90477 International Comparison Program [08.01] Poverty Poverty Related Issues for the 2011 ICP round Angus Deaton 2nd Technical Advisory Group Meeting February 17-19, 2010 Washington DC Table of Contents I. Where we are now .........................................................................................................3 II. Issues for the future........................................................................................................3 III. Updating.........................................................................................................................3 IV. Spatial pricing ................................................................................................................4 V. China ..............................................................................................................................4 VI. Housing ..........................................................................................................................4 VII. Prices paid by households for final consumption ..........................................................4 VIII. Actual consumption: health and education ....................................................................5 IX. Improving the quality and comparability of household consumption surveys ..............5 2 Poverty related issues for the 2011 ICP round Angus Deaton, October 2009 1 Where we are now In conjunction with the TAG in the last round, Olivier Dupriez and I developed a methodology for combining basic head parities for consumption from the ICP with household survey data so as to derive a set of PPPPs, P4s, or poverty-weighted purchasing power parity exchange rates. These combine the parities from the ICP with survey weights to give a set of P4s with global poverty line weights for all of the countries of the world that are regularly included in the World Bank’s global poverty counts. No numbers from rich countries are included, and the multilateral indexes are recomputed to exclude them. A long paper, including the methods and results, is available on the Bank’s website at http://siteresources.worldbank.org/ICPINT/Resources/global_Poverty_Prices.pdf This is currently being split up into two more digestible papers for publication. The P4s are somewhat different from the P3s, though for different reasons than we originally expected. The reweighting to the poverty line turns out to be relatively unimportant. Much more important are the inconsistencies between household survey and national accounts data, even in aggregate. As always, these differences are partly caused by the surveys, and partly by the national accounts. Both have flaws. 2 Issues for the future Three areas seem important: (a) Ensuring that we can update and improve the results when we have the 2011 results from the ICP. (b) Addressing the issues that were problematic in the 2005 round. Some are general to the ICP, e.g. housing, but some are more specific, or are less important for the ICP in general than they are for the poverty work, e.g. spatial pricing. Some issues, like construction, are important for the ICP, but hardly at all for the poverty work. (c) Improving the quality and comparability of household consumption surveys. This is important for the P4s, but has payoffs elsewhere in improving poverty data, national accounts, and ICP-CPI. 3 Updating Underlying the poverty program is the development of a set of “consistentized” household surveys that need to be updated. In the last round, we had 62 countries whose surveys we could use and match to the ICP. There is regular updating of those that needs to be done, as 3 well as improvement, both in number of surveys to be covered, and in greater consistency of coverage for existing surveys. We expect substantial improvement over the next five or six years in both dimensions. 4 Spatial pricing A major problem with the ICP for poverty work is the lack of good information on spatial pricing. Many large countries have separate urban and rural poverty lines, sometimes quite far apart, and the 2005 ICP did not allow separate prices to match. As we know, this raises serious problems for the ICP itself, particularly but not exclusively for China, and for the comparison of small and large countries, but the poverty work needs different price levels. This time round it was done very crudely, by assuming that the ratio of local urban to rural poverty lines accurately represented the ratio of price levels (except for India) but we can surely do better than this. For the ICP itself, it would be useful to have more extensive rural price collection, and it would be useful if the points of collection were geo-coded in some way. From the survey perspective, we plan to make better use of the unit values that are contained in some (not all) of the surveys, covering food and sometimes other items such as fuel, alcohol, and tobacco. These can be used to document spatial differences in prices for a subset of consumption, and also to compare with the parities from the ICP. This last has been on the agenda for some years now, and was previously endorsed by the TAG. The availability of unit values together with geo-coded price data (by product/service and type of outlet) would allow us to address a key outstanding issue for which conflicting and non-convincing “evidence” was provided in the past. 5 China We need rural prices for China. I suspect the fixes in the poverty work were not too bad, so this is possibly more of a problem for the ICP itself than for the poverty work. 6 Housing We omitted housing from the poverty P4s. That was certainly better than using the parities for housing that came out of the ICP. It may remain the best option next time, depending on how much improvement there is in the national accounts and the ICP procedures. Given that the poverty work excludes the OECD and other wealthy countries, it may be hard to beat just dropping the category for comparisons between poor economies. Nevertheless, it remains an issue. 7 Prices paid by households for final consumption This is the “pharmaceuticals issue,” that the ICP is collecting prices for “actual consumption” which are the prices the government would pay Merck for pharmaceuticals to be passed on at lower prices. For poverty work, these kinds of subsidies can be thought of as poverty policy, and prices should be the prices people pay. Perhaps this is not a major issue for two reasons: (a) I suspect there are not many examples other than pharmaceuticals, and if so (b) these subsidies are mostly an issue for rich countries, with which we are not concerned. 4 8 Actual consumption: health and education The Bank’s poverty counts do not take account of health and education provided by the state. This leads to inconsistencies across countries where the state and the market have different shares of health and education. A good longer term objective would be to think about whether there is some way of improving this. 9 Improving the quality and comparability of household consumption surveys The common ground between poverty research work and ICP provides a unique opportunity to build capacity and support countries in the improvement of their household survey methods. The PPPP work undertaken on the side of the ICP 2005 has identified many issues in survey questionnaires and methods, many of which could be relatively easily solved. These issues are related to the scope of survey questionnaires, to consumption aggregation methods, to the non-compliance with international classifications, to occasional very low correlation with national accounts data, etc. They may result in inaccuracy of poverty and inequality measures, in unreliable data on consumption patterns, and on irrelevance of survey data for some specific needs (including national or health accounts, ICP/CPI and others). A concerted effort to help countries improve their household surveys should be part of the agenda of ICP and poverty research. 5