Constructing and updating a cost-of-living index for the purpose of ensuring purchasing power parity of the remuneration of expatriate officials

Transcription

1 Int. Statstcal Inst.: Proc. 58th World Statstcal Congress, 2011, Dubln (Sesson CPS062) p.5485 Constructng and updatng a cost-of-lvng ndex for the purpose of ensurng purchasng power party of the remuneraton of expatrate offcals Dr. Yansaneh, Ibrahm S. Internatonal Cvl Servce Commsson, Unted Natons 2 Unted Natons Plaza, 10 th Floor New York, NY 10017, USA E-mal: yansaneh@un.org Mr. Pagan, Roberto Internatonal Cvl Servce Commsson, Unted Natons 2 Unted Natons Plaza, 10 th Floor New York, NY 10017, USA E-mal: pagan@un.org Introducton The post adjustment system (PAS) refers to the set of rules and procedures governng the remuneraton of offcals of the Unted Natons (UN) system. Its overall goal s to equalze the purchasng power of the remuneraton of UN offcals servng n varous locatons around the world wth the salares of ther counterparts n New York, whch s the base of the system. It s desgned to reflect the nternatonal character of the UN staff populaton and to be robust enough to be applcable to 200-odd duty statons wth wdely varyng levels of general economc development, stablty of such economc ndcators as nflaton and exchange rates of local currences relatve to the US Dollar; as well as the number, composton, and turnover of staff. Duty statons that are smlar to the base of the system n terms of soco-economc development, typcally n North Amerca and Europe, are classfed as Group I duty statons. Those n the rest of the world, are classfed as Group II duty statons. (See the dstrbuton of the UN staff below). Dstrbuton of target staff populaton by regon and type of duty staton As of 31 December 2009 Type of duty staton Regon Group I Group II Total Afrca - 7,193 7,193 Asa 112 3,682 3,794 Australa and Oceana Europe 11, ,525 North Amerca 5, ,663 South Amerca and the Carbbean - 1,344 1,344 Total 16,812 13,023 29,835

2 Int. Statstcal Inst.: Proc. 58th World Statstcal Congress, 2011, Dubln (Sesson CPS062) p.5486 The Internatonal Cvl Servce Commsson (ICSC), an ndependent expert body establshed by the General Assembly of the Unted Natons, s responsble for the operaton of the PAS, wth statstcal/techncal assstance provded by the Advsory Commttee on Post Adjustment Questons (ACPAQ), an expert panel comprsed of experenced prce statstcans. The post adjustment ndex The post adjustment ndex (PAI) for a duty staton reflects a smultaneous spatal and a temporal comparson of cost of lvng between the duty staton and New York. It conssts of the followng components: In-area, excludng housng - relates to lvng costs ncurred locally. It s structured accordng to the Classfcaton of Indvdual Consumpton Accordng to Purpose (COICOP) and the cost-of-lvng relatvtes of ts 84 basc headngs are calculated on the bass of prce data for a basket of around 300 goods and servces collected at the duty staton and n New York; Housng - relates to rent and other housng-related costs, ncludng costs for mantenance, utltes, and other housng costs. Ths nformaton s obtaned mostly va a housng questonnare. Penson Contrbuton - relates to the amount of penson contrbuton pad by staff, obtaned from admnstratve sources. Medcal Insurance - relates to the amount of nsurance premums pad by staff, obtaned from admnstratve sources; and Out-of-Area - relates to expendtures outsde the country of the duty staton. Whle the weght of ths component s determned from out-of-area expendtures reported by survey respondents, the change n cost of lvng for ths component s approxmated by an out-of-area ndex, calculated as a weghted arthmetc average of US dollarzed general CPIs from 26 selected countres. At any pont n tme, the PAI s calculated as an adjusted, modfed-walsh ndex, gven by the followng formula: where PAI A RF P h RF H h C PC C I MI I O OA O m 1 v A, H, PC, MI and OA denote the relatve weghts of the n-area (excludng housng), housng, penson contrbuton, medcal nsurance and out-of-area components; m denotes the number of basc headngs n the n-area (excludng housng) component; P and v denote, respectvely, the prce rato and proporton of expendtures of the -th basc headng n the n-area (excludng housng) component; RF s the general rebasng factor, whch converts nformaton about the cost-of-lvng ndex nto nformaton about salary requred to mantan purchasng power party wth New York. RF s an adjustment of RF for basc headngs contanng electronc and hgh-technology tems (a more costeffectve alternatve to the hedonc technques); h and h 0 denote, respectvely, housng-related expendtures at the locaton and n New York;

