A System of Leading Indicators for Russia

Transcription

1 26 h CIRET Conference, Taipei, Ocober 2002 Session: B2 Cyclical Indicaors for Macroeconomic Analysis - Trends, Cycles and Turning Poins A Sysem of Leading Indicaors for Russia Consrucion and Two-Years Experience of Usage Serguei V. Smirnov Absrac A sysem of leading indicaors was never compiled for pos-sovie Russia. The aricle describes he consrucion and wo-years experience of usage of he firs version of such a sysem. For daing urning poins we used a seasonally adjused indusrial producion index as a reference series. Then fory indicaors were examined, in order o decide wheher hey are or are no he leading ones. Afer esing hem agains urning poins of indusrial producion, we included seven series in our sysem of leading indicaors for Russia: [1] Effecive demand (assessmens), diffusion index (IET Surveys); [2] Socks of finished goods (assessmens), diffusion index (IET Surveys); [3] Crude oil price (Urals CIF Med); [4] Real exchange rae of he ruble; [5] Growh of inernal financial resources (assessmens), diffusion index (CEA Surveys); [6] Sock price index ( Moscow Times dollar index); [7] Real ineres rae (MIACR-overnigh). Finally we calculaed a composie leading index and a diffusion leading index. Two years experience showed ha heir behaviour as leading indicaors is quie saisfacory since January 1994 (he momen when he slump, which had been relaed o he ransformaion from a planned o a marke-oriened economy, was over). The ways for furher refining he procedure are quie obvious, oo. They are [1] consrucion of coinciden and lagging composie indices and using he coinciden index as a reference series; [2] addiion some new componens (may be wih more publishing lag) o he composie leading index. Key Words: Cyclical Indicaors, Composie and Leading Indicaors, Russia Developmen Cener, Moscow, Russia

2 26 h CIRET Conference, Taipei 2 1 Inroducion For years, no sysem of leading indicaors was compiled for pos-sovie Russia. The single and sufficien reason for his shorcoming was he coninuing indusrial recession ha looked as if would never boom ou. This recession was mainly he resul of ransformaion, or he ransiion from a planned o a markeoriened economy. I would have been odd o look for cyclical urning poins agains such a background. Moreover, he ime span of several years (no decades) was oo shor o make a disincion beween a long-erm rend and cyclical or shor-erm flucuaions. Afer 1998 financial crisis and he following economic expansion, he siuaion had changed. In he beginning of 2000, i became quie obvious ha he ransformaion had been compleed in general, and a srucural break-up was no longer a major facor in deermining economic dynamics. Long-erm rends became visible, and a cyclical reieraion of conracions and expansions became visible, oo. Therefore, we faced a se of new and inerrelaed asks, which had been impossible o solve before: o reveal expansion and conracion phases in he Russian economy afer he reforms; o precisely dae urning poins of is cycle; o consruc a sysem of leading indicaors; o calculae a composie leading index (a cerain weighed average of iniial series) and a diffusion leading index (equal o he weigh of he series wih posiive dynamics). This aricle describes in some deails he procedure of consrucion of he firs sysem of leading indicaors for Russia and sums up he experience of is wo-year regular pracical usage. 1 1 Since Sepember 2000, we have prepared and published via Inerne (see Smirnov (2000)) a monhly newsleer on he dynamics of he Composie Leading Index for Russia and all is componens. Moscow Narodny Bank (London) began o publish is purchasing managers index (PMI) for Russian manufacuring hree years before (see MNB (1997)), bu his index is no a leading one in he sense ha i doesn ge o is urning poins earlier han he economy in general does. Like all PM-indexes, he index for Russia provides a single-figure snapsho of business condiions in manufacuring wo or hree weeks earlier han official daa on indusrial oupu for he monh appear. For hose who keep in mind business cycles bu no daes of publishing, he PMI is a coinciden (no a leading) index. In April 2002, he IMF published an issue, which applies some leading indicaors for Russia (Savrev (2002)). In he meanwhile, hese calculaions have been raher occasional.

