Transcript Document
The Ring Comparison ADB workshop 26–29 June 2006 Paul McCarthy, Prices Branch ABS To get a world-wide comparison • The ICP has been carried out on a regional basis with 6 separate regions involved – – – – – – Asia/Pacific OECD/Eurostat West Asia Africa CIS Latin America • Need to combine the regional outputs into a world-wide comparison 2 World-wide comparisons 3 Linking regions • The previous ICP used a single “bridge country” approach • Linking regions through a single country is risky – any errors in the link country will affect the whole region’s comparability with the rest of the world • Linking through several countries is safer – how are any inconsistencies handled? 4 Linking regions (continued) • A new methodology has been developed to link regions in the 2005 ICP – the “Ring country” comparison • Selected countries from each region collect prices using a product list reflecting the world as a whole 5 Bilateral comparison • Links based on a comparison between pairs of countries – could use one pair of countries between regions or several pairs of countries from each pair of regions • Products matched between each pair of countries • Need to identify the countries most likely to provide a broad range of product matches – a special product list would be set up to fill the gaps that are left after initial product matching 6 “Chain link” 7 All regions link to one region 8 Mixed approach 9 “Tight” bilateral link 10 Bridge versus core • Bridge countries – countries which participate in 2 regional comparisons to provide a link between them • method used in previous ICPs • Core countries – countries that participate in a separate comparison (either bilateral or multilateral) organised specifically to provide a link between regions • the ring comparison is an example • Either approach can be used for either multilateral or bilateral comparisons 11 Pros of a single bridge country • The linking procedure is straightforward • Calculating the final PPPs between countries in different regions is fairly simple • This procedure has been used many times in the past and so there are quite a number of PPP compilers who are very familiar with it • Overall setup costs are low 12 Pros of a single bridge country (cont) • Costs incurred by the bridge country are relatively low because there is not a lot of work associated with developing the list of products to be priced • Regional products can be matched by the country through SPDs but it is not essential to do so 13 Cons of a single bridge country • The quality of the link between all countries in any pair of regions directly linked is completely dependent on a single country • The market conditions and structures of the bridge country completely (largely?) define the link • If a bridge country fails to deliver for some reason, no overall world-wide ICP comparison is possible 14 Cons of a single bridge country (continued) • If regions are successively linked together in a chain then a problem arising in the process of linking between a pair of the middle regions affects the comparison between all the regions on either side of this link • The method has been criticised for its shortcomings in assessments of past rounds of the ICP 15 “Chain link” 16 Pros of a tight bilateral • Product list can be targeted to both countries involved • Sounder results than for either the simple or multiple bridge country approach – product lists can be specifically targeted so the results are probably better than those from the ring • Linking procedure relatively straightforward for the simpler versions – however, if more than a single pair of countries prices the products then need to work out how to weight together the results for all the countries involved 17 Cons of a tight bilateral • Higher costs than those for either the simple or multiple bridge country approach (for any similar number of link countries involved) – a separate set of product lists has to be developed for each pair of countries • Need to find pairs of “similar” countries between diverse regions to set up a comparison other than a simple chain – ideal pairs are known only after the comparison is completed 18 Cons of a tight bilateral (continued) • Time taken to set up the product lists could be fairly lengthy, particularly if a country pair turns out not to be able to match products as well as had been envisaged when initially selecting them 19 Multilateral comparisons • The “Ring comparison” is a multilateral approach • The results should be better – not subject to the vagaries of a single country providing the link between any pair of countries • In practice, more complicated than previous procedures – dependent on a good range of “out of region” products being priced by all ring countries 20 Multiple bridge countries • Using multiple bridge countries for a pair of regions is a variation on the method used in previous ICPs • Reduces the effects of a bridge country coming up with poor results or a complete failure on the part of one country • Aim would be to use an average of the results of all the bridge countries between each pair of regions – if the results from one country were unusable then they could be discarded but a robust link could still be made via the remaining countries 21 Pros of multiple bridge countries • Despite setup costs being higher than for a single bridge country, they are still relatively low • The costs for each bridge country are low because there is not a lot of work to develop the product list • Linking procedure is fairly straightforward – main issue is how to weight together the results for all the bridge countries involved between each pair of regions to enable a single link factor to be calculated for each pair of regions 22 Pros of multiple bridge countries (cont) • The quality of the link is not dependent on a single country between a pair of regions • Calculating the final PPPs between countries in different regions is fairly simple 23 “Ring link” Asia/Pacific OECD/ Eurostat Africa Latin America West Asia CIS 24 Multilateral link 25 Ring link Asia/Pacific OECD/ Eurostat Africa Latin America West Asia CIS 26 What is the ring comparison? • The ring comparison is a multilateral approach used to link regional PPPs to produce world-wide PPPs – this approach requires a selected number of countries from each region to participate in a separate comparison • The ‘ring countries’ will price a common global ring product list in addition to their individual regional lists 27 Criteria for a ring country • Each participates fully in its own region – implies reasonably good expenditure-based accounts and prices statistics • Willing to act as a ring country • Have developed markets and an open economy • Has a price structure that is not distorted by high tariffs or high subsidies • Products to be priced should not be affected by a “prestige goods effect” 28 Criteria for a ring country (continued) • Able to derive national annual average prices • Have a wide range of goods and services likely to be found in other regions • Have time series price indexes to enable reported prices to be “standardised in time” • Have good data for expenditure weights • Can participate in the full GDP comparison • Able to conduct the rental survey 29 The ring countries Africa Egypt, Kenya, Senegal, South Africa, Cameroon Asia Hong Kong, Sri Lanka, Malaysia, Philippines CIS Russia, Kazakhstan Eurostat/OECD UK, Japan, Estonia, Slovenia West Asia Oman, Jordan Latin America Brazil, Chile 30 General issues • Assume linking will be carried out at the BH level • The PPPs have to be established at the individual product level within each BH and then averaged to obtain the BH PPP – applies for either a core or bridge approach – applies for either a bilateral or multilateral comparison • Importance of data consistency – within region, and regional to ring lists 31 Steps for producing ring PPPs • Prepare regional multilateral PPPs for all countries in each region • Collect prices for ring countries – the “ring list” • Divide ring prices by each country’s regional PPP • Compute ring PPPs across 6 regions • Ring PPP x regional PPP = global PPP 32 Fixity • The relative position of countries can change as the composition of the group of countries being compared changes • Fixity becomes an issue when there are two groups of countries, one smaller than the other, with the smaller group being a sub-group of the larger group • There are 2 sets of PPPs for the smaller group – the first set is that calculated for the group on its own, while the second set is that calculated for the group as a sub-group of the larger group 33 Fixity (continued) • Fixity requires that the first set replaces the second set in the broader comparison • In practice, fixity is obtained by taking the PPPs calculated for a specific sub-group of countries, substituting them for the PPPs calculated for the sub-group in a comparison covering a larger group of countries, and linking the substitute PPPs with the PPPs for the other countries included in the comparison 34 Fixity (continued) • The procedure preserves the relationships between the countries in the sub-group – it also preserves the relationship between the other countries and the sub-group as a whole. • Fixity enables regions to finalise their results prior to the ring comparison being run 35 36