3 Int. Statstcal Inst.: Proc. 58th World Statstcal Congress, 2011, Dubln (Sesson CPS062) p.5487 C and C 0 denote, respectvely, staff members average net contrbuton to the UN Jont Penson Fund at the duty staton and n New York; I and I 0 denote, respectvely, staff members average net contrbuton to the medcal plans avalable at the locaton and n New York; and O and O 0 denote, respectvely, the out-of-area ndex at tmes and 0, when the cost-of-lvng surveys were conducted at a specfc locaton and n New York respectvely. The out-of-area ndex s common to all locatons, ncludng New York, as t measures the change over tme of expendtures for purchasng a varable volume of US Dollars n a basket of 26 markets, wth dfferent currences and nflaton condtons, n countres around the world where most of the out-of-area expendtures of UN staff members are ncurred. The weghts of each of the components (A, H, PC, MI and OA) are determned for a duty staton at the tme of a cost-of-lvng survey. They are expressed n US Dollar nomnal expendtures for an average staff member at the duty staton. Such expendtures are known n advance for PC, whereas for the other components t s determned at the tme of the survey by usng the results of the household and housng expendture nformaton collected from the staff. MI s calculated as the expendture ncurred by the average staff member at the duty staton n terms of the net premums applcable to the medcal plans avalable at the duty staton. For Group II duty statons, OA s derved from the central value of the band contanng the actual percentage out-of-area expendture, whereas for Group I duty statons, OA s an admnstratvely specfed value. H, the weght for housng, s computed from the average net expendture for rent, mantenance, utltes and other housng expenses. A, the weght for the n-area (excludng housng) component, s obtaned by subtractng all the other weghts from the net salary of the average staff member. Although varable from duty staton to duty staton, the above components have the followng approxmate weghts, as a system-wde average across all duty statons: Out-of-Area - 33%; In-area (excludng Housng) - 31%; Housng - 22%; Penson Contrbuton - 10%; and Medcal Insurance - 4%. A set of common expendture weghts s used for the aggregaton of the basc headngs belongng to the n-area (excludng housng) component accordng to a weghted geometrc average. Common expendture weghts are the same for all duty statons, and were ncorporated nto the PAS methodology as a soluton to the problems of nstablty over tme and low relablty of expendture weghts when estmated for small duty statons, often subject to hgh turnover of staff and wth very heterogeneous expendture patterns. The n-area (excludng housng) ndex s smlar to a typcal CPI n terms of the aggregaton process, the use of the same nternatonal classfcaton and n practces of data collecton. However, the most obvous dfference s that, whle a CPI measures change over dfferent ponts n tme n the level of prces of a common basket of goods and servces, the n-area (excludng housng) ndex measures such change over dfferent ponts n tme and space smultaneously and, of course, does that wth reference to a target populaton dfferent from that of the CPI for a country. The PAI s thus calculated n accordance wth the methodology as broadly descrbed above. Other agences have more or less smlar approaches n determnng the ndces for ther own salary adjustment. For example, Eurostat (the statstcal offce of European Commsson), uses correcton coeffcents as a bass for the remuneraton of offcals of the European Unon. These coeffcents are calculated accordng to a Fsher formula, whch can decrease the reference base salary pad n Brussels, the base cty for European Unon comparsons. Statstcs Canada calculates the Post Allowance to