3 26 h CIRET Conference, Taipei 3 2 Key Mehodological and Saisical Issues 2.1 Choice of a Concep of Economic Cycles Consrucion of a sysem of leading indicaors may be based on wo differen conceps of economic cycle ha canno be brough ogeher. The firs deals wih he direcion of economic dynamics (going up or down, expansion or conracion), he oher deals wih is empo (going faser or slower, acceleraion or deceleraion). The firs corresponds o he classical concepion of cycles of business aciviy; he second is based on he idea of growh cycles. In he Unied Saes, urning poins are evaluaed in erms of growh and decline (in a radiion going back o he Grea Depression). Conrary o his, he OECD sysem moniors changes from periods of acceleraed growh o periods of slowing down, and vice versa (he basic idea is ha a endency owards more or less seady growh is prevailing in he world, while classical recessions marked wih decline in absolue erms are losing heir former duraion and deph). The American experience of he Grea Depression is, of course, closer o he Russian economy in is presen sae. Indeed, such quesions as a wha poin did he iniial, raher modes decline urn ino a landslide, or a wha ime did he economy ge ou of a nosedive and urn o sagnaion, are maers of ineres as such. However, now i is more han ever clear in Russia ha all he above sages are jus inner phases of he crisis. In Russia, a real (significan by inuiion) urning poin can be nohing else bu a shif from conracion o growh (or from growh o decline), no jus a slowdown or acceleraion of decline Daing of Turning Poins (Choice of a Reference Indicaor) Boh formal and informal mehods may be used for daing urning poins. An informal mehod is used in he Unied Saes. There is he Business Cycle Daing Commiee a he Naional Bureau of Economic Research, wih many leading academic scholars siing on i. The Commiee defines recession as a a period of significan decline in oal oupu, income, employmen, and rade, usually lasing from six monhs o a year, and marked by widespread conracions in many secors of he economy. The Commiee sessions decide on precise daing of a regular peak or rough in overall aciviy, upon a qualiaive analysis of all available 2 Noice ha empirical examinaion of urning poins can be done reasonably apar from heoreical and applied models, which describe he cyclical mechanism (such as waves of reiremen of fixed capial sock, Kondraiev waves, ec.). For his reason, i should be more correc o speak here no abou cycles ha imply cerain periodiciy, bu simply abou urning poins in economic siuaion.

4 26 h CIRET Conference, Taipei 4 informaion. 3 The Commiee is never oriened by he dynamics or level of any single indicaor. In paricular, defining he dae of a beginning of recession, he Commiee does no use he well-known empirical rule of wo-quarers decline in he GDP. 4 On he conrary, he OECD mehod is based on he idea of a reference series, and monhly GDP is valued as he ideal reference series. I is assumed ha a business cycle compleely reveals iself in he movemen of his indicaor, and no oher informaion is needed a all o define urning poins and phases of he cycle. However, since GDP is no compiled on a monhly basis, here is a need for some oher, proxy measure, which is moving close o he GDP. For pracical needs, he OECD uses he index of indusrial producion as a reference series. Peaks and roughs are defined in erms of his paricular indicaor, and he whole sysem of leading indicaors is consruced in correlaion wih his reference series. A he OECD, he procedure of he daing of urning poins is raher formal. I is based on comparison of acual values of he indusrial producion index wih is rend values. The poin is considered a peak when he relaion of acual value o he rends reaches a local maximum, and he poin of local minimum relaion is considered a rough. 5 Neiher he American NBER approach in is pure form (for i is hardly possible o esablish a council of auhoriaive and independen expers), nor he OECD Saisics Direcorae (because analysis of growh cycles is far from being an urgen ask) is fi for Russia. A cerain compromise is probably he bes way ou: he NBER concep of a business cycle should be applied o he OECD concep of a reference series. Since he weigh of indusry in Russian GDP is quie high (abou 25-30%), i is naural o use a seasonally adjused indusrial producion index as a reference series. 6 Local maximum poins (agains six neares monhs on boh sides) of his series may be considered peaks, and local minimum poins may be reaed as For he laes example see: NBER (2001). Indeed, wha could a commiee of economic gurus deserve if i confined is aciviy o monioring he GDP dynamics and o proclaiming ha he real GDP was declining for wo consequen quarers? The beginning of las recession in US was daed by NBER as March In he meanwhile, real GDP increased no only in he firs bu also in he second quarer of This calculaion seems o be wihin he powers of any amaeur. Well, i is no. A he OECD, hey use a special ieraive rend separaion algorihm, which is designed o handle he series wih a rend and a cyclical componen. I is impossible o be reconsruced a home. Some resricions on his approach for Russia have been revealed from wo-years experience. For deails, see Secion 4.

5 26 h CIRET Conference, Taipei 5 roughs. 7 I is also reasonable o assume ha here mus be a succession of peaks and roughs, and he ime span beween hem mus exceed half a year Selecion of Iniial Leading Indicaors In Russia, an indicaor can become a leading one in a case if, i changes he aggregae economy dynamics by shifing he demand-supply balance; i displays he expecaions of economic agens; i responds o changes in business aciviy earlier han he overall economy; i has gained recogniion as a leading indicaor in oher counries. From a pracical sandpoin, indicaors mus mee he following plain requiremens: heir flucuaions mus be cyclical (ha is, here mus be a succession of growh and decline periods); here mus be no sharp and incomprehensible jumps; he series mus be sufficienly reliable and comparable during he whole period in quesion; he informaion mus be renewed on ime (monhly and wih minimum delays agains a calendar monh). For each indicaor, which more or less mees hese requiremens, urning poins mus be found and compared wih urning poins in he economic aciviy. Noice ha no only direc bu also indirec esimaes can be used for assessmen of demand-supply balance or of expecaions. For example, demandsupply balance is direcly presened in polling liss in he daa of demand level, order books, and invenories of finished goods, and indirecly, in world oil prices, real exchange raes, curren financial siuaion of enerprises, and so on. Direc assessmens of expecaions include various survey daa, while indirec assessmens are represened in sock indices, ineres raes, numbers of newly 7 Six monhs is a cerain sandard ime span, which is required o be sure ha he observed recession (from a peak) or growh (from a rough) is serious enough o consiue a separae phase of a cycle. There should be confidence ha he nex growh (decline) is no a coninuaion of a former rend. Indeed, half of a year is nohing more han an approximae landmark.

