EUROSTAT - OECD PPP PROGRAMME 1999 THE CALCULATION AND AGGREGATION OF PARITIES

Size: px

Start display at page:

Download "EUROSTAT - OECD PPP PROGRAMME 1999 THE CALCULATION AND AGGREGATION OF PARITIES"

Mabel Todd
5 years ago
Views:

1 EUROSTAT - OECD PPP PROGRAMME 1999 THE CALCULATION AND AGGREGATION OF PARITIES INTRODUCTION The calculation and aggregation of parities requires from each country participating in a comparison, either bilateral or multilateral, the following: a set of national annual average prices; a breakdown of national expenditure. The prices have to be consistent with those used to estimate the expenditures. This note follows a worked example to show how the national annual average prices are converted into parities and how these parities are aggregated using national expenditure weights in Eurostat-OECD multilateral comparisons. The worked example is in two parts. The first part describes how parities are calculated for a basic heading. The second part explains how the parities for a basic heading are combined with those of other basic headings to provide weighted parities or PPPs for each level of aggregation up to the level of GDP. THE CALCULATION OF PARITIES FOR A BASIC HEADING Basic headings and representative products Prices are collected and reported at the level of the basic heading. In principle, a basic heading consists of a group of similar well-defined commodities for which a sample of products can be selected that are representative both of their category and of the participating countries. In practice, a basic heading is defined by the lowest level of expenditure category for which explicit weights can be estimated. Thus, for example, cheese is a basic heading and cheddar, camembert, feta, gorgonzola, gouda, are individual products within it. Expenditure on cheese is known, but expenditures on specific cheeses are not. By definition, expenditure weights are not used below the basic heading level. Since, however, in Eurostat-OECD comparisons, countries price not only items that are representative of their national market but also items that are representative of the national markets of others, the representativeness of the goods and services priced needs to be taken into account when deriving parities at the basic heading level. Hence, when reporting prices, countries are required to indicate whether or not the products they priced are representative of their national markets. In this context, a product is said to be representative if it is purchased in sufficient quantities for its price to be typical for that type of product in the national market. For instance, in the cheese example, cheddar is obviously representative of the United Kingdom, camembert of France, feta of Greece, gorgonzola of Italy and gouda of the Netherlands. However, cheddar is sold in sufficient quantities in France and the Netherlands that it is representative of these countries as well. Similarly, camembert is also representative of Germany, Norway and Sweden, and gouda of Greece, Spain and Portugal. 1

2 Method of calculation There are five stages: Calculation of "Laspeyres" and "Paasche" price ratios. Calculation of "Fisher" price ratios. Completing the "Fisher" matrix. Building the EKS matrix of transitive parities. Standardising the EKS matrix of transitive parities. Price matrix The price matrix consists of national annual average prices in local currency. The representative products for each country are indicated by an asterisk. Each country has at least one representative product which is priced in at least one other country. Prices for products 2 and 3 are not available for country D and country A respectively; product 5 is not priced in either country A or country C. Product Country 1 P 1a 3.43 P 1b 17.04* P 1c 633 P 1d 9.57* 2 P 2a 1.27* P 2b 15.67* P 2c 588* P 2d P 3a - - P 3b P 3c 443* P 3d 9.95* 4 P 4a 2.25 P 4b P 4c 755 P 4d 10.22* 5 P 5a - - P 5b 15.75* P 5c - - P 5d 11.32* 2

3 "Laspeyres" price ratio matrix When calculating the "Laspeyres" price ratios only the representative products of the base country are involved irrespective of whether or not the products are representative of the partner country. Hence, when A is the base country, the ratios for product 2 are calculated; when B is the base country, the ratios for products 1, 2 and 5 are calculated; and so on. When a country has more than one representative product a simple geometric mean of the price ratios is taken. L A/A L A/B L A/C L A/D L B/A L B/B L B/C L B/D L C/A L C/B L C/C L C/D L D/A - - L D/B L D/C L D/D

