A Framework for PovertyPPP Compilation Proposed Methodology

1
A Framework for Poverty-PPP Compilation Proposed
Methodology

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

2
PPPs and Global Poverty The World Bank Approach

• Regular estimates of global and regional poverty
• Uses 1-a day and 2- a day as international
  poverty lines
• The -poverty line is converted into local
  currency using PPPs from the ICP
• Estimate poverty incidence using national
  accounts and HES/LSMS data
• Approach since the 1990 World Development Report

3
Which conversion factors to use?

• Exchange rates
• PPPs from the ICP for the GDP as a whole used
  in real per capita income comparisons
• PPPs for consumption
• PPPs for food and beverages

4
Why ICP PPPs are not appropriate?

• Items priced are usually representative of the
  consumption of higher income groups mainly due
  to adherence to comparability
• Location of price surveys
• Urban/cities/rural
• Past ICP price surveys mainly located in
  metropolitan areas
• Outlets for price surveys
• Make sure the service component of outlets is
  comparable
• Inappropriate expenditure-share weights
• National average weights versus weights
  representative of consumption pattern of poor

5
Objective Compilation of Poverty-specific PPPs

• PPPs for converting poverty lines expressed in
  different national currencies
• Similar to price level adjustments for regional
  poverty lines
• Resulting PPPs may be used for purposes of
  converting the World Bank 1 and 2-a-day into
  various national currency units.
• Aims of the paper
• Provide background material for discussions by
  the PAG and RAB members
• Discuss a few options and put forward a suggested
  methodology for the construction of Poverty-PPPs.

6
Similarity with the ICP work

• Project is very similar to the ICP but differs
  significantly in terms of detail.
• ICP covers the whole of GDP while poverty-PPPs
  refer only to consumption.
• ICP covers the whole population this is a
  definite advantage.
• Poverty-PPPs refer to the poor but how do we
  know the reference population?

7
Iterative method to identify the reference
population

• Start with a poverty line a selected country.
• Convert it into other national currency units
  using PPPs from the PWT or the World Bank
• Use household expenditure survey data on prices
  (unit values) or some other price data and
  expenditure shares of households around the
  poverty line to derive a new set of PPPs.
• Repeat the process until the PPPs converge and
  use these PPPs to define the reference population

8
Identifying the reference population various
alternatives

• Poor as identified through country-specific
  poverty lines and methodology
• Use the existing estimates of poverty incidence
  from the World Bank based on 1/day international
  poverty lines
• Use an arbitrary percentage of the population
  the poorest 30 from each country
• Use an iterative scheme similar to the method
  used in Deaton (2004)

9
Iterative method to identify the reference
population - continued

• Need household expenditure surveys for all the
  countries included in the comparisons
• Need further research on
• The invariance of the reference population group
  and PPPs to the starting values used.
• Investigate the base invariance property of the
  method are the results sensitive to the choice
  of the base country?

10
Country coverage and a regional approach

• Poverty PPP work to be undertaken on a regional
  or sub-regional basis
• Consumption patterns of poor are likely to be
  influenced by local products and customs
• Access to markets by the rural and urban poor may
  be restricted transport costs?
• Coverage of countries to include only countries
  with a sizeable poor populations drop those
  countries with insignificant poor populations on
  the basis of the existing World Bank estimates

11
Sources of price data Option 1

• Use price data collected as a part of the ICP
• Make use of data on consumption items from the
  regional lists
• Easy to implement no additional survey costs
• But ICP product lists may have little in common
  with the consumption items of the poor.

12
Sources of price data Option 2

• Supplement/replace ICP lists with goods and
  services considered representative of the
  consumption of the poor
• Follow an approach similar to that of the ICP
• Start with product lists from the countries
• Use local knowledge and expertise of national
  price statisticians and involve poverty
  researchers
• This may have implications for the aggregation
  method used
• May need spatial chaining procedures

13
Sources of price data Option 2

• Advantages
• Product lists more relevant to the poor
• Approach consistent with the ICP approach
• Product lists can be used for further use in
  compiling temporal price indices for poorer
  households
• Cover both food and non-food items
• A proper account of the outlets used by the poor
• Disadvantages
• Costs associated with the process
• Problems encountered in preparing the lists
• Circularity involved
• Reliability of judgement samples

14
Sources of price data Option 3

• Use unit values from household expenditure
  surveys
• Utilise existing data on household expenditure
  and quantities for different consumption items
• Preliminary research shows the feasibility of
  using HES data for regional price comparisons
• Coondoo et al (2004), Deaton (2004) Atten (2002)
  and Rao (2003)

15
Sources of price data Option 3

• Advantages
• No additional resources required for price
  surveys
• This approach is consistent with work on poverty
  line determination
• Process of identifying the poor and their
  consumption is well integrated
• Disadvantages
• Analysis of HES could still be resource intensive
• Comparability of unit values across countries may
  be problematic
• Unit values are typically available for only food
  items
• Availability of quantity data
• Some countries do not report data on quantities

16
Linking regional PPPs to the US dollar

• Main problem Reference population (poor under
  1/day) in the US could be non-existent or very
  small
• Possible approaches
• 1. Use the ICP PPPs (expressed relative to the US
  dollar) for consumption from the ring-country
  comparisons to link the regional poverty PPPs to
  the US dollar.
• 2. Use basic-heading PPPs from ring country
  comparisons and derive PPPs for selected
  countries using expenditure weights of the poor
  in the link country country.
• Use the HES data to derive conversion factors
  based on price comparisons between higher and
  lower income groups.
• This approach can be used only for food items.

17
Household Expenditure Data

• Expenditure Share weights
• For the poor average weight for each commodity
  for all those under poverty line
• Weights at the poverty line Average expenditure
  share weights for those who are on the poverty
  line (within certain bounds around the line).

18
Aggregation methods

• Binary methods
• Multilateral methods
• Spatial linking methods
• Hill (1999) approach based on minimum spanning
  trees and distance measures
• Alternative linking based on the overlap of item
  lists and coverage Rao and Timmer (2004)

19
Proposed Methodology

• Country coverage Exclude all countries with
  small or negligible poverty incidence
• Use regional and sub-regional approach
• Price survey methodology
• Use the product list approach and cover food and
  non-food items
• Use HES data where appropriate and necessary
• Reference population and expenditure weights
  Use Deaton (2004) approach
• Aggregation methods Use a method that can make
  use of all the price data collected
• Linking regional parities to the US dollar
• Use ICP parities at the basic heading level from
  the ring country comparisons
• Use only link country weights in deriving PPPs
  w.r.t. the US dollar