6 26 h CIRET Conference, Taipei 6 esablished businesses, ec. The complee lis of series we have ried is given in Appendix A Calculaion of Composie Leading Index and Diffusion Index The worldwide experience shows ha a sysem consising of several leading indicaors usually displays quie an irregular picure. For his reason, here is a need o creae all-inclusive indicaors. Usually, wo indicaors are made: a composie leading index and a diffusion leading index. To unify various leading indicaors, series in a mixure of unis (billions of rubles, percens, dollars per barrel and so on) mus be given on a unified scale. For his purpose, we have used he American NBER procedure wih some modificaions. The main poin of his procedure is o equalize he differences in scale and variabiliy of iniial indicaors, and o mach he variabiliy of he composie leading index wih he variabiliy of he indusrial producion index. See a complee descripion of he calculaion procedure of he composie leading index in Appendix A3. Calculaion of a diffusion index is simple. Is value in any given momen is equal o a raio (percen) of he number of he series ha have urned for he beer in he monh in quesion o he oal number of series, which are included ino an early warning sysem. A diffusion index shows wheher growh or decline is universal, wheher i covers he overall economy or no. 3 Main Resuls for he Tuning Period 3.1. Turning poins in he Russian economic dynamics Figure 1 gives wo series: a) he iniial basic indusrial producion index (IP, NSA) calculaed from monhly chain indexes by he Federal Goskomsa 9 ; b) he seasonally adjused indusrial producion index (IP, SA). The laer indicaor is 8 9 Our paper, from he ouse, has been cenered on consrucion of a single (composie) indicaor ha should anicipae he changes in direcion of he economy. For his reason, we did no deal wih any paricular indicaor ha is coinciden or lagging a priori. The 1990 monhly daa are exrapolaed wih he index of inensiy of indusrial producion, which is calculaed by he Cenre of Economic Analyses (CEA) of he Russian Governmen.

7 26 h CIRET Conference, Taipei 7 used furher as a reference series. 10 Boh ime-series are given for he uning period (January 1990 Augus 2000). Figure 1: Turning poins for Russian indusry (he uning period) = Trough 11/92 (inermediae) 01/94 Peak 06/93 (inermediae) Peak 10/97 Trough 09/98 Trough 11/ IP, S A IP, NSA Source: Goskomsa of Russia; Developmen Cener Figure 1 also presens urning poins ha are found on formal premises, which are given in Secion 2.2. As a resul, we believe ha developmens in he Russian economic dynamics in he pas en and over years (or more precisely, during he uning period) can be divided ino he following periods: We have used he simples mehod of adjusmen: divided he iniial series values by coefficiens, which are equal o he raio of an average value of a relevan monh o he annual average. If seasonal facors are irregular, his mehod will produce incorrec resuls. For his reason, we made an evaluaion of seasonal facors using he January 1994 December 1999 daa, when seasonal facors exhibied cerain regulariy, and indusrial dynamics didn display eiher a clear upward or a clear downward rend. A shor-erm growh in December 1992 June 1993 is, mos probably, a kind of echnical correcion. Therefore, we suppose ha essenially, his period should be reaed as a random flucuaion agains a background of a longer recession.

8 26 h CIRET Conference, Taipei 8 Table 1: Business cycle reference daes for Russia (he uning period) Cycle phase Beginning End (urning poin) Duraion in monhs Indusrial producion, average monhly % change Conracion February ( November Slump February 1990 January Sagnaion February 1994 November Expansion December 1996 Ocober Conracion November 1997 Sepember Expansion Ocober ( The beginning of his conracion is no known exacly. I may have happened as early as in he firs monhs of Unil Augus 2000 Source: Developmen Cener During January 1999 Augus 2000, here were wo roughs (November 1996 and Sepember 1998) and one peak (Ocober 1997). These urning poins correspond o wo periods of conracion and wo periods of expansion. Toal duraion of conracions is 93 monhs; oal duraion of expansions is 34 monhs. 12 Time span of he firs Russian recession (a leas 82 monhs since February 1990 ill November 1996) is almos wice as long as he Grea Depression of in he Unied Saes (42 monhs), and a quarer longer han he absolue American record in he 1870s (64 monhs). The maximum conracion of indusrial oupu in Russia during his recession was 54% (exacly he same as during he Grea Depression in US). The main cause of his conracion was ransiion from a planned o a marke economy. The second Russian conracion was already marke by is naure. I was riggered by he world ( Asian ) crisis and ended in a financial collapse in Augus- Sepember The level of indusrial producion in Sepember 1998 was only 39% of January Selecion of Indicaors for he Sysem of Leading Indexes We examined fory indicaors divided ino 17 caegories (groups), in order o decide wheher hey are or are no he leading indices 13. Ou of hem, we had o rejec seven for he reason ha here was no informaion on hem excep for These figures refer o he uning period (January 1990 Augus 2000). See Appendix A1.