4 The elements of the "Laspeyres" price ratio matrix are calculated as follows: Base A: L A/A = P 2a /P 2a = 1.27/1.27 =1.000 L B/A = P 2b /P 2a = 15.67/1.27 = L C/A = P 2c /P 2a = 588/1.27 = Base B: L A/B = [(P 1a /P 1b (P 2a /P 2b ] 1/2 = [(3.43/17.04(1.27/15.67] 1/2 = L B/B = [(P 1b /P 1b (P 2b /P 2b (P 5b /P 5b ] 1/3 = [(17.04/17.04(15.67/15.67(15.75/15.75] 1/3 = L C/B = [(P 1c /P 1b (P 2c /P 2b ] 1/2 = [(633/17.04(588/15.67] 1/2 = L D/B = [(P 1d /P 1b (P 5d /P 5b ] 1/2 = [(9.57/17.04(11.32/15.75] 1/2 = Base C: L A/C = P 2a /P 2c = 1.27/588 = L B/C = [(P 2b /P 2c (P 3b /P 3c ] 1/2 = [(15.67/588(27.27/443] 1/2 = L C/C = [(P 2c /P 2c (P 3c /P 3c ] 1/2 = [(588/588(443/443] 1/2 = L D/C = P 3d /P 3c = 9.95/443 = Base D: L A/D = [(P 1a /P 1d (P 4a /P 4d ] 1/2 = [(3.43/9.57(2.25/10.22] 1/2 = L B/D = [(P 1b /P 1d (P 3b /P 3d (P 4b /P 4d (P 5b /P 5d ] 1/4 = [(17.04/9.57(27.27/9.95(20.93/10.22(15.75/11.32] 1/4 = L C/D = [(P 1c /P 1d (P 3c /P 3d (P 4c /P 4d ] 1/3 = [(633/9.57(443/9.95(755/10.22] 1/3 = L D/D = [(P 1d /P 1d (P 3d /P 3d (P 4d /P 4d (P 5d /P 5d ] 1/4 = [(9.57/9.57(9.95/9.95(10.22/10.22(11.32/11.32] 1/4 =

5 Paasche price ratio matrix P A/A P A/B P A/C P A/D - - P B/A P B/B P B/C P B/D P C/A P C/B P C/C P C/D P D/A P D/B P D/C P D/D The "Paasche" price ratios are calculated by taking only the representative products of the partner country regardless of whether these products are also representative in the base country. They are obtained by taking the reciprocals of the transposed "Laspeyres" price ratios, as follows: Base A: P A/A = P 2a /P 2a = 1/L A/A = P B/A = [(P 1b /P 1a (P 2b /P 2a ] 1/2 = 1/L A/B = P C/A = P 2c /P 2a = 1/L A/C = P D/A = [(P 1d /P 1a (P 4d /P 4a ] 1/2 = 1/L A/D = Base B: P A/B = P 2a /P 2b = 1/L B/A = P B/B = [(P 1b /P 1b (P 2b /P 2b (P 5b /P 5b ] 1/3 = 1/L B/B = P C/B = [(P 2c /P 2b (P 3c /P 3b ] 1/2 = 1/L B/C = P D/B = [(P 1d /P 1b (P 3d /P 3b (P 4d /P 4b (P 5d /P 5b ] 1/4 = 1/L B/D =

6 Fisher price ratio matrix A "Fisher" price ratio is the unweighted geometric mean of the "Laspeyres" price ratio and the "Paasche" price ratio. Hence: F A/A = [L A/A.P A/A ] 1/2 = [L A/A /L A/A ] 1/2 = [1.000/1.000] 1/2 = F B/A = [L B/A.P B/A ] 1/2 = [L B/A /L A/B ] 1/2 = [12.339/ ] 1/2 = F A/B = [L A/B.P A/B ] 1/2 = [L A/B /L B/A ] 1/2 = [ /12.339] 1/2 = F C/A = [L C/A.P C/A ] 1/2 = [L C/A /L A/C ] 1/2 = [462.99/ ] 1/2 = F A/C = [L A/C.P A/C ] 1/2 = [L A/C /L C/A ] 1/2 = [ /462.99] 1/2 = F C/B = [L C/B.P C/B ] 1/2 = [L C/B /L B/C ] 1/2 = [37.335/ ] 1/2 = F B/C = [L B/C.P B/C ] 1/2 = [L B/C /L C/B ] 1/2 = [ /37.335] 1/2 = F D/B = [L D/B.P D/B ] 1/2 = [L D/B /L B/D ] 1/2 = [ /1.9310] 1/2 = F A/A F A/B F A/C F A/D - - F B/A F B/B F B/C F B/D F C/A F C/B F C/C F C/D F D/A - - F D/B F D/C F D/D Note that F B/A.F A/B = 1, F C/A.F A/C = 1, etc; that is, the above "Fisher" price ratios satisfy the country reversal test. The elements of the above matrix, however, are not transitive; that is, F B/A /F C/A =/ F B/C, F A/B /F C/B =/ F A/C, Neither is the matrix complete. 6

