METHODOLOGY AND RESULTS

Transcription

1 Methodological Docuent METHODOLOGY AND RESULTS PRODUCER PRICE INDEX IN BOSNIA AND HERZEGOVINA Bosnia and Herzegovina Agenc for Statistics of Bosnia and Herzegovina Federal Office of Statistics of Federation of Bosnia and Herzegovina Republika Srpska Institute of Statistics This project is funded b the European Union Istat This project is ipleented b ISTAT

2 This publication has been produced with the assistance of the European Union. The contents of this publication is the sole responsibilit of ISTAT, as Ipleenting Agenc of the project EU Support to the Statistics Sector of Bosnia and Herzegovina Phase III, CARDS Twinning Project BA05-IB-ST-01. This publication can in no wa be taken to reflect the views of the European Union.

3 TABLE OF CONTENTS FOREWORD 5 INTRODUCTION 7 1. METHODOLOGICAL TOPICS Preliinar rearks Producer price definition Price deterining characteristics Saple design topics Weights sste Indices structure Chained indices eleents Decoposing the Index rates of change Surve questionnaire Qualit data checking Exaple of surve docuentation Disseination practices Press Release Concluding coents BIH PRODUCER PRICE INDICES (DECEMBER 2006 JUNE ) Main criteria for checking data qualit General rearks concerning PPI Graphical analsis b entities (period 2006 ) SOFTWARE USER S GUIDE Main Panel Description tables Action tables and facilities Data entr The Indexes Calculation Panel Software for BHAS 119 ANNEX - STATISTICAL TABLES 124 GLOSSARY 138

4

5 Foreword In the process of accession to the European Union and haronization with the EU standards and practice, the statistical sste of BiH is expected to haronise its activities with the requireents of the European Statistical Sste. The transition period is characterised b significant changes in politics, legislation and econoic sste and it deands adoption of new statistical standards and ethods. Accordingl, the statistical sste has to ensure high qualit and tiel following the changes occurring in the new circustances and to allow international coparison of statistical data. The objective of the CARDS Twinning Project EU Support to the Statistics Sector of Bosnia and Herzegovina- Phase III, funded b the European Union, was to support Bosnia and Herzegovina in carring out this iportant obligation. In the scope of aforeentioned Project, Coponent Business Statistics with Subcoponent Producer Price Index was ipleented. Publication Methodolog and results Producer Price Index in BiH is result of the joint efforts of experts fro Italian National Institute of Statistics (ISTAT) and three statistical institutions fro Bosnia and Herzegovina. We would like to express our gratitude and thanks to the European Union, the Delegation of European Coission to Bosnia and Herzegovina and Eurostat for their efforts in the ipleentation of this project and providing financial, adinistrative and technical assistance. Furtherore, we would like to express our special thanks to the expert tea fro the Italian National Institute of Statistics ISTAT, led b Mr. Gian Paolo Oneto, Tea Leader of the Coponent Business Statistics, for their open and professional efforts to contribute to the developent of new ethodolog copliant with European standards, and designing the IT tools for producer price index copilation. Special thanks go to Ms. Cecilia Pop, Resident Twinning Advisor, and her tea for highl professional and coitted support to the BiH statistical sste. We are ver glad to offer to all our users the Methodolog for copilation of BiH producer price index in line with the EU standards and regulations. 5

6

7 Introduction The Producer Price surve in Bosnia Herzegovina (BiH) is carried out b the Entities (Federation of Bosnia Herzegovina, FBiH, Republika Sprska, RS, and District of Brcko) through a coon set of ethodologies and a software tool. Both the ethods and the software were released b the Italian statisticians during the cooperation activities. The contents of the presentations and the software setting up work were agreed aong the experts of the Italian and Entities teas. The sae set of tools ensures hoogeneit in data production. Of course, each Entit carries out its own surve independentl fro the others. The BiH producer price index is the short-ter statistical indicator that shows the dnaic of transaction prices of goods aong enterprises, within each Entit and aong the (State level indices). After data collection, the PPI copiling procedures start with the snthesis (via siple geoetric ean) of price relatives (ratios between current and basic prices) up to the overall index (Lasperes aggregate indices). As a result of the cooperation project, ethods and techniques are in line with the European standards. The Producer Price surve is carried out b updating, annuall, a saple of products and the linked list of enterprises, whose updating depends on the business register provided b the Institute of Statistics. The chained index structure allows to anage a continuousl representative saple of units (products and enterprises) and the corresponding follow-up in ters of (good) qualit of the estiates (indices). One of the crucial tasks of the cooperation activities was the identification of a coon period to refer eleentar prices, indices and weights. This point presented a ver high priorit in the developent of a new surve at state level: the Entities agreed on providing a onthl data set for the period Deceber 2006 onwards, both for the eleentar data (quotation prices) and weights. Concerning the final results presented here, the liited length of the data collection period (Deceber 2006 June ), allows to have a short spectru of onthl indices with their rates of change (respectivel, for the onth to onth and the twelve-onth rates of change, 18 and 6 ratios). Such a set of indices is rather short for carring out a detailed data analsis, but its inforation content is sufficient to show that the sub-coponent on PPI has entirel achieved its ai. On the other hand, the situation at the beginning of the technical assistance activit was such that onl one Entit provided producer price indices, but the surve architecture was not in 7

8 line with Eurostat standards. Surel, the software release, was a ajor contribution to the success of the Sub-coponent. This application has plaed a central role in allowing the Entities to anage effectivel the surve. In fact, the results contained in this publication are also based on the desk work the Entities carried out b using the first software release (therefore, after the end of the last eeting), purposel set up b the Italian tea experts. The present publication is not a review of theoretical tools or a surve anual on PPI. Rather, it is a anual of surve practices with a twofold ai: to have the users acquainted with new ethods to be adopted in setting up the surve and the index nubers copilation; to show the final results of the cooperation work. It s worth noting that all final results are based on the eleentar data as the were provided b the Entities and on a coplex set of qualit check operations (carried out b the Italian statistician in the desk work session) on these eleentar data based on hpothesis the Entities experts shared. This volue is ade up of three chapters. The first one, Methodological topics, concerns surve ethodologies and techniques considering all the ain thees treated during the project. The second chapter, BiH producer price indices, discusses the results of the cooperation work, focusing on the qualit check work. Graphics and data tables in the Appendix coplete the presentation of results. Finall, the third chapter, Software Users Guide, presents a shortl description of the software release perforances. Within the IT section, training course eetings for the Entities statisticians were carried out. Such training on the job experiences allowed the Entities experts to anage the software providing indices for the period Januar June. The present Final Report, coordinated b Valerio De Santis, is a tea work. In particular, contributes were provided b Valerio De Santis (paragraphs , , 1.14, 2.2 and the Glossar), Francesca Monetti (paragraphs , 2.1, 2.3 and annex - statistical tables), Tiberio Daiani (paragraphs 1.5, 1.12, 1.13, 2.1, 2.3 and annex - statistical tables) and Massio De Cubellis (Chapter 3). 8

9 1. Methodological topics 1.1 Preliinar rearks The content of these ethodological notes focuses on the ain topics of the producer price indices as the were shown during the eetings held in Sarajevo and Banja Luka. The technical assistance activities concerned the following topics: ain criteria for the selection of products and enterprises; definition of eleentar ites; definition of transaction (producer) price; applied chain indices topics; desk work on qualit checks; desk work the estiation of weights. Eleentar (producer) prices in BiH are collected directl b the enterprises. The respondents are required to fill in the surve questionnaire, specifing the ain ites of their production that are coherent with (the denoination of) the product provided b the NSI. The index copiling procedure links products, ites and enterprises. Products and enterprises are selected b the NSI, while the ites are selected b the respondents. Given a generic product, its corresponding index is calculated b aking a snthesis (via the siple geoetric ean) of the ite prices onthl collected. In fact, the siple geoetric ean is copiled on the price relatives or icro-indexes; a price relative is, b definition, the ratio between the ite price at the current tie and its level at tie t = 0 (the ite price of Deceber of the previous ear, if the index is chained, or the average of the twelve onths ite prices of the basic ear when the index is a fixed base one). Therefore, the overall index is the result of suarising the inforation set derived fro N enterprises providing (onthl) M ite quotation prices corresponding to K products (in ters of NP PRODCOM). The relationship aong these entities is such that the following alwas holds: [1] K N M Of course, the ain index base criterion is that the nuber of ites and products (respectivel M and K) ust be kept fixed for all the period in which the (calculation) base holds. In theor, 9

10 the nuber of enterprises could change (enterprises replaceent) during the ear but when indices are chained the updating of the enterprises list is carried out once a ear, so that such a need is not a priorit outside the annual base updating procedures. The results of the activities can be sued up considering the overall PPI surves profile in ters of products, ites and enterprises (Tables 1.1 and 1.2). These results are net of the qualit check 1 work carried out b Italian and Entities experts. Table 1.1 BiH analsis and reporting units Year BiH FBiH RS DB Products ( * ) Ites Enterprises (*) 8 digit Products Table 1.2 BiH analsis and reporting units Year BiH FBiH RS DB Products ( * ) Ites Enterprises (*) 8 digit Products 1.2 Producer price definition The producer price is the transaction price between two enterprises: the dealer (producer) and the purchaser. Fro the PPI surve point of view, the anufacturing enterprise who sells its products is the reporting unit. It s worth noting the difference between the ters product and ite. The producer price excludes, b definition, VAT and excises. The definition of product coes fro the PRODCOM surve, while the ite has to be specified b the enterprise or respondent unit. In the PPI surve questionnaire, the definition and code of 1 The qualit check work for the period Deceber 2006 Deceber was carried out b the Italian statisticians. Consulting assistance was provided to the Entities statisticians for carring out qualit checks for the period Januar June. 10

11 the product is decided b the Institute of Statistics. The enterprise will choose the ost representative goods (ites) in its own anufacturing process and will transit onthl their prices. The ites ust not to be tailored anufacture because in this case their prices would be unique prices, not coparable in tie. In fact, the ai of the surve is to easure the (onthl transaction) price developent b aintaining as uch as possible fixed the transaction conditions. Transaction price refers to transactions that take place in the reference onth. Reporting unit are requested to provide the Institute of Statistics with (ite) producer prices referring to their ain goods. The enterprise singles out the ain ites of its anufacture and transits their prices concerning the ost significant transactions (in ters of turnover) that took place in the reference period (contracts drawn up regarding orders booked during the reference onth). A good practice is to choose, for each ite, the ain transaction that took place during the reference onth and transit its corresponding producer price. Therefore, supposing N orders booked during the reference onth for the sae ite singled out b the enterprises, the price collected has to refer to the ain (aong N) transactions that took place during the reference onth. The ter actual used above, eans that the transaction price includes discounts, rebates and surcharges. This eans that respondents have to identif their ites keeping in ind that price quotations can var over tie, and that the contract conditions have not to affect price quotations level. In other words, the surve is intended to easures the pure price variation. For this reason, it is foreseen that the price deterining characteristics can change over tie. When changes occur, price change has to be adjusted soehow in order to identif the true (pure) price variation Price deterining characteristics The price deterining characteristics are a set of conditions that affect the transaction between producer and purchaser and the (ite) price. When one or ore of the following price deterining characteristics phsical characteristics (qualit) of goods; unit of quantit; easure unit used; 11

12 paent and deliver conditions (paent, packing, transport costs). change, the reporting unit has to eliinate their effect on the ite price. The reason for such qualit adjustent is to easure onl the pure ite price b reoving an other eleent that could affect its level. Furtherore, there are other cases that can be treated b adopting a qualit change. In fact, a qualit adjustent a also occur when: i) a coodit (ite) ceases to be produced (and the observation unit is able to replace it with a new ite); ii) an enterprise A ceases its (anufacturing) activit and if its substitution with a new one (B) is possible, the new enterprise B enters in the list of respondents with its own ites. These will replace the ites of the ceased enterprise A. Technicall, the standard ethod used worldwide for anaging qualit adjustents is the wellknown overlap algorith. How does the overlap work? Let s suppose the reporting unit stopped the production of the ite i=1 since onth -1 onwards. At tie its ite price quotation refers the new ite i=2. This situation a be better analsed b the following three tables where prices, price relatives and rates of change are reported. Table 1 shows the price quotations of a couple of ites, the ite 1, whose anufacturing is supposed to be ceased and the (substituting) ite 2. Table 2 focuses on the corresponding price relatives or icro-indices. Finall, Table 3 shows the price relatives rates of change before and after the onth. 12

13 Table Qualit change: ite prices Year ( - 1) Year () Ites = ,12, 1 Ite 1 p p , 1 Ite 2 x p p,, 1 p Table Qualit change: price relatives Year Year () Price ( 1) relatives = Micro index 1 Micro index 2 I 1,12 1 -, 1 I I, 1 2 I, 2 I, 1 2 Table Qualit change: onth to onth price relatives rate of change Year Year () Rate of ( 1) change = index 1 D 1,12 1,11;1, 1 D - -, 2;1, 1 index 2 - D D,, 2; 2, 1; 2 D, 1, ; 2 13

14 In Table 1.3.3, for instance,,, 1, I 2 I D 2, 1;2, 1 I 2 is the onthl rate of change of the ite 2 (or, better, the rate of change of its icro-index). The unknown variable in the present context is the calculation base for the new ite (ite 2), the replacing one that will substitute the old ite 1 (because, for instance, no ore anufactured and/or sold) fro that tie onwards. This eans the denoinator of the price relative of the ite 2 has to be estiated. For the estiation of the unknown ter, the algorith of the overlapping is the ost used at international level. The unknown variable is estiated b stating the following proportion [1] p 1,12, 1 : x p : 1 1 p, 1 2 so that the solution is [2] If x p 1,12 p 1, p p, 1 2 1, 1, 1 p is unknown, the surve practice suggests to pose 2, involves that, looking respectivel at Tables 2 and 3, p, 1 2 p, I and, 0 2 D., 1;2. This hpothesis 1.4 Saple design topics The producer price surve is based on a saple of products and involves a list of producer enterprises, i.e. anufacturing units whose plants are located within the national borders and whose products are sold on the doestic arket. In BiH the products are identified b the national version of the PRODCOM classification i.e. the PRODCOM NP at 10-digits. Eleentar data (ite prices) are collected b the Entities (FBiH and RS) and the District of 14