9 26 h CIRET Conference, Taipei 9 during a very shor period (since early 1997, or even laer), and wo ohers (he expeced growh and he expeced increase in demand, from he CEA surveys) because hey apparenly had no cyclical componens. For all he res, we defined urning poins, which were compared wih peaks and roughs of he indusrial producion index. In Appendix A2, we presen differenials (in monhs) beween urning poins of he indusrial oupu ha was fixed in he previous secion, and urning poins of all oher indicaors. On comparison of urning poins wihin each group, we chose he indicaor, which was moving ahead of he overall economic cycle in he bes possible way (we ook ino accoun pair correlaions as an addiional facor). Firs of all, le us noe ha ou of seveneen chosen groups, only eigh have he indicaors, which can be reasonably used as leading signals. I is remarkable ha all direc esimaes of any possible expecaions (here are many indicaors of his kind among survey daa) in fac, are leading nowhere. In oher words, in Russia expecaions (as far as hey are revealed by surveys), as a rule, do no come rue. A he same ime, indirec measures of expecaions (such as ineres rae and sock exchange index) have proved heir validiy as leading indicaors in acual pracice. A more careful examinaion of he chosen series shows ha dynamics of assessmens of effecive demand (IET surveys) and order books (IMEMO surveys) are quie close o each oher (no surprising since orders are one of he ways o develop demand). I should be reasonable o selec one of hese indicaors for our composie leading index. Upon some hesiaion, we decided o choose he assessmens of effecive demand. Firsly, his series has a much closer correlaion wih he indusrial producion index. Secondly, here is a lile publicaion lag for his indicaor, while he resuls of he IMEMO surveys come ino view only 40 days afer he end of each calendar monh. The laer is, honesly speaking, a purely echnical facor, bu i is criically imporan for our ask (which is o warn abou a forhcoming urn of rends as early as possible). Finally, well-imed publicaion of daa was our crierion of choice of series on such caegories as socks of finished goods and curren financial condiion. In boh caegories, we preferred he IET and CEA survey series o he indicaors of he Russian Economic Baromeer, hough he former may be no beer in oher respecs han he laer The Sysem of Leading Indicaors for Russia Finally, we included seven series in our sysem of leading indicaors for Russia (see Table 2). Their behavior as leading indicaors is quie saisfacory since January 1994 or since he momen when he slump, which had been relaed o he ransformaion from a planned o a marke-oriened economy, was over.

10 26 h CIRET Conference, Taipei 10 Table 2: The Sysem of Leading Indicaors for Russia Assessmens of effecive demand (% of normal level), diffusion index (IET) (5 Assessmens of socks of finished goods (% of normal level), diffusion index (IET) (5,6 Crude oil price (Urals brand, CIF (Medierranean), $/barrel Real exchange rae of he ruble (1996 ruble/$) Assessmens of growh in inernal financial resources of indusrial enerprises, diffusion index (CEA) Sock price index («Moscow Times» dollar index) Real ineres rae (MIACR-overnigh), % per annum (6,9 Composie leading index (CLI) Iniial poin Publicaion lag (days) (1 «Exra» urning poins (numbe r) (2 09/ T-1 P-1 01/ T-1 P-2 05/90 0 T-0 P-0 01/ (7 T-0 P-1 05/95 0 T-0 P-0 09/94 0 T-0 P-1 10/ (7 T-0 P-1 10/ (7 T-0 P-0 «Lacking» urning poins (number) (2 T-0 P-0 T-0 P-0 T-0 P-0 T-1 P-1 T-0 P-0 T-0 P-0 T-0 P-0 T-0 P-0 Average leading ime (monhs) (2,3 «Weigh» of 1% symmeric incremen in CLI (4 3 0, ,213 (8 15 0, , , , ,031 (8 4 - «T» - rough, «P» - peak. 1. Number of days from he end of a calendar monh o a publicaion of daa. The publicaion lag of he indusrial producion index is days. 2. A he January 1994 Augus 2000 period or since he sar of publicaions. 3. Calculaed from Appendix A2. A he specified average, lags of some indicaors are widely dispersed. 4. For definiion see Appendix A3. Weigh of each indicaor in he composie index was calculaed as inverse value of he sandard deviaion of symmeric incremens (for socks of finished goods and he real ineres rae, wih negaive sign). The sandard deviaion was evaluaed a he January 1995 December 1999 inerval. 5. The IET daa are recalculaed ino diffusion indices, for he sake of comparabiliy wih oher survey daa, and also in order o avoid negaive values. 6. To idenify peaks and roughs, he indicaor was aken wih a reversed sign. 7. Equal o he publicaion lag of he CPI. 8. The negaive weigh corresponds o he fac ha growh (decline) in he indicaor precedes he decline (growh) in indusrial producion. 9. To avoid negaive numbers, which make he calculaion of symmeric incremens senseless, we added a consan equal o 350 o he iniial daa. Source: Developmen Cener