7 Completed "Fisher" price ratio matrix Since some prices are missing, some "Laspeyres" and some "Paasche" ratios cannot be calculated. Consequently, the respective "Fisher" ratios also cannot be produced. The missing "Fisher" ratios have therefore to be estimated by calculating the geometric mean of all the available indirect "Fisher" ratios connecting (or bridging the countries for which the ratios are missing. Hence: F D/A = [(F D/B /F A/B (F D/C /F A/C ] 1/2 = [(0.5736/ ( / ] 1/2 = F A/D = [(F A/B /F D/B (F A/C /F D/C ] 1/2 = [( /0.5736( / ] 1/2 = F A/A F A/B F A/C F A/D F B/A F B/B F B/C F B/D F C/A F C/B F C/C F C/D F D/A F D/B F D/C F D/D

8 EKS matrix of transitive parities Although certain elements of the completed "Fisher" matrix are transitive because of the way they are estimated, the elements of the original "Fisher" matrix are not. In order to ensure overall transitivity, the EKS (Eltetö-Köves-Szulc method is used to obtain the final balanced parities. Transitivity is achieved by replacing each direct ratio by the geometric mean of itself and all the corresponding indirect ratios obtained by using each of the other countries as a bridge: Hence: EKS A/A = F A/A = EKS B/A = [(F B/A /F A/A (F B/B /F A/B (F B/C /F A/C (F B/D /F A/D ] 1/4 = [(F B/A 2 (F B/C /F A/C (F B/D /F A/D ] 1/4 = [( ( / (1.7434/ ] 1/4 = EKS C/A = [(F C/A 2 (F C/B /F A/B (F C/D /F A/D ] 1/4 = [( (30.362/ (51.748/ ] 1/4 = EKS D/A = [(F D/A 2 (F D/B /F A/B (F D/C /F A/C ] 1/4 = [( (0.5736/ ( / ] 1/4 = EKS A/B = [(F A/B 2 (F A/C /F B/C (F A/D /F B/D ] 1/4 = [( ( / ( /1.7434] 1/4 = EKS C/B = [(F C/B 2 (F C/A /F B/A (F C/D /F B/D ] 1/4 = [(30.362²(462.98/9.8286(51.748/1.7434] 1/4 = EKS D/B = [(F D/B 2 (F D/A /F B/A (F D/C /F B/C ] 1/4 = [( ²(7.1013/9.8286( / ] 1/4 = EKS A/A EKS A/B EKS A/C EKS A/D EKS B/A EKS B/B EKS B/C EKS B/D EKS C/A EKS C/B EKS C/C EKS C/D EKS D/A EKS D/B EKS D/C EKS D/D

9 It can be easily shown that the EKS matrix is transitive because the ratios between the corresponding items in each respective country (that is, the ratios of the respective elements between any given pair of columns are all equivalent. EKS A/C = EKS A/B / EKS C/B = / = EKS A/C = EKS A/D / EKS C/D = / = EKS B/C = EKS B/A / EKS C/A = / = EKS B/C = EKS B/D / EKS C/D = / = EKS D/C = EKS D/A / EKS C/A = / = EKS D/C = EKS D/B / EKS C/B = / = Thus, when all the elements in column A, for example, are multiplied by (EKS A/B, they turn out to be just the same as the respective elements in column B (subject to minor rounding errors. Similarly, when the elements in column B are multiplied by (EKS B/C and the elements in column C are multiplied by (EKS C/D, they give rise to the equivalent elements in column C and D respectively. EKS A/B = EKS A/A x EKS A/B = x = EKS B/B = EKS B/A x EKS A/B = x = EKS C/B = EKS C/A x EKS A/B = x = EKS D/B = EKS D/A x EKS A/B = x = EKS A/C = EKS A/B x EKS B/C = x = EKS B/C = EKS B/B x EKS B/C = x = EKS C/C = EKS C/B x EKS B/C = x = EKS D/C = EKS D/B x EKS B/C = x = Moreover, the factors used to convert B to C and C to D can be used indirectly to convert B to D and generate the same results as those obtained by using the direct coefficient. EKS A/D = EKS A/B x EKS B/D = x = EKS B/D = EKS B/B x EKS B/D = x = EKS C/D = EKS C/B x EKS B/D = x = EKS D/D = EKS D/B x EKS B/D = x =