15 Brcko (DB). At state level, the Agenc for Statistics (BHAS) collects a onthl set of eleentar product indexes and provides users with aggregate indices at state level. Eleentar product indexes are copiled firstl at 10-digit and then at the 8-digit level. Therefore, in the longitudinal structure of the PPI, there are two product aggregates with their own indexes; this characteristic has been aintained and allows the Entities to use their (old) national PRODCOM classification. A ajor work developed during the project has been the definition of the data collection eleentar unit. The ite is a specific product: its identification is provided directl b the respondent unit (the enterprise), when it fills in the for for the first tie. In other words, the respondent adjusts the product definition provided b the NSI to its own anufacturing process. The outcoe is the identification of one or ore appropriate (in the sense of the PRODCOM definition) ites whose prices will be onthl surveed. The surve questionnaire, in fact, erges products and anufacturing enterprises. Aong the goods anufactured, the enterprise chooses those ites that are representative of its production in ters of the product definition assigned b the NSI. The PRODCOM list is drawn b the structural (annual) business surve and its questionnaire is identified as odel IND21 for both the Entities. In the PPI applied context, the technique for sapling products and industries is alwas a coproise between ethodolog and practice. Products are generall sapled firstl; then, once the basket of products has been defined, the list of enterprises is selected through the business register. The PRODCOM NP list provides the population of products anufactured, i.e. for each code of the classification it gives the value of production sold. It s worth noting that this inforation (coing fro the annual industr surve, odel IND-21) usuall does not allow to ake a distinction between goods (anufactured) sold on doestic and non-doestic arket (this inforation depending on the level of detail of the annual questionnaire). Then the doestic value of production sold has to be soehow estiated (i.e. using an appropriate algorith). 15

16 What allows to erge products and enterprises is the nested criterion of the classification used. The NACE classification is ade up of 8 digits: the first four digits identif the (ain) kind of econoic activit (class level), while the last four ones provide the product identification. The BiH list of products is ade up of 10-digit: the last two characterise the inventor of product in the national version. The Table 4 shows, in a ver snthetic anner, the sapling procedure carried out once a ear when the index is chained. Table Sapling activities STEPS ACTIVITIES Step 1 Step 2 Step 3 Step 4 Step 5 identification of the products population definition of the products saple identification of the reporting units population; definition of the enterprises list joining the products saple and the enterprises list Step 6 data collection operations Step 7 definition of the actual saples of products and enterprises The ain criterion for reducing a population vector to a saple one is to aintain fixed the total aount of values so that what actuall is reduced is the nuber of units. Each ite within its population (and after sapling, in its theoretical saple) has its own (absolute) weight: when the ite is a product, this weight is the PRODCOM value; when the ite is an enterprise, this weight is the turnover (sales of anufactured goods). The relative weight is given b the ratio between the (absolute) value and the su of all ites values. Therefore (b construction) the su of all the relative weights is 1 or 100 (it depends on the adopted scale for deriving the weights relative). Whatever sapling ethod used, the reduction to the saple iplies a weight collapsing procedure so that in ters of absolute (and relative) values (weights), the saple su atches with the population one. After sapling, in fact, we 16

17 distinguish between sapled and non-sapled units; the weights of these last have to be reallocated aong the sapled units. For this reason the weights reallocation does not affect the sapling technique used. 1.5 Weights sste This paragraph deals with the applied procedure for deriving the weights sste for BiH. Such a procedure has two phases: the first concerns the definition of three vectors of weights, one per Entit (FBiH, RS and DB), directl drawn b the structural business statistics (SBS) source (data coing fro the for IND-21). The second phase consists in setting up the sste of weights at state level, b joining the three vectors in ters of a weighted arithetic ean. So doing, the weighted sste is, in ters of a nested classification, a coherent set of values. Of course, the weighted arithetic ean is based on the absolute values of the Entities weights, while the producer price indices are alwas copiled using the weight relatives. Therefore, starting fro the absolute values of weights, within each Entit the weight relatives are ratios (or coefficients) between the absolute values of an aggregate A and the total value of all the aggregates of the overall index. The weight of an aggregate at state level is equal to the su of the absolute values of the Entities weights. The weight relative, for a given aggregate, for BiH is defined b the ratio between the absolute value of the aggregate and the su of the values of all the aggregates. To su up, let s consider a generic aggregate A such that W(A FBiH), W(A RS), W(A DB) are the absolute values of its weight in each Entit. Then, the following equations hold [3] w(a FBiH) = W(A FBiH) A W(A FBiH) A w(a FBiH) = 1 w(a RS) = W(A RS) A W(A RS) A w(a RS) = 1 w(a DB) = W(A DB) A W(A DB) A w(a DB) = 1 W(A BiH) = W(A FBiH) + W(A RS) + W(A DB) w(a BiH) = W(A BiH) A W(A BiH) A w(a BiH) = 1 Within each Entit, producer price indices are copiled using the own vector of weight relatives. This is true for the state level too but the weight relatives are defined after suing the Entities absolute (aggregates) weights. 17

18 In BiH (and in the Entities too), the hierarchical architecture of the classification of products is ade up of seven aggregates Table BiH national products classification LEVELS AGGREGATES 10-DIGIT NP PRODUCT 8-DIGIT PRODCOM PRODUCT 4-DIGIT NACE CLASS 3-DIGIT NACE GROUP 2-DIGIT NACE DIVISION 2-LETTER NACE SUB-SECTION 1-LETTER NACE SECTION - OVERALL The nested structure of this classification allows to anage the overall weight of products at an level of aggregation. In other words, for the equations [3] it alwas holds that given a level of aggregation, sa A, the su of the weight relatives is 1. As concerns the weights source, in FBiH, RS and DB the ain one is the annual industrial surve. Both PRODCOM and the structural business statistics are contained in the for IND- 21. The advantage of using the SBS statistics is first and foreost that data are generall consistent aong Meber State, since the are based upon the Council Regulation on Structural Business Statistics. PRODCOM data are arranged with SBS data: the saple of products is firstl based on the inforation derived b this source. Secondl, fro the SBS source, the aggregate at 4-digit level (classes) are selected. For erging the products values (in ters of PRODCOM) and those of the classes (SBS), the products weight relatives within classes are calculated. Then, the value of an class is detailed at product level, using the weight relatives calculated above. Such a procedure allows to define a coherent and nested sste of weights. The procedure can be suarized with this forula: 18

19 [4] W p = ( Prod R p / Prod R c ) SBS T c where: R = classes weight fro Prodco source; T = classes weight fro SBS source; W= product weight. The ain benefit of this choice is that it aintain the data configuration of the SBS classes preserving the class s structure fixed fro the Prodco source. The SBS variable used to deterine the PPI weight sste is the doestic turnover Theoretical and actual weights sste: the proble of the redistribution of weights The basic concept of sapling (product and enterprises) is that while the saple contains a reduced nuber of units (copared to the population) in ters of total value associated to the reduced set of units, saple and population total values ust be the sae. Such a criterion iplies the redistribution of the non-sapled units values. The sae approach holds when coparing the theoretical saple and the actual one, i.e. the saple before and after the data collection operations (issing units). Let us suppose that the aggregate A is one of the units in the (theoretical) saple (i.e. foreseen b the saple design) but not in the actual one. In ters of actual saple, the aggregate A is a issing unit and its corresponding value is a issing value. Such a situation iplies the redistribution of the value of A aong those included in the actual saple. The ai of redistribution is to aintain the total value of the actual saple equal to its theoretical design. The proble of how to redistribute the issing value of the (non-collected) aggregate A a be approached in a nuber of was. In this context, it was decided to adopt the criterion of closeness in ter of classification. In other words, if A is a issing product, its (theoretical) value will be redistributed aong all the other products within the sae NACE class. Therefore, given a generic aggregate, its place within the classification autoaticall allows the identification of the target (cluster) aggregates. Fro a practical point of view, three cases can be distinguished: 19

20 1. More than one target cluster. In this case the redistribution is ade proportionall to their relative share of values according to a specific calculation described below; 2. One target cluster onl. In this case the effective redistribution consists in adding the value of the issing cluster to the value corresponding to the target cluster; 3. No target clusters available (cluster collapsing). In this case the redistribution is ade in the nearest upper cluster. Exaple Weight redistribution situation Let us suppose the Class issing. Then: Case 1: if there is no cluster for NACE class and in the group 15.1 there are two classes with products collected (class and class 15.12), the weight of class should be distributed between classes and 15.12, not sipl b assigning half of the weight to each of these two classes, but b dividing the weight of Class according to the weight relatives of classes and The index for Group 15.1 is then copiled fro the adjusted weights of classes and Case 2: if there is no cluster for NACE class and in the group 15.1 there is onl one class with products collected (class 15.11), the weight of class should be added to the weight of Class Case 3: if there is no cluster for NACE class and this class represents the onl class for group 15.1, the weight of class should be added to the relative weights of the group The ratio between the value corresponding to a cluster and that corresponding to the hierarchical higher cluster is defined as effective cluster. The first step of the weights redistribution procedure is to calculate the relevant coefficients for all the aggregates. These coefficients represent the division s quota referred to a fixed aggregate. With reference to a generic aggregate, these coefficients will be a part of one cluster coefficient that represents the final division coefficient. This relevant coefficient represents the ultiplicative factor that pairs with 20

21 the total value corresponding to all the aggregates. The result is a new recalculated value to be joined to the aggregate. After redistributing the weight sste can be copiled. Fro the lowest level (the class) the weight s coefficient is equal to the ratio between the value corresponding to the cluster s product (obtained as a result of the previous steps) and the total value of the products. The final step consists in a further distribution of the weight estiates to the products within the classes according to the original ratio percentages of the classes structure (class relative weight). Finall the final weight is perfored b noralizing the total value of the product equal to (b ultiplication of with the weight s coefficients). The process called weight noralization is used to convert each set of original values into a standard scale. In this case the scale used is: total product values = In the sae wa it is possible to calculate the noralized weight s coefficients on the upper aggregation level as indicated above about classes and their upper level of aggregations. In this phase the total su of the noralized weight s coefficients, referred at a certain level, could not be equal to , owing to rounding off probles. This is caused b the approxiate ethod used to clustering , based on 1 unit digit (while in the previous phases we have planned an approxiate ethod based on 15 decial digits). In this case a put and takes technique has been used, based on the relevant of the product s aggregations, in order to adjust the noralized weight distributing the rest derived fro the noralization BiH weights estiation criteria For both the Entities, the data source is the annual surve on industr. Fro the Entities database the total sales values (at 10-digit level, NP PRODCOM) and the turnover values (at 4- digit level) were selected. The first target was the estiation of the doestic product weights. Product weights can be perfored b adjusting the products values b classes fro the 10 digits database, according to the export s shares derived fro the class aggregates, in order to take out the products doestic value of sales. The steps of the procedure are the following: 1. estiation of the classes' export share fro the total sales at the 4-digit aggregates; 21

22 2. arrangeent of the 10-digit aggregates b subtracting the export share estiated b ultipling the product values with the class values, calculated above, referred to the export sales; 3. definition of the share of the new product value on its own class; 4. arrangeent of the new product weights b ultipling the ratio calculated at the previous step with the class value, referred to doestic sales, of the corresponding class; 5. final adjustent of the weight values at product level. In details, the weights estiation criteria run as follows Step1 Fro i =1 to n (n = nuber of classes) [5] 4-digitsC i(d) = ( 4-digit T i(nd) / 4-digit T i(tot) ) where: T = classes weight; C= classes doestic ratio; Step 2 For i =1 to n (n = nuber of classes) and j =1 to ( = nuber of products) [6] 10-digisR j,i(d) = 10-digis R j,i(tot) ( 10-digis R j,i(nd) 4-digit C i(d) ) where: R = products weight; C= classes doestic ratio;. Steps 3 and 4 For i =1 to n (n = nuber of classes) and j =1 to ( = nuber of products) 22

23 10-digitS j(d) = ( 10-digit R j(d) / 10-digit R i(d) ) 4-digit T i(d) where: T = classes weight; R = products weight; S= new product value (10 digits). Step 5 For j =1 to ( = nuber of products) 10-digitj(d) = ( 10-digit S j(d) / j 10-digit S j(d) ) 10-digit S j(d) where: S = new product value; = product weight (10 digits). These product weights will be adjusted for the final estiation of the weights schee. The ain benefit of this choice is to aintain the frae data as it derives at the 4-digit level, preserving at the sae tie the products structure estiated at the 10-digit level. In this wa, the estiates are obtained b a reasonable calibration of the inforation drawn both at 4-digit and 10-digit level. As concerns the weights structure of the District of Brcko, the estiation procedure foresees two steps: 1. to find out, for all products, the share of the doestic product s value b subtracting the export share b division fro the 4 digit; 2. to adjust these values according to the product ratios. The final adjustent regarded the weights reference ear according to the ear related to the sources. Chained indices foresee the annual updating procedure for weights. This eans that for 23

24 ear, the weights vector coes fro ear 2006, while for ear it is derived b ear. Further, price relatives of ear are referred to Deceber 2006 and those of ear are referred to Deceber. For the District of Brcko data referred to 2005; two adjustents occurred under the assuption that the value of products rise such as the PPI percentage ratio fro ear to ear. This adjustents consists in updating the final vector of weights fro the previous ear using a spread, calculated b NACE division, related to the percentage ratio of the PPI fro the Deceber of the current ear to Deceber of the previous ear (equal to the weights reference ear). Finall, the basket of products for the ear was updated. Such operation was ade possible b the desk work that Entities statisticians cancel out b using the software application. The results are the following: Table Products and ites coparison - YEAR PRODUCTS ITEMS FLOWS IN OUT FBiH RS Brcko FBiH RS Brcko FBiH RS Brcko FBiH RS Brcko Furtherore, after perforing the weight adjustent procedure, three noralized vectors of weights are obtained (one per Entit). The weights noralization within the Entities shows the weight relatives or the (weight) coefficients that are actuall used for copiling the PPI indices. The weights noralization aong the Entities, i.e. the BiH vector of weights, derives fro the su, for an aggregate, of the absolute weight in FBiH, RS and DB. Then, there are set up the weight relatives (or weight coefficients) b dividing each aggregate absolute weight b the su of all the absolute weights. 24

25 Table PPI and weight relatives coparison. Main NACE aggregates per Entit. NACE ain aggregates FBiH RS DB Total C CA CB D DA DB DC DD DE DF DG DH DI DJ DK DL DM DN E EA Table PPI and weight relatives coparison. MIGs aggregates per Entit. MIG FBiH RS DB Total Interediate goods Capital goods Consuer durables Consuer non durables Consuer goods Energ