11 26 h CIRET Conference, Taipei 11 Effecive demand, % of normal level (IET Surveys) is perhaps he bes of all indicaors o perform he funcion of a leading index. I is moving very closely o he indusrial producion index, bu several monhs ahead of he laer a urning poins. On he conrary, almos no direc correlaion of indusrial producion and world oil prices is acually observed. For example, in , oil prices were going up, while indusrial oupu was going down, and in 1997, he opposie was rue. On he oher hand, i s a common view ha he level of oil prices has i effec on he Russian economy in he longer run (wih greaer lags). For his reason, we did include his indicaor ino he sysem of leading indicaors The Composie Leading Index for Russia The resuls of calculaion of our composie leading index for he uning period are oulined in Figure 2. Figure 2: Index of Indusrial Producion (IP) and Composie Leading Index (CLI), January 1995 Augus / = /97 07/ /96 Trough 11/96 Peak 10/97 Trough 09/ CLI IP, SA Source: Developmen Cener

12 26 h CIRET Conference, Taipei 12 The major resul is ha he composie leading index, which we have offered, in fac goes ahead of he business cycle. 14 The value of he lead (someimes only wo monhs) is no so grea. However shorened ime horizons are very much ypical of he presen-day Russian economy. 15 And more, i is no jus a lag or a lead a urning poins ha maers here. For example, he peak of he indusrial producion index in 1997 lagged behind he peak of he composie leading index by jus wo monhs. Wihou a doub, a ha ime, in he fall of 1997, i was hardly possible o precisely dae he peak, or he urning poin o he nex phase of indusrial conracion. Bu le us look a he siuaion as i appeared, say, in April Then, he indusrial producion index of he las six monhs wen down from is peak by jus 2.8%, while he composie leading index declined by 10.6%. In oher words, dynamics of indusrial producion gave no definie sign of he forhcoming downurn, while dynamics of composie leading index showed ha he downurn was inescapable. More or less he same is rue abou he rough of Sepember A ha momen, indusrial oupu wen down, and a furher conracion could be expeced. However, he composie leading index wen up (for he second consecuive monh, afer a lenghy decline), showing ha he urn o growh was jus around he corner. Since spring 2000, coninued growh in he CLI has salled, suggesing ha high-growh phase in indusrial producion may soon come o an end. Unil auumn 2000, he dynamics of indusrial producion by iself gave no grounds for such pessimism Diffusion Leading Index for Russia A char of our diffusion leading index shows ha, since February 2000, as a rule, less han half of he leading indicaors improved every monh. A comparison of he diffusion index wih he incremen in he composie leading index expressed as a percenage shows ha boh indicaors measure approximaely he same hings. Indeed, all oher facors being equal, he more indicaors show ha he siuaion is improving, he faser he composie index grows Till May 1995, he composie leading index was consruced from six series only (CEA surveys didn exis). In he inernaional pracice, a hree-monh inerval is considered a shor erm, while in Russia, i is raher a long erm.

13 26 h CIRET Conference, Taipei 13 Figure 3: Diffusion Leading Index (DLI), January 1995 Augus Percens Source: Developmen Cener 4 Two-years Experience of Pracical Usage Almos wo years have passed since Augus 2000, and we have accumulaed sufficien empirical daa o assess he pracical applicabiliy of our composie leading index for forecasing urning poins for he Russian economy. Moreover, cerain conclusions can be drawn abou he difficulies encounered in he on-line applicaion of his insrumen. Among undeniable meris of he composie leading index, we shall menion ha i has clearly designaed wo raher shor conracions in he Russian indusrial producion in Augus 2000 December 2000 and in Augus 2001 February The firs was never deeced even by preliminary direc daa on indusrial producion compiled by Goskomsa. This conracion became visible only upon he annual revision of he Goskomsa daa in May 2001, five monhs afer i had been over. The second conracion ha sared in he second half of 2001 was also pracically invisible by iself unil December A ha ime, a decline in he composie leading index pu emphasis on he downward rend in indusrial oupu. Among he demeris of he composie leading index, we shall refer o is failure o cach a period of indusrial expansion in January Augus The value of he composie leading index kep acually unchanged, while he volume of indusrial producion expanded by 16.7% in his period.