10 EKS matrix of standardised parities In the EKS matrix of transitive parities, the parities in each column are expressed with the corresponding country as a base. To obtain a set of standardised parities -- that is with the group of countries as a base -- each element of the matrix is divided by the geometric mean of its column s elements. Hence: EKS A = EKS A/A / (EKS A/A x EKS B/A x EKS C/A x EKS D/A 1/4 = / (1.000 x x x /4 = EKS B = EKS B/A / (EKS A/A x EKS B/A x EKS C/A x EKS D/A 1/4 = / (1.000 x x x /4 = EKS C = EKS C/A / (EKS A/A x EKS B/A x EKS C/A x EKS D/A 1/4 = / (1.000 x x x /4 = EKS D = EKS D/A / (EKS A/A x EKS B/A x EKS C/A x EKS D/A 1/4 = / (1.000 x x x /4 = EKS A EKS B EKS C EKS D This provides the following vector of standardised parities. EKS

11 THE AGGREGATION OF BASIC HEADING PARITIES Method of calculation There are four stages: Calculation of Laspeyres and Paasche PPPs. Calculation of Fisher PPPs. Building the EKS matrix of transitive PPPs. Standardising the EKS matrix. Parity matrix and value matrix The parity matrix consist of the parities calculated by the EKS method for each of the basic headings as described in Part One. Each row comes from a separate standardised EKS vector. The first row comes from the standardised EKS vector in Part One. Basic Country Heading v P va P vb P vc P vd w P wa P wb P wc P wd x P xa P xb P xc P xd y P ya P yb P yc P yd z P za P zb P zc P zd The value matrix contains expenditure values in national currencies by basic heading and by country. Basic Country Heading v V va 5 V vb 110 V vc 2000 V vd 120 w V wa 20 V wb 240 V wc 5300 V wd 180 x V xa 15 V xb 300 V xc 3500 V xd 200 y V ya 35 V yb 450 V yc V yd 250 z V za 25 V zb 500 V zc 6500 V zd 250 The following shows how PPPs are calculated for three aggregates: Aggregate 1 = v + w Aggregate 2 = x + y + z Aggregate 3 = v + w + x + y + z (overall PPPs 11

12 Laspeyres PPPs The Laspeyres PPPs for Aggregate 1 are calculated as follows: Base A: L1 A/A = [(P va /P va V va + (P wa /P wa V wa ] / (V va + V wa = [(0.0746/ (0.0731/ ] / ( = L1 B/A = [(P vb /P va V va + (P wb /P wa V wa ] / (V va + V wa = [(0.8657/ (0.9504/ ] / ( = L1 C/A = [(P vc /P va V va + (P wc /P wa V wa ] / (V va + V wa = [( / ( / ] / ( = L1 D/A = [(P vd /P va V va + (P wd /P wa V wa ] / (V va + V wa = [(0.5298/ (0.6945/ ] / ( = Base B: L1 A/B = [(P va /P vb V vb + (P wa /P wb V wb ] / (V vb + V wb = [(0.0746/ (0.0731/ ] / ( = L1 B/B = [(P vb /P vb V vb + (P wb /P wb V wb ]/(V vb + V wb = [(0.8657/ (0.9504/ ]/( = L1 C/B = [(P vc /P vb V vb + (P wc /P wb V wb ] / (V vb + V wb = [( / ( / ] / ( = L1 D/B = [(P vd /P vb V vb + (P wd /P wb V wb ] / (V vb + V wb = [(0.5298/ (0.6945/ ] / ( = Base C: L1 A/C = [(P va /P vc V vc + (P wa /P wc V wc ] / (V vc + V wc = [(0.0746/ (0.0731/ ] / ( = L1 B/C = [(P vb /P vc V vc + (P wb /P wc V wc ] / (V vc + V wc = [(0.8657/ (0.9504/ ] / ( = L1 C/C = [(P vc /P vc V vc + (P wc /P wc V wc ]/(V vc + V wc = [( / ( / ]/( = L1 D/C = [(P vd /P vc V vc + (P wd /P wc V wc ] / (V vc + V wc = [(0.5298/ (0.6945/ ] / ( =