26 Table PPI and weights coparison. Main NACE aggregates per Entit. NACE ain aggregates FBiH RS DB Total C CA CB D DA DB DC DD DE DF DG DH DI DJ DK DL DM DN E EA Table PPI and weight relatives coparison. MIGs aggregates per Entit. MIG FBiH RS DB Total Interediate goods Capital goods Consuer durables Consuer non durables Consuer goods Energ

27 1.6 Indices structure The PPI is a weighted arithetic eans of siple geoetric eans. B using a different terinolog, the PPI is a nested Lasperes index based on the Jevons ones. The first step of the procedure for copiling the PPI deals with the calculation of a series of ratios each one being the quotient between two prices: the current price (nuerator) and the price base (denoinator). When indices are chained, the price base is that of the onth of Deceber of the ear -1. The quotient of such a ratio is also known as price relative (of the ite k). In forula [2], I k, p k 1,12 p k The second step of the copiling procedure is the snthesis of the inforation collected at the ite level. For a given product, K, this snthesis is obtained b using the siple geoetric ean (or Jevons eleentar price index) of all and onl those price relatives that can be associated to it. It s worth noting that while the first step of the procedure involves the calculation of M price relatives (M = total nuber of ites), the second step concerns the calculation of K siple arithetic eans, one per product. [3], I K, ( I kk i ) 1/ kk The reason for using a siple (geoetric) ean for aggregating the price relatives is that there is no inforation to set up a weights structure at ite level. The hpothesis underling is that price relatives have all the sae weight in deterining their correlated product indexes. The final step of the copiling procedure is the calculation of the snthetic indexes, i.e. the upper level indexes in ters of NACE classification. The aggregation criterion is a weighted arithetic eans, the well-known Lasperes forula. Fro the 4-digit level up to the overall index, the Lasperes forula is alwas used. 27

28 ,, [4] I A I K K A w A where [5] w A W K A W K K i.e. the weight relative of the aggregate A within the classification used. The equation [5] is the ratio between the absolute weight of A (given b the su of the absolute weights of all the products that belong to A) and the total weight or the overall index weight, equal to the su of all the product absolute weights in the index breakdown. The longitudinal structure of weights is therefore nested: given a generic aggregate, A, its weight in ters of NACE classification is equal to the su of all the lower (weights) aggregates belonging to A. In other words, if A is 3- digit level (group) in the NACE classification, its weight is equal to the su of all the 4-digit level (classes) aggregates belonging to A (i.e. all the 4-digit level whose first 3 digits are the sae of A). On the other hand, the su of all the aggregates at 4-digit level is equal to the su of all their upper classes. Here is the other aspect of the nested structure: chosen a generic aggregate level, for instance the 4-digit level, the su of the 4-digit level aggregates is equal to the sae scalar: 1 or its ultiples. The weights are nested because, for each aggregate, the su of weights is alwas equal to the sae scalar: 1 or its ultiples. For what concerns the MIGs indices, the aggregation holds b considering groups of products. Therefore, each MIG index is carried out aggregating the eleentar product indices that belong (in ters on PRODCOM) to it. The forula used is alwas the Lasperes one. 1.7 Chained indices eleents The ai of this paragraph is to briefl provide details about the eaning of chained indices. When we sa that an index nuber is chained, we refer to an index annuall chained on a onthl base. The onthl base coonl chosen is the onth of Deceber. Supposing to be in the ear, the chained index of the onth is the ratio between the price collected at tie 28

29 and the price of Deceber of the ear 1. This last price is the price base also known as calculation base. Therefore, at the ear, price relatives are calculated b dividing each ite price at the (current) onth b its corresponding value of Deceber of the previous ear, 1. Forulas, [5] [6], I 1,12; A, I B; A show, respectivel, the calculation base index and the reference base index (B), both copiled for the aggregate A in the onth of the ear. Of course, new ites (i.e. selected for the first tie) enter in the saple in Deceber of the ear -1 and their price relatives will be calculated starting fro Januar of the ear onwards. The reference base index, for a given aggregate A and at tie (, ), is the product of two ters: the calculation base index and the linking coefficient.,, 1 j,12 [7] I B; A I 1,12; A I j,0; A j B 1 where B = 0, so that B + 1 = 1. Exaple calculation and reference base indices Looking at Table 1.7.2, let s focus on the reference base index at tie = 5, = Ma. In Table the corresponding calculation base indices are shown. The reference base index at tie ( = 5, = 5) is the result of an equation like [7]; in particular, the following equalities hold: 1,12 I 1,0 1,002, 2,12 I 2,0 1,015, 3,12 I 3,0 1,019, 4,12 I 4,0 1,010 and 29

30 5,5 I 1,004 4,0 Therefore 2, 5,5 I 1,051 B 1,004 1,002 1,015 1,019 1,010 Note that while in the Tables 1.7. indices are expressed in per cent, in the calculations above the don t. Exaple towards a new reference base: the re-scaling procedure Let us suppose to be at tie = 5 and to decide to update the reference base. We know that when indices are chained, the (calculation) base is updated annuall. Therefore the reference base is alwas updated and its change is onl an algorithic operation: the old reference base is replaced with the new one. Such an operation is sipl a re-scaling of the indices. Technicall, the proble of substituting the old reference base = 1 with the new one, = 5, is solved b rescaling the reference base indices (of the ear = 5). The twelve reference base indices are divided b their siple arithetic ean. The result is a new set of indices for the sae ear = 5 but now referring to the new reference base: the ear = 5 (see Table 1.7.3). Then, , 5, 5, 5, I, 5 I I I I B B B 0 B 5 B 0 I B 0 1 i.e. being, 2 Actuall, when dealing with such a calculations, the rounding off criterion used should be specified, i.e. at what digit data (indices) are rounded off. A good practice, in particular when anaging chained indices, could be the rounding off at the sixth digit, both for the calculation indices and their ratios (reference base indices). Then, cutting indices (or per cent indices) at the third (first) digit rounded off. In these exaples however, we didn t appl such good rule and verif the results b coparing the Tables 1.7. Therefore, soe results could be different owing to the were derived rounding off with a lesser nuber of decials. 30

31 I 5 B , I B 0 1,05 1 it holds that, 5,1 I B 5 1,048 5,2 0,998 ; I 1,05 B 5 1,045 5,12 0,995 ;...,; 1,05 I B 5 1,004 1,054 1,05 Finall, in ters of indices tie (basic reference) series, once the reference base has been updated (in the sense of a re-scaling), the backwards reconstruction of the (new) reference base indices, actuall copletes the re-scaling procedure. Table shows the (new) reference base indices. Of course, this calculation follows exactl the sae ethod used for deriving the new reference base indices. In other words, all the old reference base indices are re-scaled b dividing each one b the arithetic ean of the ear = 5. For instance, focusing the attention on onth = 9 (Septeber) for the ears j (j = 1,..,4), the coparison between the ninth coluns of Tables and can be analsed looking at the following: 1,9 I B 5 1,001 2,9 0,953 ; I 1,05 B 5 1,028 4,9 0,977 ;...,; 1,05 I B 5 0,996 1,046 1,05 Further, it holds that,, D D, 1; B 0, 1; B 5 and,, D D 1, ; B 0 1, ; B 5 i.e. the re-scaling procedure does not odif the index rate of change (both onth to onth and twelve-onth). In fact, let us consider the ear = 3 and the onth of June (see Tables and 1.7.4). The indices onth to onth and twelve-onth rates of change are respectivel: 31

32 3,6 1,026 0,977 3,6 D 1,003 D 3,5; B 0 1,023 0,974 3,5; B 5 3,6 1,026 0,977 3,6 D 1,014 D 2,6; B 0 1,012 0,964 2,6; B 5 Coing back to equation [7] it can be seen that there is another wa of deriving the reference base index. [8], I B; A, I 1,12; A 1,12 I B; A For instance, the sae result in Exaple 1 can be derived b the following 5,5 I 1,051 B 1,004 1,046 32

33 Table Calculation base indices Years Months Jan Feb Mar Apr Ma Jun Jul Ago Sep Oct Nov Dec = 1 100,4 100,5 100,9 100,1 98,9 99,0 99,7 100,2 100,1 100,4 99,6 100,2 = 2 100,2 100,5 100,9 100,6 100,8 101,0 100,5 102,3 102,4 103,0 102,1 101,5 = 3 100,1 100,5 100,1 100,0 100,6 100,9 99,9 101,0 101,5 101,8 101,9 101,9 = 4 100,3 100,4 100,8 100,8 100,2 100,4 100,7 100,8 100,1 100,5 100,9 101,0 = 5 100,1 99,8 99,9 100,3 100,4 100,7 100,5 100,3 99,9 100,1 100,5 100,7 = 6 103,8 104,8 104,7 105,4 106,2 106,3 106,6 107,4 105,9 106,4 106,7 107,4 =7 99,8 99,4 100,0 100,2 100,7 101,6 101,2 100,7 100,3 99,9 101,2 101,4 Table (old) Reference base indices (B = 0) Years Months Jan Feb Mar Apr Ma Jun Jul Ago Sep Oct Nov Dec = 1 100,4 100,5 100,9 100,1 98,9 99,0 99,7 100,2 100,1 100,4 99,6 100,2 = 2 100,4 100,7 101,1 100,8 101,0 101,2 100,7 102,5 102,6 103,2 102,3 101,7 = 3 101,8 102,2 101,8 101,7 102,3 102,6 101,6 102,7 103,2 103,5 103,6 103,6 = 4 103,9 104,0 104,4 104,2 103,8 104,0 104,3 104,4 103,7 104,1 104,5 104,6 = 5 104,8 104,5 104,6 105,0 105,1 105,4 105,2 105,0 104,6 104,8 105,2 105,4 = 6 104,2 105,2 105,1 105,8 106,6 106,7 107,0 107,4 105,9 109,4 106,7 107,4 =7 107,6 107,2 107,8 108,1 108,6 109,6 109,1 108,6 108,1 107,7 109,1 109,3 Table Re-scaling indices Months Year Jan Feb Mar Apr Ma Jun Jul Ago Sep Oct Nov Dec = 5 99,8 99,5 99,8 100,0 100,1 100,4 100,2 100,0 99,6 99,8 100,2 100,4 Table (new) Reference base indices (B = 5) Years Months Jan Feb Mar Apr Ma Jun Jul Ago Sep Oct Nov Dec = 1 95,6 95,7 96,1 95,3 94,2 94,3 95,0 95,4 95,3 95,6 94,9 95,4 = 2 95,6 95,9 96,3 96,0 96,2 96,4 95,9 97,6 97,7 98,3 97,4 96,9 = 3 97,0 97,3 97,0 96,9 97,4 97,7 96,8 97,8 98,3 98,6 98,7 98,7 = 4 99,0 99,0 99,4 99,2 98,9 99,0 99,3 99,4 98,8 99,1 99,5 99,6 = 5 99,8 99,5 99,6 100,0 100,1 100,4 100,2 100,0 99,6 99,8 100,2 100,4 = 6 104,2 105,2 105,1 105,8 106,6 106,7 107,0 107,4 105,9 109,4 106,7 107,4 =7 107,6 107,2 107,8 108,1 108,6 109,6 109,1 108,6 108,1 107,7 109,1 109,3 33

34 1.8 Decoposing the Index rates of change First of all, when talking about onthl rate of change of the index, it is coon to refer to the rate of change between the index at tie and at tie -1, while for the twelve-onth rate of change the ratio is copiled b the index at tie t (for the ear ) and the corresponding index at tie t (in the ear -1). The following forulas are index rates of change in ters of reference base [9] 1 1,, 1, B I B I B D [10] 1 1,, 1, B I B I B D Of course the equations [9] and [10] can be tailored for an sub-aggregate k of the overall index: [11] 1 1, ;, ; 1, ; k B I k B I k B D [12] 1 1, ;, ; 1, ; k B I k B I k B D When dealing with chained indexes, it s worth noting that their non-additivit in ters of basic reference ear iplies soe expedients to be used concerning the rate of change calculation and its (correct) interpretation. The onth to onth and the twelve-onth rate of change (of the total index) are defined (in ters of calculus base) respectivel b the following forulas: 34

35 , I,, 0 [13] D, 1 1, 1 I, 0 1,12 I, 1,0, [14] D 1, I 1,0 1 1, I 1,0 Once defined how to calculate the index rate of change in ters of i) onthl and twelve-onth rate of change; ii) basic reference ear and calculus base; iii) overall index and its subaggregates k, the data analsis suggests a crucial question: how to easure the contribution of each aggregate k in deterining the overall index rate of change. The answer of Ribe (1999) gives a useful solution. Following the Ribe proposal, both the onth to onth and the twelveonth rate of change can be decoposed in coherent (see additive) sub coponents, each of the easuring the actual contribution in deterining the overall rate of change. In fact, supposing the overall index with a breakdown in k sub-aggregates or groups (of products) its forula is, [15] I, 0 w, 0; k k, I, 0; k being w 0; k, the sub aggregate k weight. The basic idea of the Ribe decoposition is that the su of the effects of the aggregates rates of change is equal to the total index rate of change. Therefore, given a generic sub aggregate, k, the rate of change of k a be regarded as its actual contribution to the total index rate of change, both onthl and in ters of the twelve-onth index rate of change. In forulas, this eans respectivel for the equation [9] and [10] that,, [16] C, 1 C, 1; k k 35

36 ,, [17] C 1, C 1, ; k k where, w,0; k,, 1 [18] C I I, 1; k, 1,0; k,0; k I,0 1,12 I, 1,0, [19] C w I 1 1, ; k,0; k 1,,0; k I 1,0 w 1,0; k 1, I 1,0 1,12 I 1,0; k 1, I 1,0; k Furtherore, for the twelve-onth rate of change, such a decoposition has a second relevant propert: it shows that the (total and the sub aggregate) index rate of change can be decoposed in two additive parts: the previous and the current ear effect. B adding and subtracting I 1,12 1,0 in equations [10], we derive the following equation in which the first addend on the left is the current ear effect, while the second one is the previous ear effect,12,12 I I, 1,0, 1,0,, [20] C ( I 1) 1 C ( ) C ( ) 1,,12 1,0,12 1, 1, I I 1,0 1,0 current ear effect previous ear effect 36