14 26 h CIRET Conference, Taipei 14 Figure 4: Revisions of Indusrial Producion Index and Daing of Turning Poins, January 1999 May Peak 08/ = Peak 08/00 Trough 12/00 Trough 02/ CLI IP afer revision in May 2001 IP before revision in May 2001 Source: Developmen Cener On he oher hand, here is no reason a all in aemps o make he composie leading index as close as possible o he dynamics of indusrial producion. Pracical experience has already shown ha a reasonable on-line daing of urning poins on a sole base of index of indusrial producion (which is our reference indicaor) is a very difficul ask, because Goskomsa ends o make large-scale a poseriori revisions of he daa. As seen from Figure 4, annual revisions change he characerisics of indusrial oupu ime-series quie subsanially, affecing even he presence or absence of urning poins. The way ou of his dead end can be found in compilaion of a sysem of coinciden indicaors and in applicaion of he composie coinciden index as a reference indicaor. As for componens of he composie leading index, none of hem has proved o be unnecessary bu some of hem may be reasonably correced. For example, in a case we rely on an index Share of companies having socks above normal level insead of he diffusion index Socks, % of normal level, we could beer inerpre he correlaion beween changes in invenories and in volumes of indusrial producion. A seasonal adjusmen of indicaor Growh in inernal financial resources of indusrial enerprises could allow us o assess curren financial condiion of enerprises wih higher degree of cerainy.

15 26 h CIRET Conference, Taipei 15 We also believe here is a need o expand he number of componens in he composie leading index. Generally speaking, seven componens are no so few, bu in recen years, wo of hem (he real exchange rae of he ruble and he real ineres rae) were making a nex o zero conribuion o he composie leading index. This is a reflecion of economic realiy (he rae of real appreciaion of he ruble is acually quie low, and ineres rae flucuaions are negligible by hisorical measure); bu his makes he variabiliy of he diffusion leading index oo low. As a resul, pracice of applicaion of our diffusion leading index for exploraion of urning poins in economic aciviy is problemaical. 5 Conclusions In is presen form, he composie leading index for Russia allows o foresee changes in he economic aciviy, hough in a very general shape. However, is abiliy o forecas is far from being accurae enough no o call for improvemens in he echnique of is calculaion. Main direcions for furher developmen of he sysem of leading indicaors are also quie clear. In he firs place, a sysem of leading indicaors should be supplemened wih a sysem of coinciden and lagging indicaors. In fuure, a composie coinciden index will be a beer opion as a reference indicaor ha he index of indusrial producion. The index of indusrial producion, as Goskomsa calculaes i, is prone o subsanial revisions ha someimes may even change he qualiaive picure of ongoing changes (o say nohing of such deails as accurae daing of urning poins). In he second place, along wih he developmen of coinciden and lagging indexes here is good reason o upgrade he composie leading index, mosly wih inclusion of some new componens and wih a cerain refining in he old se of leading indicaors. In choosing he array of indicaors, more accuracy in he forecasing of urning poins may be a beer crierion han less publishing lag. I would be worhwhile o use he experience of Japan where imely provisional esimaion of a composie leading index is based on a par of is componens and hen is updaed as soon as he daa for oher componens arrive. Finally, all he ime-series wih a seasonal componen should be adjused for seasonal flucuaions wih a sandard and use-proved echnique (such as American X-12-ARIMA or European Tramo/Seas). We can hope ha upon his rearrangemen, our composie leading index for Russia will become a sill more valuable ool for foreseeing he coming urns in economic aciviy.

16 26 h CIRET Conference, Taipei 16 References Moscow Narodny Bank (1997): Purchasing Managers Index. Monhly Newsleer. hp://www2.mosnar.com/dynamic/pmi.asp?sec=1&crl_mode=archive Naional Bureau of Economic Research (2001): The Business-Cycle Peak of March hp:// Smirnov, Serguei (2000): Composie Leading Index for Russia. Monhly Newsleer. hp:// Savrev, Emil (2002): Growh Afer he 1998 Crisis. In: IMF. Russian Federaion: Seleced Issues and Saisical Appendix. hp://

