1. Angus Deaton, Princeton University WORLD STATISTICS DAY: ICP USERS MEETING

2. The BIG questions

3. The shape of the world  Who is poor and who is rich?  How many poor people are there in the world?  How can we measure progress on income poverty for the MDGs?  How do the poor live? What is life really like in the poorest places in the world?  How big are the differences?  What is the ratio of American to Indian income?  How do we describe the living standards of poor people to people in the rich world?  The global distribution of income?  Over countries  Over the citizens of the world 3

4. NONE can be answered without PPP exchange rates

5. Where do PPPs come from?  Ultimately from the International Comparison Program (ICP)  Though the CIA fact-book may be the most heavily used immediate source  For academic users, perhaps the Penn World Table  Or the World Development Indicators  ICP collects prices on comparable goods & services in many countries  To construct multilateral price indexes for each country relative to a base, such as the US  For consumption, investment, GDP, etc  Used to deflate nominal local currency amounts to give “real” common unit international PPP measures 5

6. History  ICP is like the Olympic Games, though somewhat less regular  First were just a few  Amateurs  Over time, professionalized, lots of training  Huge improvement in technique  Regularly held  First in 1960 & 1970s, U. Penn plus UN  Six countries in 1967  Four more in 1970  Prices for relatively small number of goods and services  Extended to other countries using interpolation  1978 results for more than 100 countries

7. ICP 1993  Before 2008, PPPs used price data collected in 1993, updated for inflation rates since then  Important missing (or partially missing) countries, including India and China, both imputed based on old or incomplete data  A regional system with each region collecting prices on its own, and calculating its own PPPs with regional numeraire  Weak center with ad hoc links between regions  Between regional links are Achilles heel of ICP  Involve hard comparisons between countries with different patterns of demand and relative prices  Think of comparing a Bihari laborer who eats only rice with a Congolese farmer, or Japanese factory worker  UN (1997) report concluded that the ICP 1993 had lost credibility  Yet these numbers are encoded in the poverty MDG  Academic users treat Penn World Table (1993 based) with abandon 7

8. ICP 2005  Did much better: global office housed by World Bank  146 countries  Including India and China  Many African countries never previously included  Regional structure again, each region pricing its own regional list  Makes sense, but some regions very diverse  A “ring” of 18 countries, at least 2 in each region  Ring countries priced a special ring list of more than 1,100 commodities  These prices were then used to link the regions  Calculating price indexes for whole regions relative to one another 8

9. Did it make a difference? (or just same old, same old?)

10. Headline result  Per capita GDP of both India and China both much reduced using the new data  Using 2005 international dollars  China in 2005 from $6,757 to $4,088  India in 2005 from $3,452 to $2,222  Note that the US is numeraire  So we could just as well say that the US got richer  Essentially, India and China moved further away from the US and other rich countries  Their PPPs relative to the US increased, so “real” amounts fell  Not only India and China 10

11. .5 1 1.5 2 2.5 Ratio of new to old PPP for 2005 67891011 Logarithm of per capita GDP in 2005 international $ Congo, DR Burundi Sao Tome & Principe Cape Verde Lesotho Guinea Ghana Cambodia Togo Guinea Bissau India Philippines ChinaNamibia Tonga YemenCongo, RLebanon Gabon Kuwait Fiji Nigeria Tanzania Angola Bolivia Ethiopia Vietnam Bangladesh 11

12. .5.52.54.56.58.6 19701980199020002010 year Post 2005 ICP Pre 2005 ICP Gini coefficient for per capita GDP, weighted by population 12

13. .45.5.55.6 196019701980199020002010 year WDI 2008, 2005 prices WDI 2007, 1993 prices PWT 5.6, 1985 prices PWT 6.2, 1993 prices Gini coefficient for per capita GDP, weighted by population 13

14. Extensions: where from here?  Gradual process of technical improvement  Those who work on this could give a long list!  Government services: health, education  Construction  Improving national accounts  International $ accounts use both national accounts and PPPs, and are only as good as weakest of these  Becoming a high priority  Broadens the range of ICP and new partners  Linking the regions  Technical and conceptual problems here  Important that users be involved  Academics, for example, are not very well informed on strengths and weaknesses

15. Prices and quantities  National income accounts are based on collection of both physical volumes and prices (or volumes and expenditures)  ICP is different, collects only prices  Expenditures are collected by local statistical offices as part of their national accounts  Ideally, the ICP could collect volumes as well as prices  Beginning to do so: e.g. education, or housing  Again, the long term aim is integration of national accounts and ICP  Long term aim, but should be kept in mind

16. Linking between rounds  Past rounds have been different from one another  In country coverage and technical improvement  Made little sense to reconcile them with previous rounds  E.g. 2005 with 1993  From now on, more regular, higher quality  How to blend old information with new?  Avoid discontinuous jumps  Updating between rounds?  Long term goal is to integrate ICP with domestic price collection  Many challenges associated with this  Otherwise we have to explain CPI versus ICP differences

17. New uses of ICP data  Gallup World Poll uses PPPs in their data  More than 155 countries, random national samples  Surveying the population of the world every year!  Gallup collects income data  Single question but matches other information  Includes in their numbers incomes for individuals in PPPs  These numbers are valuable to their clients  Perhaps we will hear more about these uses  Possible uses of Gallup data collection back into ICP?  Their regularity could conceivably help with updating  Again, blue sky at this point

18. New uses of ICP information  ICP collects millions of price quotes around the world  It then turns them into a set of index numbers (PPP exchange rates) which are published  But the prices themselves could have many other uses  Others may want their own indexes, with different weights for different purposes  Sectors: e.g. much interest in health, and prices of health related items, pharmaceuticals, or procedures  International patterns of malnutrition: prices of milk, cereals, etc.  Prices for providing safe water in countries around the world  Some of the prices (ring prices) are available to researchers  But many prices are not currently available  Another area where working with users, and with countries, ICP could produce and publish more, and more useful information

19. Thank you!