37 [21], C 1, ; k 1,12 I 1,0 w, I,0; k 1 k 1, I,0; 1,0 current ear effect w 1,0; k 1,12 1, I I k k 1, I 1,0; 1,0; 1,0 previous ear effect, C ( ) 1, ; k, C ( ) 1, ; k Exaple Additive decoposition of the index rate of change Let us suppose to anage a chained index price whose breakdown foresees onl two subaggregates, A and B. Furtherore, we will analse the situation of indices and their rates of change in a restricted tie series interval: two ears ( and 1) with onthl data. Let the following be the sste of weight relatives -1 A 0,85 0,60 B 0,15 0,40 G 1,00 1,00 In the Tables and indices in calculation and reference base are shown respectivel. G is the overall - or total - index, while A and B its sub-aggregates. Using forulas [9] [12] we derive the results of colun [3] in Table and colun [3] in Table i.e. respectivel the twelve-onth and onth to onth (reference base) indices rate of change. In fact, as concerns Table 1.8.3, colun [3] we have [22] A (1,086 0,975) 1 = 0,1138 B (1,091 1,040) 1 = 0,0490 G (0,991 1,090) 1 0,

38 while, in Table 1.8.5, colun [3] we have [23] A (1,098 1,086) 1 = 0,0110 B (1,096 1,091) 1 = 0,0050 G (1,100 1,090) 1 0,0092 Topic 1. Decoposition of the twelve-onth rates of change within sub-aggregates Given the twelve-onth aggregate rates of change, which is the effect (contribution) due to the current and the previous ear? In other words, the proble is how to decopose the twelveonth rate of change in two additive sub-coponents: the first is the contribution (to the twelve-onth rate of change) of the index dnaics since the beginning of the ear. The second is the contribution due to the past index dnaics, or what has been inherited fro the previous ear -1. Therefore (see Table 1.8.3), b appling, for each aggregate (A, B and G), the equation [20], it follows,,, [24] A C C ( ) C ( ) 1, ; A 1, ; A 1, ; A = (1,116 1,011) (1,009 1) + [(1,116 1,011) 1] = 0, , = 0, ,1138,,, B C C ( ) C ( ) 1, ; B 1, ; B 1, ; B = (1,072 1,021) (0,999 1) + [(1,072 1,021) 1] = - 0, , = 0, ,0489,,, G C C ( ) C ( ) 1, ; G 1, ; G 1, ; G = (1,109 1,013) (1,005 1) + [(1,109 1,013) 1] = 0, , = 0, ,

39 Topic 2. Decoposition of the twelve-onth rates of change aong the sub-aggregates Now, let us look at the aggregates A and B and their respective twelve-onth rates of change. The proble consists in analsing their contribution to the twelve-onth rate of change of the ain aggregate, i.e. G. If we look at the first colun of Table 1.8.4, we note that the twelveonth rates of change are not additive: the rate of G is not equal to the su of those of A and B. Therefore, the ai of the twelve-onth rate decoposition is to explain in ters of additive sub-coponents how the overall rate of change can be actuall seen as the su of two subcoponents, each one corresponding to its own sub-aggregate. Further, we will be able to easure not onl the contributions of the sub-aggregates but also which part of these contributions is due to the current sub-coponent and the previous one. The decoposition proble, as provided b the equation [21], can be faced as follows:,,, [25] A C C ( ) C ( ) 1, ; A 1, ; A 1, ; A = 0,6 (1,109 1,013) (1,009 1) + (0,85 1,013) (1,116 1,011) = 0, , = 0, , ,0881 = 0,0940 A + A = 0,0940,,, B C C ( ) C ( ) 1, ; B 1, ; B 1, ; B = 0,4 (1,109 1,013) (0,999 1) + (0,15 1,013) (1,072 1,021) = - 0, , = 0, , ,0076 = 0,0076 B + B = 0,0076 These forulas eet a twofold goal (see Table 1.8.4): on the one side, it allows to easure the effect on the overall index (twelve-onth) rate of change that is due to its own subcoponents, A and B; on the other side, it is also possible to assign to each sub-aggregates how uch of their contributes are due to the current ear and how uch is due to the previous one. The equation [21] gives an additive solution to this thee; in fact, the following holds: 39

40 Current ear sub-coponent: [26] ) (, ; 1, ) (, ; 1, ) (, ; 1, B C A C G C 0, (- 0, ) 0, Previous ear sub-coponent: ) (, ; 1, ) (, ; 1, ) (, ; 1, B C A C G C 0, , , Overall index rate of change decoposition ) (, ; 1, ) (, ; 1,, ; 1, G C G C G C 0, , , ,1011 Contribute of A in deterining the overall index rate of change decoposition [27] B C A C G C, ; 1,, ; 1,, ;, 1 ) (, ; 1, ) (, ; 1, ) (, ; 1, ) (, ; 1, B C B C A C A C = (0, , ) + (- 0, , ) = 0, , = 0, ,1011 Topic 3. Decoposition of the onth to onth rate of change Finall, forula [18] (and its results in Table 1.8.5) allows to coplete the decoposition thee concerning the indices rate of change. Here the proble concerns the onth to onth index rate of change. [28] 1,,0;,,0; 1,,0,0;, 1;, A I A I I k w A C = (0,6 1,028) (1,020 1,009) 0,

41 1,,0;,,0; 1,,0,0;, 1;, B I B I I k w B C = (0,4 1,028) (1,040 0,999) 0, ,,0,,0, 1;, I I G D = (1,028 1,005) 1 = 0,0023 Here is the final result: B C A C G D, 1;,, 1;,, 1;, = 0, ,0159 = 0,

42 Table Calculation base indices Aggregates w Weights 1 w Calculation base indices 1, I 1, 0 I,, A ,011 1,017 1,116 1,009 1,020 B ,021 1,023 1,072 0,999 1,040 G ,013 1,018 1,109 1,005 1,028 Table Reference base indices Aggregates w Weights 1 w 1, I B Reference base indices I, B A ,975 0,980 1,076 1,086 1,098 B ,040 1,042 1,092 1,091 1,096 G ,991 0,996 1,085 1,090 1,100 Table Twelve-onth rates of change decoposition ( = Januar) (Contribution of the current and previous ear sub-coponents on the aggregates rates of change) Aggregates 1, I B I, B C,, 1, 1, ( ), C ( ) C 1, [1] [2] [3] [4] [5] A 0,975 1,086 0,1138 0,0099 0,1039 B 1,040 1,091 0,0490-0,0010 0,0499 G 0,991 1,090 0,0999 0,0005 0,

43 Table Twelve-onth rates of change decoposition ( = Januar) (Contribution of the sub-aggregates A, B in deterining the overall index rate of change) Aggregates D,, 1, 1, ( ), C C 1, ( ), ( ) C 1, [1] [2] [3] [4] A 0,1138 0,0059 0,0881 0,0940 B 0,0490-0,0004 0,0076 0,0071 G 0,0999 0,0055 0, Table Month to onth rates of change decoposition ( = Februar) (Contribution of the sub-aggregates A, B in deterining the overall index rate of change) Aggregates, 1 I I, D,, 0, 0, 1 C,, 1 [1] [2] [3] [4] A 1,009 1,020 0,0110 0,0064 B 0,999 1,040 0,0410 0,0159 G 1,005 1,028 0,

44 1.9 Surve questionnaire Data collection is carried out through a onthl paper for. A good design of the questionnaire is essential for the success of the price data collection and for the release of accurate and reliable results. It is iportant that the respondent unit (enterprise) understands what the questionnaire is asking; the forat and laout should facilitate the National Institute of Statistics (NSI) in the data extraction (price, ite description, etc) for effective qualit assurance. To eet these objectives, the questionnaire should: provide clear instructions on what the respondent is required to do; provide a clear definition of the product requiring data; show how to change the description of the ite (qualit changes); ensure supporting notes for each ite of data to be collected; request reasons for price changes; enable respondents to coplete the for quickl and accuratel; use siple and clear language; clearl identif the organization responsible for the surve, provide a contact point and telephone nuber for enquiries and further inforation. Paragraph 1.11 shows an exaple of the surve questionnaire that the NSI should send to reporting units at the beginning of each ear together with the introductor letter presenting the surve, the guide for a correct copilation and a blank odel (surve docuentation). The questionnaire displas all inforation contained in the software application for data entr (see Chapter 3 ) and it is structured into the following sections: 1. Heading 2. Personal data and tie period 3. Product description 4. Ite description 5. Monthl inforation 44

45 Sections fro 1 to 4 are pre-filled with the inforation inserted in the database, given b the reporting units both for defining the base of the index and with those provided during the ear to specif qualit changes or additional new ites. The pre-filled for enables respondents to coplete the for ore quickl and accuratel and helps to avoid istake. Section I Heading LOGOS OF THE INSTITUTE OF STATISTICS MONTHLY SURVEY ON PRODUCER PRICES OF INDUSTRIAL PRODUCTS SOLD ON DOMESTIC MARKET Mod. C-41 Section I contains the logos of the NSI, the nae of the surve and the for code. Section II - Personal data and tie period Section II contains personal data and it is divided into the following parts: the first one includes e-ail, address, phones and fax of the NSI. A contact point (telephone and fax nubers) is also indicated, so that respondents can get in touch to resolve possible probles; the second part contains the enterprise code and the personal data of the responsible person. Beside the personal data, there is the tie period part: the respondent has ark with a X the cell corresponding to the current onth. Section III - Product description The inforation contained in the third section concerns the product code and the definition (according to the PRODCOM classification) for which the reporting units have been asked the 45

46 identification of the ost representative ites produced and sold. The products have been assigned to enterprises b the unit/product sapling (see 1.4). Section IV - Ite description Section IV contains the inforation on price deterining characteristics (ite description, unit easure and quantit per unit) that can be changed b respondent onl aking a qualit change b using a blank odel (see 1.3 for rules to anage correctl qualit changes). The ites - selected b enterprises (when necessar, with the assistance of NSI staff) - should: be the ost representative ite of the enterprise s doestic sales of the eleentar product; reflect changes over tie in the average prices of the eleentar product; be produced and sold on a regular basis; be full and clearl described. Accurate price transaction descriptions are critical in ensuring price transaction continuit (the continuit propert ensures that the sae price transaction is priced ever onth so that a picture of the price changes is established). To aintain this principle, a detailed description of the price transaction fro respondents is needed. In other words, units should record all inforation uniquel defining the price transaction selected to ake sure that the sae price transaction is priced onthl. An accurate description is also useful to control the qualit changes: when the description of an ite changes, a check has to be ade to verif if the sae ite has been replaced or not. In the latter case, the change ust be ade. be as continuous as possible over tie in order to ensure prices coparabilit over tie (excluding the products tailor anufactured because their price would be a unique price, not coparable over the tie). 46

47 Exaple Ite description: Assuing the Prodco Sweet biscuits (excluding those copletel or partiall coated or covered with chocolate or other preparations containing chocolate) : correct description: Danish butter biscuits tin packet weight: 300 gra; wrong description: Danish biscuit. The second description is insufficient, as it is not detailed enough to identif a specific ite for which the price has been established. In fact, the onthl price provided b the enterprise could be applied to an different packages, deterining price changes that are not pure. Instead, the individual characteristics listed in the correct description (e.g. ingredient, package and weight) ensure that the returned price is consistent fro period to period, thus allowing changes of qualit to be controlled. Section V - Deadline reinder PLEASE RETURN THE QUESTIONNAIRE 10 DAYS BEFORE THE END OF THE CURRENT MONTH Section V is an deadline reinder that has the purpose to reind respondents to transit the fors within 10 das fro the end of the reference period. This reinder should help the NSI to reach an high response rate (peritting good qualit in the indicator products and less enterprise to follow up on) and to eet the deadline for data transission (1 onth and 15 calendar das) as required b Reg. STS 1158/05. Section VI Monthl price Section VI concerns the inforation that has to be provided onthl b the respondent: i) sale ade: in this field the respondent has to ark es if the enterprise has realized the transaction or no if the enterprise had no an ite sale (in the last case the units shall not fill in the other fields). This inforation is useful to the NSI to identif two tpes of enterprises: 47

48 tpe 1, the enterprise, having sold an ite, fulfills the sale ade field (respondent unit); tpe 2, the enterprise, having not sold ite, doesn t fulfill in the sale ade field (non-respondent unit). In the first case, the unit is not subject to the follow-up procedure. In the second case, the enterprise has not sold goods but doesn t have to provide an inforation. In this case a followup procedure is necessar. ii) price of the current onth and the previous one: in this field the enterprise ust indicate the transaction prices referred to the ain transaction (in ters of turnover) that took place in the reference period and in the previous one. Entering double price allows both respondents and NSI to have an iediate check of the coherence between the previous onthl price and the current one; iii) reasons for price change: when the current and the previous onth price differ, the respondent is asked to ark with x the cell corresponding to the ain cause for such change. The list of causes is broken down in 7 positions: 1. noral arket perforance: the price change is irrelevant or it is ascribed to the noral arket trend; 2. reduction in price/prootion: the price change is due to rebates or prootions; 3. custoer changes: the current onthl price changes according to the quantit of goods ordered b the custoers; 4. copetitive factors: the current onthl price is due to arket copetition, i.e. when the enterprise wants to get a new share in the arket (new clients) or aintain the old shares (old clients) or beat the copetitive pressure fro other countries; 5. aterial cost changes: changes affect the current price; 6. updating reference price list: the current price is updated on the basis of the new list of prices; 7. end of reduction and prootion: the transaction price coes back to the level existing before the reduction in price capaign. Reasons for price change are useful for pre-validating data and for reducing re-follow up procedures with the respondents Qualit data checking Qualit plas a fundaental role in the data production process context; high-qualit standards allow the NSI to be confident with indices it produces and ensure that the observed price changes are genuine and not resulting fro errors. Data editing procedures ust be developed 48

49 to ensure that a high standard of data collection is aintained for each collection period. Data editing involves checks for studing the qualit of data in ters of copleteness, coherenc and continuit of the basic inforation provided b respondents. It should be perfored during the anual checking, data recording and validation phases. The anual checking of the questionnaires should be carried out before data entr b the office personnel that onthl is charged of the following activities: prices are sent in due tie. Otherwise, it would be necessar to take appropriate follow-up action (tiel checks); for has copletel been filled-in, i.e. that the required fields have not been left blank (copleteness checks); all prices deterining characteristics refer to the sae characteristics of qualit, quantit and ters of sale to ensure pure price change over tie. This tpe of control is useful to identif change of qualit not reported b enterprises (coherent checks); there is coherence between the current onthl price and the one received in the previous onth (procedure of anual non statistical checking of price). If the questionnaires don t pass the test of anual qualit check, the staff should contact the respondent. This happens, for instance, when one or ore of the following situations occur: a blank for: it needs to be understood if the blank for eans no sale in the current onth. In this case the respondent should have fill in the specific field no sale indicating no ; for not copletel filled in (incopleteness errors): it occurs when, for instance, the respondent provides price for two ites and gives the price onl for one; anoalous inforation: - price is not coherent with the ite characteristics reported on the for (ite description, easure unit and quantit per unit). Such a case occurs when the respondent akes qualit changes without reporting the; it needs to receive the correct inforation; - the respondent akes qualit changes but indicates no sale. The current onthl price and the previous one for the new ite (or for the new easure unit, quantit per 49