17 26 h CIRET Conference, Taipei 17 Appendices A1 Reference Series and Poenial Leading Indicaors Indicaor (source) 1 Possible reasons for moving ahead of he overall economic cycle REFERENCE SERIES Indusrial producion: Indusrial producion index (seasonally adjused), 1996 = 100 (GKS) DEMAND AND SUPLY: direc esimaes and indicaors Measuremen of produc demand: A) Growh in effecive demand (IET) 2,3 B) Growh in demand (CEA) 2 C) Effecive demand (assessmens), % of normal level (IET) 2,3 D) Demand, % of normal level (CEA) 2 Consumer demand: A) Index of curren condiions (CSI Fund) 4 B) Number of new passenger cars regisered in Moscow ( Segodnya daily newspaper) Reference series, presumably synchronized wih he overall economic cycle. In he absence of resource consrains, changes in demand lead o growh in supply. However, his does no happen immediaely, bu wih a cerain ime lag (due o echnological ineria in producion) Revival of demand for consumer durables, for cars in paricular, leads o growh in supply of consumer goods, and laer, on all echnological feedbacks, o a overall economic revival. Orders for indusrial producs: A) REB-7: Indusry order-book level, % rising over 1-monh span (IMEMO) 2 B) REB-17: Indusry order-book level, % of normal level (IMEMO) C) Provision of enerprises wih orders, monhs (GKS) Exra orders for indusrial producs show ha demand is growing, o be followed wih increase in oupu Socks of finished producs: A) REB-8: Socks of indusrial producs, % rising over 1-monh span (IMEMO) 2 B) REB-16: Socks of indusrial producs, % of normal level (IMEMO) C) Socks, % of normal level (IET) 2,3 D) Socks, % of normal level (CEA) 2 E) Socks of finished goods, % of commodiy resources (GKS) Growing socks of producers goods show ha sales are difficul. In his siuaion, producers should adjus heir oupu downward. Hence, here is an inverse relaion wih oupu: growing socks of finished goods precede a decline in oupu.

18 26 h CIRET Conference, Taipei 18 DEMAND AND SUPPLY: indirec esimaes and indicaors Crude oil prices: Urals brand (CIF, Medierranean), $/barrel (Reuers) Large-scale inflow of oil dollars leads o higher effecive demand of governmen, producers and consumers (along wih rising wages). On he conrary, declining earnings from expors leads o lower effecive demand. Real exchange rae of he ruble: Official exchange rae of he ruble in 1996 prices (1996 ruble/$). (CBR, GKS) 5 Appreciaion of he ruble has an adverse effec on expor performance and increases impor compeiion. On he conrary, depreciaion of he ruble encourages domesic producers and increases domesic demand for home producs. Money supply: A) Money sock M 2, billion rubles a 1996 prices (CBR) 5 B) Reserve money, billion rubles a 1996 prices (CBR) 5 Increased (o a cerain exen) supply of money encourages producer and consumer demand. Curren financial condiion: A) REB-19: Share of enerprises in good or normal financial condiion, % (IMEMO) B) Profis growh (CEA) 2 C) Growh in inernal financial resources of enerprises (CEA) 2 D) Moneary resources of indusrial enerprises, billion rubles a 1996 prices (GKS) 6 E) Deposis of enerprises (nominaed in rubles), billion rubles a 1996 prices (CBR) 6 F) Non-cash money (M 2 -M 0 ), billion rubles a 1996 prices (CBR) 6 G) Profis, % of oupu (indusry), (GKS) H) General assessmen of economic siuaion (CEA) 2 Financial well-being leads o exra demand for indusrial producs. Improvemen of financial condiion can precede he general economic revival (for example, due o increases in creaion of money by moneary auhoriies).

19 26 h CIRET Conference, Taipei 19 EXPECTATIONS: direc esimaes and indicaors Consumers expecaions: Index of consumers expecaions (CSI Fund) 4 Displays consumers expecaions. If hey are followed wih acual changes in consumer demand, furher adjusmen of oupu is ineviable. Expeced producion growh: A) REB-26: Anicipaed growh in oupu (over 3-monhs spans), (IMEMO) 2 B) Anicipaed growh (over 2-3-monhs spans) (IET) 2,3 C) Anicipaed growh (CEA) 2 Displays producers expecaions. This indicaor, so o say, mus lead acual changes in oupu volume by definiion. Expeced change in financial condiion: A) REB-28: Anicipaed improvemen in financial siuaion (over 3 monhs spans), indusry (IMEMO) 2 B) Anicipaed increase in profis (CEA) 2 C) Anicipaed growh in inernal financial resources (CEA) 2 Displays producers expecaions for macroeconomic changes. Esimae of anicipaed demand: A) Anicipaed change in demand (over 2-3- monhs spans) (IET) 2,3 B) Anicipaed growh in demand (CEA) 2 Displays producers expecaions for poenial sales. EXPECTAIONS: indirec esimaes and indicaors Sock price index: «Moscow Times» dollar index (Reuers) 7 Displays invesors (chiefly foreign) expecaions of macroeconomic changes. Moreover, leaps in marke capializaion of companies may provoke adjusmens of invesors decision, which in urn, lead o changes in volumes of oupu. Foreign exchange reserves: Foreign exchange reserves a he Bank of Russia (gold excluded), $ billion (CBR) A sharp decline in foreign exchange reserves, backed by negaive expecaions of foreign invesors, can lead a decline in he real secor, for financial markes have less ineria.