4 Int. Statstcal Inst.: Proc. 58th World Statstcal Congress, 2011, Dubln (Sesson CPS062) p.5488 adjust salares of Canada s dplomatc and other staff servng around the world. In the Unted States a Cost-of-Lvng Allowance (COLA) s calculated for ths purpose. In addton to governmental and nter-governmental agences, prvate sector companes, such as Mercer, operate n the feld of cost-oflvng measurement for the purpose of salary adjustment of offcals servng at dfferent locatons. Ensurng purchasng power party of remuneraton Cost-of-lvng surveys are conducted regularly to establsh the PAI for each of the duty statons. The surveys comprse: a prce survey to collect data on prces of a basket of about 300 goods and servces (n some cases, for example for the tem car, 19 possble dfferent brands and car models are consdered n the basket, brngng the number of surveyed tems well over 300); and expendture surveys va self-admnstered questonnares, to collect nformaton about household and housng expendtures drectly from UN staff, used to determne the weghts appled n the aggregaton of the varous components of the PAI. Whle most of the organzatonal and logstcal aspects of the surveys are centrally managed by the ICSC, some aspects of the data collecton and logstcal admnstraton of the surveys s delegated to local UN offces. The result of a cost-of-lvng survey s an ndex, the PAI, used to establsh a benchmark level of salary at the duty staton, ensurng purchasng power party wth the level of salary n New York, at a gven pont n tme. Snce these surveys are conducted wth vared frequences, dependng on the type and crcumstances of each duty staton, the PAI needs to be updated on a monthly bass between surveys so that purchasng power party s mantaned over tme. Thus, each month, exchange rates of the local currency aganst the US Dollar, as well as updated local, dsaggregated CPI seres, are used to approxmate changes n the relatvtes of the cost of lvng at the duty staton wth New York. Snce both local nflaton and currency exchange rates have an mpact on the updatng of the ndex, ther movements may renforce, but also offset, each other. However, snce some ndex components, notably the penson contrbuton, the medcal nsurance and the out-of-area, do not depend on local condtons, changes for these components are less frequent and are accounted for by approprate ndcators derved from other non-local sources of nformaton. The establshed updatng mechansms are so senstve to evolvng economc crcumstances that very often new PAI benchmark results derved exclusvely from the survey process are very close to the PAI resultng from the updatng mechansms. Although the ndex s updated on a monthly bass, not all of ts updates translate nto monetary changes of the post adjustment. Derved from the PAI s the post adjustment multpler, whch represents a number of percentage ponts of the common base salary. The multpler also called the post adjustment classfcaton (PAC), changes accordng to operatonal rules that add admnstratve consderatons and objectves, chefly the stablty of net remuneraton, to the ndex. Only a pure measurement of an ncrease n the cost-of-lvng, ether through a survey or establshed updatng mechansms, may trgger a change n remuneraton. For Group I duty statons, the PAC s adjusted every month to account for exchange-rate movements n order to keep the amount of salary n local currency wthn a reference band. Furthermore, t s adjusted once a year to possbly ncrease salares, and to reset the reference band, by compensatng for local nflaton and ncreases n cost-of-lvng accountable to other non-local components. Group II duty statons, genercally referred to as feld duty

5 Int. Statstcal Inst.: Proc. 58th World Statstcal Congress, 2011, Dubln (Sesson CPS062) p.5489 statons, have ther PACs updated every four months, to account for exchange-rate movements, local nflaton, and other ndcators referrng to the non-local components. Another set of rules, desgned to protect net salares aganst abrupt and sgnfcant changes to local nflaton and exchange rates, becomes applcable, whenever necessary, untl the economc condtons permt the measurement of the cost of lvng by the approprate ndcators. Concludng remarks The PAS and ts methodology are subject to a contnuous revew and are subject to change n accordance wth a 5-year cycle of survey rounds. In the most recent past, solutons were found to dffcult problems, such as that of rapdly changng specfcatons and qualty of electronc and hghtechnology tems, whch now are treated accordng to the real-tme prce-comparson approach. However, several other problems emerged n varous areas, from data collecton and processng. For example, the comparablty of tems and outlets, the detecton and treatment of outler prces n a stuaton where sometmes only too few prce quotatons are avalable. As wth the methodology underlyng cost-of-lvng, the applcaton of the PAI n the formulaton of the salary-settng polcy of the UN s subject to the normal trade-off between precson of the ndex and stablty of remuneraton. Whle the ICSC understands the mplcatons of these two opposng factors, and provde statstcal support and clarfcatons about them, the decson regardng the optmal balancng between the two s non-statstcal n nature and ultmately rests wth the UN General Assembly. The ICSC engages n cooperatve relatonshps wth other agences nvolved n cost-of-lvng measurement for remuneraton of expatrate offcals. For nstance, Eurostat and the Inter-Organsatons Study Secton on Salares and Prces of the OECD (IOS), operate smlar systems for ther offcals. Dfferences n the overall compensaton packages as well as target consttuences, dctate necessary dfferences n methodologes. However, there s certanly fertle and establshed ground for cooperaton among these nsttutons, especally n the area of exchange of statstcal nformaton (statstcal data and methodology, etc.). Under an ongong memorandum of understandng wth Eurostat and IOS, ICSC uses European Comparsons Programme prce data for valdaton, whereas Eurostat uses ICSC prce data for ther Extra-EU duty statons. Furthermore, IOS produces the rent data used by the ICSC n the calculaton of rent ndces for Group I duty statons under preferental fnancal arrangements. All three agences cooperate n methodologcal work and harmonzaton of survey nstruments. The Internatonal Comparson Programme, managed by the World Bank, also shares several aspects of methodologcal smlarty and mght offer further possbltes for data exchange and postve cross-fertlzaton of deas and methodologes n the future. The work of ICSC n estmatng dfferentals of cost of lvng for nternatonal offcals s prmarly addressed to the consttuent organzatons of the UN system. However, there s a plethora of other nsttutons, ncludng the dplomatc servces of several countres, that use ICSC ndces for ther own compensaton purposes. In some cases, these nsttutons have adopted wholesale the UN salary structure and cost-of-lvng adjustment system. In other cases they have customzed the PAS to ther needs. It s especally for ths second group that we advocate the use of our Retal Prce Indces (RPIs), another mechansm for cost-of-lvng comparsons based on ICSC data. The RPIs are calculated on the