50 unit) are needed for the construction of the index. Therefore the respondent can ake the qualit changes onl when there has been a transaction of the ite; - the respondent has ade a qualit change or added a new ite, indicating a ite description non coherent with the assigned product; outlier price and incopleteness errors: there is no coherence between the current price and previous one and the enterprise has not provided the reason for this change. After clearing up all errors that occurs during the anual checking, the inforation are iputed into the database, using a guided and checked data-entr procedure. It is integrated into the software application that was developed for anaging the PPI surve process (see Chapter 3). In particular, this tool allows to eliinate incopleteness errors that should be verified during the recording phases and to identif the outlier prices. The first tpe of error is eliinated through andator fields, so that operators are obliged to give the requested inforation to coplete the data registration. In the case of no sale in the current onth, the sste requires ticking out a specific box; the outliers prices are identified b a filtering ethod according to whether the price changes (b coparison between the current price and last onthl validated price) fall outside a predefined range (acceptance interval), such as ±10 % or even 50 %. If outliers occur, the operator is forced to indicate the reason for price changes using a cobo box that reports the sae list of causes indicated on the questionnaire. The precoding and the recording of the outliers perit, in the subsequent phase of validation, to investigate onl the variation falling out of the acceptance interval and not all the prices provided b the enterprise. The filtering ethod is useful to capture tping errors, cases where a respondent has erroneousl reported on a different product or has ade a qualit change without reporting the. The acceptance range should be set independentl for each product group. For products that have volatile prices, such as oil or seasonal ites, it is recoended to have quite wide verification tolerances. Other products a have ore stable prices, and narrower tolerances would be ore appropriate. To set verification tolerances for a particular product, price changes over a period of tie (two or ore ears) need to be analsed. Once prices are recorded, the validation procedures (checks) should be run: the operators should detect all errors or outliers identified during the data entr phase b the filtering ethod. Practicall, when a large price change has been identified the operators should alwas check it again. This activit can be carried out: 50

51 coparing the data entered in the database with those reported on the questionnaire to verif possible tping errors; checking if the explanation given b the respondent adequatel describes the divergent price behaviour; asking the respondent to check if the anoalous price is an error or not. In the first case, the respondent has to provide the correct price. If, on the contrar, the price is confired, further explanations about the divergent price developent ust be sought; coparing the current price change with the price change of the sae or of a siilar product to verif if the price behaviour is the sae; analsing the tie series index series of the aggregate (fro eleentar to overall index). In order to eet the deadlines, it is preferable to place ephasis on those outliers that a have a significant ipact on the final result. For exaple, let s consider an eleentar aggregate (product) with a weight of 2 per cent contains 10 prices, and another eleentar aggregate of equal weight referred to 100 prices: obviousl, an error in a reported price will have a uch saller effect in the latter, where it a be negligible, while in the forer it a cause a significant error in the eleentar aggregate index and even influence higher-level indices. Monthl, in addition to the above entioned checks, other qualit reports should be produced to inspect: the respondent doesn t answer for ore than three onths (non-respondent unit). Non-response a be due to: relocation of the establishent in foreign countries; discontinuance of the output activit; change of the reference arket: the enterprise has left the doestic arket and sells its products ainl, and in peranent wa, on foreign arkets; refusal to co-operate; cease of the production activities; lack of awareness of the iportance of statistics. 51

52 In this case it is recoended to verif the reason for the non-response, then to replace the previous respondent with another one able to provide the price for a siilar ite; the respondent provides no sale for ore than three onths; the respondent provides the sae price over a long period of tie. In these cases it is useful a reinder telephone call with the ai of checking whether the ite is no longer representative or the fir has stopped its production. In the forer case a qualit change is needed, in the latter a new copan will be searched as a substitute. After the validation phase, the registered data have to be subitted to autoatic procedure for treating non-responses, qualit changes and eventuall correcting errors and outliers. Subsequent to the index copilation on the basis of validated data, a further anual checking should be ade, aied at locating possible inconsistencies in the trend of indices: firstl, a coparison should be ade between the percentage change on the previous onth and the percentage change on the corresponding onth of the previous ear of the ite indices (the icro-indices); then, siilar product indices should be copared; finall, a coparison should be carried out between the trend of product indices (or group indices) and alternative price index as consuer prices for the sae groups or products. In addition, it is recoended to inspect periodicall the surve qualit indicators for i) assessing the qualit of the surve process; ii) identifing the necessar action for iproving production processes; iii) checking the efficac of the action taken for iproving the qualit of the surve phases. The qualit indicators should be ipleented throughout the different phases of the process. In this context are presented and built in the software (see Chapter 3) a set of standard qualit indicators on data collection (referred both to enterprises and price quotations), editing and iputation phases. In future it could be useful to ipleent other indicators to onitor specific surve aspects or to eet users requireent. The data collection indicators have been defined with the ai of evaluating the success - or the failure in obtaining inforation fro the units (enterprises) during the onthl data collection. In particular, three different rates are available: total response rate, partial response rate and total non response rate (copleentar to the response rate). The editing and iputation indicators have been developed with the ai of docuenting to what extent the original data (price) have been 52

53 validated (price validate rate) and how uch have been iputed (iputation rate). When the qualit indicators assue anoalous value (alar bells) further analsis should be perfored to discover the causes of the strange value. The definition of the quantities needed for the indicators calculation and their coputation are presented in Figures and and in Tables fro to

54 Table Data collection indicators referred to enterprises INDICATOR FORMULAS Dead units rate total dead units/in-scope units *100 Live units rate total live units/in-scope units *100 Percentage of live unit rate Total response rate total live respondents/in-scope units *100 Partial non response rate total partial non-respondents / in-scope units *100 Total non response rate total live non-respondents/in-scope units *100 Percentage of response rate Deadline response rate Following contact response rate respondents within the deadline/in-scope units *100 following contacts respondents/in-scope units *100 Percentage of dead units rate Ceased production rate units that have stopped the production of the assigned product /in-scope units *100 Cased respondents rate cased units / in-scope units *100 Refusal rate refusing units / in-scope units *100 Other reasons rate ceased units for other reasons/in-scope units *100 54

55 Table Data collection indicators referred to prices INDICATOR Collected inforation rate Non collected inforation rate Qualit change rate FORMULAS nuber of ites for which has been possible collected inforation/total ites *100. nuber of ites for which has not possible to collect inforation / total ites*100 nuber of ites involved b qualit changes /total ites *100 Percentage inforation rate no sale rate nuber of no sale /total ites *100 Prices rate nuber of ites for which has been possible collected prices/total ites *100 Percentage price rate In range price rate Out range price rate nuber of ites for which the prices are in range /total ites *100 nuber of ites for which the prices are out range /total ites *100 Percentage qualit change rate Ite substitution rate nuber of substitute ites / total ites *100 Unit easure change rate nuber of substitute ites / total ites *100 Unit quantit change rate Unit paent and deliver conditions change rate nuber of ites involved b unit quantit change / total ites *100 nuber of ites involved b paent and deliver conditions change / total ites *100 Ceased ite rate nuber of ceased quotations/total ites *100 55

56 Table Editing and iputation indicators INDICATOR COMPUTATION Price validate rate nuber of validate quotations/total prices *100 Price non validate rate nuber of non validate quotations/total prices *100 Iputation rate nuber of iputed prices/total prices *100 Figure Classification and definition of the quantities needed for data collection indicators referred to enterprises IN-SCOPE UNITS: total nuber of units belonging to the surve of interest. For saple surve, as in our case, it corresponds to the nuber of sapling units. DEAD UNITS: units that have changed status becoing out of the scope for the surve (i.e. ceased production activit, stopped the production of the assigned product, transfer of establishent to foreign countr ) or have refused to cooperate LIVE UNITS: working units RESPONDENTS: units for which it has been possible to obtain the inforation required NON RESPONDENTS: units that have been subitted to the follow-up procedure but have not provided an inforation TOTAL RESPONDENTS: units for which it has been possible to obtain the total inforation required PARTIAL RESPONDENTS: units for which it has been possible to obtain partial inforation required WITHIN THE DEADLINE: units for which it has been possible to obtain the inforation required within the deadline FOLLOWING CONTACT: units for which it has been possible to obtain inforation after the follow-up procedure. CEASED ACTIVITY: units have ceased production activit CEASED PRODUCTION: units have stopped the production of the assigned product REFUSAL: units refusing to participate to the surve OTHER REASONS: units have ceased to provide inforation required for other reasons 56

57 Figure Classification and definition of the quantities needed for data collection indicators referred to prices and editing and iputation indicators TOTAL ITEMS: total ites provided b the units belonging to the surve of interest. For saple surve, as our context, it corresponds to the nuber of sapling units. DEAD ITEM: ites referred to units dead owing to cease production activit, stopped the production of the assign product, transfer of establishent to foreign countr ) or refusal to the cooperation LIVE ITEMS: ites referred to working units ITEM INVOLVED BY QUALITY CHANGE STANDARD ITEM: ite doesn t involved b qualit changes PRICES: prices that have been collected in the current period NON COLLECTED INFORMATION: inforation that hasn't been possible collected NO SALE: inforation provided fro unit doesn't have an ite sale in the current period IN RANGE PRICES: prices that fall inside the pre-specified interval OUT RANGE PRICES: prices that fall outside the prespecified interval VALIDATE PRICES: prices that fall inside the pre-specified interval NON VALIDATE PRICES: prices that do not fall inside the pre-specified interval IMPUTATED PRICES: prices issing and/or incoherent CHANGE ITEM CHANGE IN UNIT MEASURE CHANGE IN QUANTITY PER UNIT CHANGE IN PAYMENT AND DELIVERY CONDITIONS 57

58 1.11 Exaple of surve docuentation Letter of surve introduction COMPANY NAME ATTN: PERSON NAME ADDRESS CITY, PROVINCE POSTAL CODE DATE Dear Sir/Mada, the (add the nae of the) NSI carries out the Monthl surve on producer prices of industrial products sold on doestic arket in accordance with the European Counit Legislation that regulates the short-ter statistics, with the purpose to easure the average price developent of industrial products of doestic origin at the initial stage of coercialization on doestic arket. The surve refers to a set of products of the ain goods anufactured and sold on doestic arket and it is based on the saple of producer enterprises. Herein are enclosed the questionnaire C-41bis, the ethodological note and a blank odel to specif changes or additional new ites as representative of own production. To eet the deadline for data transission (1 onth and 15 calendar das) as required b the Regulation (EC) STS 1158/05, the respondent is requested to transit the data within 10 das fro the end of reference period. It is a pleasure to reind that it is possible to obtain inforation about the work of the Institute and its services on the following web site add the web side. Thank ou ver uch for our co-operation. Best regards. Signature 58

59 Copilation guide I - MONTHLY QUESTIONNAIRE COMPILATION GUIDE (C-41bis) 1. Surve purpose The doestic output price index easures the average price developent of industrial products of doestic origin at the initial stage of coercialization on doestic arket. Its purpose is to provide inforation on business ccle oveents. It is also used as a deflator, to index contracts in the private sector and as an analtical tool for business and researchers. 2. Legal basis The producer price surve eets the requireents of the Council Regulation (EC) n. 1165/98, concerning the short-ter statistics and the Council Regulation (EC) no 1158/2005, aending Council Regulation (EC) No 1165/98. The definition of short-ter statistics variables are laid down in the Coission Regulation (EC) N 1503/ Observation unit The observation units are the industrial enterprises whose plants have to be placed in the territorial area and its output has to be sold directl on the doestic arket. Enterprises not involved in anufacturing activit are not subject to the surve (the coercial businesses are excluded). 4. Analsis unit The analsis unit is the product-ite anufactured and delivered on doestic arket. The observed ite (pre-filled on the questionnaire) has been selected b the enterprise, within the assigned product, according to the following rules. The ite has to be: 4.1 the ost representative ite anufactured and sold on arket ; 4.2 produced and sold on a regular basis; 4.3 a standard production (tpe products tailor-anufactured are excluded because their price would be a unique price, not coparable over the tie); 4.4 full identified and described in ters of the qualit and transaction characteristics; and as far as possible, their qualit and transaction characteristics should be unchanged over tie in order to ensure coparabilit of the prices. 59

60 5. Surve price The onthl observed price, relating to the product-ite, ust satisf the following rules: 5.1 it is referred to the ost usual stipulation agreeents (concerning quantit, qualit, package and paent) kept stable over tie; 5.2 it is referred to the doestic production directl delivered on the doestic arket; 5.3 it is an actual transaction price, not a list price; 5.4 it is calculated ex-factor. Ex-factor price excludes the insurances and transport costs; 5.5 it is referred to the oent of order not to the oent when the products leave the establishent gates; 5.6 it is provided in national currenc. If the enterprise has no ites sale in the reference period, it is requested to provide the inforation, pointing out no in the specific field on the questionnaire. It is reinded that the onthl prices ust be referred to the sae ite and deliver condition, in order to ensure consistenc fro period to period. Surve prices are not: 5.7 prices referring to the doestic production directl delivered on the non-doestic arket: the sales of goods anufactured b resident enterprises ust be perfored onl on the doestic arket; 5.8 prices referring to the product anufactured on the non-doestic arket; 5.9 prices referring to transaction between units belonging to the sae enterprise group (transfer price). 6. Surve period The price for each selected ite should be referred to the eans of all transactions that took place in the reference period. When such a calculation is too coplicated, the enterprise can provide the price referring to the ain transaction (in ters of turnover) that took place in the reference period. 60