13 Base D: L1 A/D = [(P va /P vd V vd + (P wa /P wd V wd ] / (V vd + V wd = [(0.0746/ (0.0731/ ] / ( = L1 B/D = [(P vb /P vd V vd + (P wb /P wd V wd ] / (V vd + V wd = [(0.8657/ (0.9504/ ] / ( = L1 C/D = [(P vc /P vd V vd + (P wc /P wd V wd ] / (V vd + V wd = [( / ( / ] / ( = L1 D/D = [(P vd /P vd V vd + (P wd /P wd V wd ]/(V vd + V wd = [(0.5298/ (0.6945/ ]/( = Aggregate 1 L A/A L A/B L A/C L A/D L B/A L B/B L B/C L B/D L C/A L C/B L C/C L C/D L D/A L D/B L D/C L D/D 13

14 The Laspeyres PPPs for Aggregates 2 and 3 are similarly calculated. Aggregate 2 L A/A L A/B L A/C L A/D L B/A L B/B L B/C L B/D L C/A L C/B L C/C L C/D L D/A L D/B L D/C L D/D Aggregate 3 L A/A L A/B L A/C L A/D L B/A L B/B L B/C L B/D L C/A L C/B L C/C L C/D L D/A L D/B L D/C L D/D 14

15 Paasche PPPs The Paasche PPPs are obtained by taking the reciprocals of the transposed Laspeyres indices: Hence for Aggregate 1: P1 A/A = 1/L1 A/A = 1/ = P1 B/A = 1/L1 A/B = 1/ = P1 C/A = 1/L1 A/C = 1/ = P1 D/A = 1/L1 A/D = 1/ = Aggregate 1 P A/A P A/B P A/C P A/D P B/A P B/B P B/C P B/D P C/A P C/B P C/C P C/D P D/A P D/B P D/C P D/D 15

16 The Paasche PPPs for Aggregates 2 and 3 are similarly calculated. Aggregate 2 P A/A P A/B P A/C P A/D P B/A P B/B P B/C P B/D P C/A P C/B P C/C P C/D P D/A P D/B P D/C P D/D Aggregate 3 P A/A P A/B P A/C P A/D P B/A P B/B P B/C P B/D P C/A P C/B P C/C P C/D P D/A P D/B P D/C P D/D 16

17 Fisher PPPs The Fisher PPPs are the unweighted geometric means of the Laspeyres and Paasche PPPs. Hence, for Aggregate 1: F1 A/A = [L1 A/A.P1 A/A ] 1/2 = [L1 B/A /L1 A/B ] 1/2 = [1.0000/1.0000] 1/2 = F1 B/A = [L1 B/A.P1 B/A ] 1/2 = [L1 B/A /L1 A/B ] 1/2 = [ / ] 1/2 = F1 C/A = [L1 C/A.P1 C/A ] 1/2 = [L1 C/A /L1 A/C ] 1/2 = [305.14/ ] 1/2 = F1 D/C = [L1 D/A.P1 D/A ] 1/2 = [L1 D/A /L1 D/A ] 1/2 = [ / ] 1/2 = Aggregate 1 F A/A F A/B F A/C F A/D F B/A F B/B F B/C F B/D F C/A F C/B F C/C F C/D F D/A F D/B F D/C F D/D 17

18 The Fisher PPPs for Aggregates 2 and 3 are similarly calculated. Aggregate 2 F A/A F A/B F A/C F A/D F B/A F B/B F B/C F B/D F C/A F C/B F C/C F C/D F D/A F D/B F D/C F D/D Aggregate 3 F A/A F A/B F A/C F A/D F B/A F B/B F B/C F B/D F C/A F C/B F C/C F C/D F D/A F D/B F D/C F D/D Note that F B/A.F A/B = 1, F C/A.F A/C = 1, etc; that is, the above Fisher PPPs satisfy the country reversal test. The elements of the above matrices, however, are not transitive; that is, F B/A /F C/A =/ F B/C, F A/B /F C/B =/ F A/C, In order to ensure transitivity, the EKS method is used to obtain the final PPPs. 18