6 Int. Statstcal Inst.: Proc. 58th World Statstcal Congress, 2011, Dubln (Sesson CPS062) p.5490 bass of the same raw data used to compute the PAIs, but n a slghtly dfferent fashon: whle the PAI compares the duty statons to New York at a fxed pont n tme, the RPI compares the duty statons to New York at the same pont n tme. Moreover, RPIs may be customzed n terms of components, weghts and base cty for cost-of-lvng comparsons. They are publshed b-monthly by ICSC wth an Excel tool that enables users to customze the system of weghts used to aggregate the major components of the ndex accordng to ther needs and crcumstances. The dssemnaton of the PAIs and of the RPIs s publc and for free through ICSC s web ste Of course, only aggregated data s dssemnated by ICSC, whle ndvdual prce and expendture data collected from cost-of-lvng surveys are protected by confdentalty rules n accordance wth the fundamental prncples of offcal statstcs. Whle the PAS s regularly revewed and ts methodology adjusted n response to changes n the realtes of the nternatonal communty, thoughts must be addressed also to long-term strategc ssues that have the potental of ntroducng radcal changes n current operatng procedures. One such ssue s a wder applcaton of web-based surveys. As a matter of fact, n ts most recent cost-of-lvng surveys at headquarters duty statons (duty statons hostng headquarters of at least one UN organzaton), the use of well-desgned web-based surveys boosted staff partcpaton n ts surveys, tradtonally very low n large headquarters duty statons, to new record levels of response rates rangng from about 50% to 90%, yeldng approxmately ten tmes as much useable data as obtaned from the last surveys fve years ago. Another strategc ssue s achevng a much hgher ntegraton wth the methodologes for the adjustment of salares of other nternatonal nsttutons. Work on ths aspect s underway wth Eurostat. REFERENCES ILO/IMF/OECD/UNECE/Eurostat/The World Bank. Consumer prce ndex manual: Theory and practce, Geneva, Internatonal Labour Offce, Internatonal Cvl Servce Commsson. The Post Adjustment System - What t s, how t works. New York, ICSC, ABSTRACT The paper presents an overvew of the methodology and procedures for creatng the post adjustment ndex, whch s desgned to equalze the purchasng power of the remuneraton of offcals of the Unted Natons system, servng around the world, n countres wth wdely varyng economc condtons, target populaton szes and compostons; and levels of development. The major components of the ndex, and procedures for measurng them, are descrbed, ncludng the use of surveys of expendtures admnstered to expatrate offcals, and of prces of a representatve basket of goods and servces. Procedures for updatng the ndex over tme, takng nto account nflaton and exchange-rate fluctuatons, are also presented. Problems encountered n constructng the ndex, and proposed solutons, are also dscussed, ncludng adjustments for changes n qualty over tme. KEYWORDS: cost-of-lvng ndex, consumer prce ndex, expendture surveys, prce surveys