61 7. Change of the selected ites When the ite is no longer representative or ceases to be produced or changes soe of its qualitative or quantitative characteristics, the enterprise ust i) counicate the new inforation b the blank odel (enclosed in the surve docuentation); ii) ust provide for the new ite the current onthl price and the previous one. The previous onth price is needed for the construction of the index. In particular, the qualit changes ust be carried out in the following cases: 7.1 ite substitution when the ite is no larger representative or ite ceases to be produced. The enterprise proposes a new ite representative of the assigned product with the sae characteristics as the previous one in order to guarantee continuit in price data. The new ite is the one to be priced fro now onwards; 7.2 change of quantit unit; 7.3 change of unit easure; 7.4 change in paent and deliver conditions (paent, packing, transport costs). 8. Reason for price change The coparison between the current and previous ite price has to be specified b choosing one of the following causes: 1. noral arket perforance: the price change is irrelevant or it is ascribed to the noral arket trend; 2. reduction in price/prootion: the price change is due to rebates or prootions; 3. custoer changes: the current onthl price changes according to the quantit of goods ordered b the costuers; 4. copetitive factors: the current onthl price is due to arket copetition, i.e. when the enterprise wants to get a new share in the arket (new clients) or aintain the old shares (old clients) or beat the copetitive pressure fro other countries; 5. aterial cost changes: changes affect the current price; 6. updating reference price list: the current price is updated on the basis of the new prices list; 7. end of reduction and prootion: the transaction price coes back to the level existing before the reduction in price capaign. 61

62 Reasons for price change are useful for pre - validating data and for reducing respondent follow-up procedures with the respondents. II- DEADLINE AND DATA TRASMISSION ATTENTION Before copiling questionnaire C-41bis, the respondent should do a cop of the questionnaire; then the cop has to be used for providing the requested inforation; finall, the respondent sends the fulfilled-in questionnaire as specified at the following part 2. 1) Deadline for onthl questionnaire transission Prices ust be transitted within 10 das after the end of the reference onth. 2) Transission odalities: e-ail: (indicate the e-ail) fax: (indicate the nuber fax) post: (indicate the address In order to ease cooperation between the Institute and the enterprise, please keep the enterprise address and the nae of the person in charge of filling the for alwas updated. 62

63 Surve questionnaire LOGOS OF INSTITUTE OF STATISTICS NAME OF INSTITUTE OF STATISTICS MONTHLY SURVEY ON PRODUCER PRICES OF INDUSTRIAL PRODUCTS SOLD ON DOMESTIC MARKET Mod. C-41 ENTERPRISE CODE YEAR PRICE STATISTICS DIVISION Personal data of respondent Month E-ail: pre-filled Surnae: pre-filled Januar Februar March April Address: pre-filled Nae: pre-filled Phones: pre-filled Phone: pre-filled Ma June Jul August Fax: pre-filled Fax: pre-filled E ail: pre-filled Septeber October Noveber Deceber PLEASE SEND THE QUESTIONNAIRE BACK 10 DAYS BEFORE THE END OF CURRENT MONTH Product code pre-filled Product description pre-filled Progressive nuber Serial code Ite description (*) Unit Measure (*) Quantit per unit (*) 1 pre-filled pre-filled pre-filled pre-filled 2 pre-filled pre-filled pre-filled pre-filled 3 pre-filled pre-filled pre-filled pre-filled 4 pre-filled pre-filled pre-filled pre-filled 5 pre-filled pre-filled pre-filled pre-filled Changes of fields indicated with (*) ust be counicate b the blank odel Progressive nuber Serial code Sale ade (indicate es/no) Price of previous onth (in KM) Price of current onth (in KM) If the onthl current price is not equal to the onthl one, ark with x the cell corresponding to the ain reason for price change, [ see the reason for price change] 1 pre-filled 2 pre-filled 3 pre-filled 4 pre-filled 5 pre-filled Reason for price change 1 NORMAL MARKET PERFORMANCE 4 COMPETITIVE FACTORS 6 UP-DATING REFERENCE LIST PRICE 2 REDUCTION IN PRICE/PROMOTION 5 MATERIAL COSTS CHANGES 7 END OF PRICE REDUCTION/PROMOTION 3 CUSTOMER CHANGES 63

64 1.12 Disseination practices Publishing data The crucial ai of disseination is to provide users with validated statistical data. A standard press release forat is aied at presenting coprehensible inforation with a level of detail not tailored on specialist users. At European level there isn t a coon set of rules to structure a press release schee, but onl soe recoendations whose ai is to help users to evaluate the correspondence between data provided b Eurostat and b the NSI. As such these requireents do not attept to align the national press releases forat aong countries. Eurostat onl recoends the use of footnotes or annexes in order to help the data users in reading the press release. The ai of the recoendations on publications is not to replace the national disseination but to assure that users can find suppleentar inforation. Moreover, the users of the data should be ade aware of the interpla of the NSI s with Eurostat in providing statistical inforation. The recoendations are the following: 1. Both Eurostat and the national press releases should give the priar focus to the highest aggregate of econoic sectors according to the aggregation of the STS Regulation and transitted b the Meber State to Eurostat. 2. When the use of aggregates in national publications differs fro the structure of the STS Regulation, such as figures for total industr including construction, these aggregations should be suppleented b figures according to the definition of sectors in the STS Regulation; 3. When national data are calculated (and disseinated in the national press release) using a different ethod with respect to Eurostat, differences aong data provided b the NSI and Eurostat should be indicated in the footnote of the national press release (for instance, when such a differences are due to seasonal adjustents ethods, the working da adjustent, etc.) 4. When data are published on the basis of different definitions copared to those of the Coission Regulations, i.e. 586/2001 and 588/2001, these differences should be stated as well. 5. A National Institute a use oving averages, such as bionthl to bionthl differences. In this case, the coparison of the onth-to-onth (quarter-to-quarter) should be given in addition. 64

65 6. For indicators for which there is no adjustent deeed useful, such as aggregated output prices, there should be growth rates for onth-to-onth (quarter-to-quarter) and ear-to-ear. 7. Year-to-ear growth rate coparisons should be used for the data series. In case oving averages are used, such as two-onths-to-two-onths differences, the onthl (quarterl) ear-to-ear coparisons should be shown in addition. 8. If MIGS are calculated in a different wa copared to the standard of the Coission Regulation 586/2001, the difference should be pointed out in the national Press Release; 9. If data refer to a national classification not atching those foreseen b the STS Regulation (NACE, CPA and CC), the fact should be clearl indicated and, if possible, the differences shown, for exaple in a ethodological annex; 10. Both Eurostat and the NSI s should start indicating in their respective publication the location of where to find additional inforation, for exaple Internet. The Eurostat and national Press Releases a show an suppleentar inforation according to their specific preferences. The press release should include a section called background notes with technical inforation such as: Specification on definitions what are producer price ; Basic qualit inforation (accurac, revisions); Methodological notes about the saple, the weights and the base used for the surve; The classification used; The publication polic; A friendl guide for the users to interpret the inforation of the tables how do I use the data Data revisions One of the ain points of criticis of the STS data is the high frequenc of revisions and the absence of a coon European revision polic. The proposed revision polic, of course, does not ai at ipeding further revisions. On the contrar its scope is to provide Eurostat (and the users too) with the necessar inforation to properl face this topic. The inforation on the revisions should be given with a high priorit. Moreover, the necessar inforation about revisions should be passed together with data and in iportant 65

66 cases beforehand. A correct inforation polic related to revisions enhances the credibilit of the data. Generall there are two reasons for revisions: 1. Revisions due to noral statistical procedures (for instance new inforation available, change in the ethodolog, change of the base ear); 2. Revisions due to the correction of errors. Eurostat suggests the following polic for dealing with revisions: 1. The need for tiel and accurate inforation on revisions priaril applies to gross data and working das adjusted data as specified in the Regulation, and for the correction of errors also to seasonall adjusted and trend-ccle data. 2. Corrections of errors should in principle be done as soon as possible after the are detected. This concerns all kinds of errors, statistical as well as data processing or data transission errors. The NSI will need to judge the iportance of the errors and decide on tiing of the national disseination for the error correction. 3. Detected errors in national data should be ade known as soon as possible to Eurostat. Ideall, this should be done iediatel after the detection. In addition, the data corrections should be counicated as soon as the have been applied to the data. 4. In case of significant errors, Eurostat ust be given the possibilit to infor its users about reasons and the nature of errors and to take appropriate actions. This background inforation on the errors and their correction should be provided together with the corrected data. Ideall, this inforation should be aligned with the national publication on revisions. 5. Changes in seasonall adjusted data or trend-ccle data due to the regular revision of seasonal or trend-ccle estiation are not part of the inforation polic on revisions. However, significant changes, for exaple of the ethod used or a fundaental readjustent of paraeters should be treated like a revision. 6. For statistical data revisions, each NSI should develop a coherent revision polic with the occasions on when to appl revisions as well as a regular tie pattern for the revision. This revision polic should be ade known to Eurostat and should appear in the release calendar for the data suppl b the NSI s. 7. Revisions appl concerning long periods should be done for entire calendar ears. Successive partial revisions should be avoided. 66

67 8. Noral statistical data revisions do not require explicit inforation unless the degree of the revision is ver high and a need to be explained to users. In that case, Eurostat should be infored about the nature of the revisions. A guideline in which case explicit inforation of Eurostat is required should be the perceived national need for inforation. When a NSI recognises such a need at national level, it should provide also Eurostat with this inforation. Nevertheless, it can also happen that Eurostat perceives a need for inforation that goes beond the national level. In that case, the NSI ust respond to a request for inforation b Eurostat within a short period of tie. 9. At an oent, the NSI a decide to carr out a special revision for an period, in addition to the correction of errors and noral statistical data revisions. It should be subject to a prior inforation fro the NSI to Eurostat and fro Eurostat to users that covers the reasons and inforation on the ipact of the revision on the data. NSI s and Eurostat should onl use this possibilit for well-founded reasons, such as change of base ear, change of ethodolog, etc. 10. The revisions should not cause differences in the data published in electronic for b Eurostat and the Meber State concerned. This iplies that revisions have to be transitted to Eurostat no later than the are released at the national level Publishing tietable In order to be aligned to the standard tieliness the NSI should provide a detailed tietable for publishing the PPI using a press release. This onthl tie table ust be planned at the end of the ear and for the whole following ear. The standard inforation about this tietable foresees the identification of the index denoination (for instance: Producer Price Index), date and tie of the releases, the publishing flag (provisional or final), the revision polic and other inforation about possible press briefing linked to the release. 67

68 1.13 Press Release A press release standard foresees a docuent ade up of two ain parts: tables of results, graphs and short coents to highlight the current onth results and their evolution easured in ters of onth to onth and twelve-onth rates of change. Moving averages, tie series indices and ethodological notes are also included in the press release. A BiH press release proposal is shown below (see p. 67). It s content concerns the ain aggregate indices for FBiH, RS and the District of Brcko. Results have been presented in ters of NACE sectors of econoic activit and b MIGs (Main Industrial Groupings) classification. 68

69 Producer Prices Index for Industrial Products June In the Bosnia Herzegovina (BiH) the output price index for hoe sales of anufactured products (base Dec =100) results 112,3 rising 2,1 per cent on previous onth and 12,2 per cent in the ear to Ma. The output price index excluding energ and petroleu rose 3,2 per cent on previous onth and 11,2 per cent in the ear to Ma. The average difference in the first six onths of the ear rising 0,4 per cent on six onths of the previous ear. Overall Index Rate of change M/M-12 In the Bosnia Herzegovina Federation (FBiH) the output 13 price index for hoe sales of anufactured products (base Dec 10 =100) results 109,8 rising 3,1 9 BiH 8 RS per cent on previous onth and 7 6 FBiH 9,8 per cent in the ear to Ma. 5 4 In the Sprska Republic (RS) the 3 output price index for hoe sales 2 1 of anufactured products (base 0 Dec =100) results 111,7 G F M A M G (Fro Januar to June ) rising 1,1 per cent on previous onth and 11,5 per cent in the ear to Ma. For Brcko district the output price index for hoe sales of anufactured products (base Dec =100) results 154,4 decreasing 1,6 per cent on previous onth and rising 54,2 per cent in the ear to Ma. Table 1. Index of Producer Prices for Industrial Products (base Deceber =100). June INDEX Percentage changes June Jun 08 Jun 08 Jan 08-Jun 08 (a) Ma 08 Jun 07 Jan 07-Jun 07 BiH ,1 + 12,2 + 8,7 FBiH 109,8 + 3,1 + 9,8 + 6,2 RS 111,7 + 1,1 + 11,5 + 8,6 BRCKO 154,4-1,6 + 54,2 + 47,3 (a) Six-onths-on-six-onths average difference. 69

70 Analsis b MIG S (Main Industrial Groupings) classification In the Bosnia Herzegovina (BiH) there isn t variation on the previous onth for energ and capital goods; there are variations for consuer goods (inus 0,1 per cent) and the interediate goods (plus 5,9 per cent). In the Bosnia Herzegovina (BiH) there are the following variation in the ear to Ma: interediate goods plus 18,2 per cent, consuer goods plus 10,8 per cent (durables plus 1,3 per cent, non durables plus 11,6 per cent), capital goods plus 2,1 per cent, and for energ plus 8,8 per cent. In the first six onths of the, the ost relevant variation on the sae period of the previous ear results fro the interediate group (plus 11,3 per cent). Table 2. BiH - Index of Producer Prices for Industrial Products b MIG S classification (base Deceber =100). June Main Industrial Groupings INDEXES Percentage changes June Jun 08 Jun 08 Jan 08-Jun 08 (a) Ma 08 Jun 07 Jan 07-Jun 07 Consuer goods 111,1-0,1 + 10,8 + 9,2 Consuer durables 102,4 0,0 + 1,3 + 1,2 Consuer non durables 111,8-0,1 + 11,6 + 9,8 Capital goods 103,0 0,0 + 2,1 + 2,0 Interediate goods 124,5 + 5,9 + 18,2 + 11,3 Energ 102,4 0,0 + 8,8 + 7,0 Overall Index ,1 + 12,2 + 8,7 (a) Six-onths-on-six-onths average difference. Analsis b sector of econoic activit The ost relevant sector in ters of rising both on previous onth and in the ear to Ma is etal and etal products (on previous onth plus 12,4 per cent, in the ear to Ma plus 29,7 per cent). In ters of decreasing on previous onth the sectors are onl wood and wooden articles (excluding furniture) (inus 0,4 per cent) and cheical products and snthetic fibres (inus 0,1 per cent). In ters of decreasing in the ear to Ma the sectors are onl textile products, clothing (inus 3,2 per cent) and cheical products and snthetic fibres (inus 2,1 per cent). The ost relevant average difference in the first six onths of the ear results b sector food, beverages and tobacco rising 14,4 per cent on six onths of the previous ear. 70

