18
HES and the Expenditure Weights

• Need to use national average weights to combine
  BH parities
  
• Weights for the poor households
• Democratic weights average of the expenditure
  weights of all the poor households
  
• Plutocratic weights expenditure weights of the
  average household this can be dominated by
  richer households
  
• Will require special tabulations of weights for
  poverty-PPP computation
  

19
Democratic and Pultocratic weights
Household 1
Household 2
Item 1 50 0.33 20 0.67
Item 2 100 0.67 10 0.33
Democratic weights Plutocratic weights
Item 1 0.5 0.38
Item 2 0.5 0.62
20
Topics for discussion

• What is the most recent HES?
• Several countries are in the process of
  conducting new surveys
  
• What is the timeframe for availability
• Availability of unit record data
• Alternative is to undertake tabulations of
  weights for poverty-PPP calculations
  
• Matching ICP price survey coverage and HES
  coverage
  
• Rural and urban or only rural or only urban