19 EKS PPPs Transitivity is achieved by replacing each PPP by the geometric mean of itself and all the corresponding PPPs obtained by using each of the other countries as a bridge as follows: Hence for Aggregate 1: EKS1 A/A = F A/A = EKS1 B/A = [(F B/A /F A/A (F B/B /F A/B (F B/C /F A/C (F B/D /F A/D ] 1/4 = [(F B/A 2 (F B/C /F A/C (F B/D /F A/D ] 1/4 = [( ( / ( / ] 1/4 = EKS1 C/A = [(F C/A 2 (F C/B /F A/B (F C/D /F A/D ] 1/4 = [( ( / ( / ] 1/4 = EKS1 D/A = [(F D/A 2 (F D/B /F A/B (F D/C /F A/C ] 1/4 = [( ( / ( / ] 1/4 = EKS1 A/B = [(F A/B 2 (F A/C /F B/C (F A/D /F B/D ] 1/4 = [( ( / ( / ] 1/4 = EKS1 C/B = [(F C/B 2 (F C/A /F B/A (F C/D /F B/D ] 1/4 = [( ( / ( / ] 1/4 = EKS1 D/B = [(F D/B 2 (F D/A /F B/A (F D/C /F B/C ] 1/4 = [( ( / ( / ] 1/4 = Aggregate 1 EKS A/A EKS A/B EKS A/C EKS A/D EKS B/A EKS B/B EKS B/C EKS B/D EKS C/A EKS C/B EKS C/C EKS C/D EKS D/A EKS D/B EKS D/C EKS D/D 19

20 It can be easily shown that the EKS matrix is transitive because the ratios between the corresponding items in each respective country (that is, the ratios of the respective elements between any given pair of columns are all equivalent. EKS A/C = EKS A/B /EKS C/B = / = EKS A/C = EKS A/D /EKS C/D = / = EKS B/C = EKS B/D /EKS C/D = / = EKS B/C = EKS B/A /EKS C/A = / = EKS D/C = EKS D/A /EKS C/A = / = EKS D/C = EKS D/B /EKS C/B = / = Thus, when all the elements in column A, for example, are multiplied by (EKS A/B, they turn out to be just the same as the respective elements in column B (subject to minor rounding errors. Similarly, when the elements in column B are multiplied by (EKS B/C and the elements in column C are multiplied by (EKS C/D, they give rise to the equivalent elements in column C and D respectively. EKS A/B = EKS A/A x EKS A/B = x = EKS B/B = EKS B/A x EKS A/B = x = EKS C/B = EKS C/A x EKS A/B = x = EKS D/A = EKS D/A x EKS A/B = x = EKS A/C = EKS A/B x EKS B/C = x = EKS B/C = EKS B/B x EKS B/C = x = EKS C/C = EKS C/B x EKS B/C = x = EKS D/C = EKS D/B x EKS B/C = x = Moreover, the factors used to convert B to C and C to D can be used indirectly to convert B to D and generate the same results as those obtained by using the direct coefficient. EKS A/D = EKS A/B x EKS B/D = x = EKS B/D = EKS B/B x EKS B/D = x = EKS C/D = EKS C/B x EKS B/D = x = EKS D/D = EKS D/B x EKS B/D = x = 20

21 The EKS matrices for Aggregates 2 and 3 are similarly calculated. Aggregate 2 EKS A/A EKS A/B EKS A/C EKS A/D EKS B/A EKS B/B EKS B/C EKS B/D EKS C/A EKS C/B EKS C/C EKS C/D EKS D/A EKS D/B EKS D/C EKS D/D Aggregate 3 EKS A/A EKS A/B EKS A/C EKS A/D EKS B/A EKS B/B EKS B/C EKS B/D EKS C/A EKS C/B EKS C/C EKS C/D EKS D/A EKS D/B EKS D/C EKS D/D 21

22 Standardised EKS PPPs In the three EKS matrices, the PPPs in each column are expressed with the corresponding country as a base. To obtain a set of standardised PPPs -- that is, with the group of countries as a base -- each element of the matrices is divided by the geometric average of its column s elements. Hence for Aggregate 1: EKS1 A = EKS1 A/A / (EKS1 A/A x EKS1 B/A x EKS1 C/A x EKS1 D/A 1/4 = / (1.000 x x x /4 = EKS1 B = EKS1 B/A / (EKS1 A/A x EKS1 B/A x EKS1 C/A x EKS1 D/A 1/4 = / (1.000 x x x /4 = Aggregate 1 EKS A EKS B EKS C EKS D

23 The calculation is repeated for Aggregates 2 and 3. Aggregate 2 EKS A EKS B EKS C EKS D Aggregate 3 EKS A EKS B EKS C EKS D The three standardised EKS matrices obtained can be reduced to three EKS vectors of standardised parities. Aggregate Aggregate Aggregate

Basic headings, representative products and quasi expenditure weights

Basic headings, representative products and quasi expenditure weights PPP methodological manual ANNEX V: CALCULATION AND AGGREGATION OF PPPS Introduction V.1. The calculation and aggregation of PPPs requires each country participating in a Eurostat-OECD comparison to provide