71 Table 3. BiH - Index of Producer Prices for Industrial Products b sector of econoic activit (base Deceber =100). June SECTOR INDEXES Percentage changes June Jun 08 Jun 08 Jan 08-Jun 08 Ma 08 Jun 07 Jan 07-Jun 07 C. Minerals 106,7 + 0,2 + 5,9 + 5,7 DA Food, beverages and tobacco 117,7 0,0 + 16,2 + 14,4 DB Textile products, clothing 95,5 0,0-3,2-4,2 DC Leather. leather and hide articles DD Wood and wooden articles (excluding furniture) 100,9 + 0,2 + 1,8 + 1,5 106,4-0,4 + 5,3 + 5,6 DE Paper and publishing 101,4 + 0,2 + 0,7 + 0,8 DF Coke, petroleu products 122,1 + 3,7 + 20,3 + 11,3 DG Cheical products and snthetic fibres DH Rubber and plastic products 99,2-0,1-2,1-1,5 101,6 0,0 + 0,9 + 2,9 DI Non etal inerals 121,3 + 3,1 + 9,9 + 12,6 DJ Metal and etal products 137,3 + 12,4 + 29,7 + 11,1 DK Machiner and echanical equipent DL Electrical achiner. electrical and optical 104,8 0,0 + 0,1 + 0,8 117,9 + 2,2 + 18,6 13,6 DM Means of transport 103,1 0,0 + 2,9 + 2,8 DN Other products or Manufacturing industries 107,6 + 2,0 + 5,6 + 2,7 (including furniture) E. Electricit, gas, water 99,8 0,0 + 9,4 + 7,0 Overall Index ,1 + 12,2 + 8,7 (a) Six-onths-on-six-onths average difference. 71

72 1.14 Concluding coents The results of this joint work clearl show that the ajor ais of the PPI Sub-coponent have been achieved. In conclusion, the BiH Agenc for Statistics is now able to provide a onthl PPI b erging the Entities indices. At the sae tie, the Entities SIs can provide their respective users a set of indices in line with the European standards. However, soe topics were not dealt with. In particular the following thees were nor treated during the project but could be appropriatel considered in the near future: 1. the changeover to NACE Rev. 2; 2. the non-doestic coponent of the producer price index; 3. backcasting topics. 72

73 2. BiH Producer price indices (Deceber 2006 June ) 2.1 Main criteria for checking data qualit In order to respect the qualit requireent planned for the surve, in ters of accurac and reliabilit of the indexes, following the collection of the data it is necessar to check the database. This phase is called data analsis and it involves an adjustent on the tie series of data collecting. This adjustent is necessar on one side to verif the results of the surve plan and correction phase; on the other side, to correct possible inconsistenc of the data with hpothesis fixed during the surve design. The data qualit check has been perfored on the data provided b each Entit SI for the period Deceber June. In particular, fro Deceber 2006 to Deceber the data qualit check has been carried out b Istat experts whether the price changes (fro the previous to the current onth) fall outside a soe predefined range (filtering ethod). On the contrar, fro Januar to June the check and validation phases have been carried out directl b each Entit SI b using the PPI Aplikacija software provided. In order to ake the checking procedure referring at each (ite) quotation prices as clearer as possible, the adopted sste of hpothesis foresees the following actions: Range 1: Monthl percentage changes = 0 Reason: prices are stable for all the investigated (inspected) period. Action: validated without corrections. Range 2: Monthl percentage changes < ±15% Reason: the price variation are ascribed to the noral arket trend. Action: validated without corrections. Range 3: ± 15%< Monthl percentage changes < ± 30 Reasons: o onl one price lies outside the range: updating reference price list or reduction in price (onl if there is a teporar reduction in price, then the price level coes back to the previous one). o ore than one prices lies outside the range: noral arket trend Action: validated without corrections. Range 4: Monthl percentage changes > ± 50% Reasons: o onl one price lies outside the range: qualit change or tping error Action: qualit change or corrections 73

74 o ore than one price lies outside the range: anoalous prices; in the case of large and recurrent variations (> ± 100), it has been decided to delete the ite. Action: it has been decided to delete the ite. The following tables show the outcoe of the analsis (in absolute figures and in percentage) in accordance with the ethods and the hpotheses described above. In particular the results for FBiH and RS are the following: Table FBiH and RS ite s percentage of flags stable price noral arket trend reduction in price updating price qualit change double qualit change qualit change + anoalous prices anoalous prices tping error eliination FBiH RS Table FBiH ite s percentage of flags b NACE DIVISION DIV stable price noral arket trend reduction in price updating price qualit change double qualit change qualit change + anoalous prices anoalous prices tping error eliination Total 10 2,25 0, , ,09 1,04-0,63 0, , ,66 4,84 0,17 2,94 2,71-0,06 0,35 0,06-21, , , , , ,54 0,17-0,23 0, , ,36 0,46-0,17 0,12 0,06-0, , ,07 2,02-0,58 0, ,12-10, ,50 0,75-0, , ,06 2, ,50 0,40-0,06 0, , , , , , ,23 1,38-0,81 0,29-0,06 0, , ,80 0,86 0,06 0,12-0,06 0,06 0, , ,01 2,94-0,75 0,06-0, , ,23 0,63-0,12 0, , ,80 0, , ,12-0,06 5, ,69 0,40-0, , , , ,29 0,06-0,06 0, , , , ,46 1,15-0, , ,65 1,73-0,58 0, , , , ,04 Total 64,15 21,61 0,23 7,72 4,50 0,12 0,17 1,10 0,17 0,

75 Table RS ite s percentage of flags b NACE DIVISION DIV stable price noral arket trend reduction in price updating price qualit change double qualit change qualit change + anoalous prices anoalous prices tping error eliination Total 10 0,10 0,21-0, , ,21 0,10 0, , ,45 0,83 0,10 0,52 1,04 0, ,10 4, ,98 10,57 0,31 4,56 5,18-0,11 0,52 0,10 0,62 29, ,41 0, , ,83 0,10 0,10 0,10 0, , ,31 0,52 0,10 0,21 0,73-0,22 0,10-0,21 1, ,62 0, ,21-0, , ,25 4,04-1,55 2,59 0,41 0,33 0,20-0,72 13, ,31 0,93-0,21-0, , ,80 1,97-0,52 0, ,10-0,10 6, ,10 0,52 0,21 0,62 0, ,41 1, ,49 1,14-0,31 0, , , ,35 0,52-0,41 0,41-0,11 0,31-0,10 3, ,97 1,97-0,73 0,62 0,21-0,10 0,21-5, ,31 0,52-0,10 0, , , ,35 2,90-0,41 0,62 0,10 0,11 0,10 0,21 0,51 6, ,93 0,31-0, , ,10 0, , , , ,04 0,31-0,41 0,41 0,10-0,10-0,10 2, , ,10 0, ,10 0, ,52 0,31-0,10 0,10 0, ,21 0,10 1, ,21 0,31-0,21 0, , , ,10 0, ,87 0,83-0,62 0,83 0,10-0,20-0,31 7, ,21 1,14-0,21 0, , ,10 0, , ,20-0,21 1,35 Total 35,23 30,78 0,93 12,12 15,44 1, ,48 0,73 3, Concerning the corrections, the following procedure has been adopted: o o Anoalous price: in this case the price carring forward the last price has been adopted. Qualit change: in this case the overlapping ethod has been applied. Finall the ite has been deleted when it was not possible to operate an adjustent (the profile of the tie series is too uch variable). Another cause of eliination of the ite refers to a isatching error fro the PRODCOM NP code of the product collected to the code of the sources used for weighting schee. 75

76 2.2 General rearks concerning PPI Fro Deceber 2006, the PPI started at Entit and State level too. The ai of this brief note is to introduce a graphic reading of the PPI Sub-coponent results. It s worth noting that the results are net of the qualit check work carried out b the Italian and the Entities statisticians (respectivel, sub-period Deceber 2006 Deceber, sub-period Januar June ). Furtherore in the first sub-period a nuber of stable prices characterised indices. The second sub-period shows a higher volatilit in the indices profile, also reflecting the results of the follow up procedures with the respondents. Such an approach was not allowed when checking data in the desk work (first sub-period). Indices fro Deceber 2006 up to Deceber are copiled adopting as a reference base Deceber Fro Januar, the indices calculation base is Deceber while the reference base is Deceber The Tables 1.1, 1.2 and show that between Deceber and Januar the units flow (products) are actuall negligible and, as such, not affecting indices and their rates of change. Looking at the graphic indices in, the BiH Sub-section DJ shows an high (twelve-onth) rate of change, surveed in June (+ 29,7%). Such a rate of change is the result (weighted according to an arithetic ean) of the Entities SIs rates of change (+ 38,3% FBiH and + 17,2% RS). It is worthwhile to note that the DJ twelve-onth rates of change show in and for both the Entities SIs an increasing onotonic trend. A siilar analsis can be ade referring the Sub-section DF, whose twelve-onth rate of change in June is equal to + 20,3% for BiH and, respectivel for FBiH and RS, + 49,2% and + 5,5%. This last result confirs, in general, that the BiH rates of change in are firstl affected b the volatilit shown b the FBiH indices. As entioned above, such a difference sees to suggest a different approach in carring out the PPI surves. 76

77 2.3 Graphical analsis b entities (period 2006 ) Graph 1 Producer Price, Overall Indices Year (onthl data, base Dec =100) 160,0 150,0 140,0 130,0 120,0 110,0 100,0 90,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 2 - Producer Price, Section C indices. Year (onthl data, base Dec. 2006=100) 118,0 115,0 112,0 109,0 106,0 103,0 100,0 97,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 77

78 Graph 3 - Producer Price, Section D indices. Year (onthl data, base Dec. 2006=100) 160,0 150,0 140,0 130,0 120,0 110,0 100,0 90,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 4 - Producer Price, Section E indices. Year (onthl data, base Dec. 2006=100) 115,0 110,0 105,0 100,0 95,0 90,0 85,0 80,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 5 - Producer Price, Subsection CA indices. Year (onthl data, base Dec. 2006=100) 125,0 120,0 115,0 110,0 105,0 100,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 78

79 Graph 6 - Producer Price, Subsection CB indices. Year (onthl data, base Dec. 2006=100) 125,0 120,0 115,0 110,0 105,0 100,0 95,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 7 - Producer Price, Subsection DA indices. Year (onthl data, base Dec. 2006=100) 170,0 160,0 150,0 140,0 130,0 120,0 110,0 100,0 90,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 8 - Producer Price, Subsection DB indices. Year (onthl data, base Dec. 2006=100) 106,0 104,0 102,0 100,0 98,0 96,0 94,0 92,0 90,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 79

80 Graph 9 - Producer Price, Subsection DC indices. Year (onthl data, base Dec. 2006=100) 104,0 102,0 100,0 98,0 96,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 10 - Producer Price, Subsection DD indices. Year (onthl data, base Dec. 2006=100) 114,0 112,0 110,0 108,0 106,0 104,0 102,0 100,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 11 - Producer Price, Subsection DE indices. Year (onthl data, base Dec. 2006=100) 108,0 106,0 104,0 102,0 100,0 98,0 96,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 80

81 Graph 12 - Producer Price, Subsection DF indices. Year (onthl data, base Dec. 2006=100) 150,0 140,0 130,0 120,0 110,0 100,0 90,0 80,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 13 - Producer Price, Subsection DG indices. Year (onthl data, base Dec. 2006=100) 115,0 112,0 109,0 106,0 103,0 100,0 97,0 94,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 14 - Producer Price, Subsection DH indices. Year (onthl data, base Dec. 2006=100) 106,0 104,0 102,0 100,0 98,0 96,0 94,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 81

82 Graph 15 - Producer Price, Subsection DI indices. Year (onthl data, base Dec. 2006=100) 130,0 125,0 120,0 115,0 110,0 105,0 100,0 95,0 90,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 16 - Producer Price, Subsection DJ indices. Year (onthl data, base Dec. 2006=100) 160,0 150,0 140,0 130,0 120,0 110,0 100,0 90,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 17 - Producer Price, Subsection DK indices. Year (onthl data, base Dec. 2006=100) 106,0 105,0 104,0 103,0 102,0 101,0 100,0 99,0 98,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 82

83 Graph 18- Producer Price, Subsection DL indices. Year (onthl data, base Dec. 2006=100) 125,0 120,0 115,0 110,0 105,0 100,0 95,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 19 - Producer Price, Subsection DM indices. Year (onthl data, base Dec. 2006=100) 104,0 103,0 102,0 101,0 100,0 99,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 20 - Producer Price, Subsection DN indices. Year (onthl data, base Dec. 2006=100) 118,0 116,0 114,0 112,0 110,0 108,0 106,0 104,0 102,0 100,0 98,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J 83

84 Graph 21 - Producer Price, Subsection EA indices. Year (onthl data, base Dec. 2006=100) 116,0 112,0 108,0 104,0 100,0 96,0 92,0 88,0 84,0 80,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 22 - Producer Price, Interediate goods industries indices. Year (onthl data, base Dec. 2006=100) 135,0 130,0 125,0 120,0 115,0 110,0 105,0 100,0 95,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 23 - Producer Price, Capital goods industries indices. Year (onthl data, base Dec. 2006=100) 105,0 104,0 103,0 102,0 101,0 100,0 99,0 98,0 97,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J 84

85 Graph 24 - Producer Price, Durable consuer goods indices. Year (onthl data, base Dec. 2006=100) 103,0 102,0 101,0 100,0 99,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 25 - Producer Price, Non-durable consuer goods indices. Year (onthl data, base Dec. 2006=100) 170,0 160,0 150,0 140,0 130,0 120,0 110,0 100,0 90,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J Graph 26 - Producer Price, Consuer goods indices. Year (onthl data, base Dec. 2006=100) 170,0 160,0 150,0 140,0 130,0 120,0 110,0 100,0 90,0 J F M A M J J A S O N D J BiH FBiH RS Brcko F M A M J 85

86 Graph 27 - Producer Price, Energ indices. Year (onthl data, base Dec. 2006=100) 115,0 110,0 105,0 100,0 95,0 90,0 85,0 J F M A M J J A S O N D J BiH FBiH RS F M A M J Graph 28 Producer Price, Overall rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 86

87 Graph 29 Producer Price, Section C rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 30 Producer Price, Section D rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 31 Producer Price, Section E rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 87

88 Graph 32 Producer Price, Subsection CA rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 33 Producer Price, Subsection CB rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 34 Producer Price, Subsection DA rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 88

89 Graph 35 Producer Price, Subsection DB rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 36 Producer Price, Subsection DC rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 89

90 Graph 37 Producer Price, Subsection DD rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 38 Producer Price, Subsection DE rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 90

91 Graph 39 Producer Price, Subsection DF rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 40 Producer Price, Subsection DG rates of change b Entities. Year (onthl data, base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 41 Producer Price, Subsection DH rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 91

92 Graph 42 Producer Price, Subsection DI rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 43 Producer Price, Subsection DJ rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 92

93 Graph 44 Producer Price, Subsection DK rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 45 Producer Price, Subsection DL rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 93

94 Graph 46 Producer Price, Subsection DM rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 47 Producer Price, Subsection DN rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 94

95 Graph 48 Producer Price, Subsection EA rates of change b Entities. Year (base Dec =100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 49 - Producer Price, Interediate goods industries rates of change. Year (base Dec. 2006=100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 50 - Producer Price, Capital goods industries rates of change. Year (base Dec. 2006=100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 95

96 Graph 51 - Producer Price, Durable consuer goods rates of change. Year (base Dec. 2006=100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 52 - Producer Price, Non-durable consuer goods rates of change. Year (base Dec. 2006=100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 96

97 Graph 53 - Producer Price, Consuer goods rates of change. Year (base Dec. 2006=100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 54 - Producer Price, Energ rates of change. Year (base Dec. 2006=100) J F M A M J J A S O N D J F M A M J BiH M_M-1 RS M_M-1 FBiH M_M-1 BiH M_M-12 RS M_M-12 FBiH M_M-12 Graph 55 Brcko Producer Price, Overall rates of change. Year (base Dec =100) 60,0 50,0 40,0 30,0 20,0 10,0 0,0-10,0-20,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 97

98 Graph 56 Brcko Producer Price, Section D rates of change. Year (base Dec =100) 60,0 50,0 40,0 30,0 20,0 10,0 0,0-10,0-20,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 Graph 57 Brcko Producer Price, Subsection DA rates of change. Year (base Dec =100) 70,0 60,0 50,0 40,0 30,0 20,0 10,0 0,0-10,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 98

99 Graph 58 Brcko Producer Price, Subsection DG rates of change. Year (base Dec =100) 14,0 12,0 10,0 8,0 6,0 4,0 2,0 0,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 Graph 59 Brcko Producer Price, Subsection DN rates of change. Year (base Dec =100) 8,0 7,0 6,0 5,0 4,0 3,0 2,0 1,0 0,0-1,0-2,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 99

100 Graph 60 Brcko Producer Price, Interediate goods industries rates of change. Year (base Dec =100) 12,0 10,0 8,0 6,0 4,0 2,0 0,0-2,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 Graph 61 Brcko Producer Price, Durable consuer goods rates of change. Year (base Dec =100) 8,0 6,0 4,0 2,0 0,0-2,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M

101 Graph 62 Brcko Producer Price, Non-durable consuer goods rates of change. Year (base Dec =100) 80,0 70,0 60,0 50,0 40,0 30,0 20,0 10,0 0,0-10,0 J F M A M J J A S O N D J F M A M J Brcko M_M-1 Brcko M_M-12 Graph 63 Brcko Producer Price, Consuer goods rates of change. Year (base Dec =100) 80,0 70,0 60,0 50,0 40,0 30,0 20,0 10,0 0,0-10,0 J F M A M J J A S O N D J F M A M J M_M-1 M_M

102 3. Software User s Guide 3.1 Main Panel The ain appearance of the software release PPI Applikacija (Figure 3.1.1) is a panel ade up of three logical coluns. In the iddle one there are the links to the Calculation Panel, the Indexes dnaic report and the Qualit indicators report. On the right the colun links contains the action buttons. On the left colun there are the links to the ain facilities of the application. the reports and actions tables. All the fors (facilities) will be shortl described in this docuent. About the aintenance of the software application, descriptions tables have to be updated ver rarel, because the inforation the contain, are set up at the beginning of the software use. The action tables need to be updated ever ear (in Deceber), after the calculation of final indexes. These tables are: BASE_PROD_SERIES, HIERARCHY, WEIGTH. During the ear, before starting the new series (ites) data collection (for the new base), it is necessar to add new units (enterprises and new products) in the appropriate tables. The for (Data Entr) is used for updating the table PRICES_COLLECTIONS. The other tables concern the calculation procedure. To access each for the user has to push the linked button. Each for contains a button naed Back to coe back at the Main Panel for. 102

103 Figure Main Panel 3.2 Description tables The for Bases (Figure 3.2.1) allows the user to entr and odif eleentar data in the bases table. The eleentar data entered concern the calculation base index being the chained Indices firstl copiled in their calculation base and then in the reference one. In the table PRICES_COLLECTIONS, the eleentar prices are stored. To add a new base the user has to input its data in the row arked b the asterisk. It s worth noting that the indexes can be calculated on the reference base starting fro the second ear onl. 103

104 Figure For Bases The for Product classification (Figure 3.2.2) allows to input and odif data in CLASSIFICATIONS tables. The classifications concern products. To add a new product classification the user has to insert its nae in the row arked b the asterisk. Figure For Product classification The for Unit of easure (Figure 3.2.3) allows the user to input and odif eleentar data in the table UNITS_OF_MEASURE. To add a new unit of easure the user has to insert its nae in the row arked b the asterisk. 104

105 Figure For Unit of easure This for Rearks (Figure 3.2.4) allows to input and odif eleentar data in the table REMARKS. WARNING: the id_reark 0 (no rearks), 1 (ceased ite) and 2 (ceased enterprise) ust not be odified since their values are those used in the VBA identification codes for ceased units (ites and enterprise). Figure For Rearks ID_REMARK REMARK_DESCRIPTION 0 No rearks 1 Ceased ite 2 Ceased enterprise The for Out of range causes (Figure 3.2.5) allows the user to input and odif eleentar data in the table OUT_OF_RANGE_CAUSES. The eaning of KIND field is to define if the 105

106 reasons or causes description refer to a price that is out of range (because greater than the axiu price set in PRICES_COLLECTIONS table KIND=High or because saller than the iniu price set Kind=Low). To add a new out of range cause the user has to select its tpe and enter its nae in the row arked b the asterisk. Figure For Out of range causes 3.3 Action tables and facilities These facilities are placed on the left frae of the ain panel. Procedure steps: 1. select the reference (i.e. the onth at who data enter operation refer); 2. set up the elaboration onth and define the range for the elaboration onth; 3. ake the data entr and checking work (Out of range Analsis); 4. calculate the indexes. The for Enterprise List (Figure 3.3.1) allows the user to insert, odif, delete enterprises. WARNING: Deleting an enterprise eans deleting all the inforation linked (in the other tables). To add a new enterprise the user has to push the button naed Add new enterprise in the botto of the for, entering data required to press the button Back or push on the Record selector bar to save the data just entered. 106

107 Figure For Enterprise List The facilit Products and ites (Figure 3.3.2) allows the user to insert, odif and delete products and their corresponding ites. The ites can be entered directl b the data entr for, after inserting the questionnaire, through the button Add new series. To add a new product the user has to push the icon at the botto of the for, then insert the product code and its description; to save the data just inserted the user has to push the button Back or push on the record selector bar. 107

108 Figure For Products and Ites The for Questionnaires (Figure 3.3.3) allows the user to join the enterprises with the products required. The Id_Questionnaire is the priar ke and allows to identif questionnaires. Such a priar ke should be printed in the paper questionnaire. To add a new questionnaire the user has to select the corresponding enterprise and product, in the row arked b the asterisk. 108

109 Figure For Questionnaires The for Bases Products Series (Figure 3.3.4) allows the user to read the products and the series list referring to each base onl. However, it does not allow the user to enter an data, as onl the IT expert assistance allows to enter soe additional data. The data of this table (BASES_PRODUCTS_SERIES) ust be inserted at the beginning of each ear. 109

110 Figure For Bases Products Series The working calendar for (Figure 3.3.5) allows the user to define, for each period (starting fro the field Begin_Data up to the field End_Data) which onths are qualified for data entr operations (data-entr, qualit change and the adding of new series). The first da of the onth the data collection operation starts with regard to the previous onth -1. Usuall, up to the first half of the onth, there are two qualified onths for data entering: the forer concerns the calculation of final indices of the onth -1 whle the latter onth, allows to enter prices concerning the provisional calculation. The for contains soe checks in order not to input incongruent data. 110

111 Figure For Working da calendar The for Consolidation and range definition (Figure 3.3.6) ust be used iediatel after the index calculation. It allows the user to consolidate the onth for which indices has been just calculated. Once provisional indices have been consolidated the calculation of final indices is allowed. Iediatel after the consolidation, the user ust execute the definition range operation that sets up the iniu and axiu price updating the data expectation range. In the above for there are soe checks for verifing that the data entered are congruent. If the user tries to input data which are not congruent, when he is going to save the, the essage below will appear. The user has to press Ok button and insert congruent data. 111

112 Figure For Consolidation and range definition 3.4 Data entr The facilit of Data Entr (Figure 3.4.1) allows the user to enter into the questionnaires in order to: 1. ake a qualit change; 2. set an ite ceasing; 3. correct tping error; 4. add new ites; 5. reset the last operation. There are two entering was: the first is direct, b selecting the id_questionnaire (that should be printed on paper for) in the second cobo box and pushing button naed Data Entr ; the second through the coplete list of the enterprise questionnaires b selecting the enterprise in the first cobo box and pushing the button naed Data Entr. In this case, a single for, will be opened and the user can ove fro a questionnaire to another using the record selector in the botto of the for. 112

113 Figure For Data Entr Access at the questionnaire Starting fro the for below (figure 3.4.2), for each questionnaire, the user can enter into each ite. The button Add / Upd Data allows, for the corresponding ite, to enter into the for in which he can enter prices or the no sale flag ; the button C. of Q. allows to enter into the Change of Qualit fors. Vive versa, the Reset button allows to reset the data for the working onth. When the user pushes this button the following essage will be displaed: The user can confir the reset operation pressing the button Yes. At the botto of the for there is also the button Add new series that allows the user to add a new series (ite) for the product displaed, starting fro the elaboration onth set. About the C. of Q. button: onths are qualified and the qualit change is allowed onl on the first onth, the elaboration one. To ake a qualit change on the second onth the user has to wait that the second onth becoes the first one. This liitation (in aking qualit change) avoids errors in case there are two qualified onths and the user is aking a qualit change in the first onth after entering prices on the second one about the old ite. This liitation avoids also a situation with two qualified onths, in which two different ites are present. 113

114 Figure For Data Entr Questionnaire level The for Add new series /figure ) allows the user to add a new ite (series). The icroindex base is autoaticall set up. The series code is autoaticall updated. The new ites are flagged as spare (i.e. the are not involved in the onthl index calculation procedure). To coplete the entering of the new series, the user has to coplete the for and press the button Save Data. If the operation has been correctl executed, the following essage will be displaed: 114

115 Figure For Add new series The for Data entr Prices Input (figure 3.4.4) allows the user to input the no sale flag and the out of range (O.O.R.) causes. At the botto of the for the data about the last thirteen onths are displaed. At the top, there are the series code selected in the previous for and the corresponding ite description. There are up to two boxes (one per onth) for entering the data. The cobo box about O.O.R. cause is loaded in a dnaic wa depending on the reason of out of range causes. If the price entered is out of range, the enu naed O.O.R. cause will be enabled and the user has to select the out of range cause. If the user doesn t specif the cause and tr to save the following essage will appear: To delete a wrong price entered the user has to select the no sale flag twice, so that the non sale flag is not selected anore and the price is equal to zero, then press the button Save Data to save the updating. To coe back to the previous for the user has to click on the button Back. 115

116 Figure For Data Entr Prices Input The change of qualit for (figure 3.4.5) is ade up of two sub-fors: the forer requires the user to select the kind of operation to do: 1. to correct isspelling (tpe error) in the description of the ite; 2. to ake a qualit change; 3. to reark a ceased ite. 116

117 Figure For Data Entr Change of Qualit 1/2 At the top of the second for (figure 3.4.6) the following ites are shown: the reason of qualit change selected in the previous for and the new values for the sae series code like the new ite description, the quantit and the units of easure. To save the data and to coplete the qualit change operations, the price of the new ite for the onth of replaceent and the price of the new ite for the onth before the replaceent are required. If the price of the previous onth is not available for the new ite the user as to tick out the flag price for onth before not available. 117

118 Figure For Data Entr Change of Qualit 2/2 3.5 The Indexes Calculation Panel This for (figure 3.5.1) allows the user to calculate indexes both in calculation and reference base. This for is ade up of two sections arked b a bold rectangle: the first section allows the onthl calculation and the next one concerns a utilit to help the user. Figure Index Calculation Panel 118

119 3.6 Software for BHAS The Software developed for the BHAS, allows the experts to calculate the indexes at state level as weighted averages on the Entities SIs indices. The ain panel contains the link to the for Indexes Calculation Panel and the link for printing, In the indexes Calculation Panel (figure 3.6.1) ear, onth and tpe of calculation (final or provisional) can be selected. The Entities SIs files have to be firstl iported and saved. Afterwards the are stored in the table AGGREGATED_INDEXES_REF b using the button Iport file. In order to calculate the indices, the user has to push the button Execute calculation. Figure Index Calculation Panel Once the indices are calculated tables can be printed according to the for below (figure 3.6.2). This for allows the user to select the period and the grouping of indexes to print. 119

120 Figure Dnaic Printing Indexes Fro a technical point of view, in the software there are two kind of operations to execute: onthl operations and annual ones. Each onth the Entities SIs export their indexes in a text file and send it to the BHAS. The BHAS: 1. saves these files in path C:\; 2. iports the in the database; 3. calculates indexes at state level. After updating these tables, a set of operations to provide the hierarch and weight structure has to be carried out. Each Entit SI, transits to the BHAS at the beginning of the ear two excel files containing: 1. the result of the quer EXPORT_ANNUAL_WEIGTH_FOR_AGENCY ; (BHAS has to cop and paste these data set in the tables PRODUCT_WEIGHT_RSIS, PRODUCT_WEIGHT_FIS and PRODUCT_WEIGHT _BRCKO. 120