20 26 h CIRET Conference, Taipei 20 Bank lending o he real secor: Claims of credi insiuions on enerprises and individuals, billion rubles a 1996 prices (CBR) 6 Volume of lending o he real secor and households displays expecaions of commercial banks for macroeconomic changes. Moreover, banks lending for invesmen programs leads o oupu growh. New sar-ups: Newly esablished enerprises regisered, by 1000 a he Unified Public Enerprise Regiser (GKS) Increase in he number of sar-ups for covering fuure demand display anicipaions of enrepreneurs. Level of real ineres raes: MIACR-overnigh real rae, % per annum (CBR) 8 Rising ineres raes show ha risk levels in he economy are growing, declining raes show ha hey are sabilizing. Decisions o increase oupu volumes can be expeced in he laer case raher han in he former case. Besides, high ineres raes hold banks lending back, also hampering expansion of oupu. Hence, here is an inverse relaion wih oupu: rising real ineres raes lead o decline in oupu. 1. Full names of sources: CBR he Cenral Bank of Russia; GKS Goskomsa (he Cenral Saisical Commiee of he Russian Federaion); CEA Cenre for Economic Analysis under he Russian Governmen; IMEMO Insiue of World Economy and Inernaional Relaions, Russian Academy of Sciences (he Russian Economic Baromeer Bullein); IET Insiue for he Economy in Transiion (he Russian Bullein of Conjuncure Surveys. Indusry); CSI Fund Consumer Senimen Index Fund. 2. Diffusion Index. 3. Iniial dae are given by IET in a forma of balance indicaors (B, a balance of posiive and negaive responds, in % of he oal number of respondens). We have recalculaed hem in he forma of diffusion indexes (D, a sum of posiive and a half of neural responds, in % of he oal number of respondens) by D = 0.5 (100+B) formula. This procedure increases comparabiliy wih oher surveys, bu wha is more, i allows avoiding negaive numbers, which lead o considerable echnical difficulies a saisical handling of ime series. 4. Since March Index of consumer senimen and he wo of is componens (Index of curren condiions and Index of consumer expecaions) are published bimonhly. For our calculaions, we obained he missing monhly daa wih linear inerpolaion. Clearly, definiion of urning poins requires being especially accurae in his case. 5. Deflaed by consumer price index (1996=100). 6. Deflaed by producer price index (1996=100). 7. Recalculaed from he ruble index wih division by he official exchange rae of he dollar. 8. Nominal rae, less he monhly incremen in consumer prices, annualized.

21 26 h CIRET Conference, Taipei 21 A2 Leading over (+) or lagging behind (-) he index of indusrial producion (January 1990 Augus 2000) Indicaors and sources 2 Iniial T 4 poin 3 11/92 P 4 06/93 T 5 - P 5 - T 11/96 P 10/97 Growh in effecive demand 06/95 n.a. n.a. n.a. X (IET) Effecive demand 09/92-6 X X (assessmens), diffusion index (IET) Demand, % of normal level 05/95 n.a. n.a. n.a. X (CEA) Consumer senimen index, 01/93 n.a. - - X curren condiions (CSI Fund) REB-7: Indusry order-book 01/ X X level (percen rising over 1- monh spans), diffusion index (IMEMO) REB-17: Indusry orderbook 01/93 n.a. 4 X X level, % of normal level (IMEMO) REB-8: Socks of finished 01/ X X indusrial producs (percen rising over 1-monh spans), diffusion index (IMEMO) 6 REB-16: Socks of finished 01/93 n.a. - X X indusrial producs, % of normal level (IMEMO) 6 Socks of finished goods 09/92 n.a. 2 X X (assessmens), diffusion index (IET) 6 Socks, % of normal level 05/95 n.a. n.a. n.a. n.a (CEA) 6 Crude oil price, Urals brand 05/ (CIF, Medierranean), $/barrel Real exchange rae of he 01/ ruble (1996 ruble/$) Money sock M 2, a /94 n.a. n.a. n.a prices Reserve money, a /95 n.a prices REB-19: Share of indusrial 01/93 n.a. 3 X X enerprises in good or normal financial condiion (IMEMO) Growh in inernal financial resources of enerprises 05/95 n.a. n.a. n.a T 09/98

22 26 h CIRET Conference, Taipei 22 (assessmens), diffusion index (CEA) Moneary resources of 01/ indusrial enerprises, a 1996 prices Deposis of enerprises (in 12/94 n.a. n.a. n.a. n.a rubles), a 1996 prices Non-cash money (M 2 - M 0 ), 12/94 n.a. n.a. n.a. n.a a 1996 prices Profis (less losses), % of 12/ X X oupu General assessmen of 05/95 n.a. n.a. n.a. n.a economic siuaion (CEA) Consumer senimen index, 01/93 n.a expecaions (CSI Fund) REB-26: Anicipaed 04/ X X indusrial oupu (percen rising over 3-monhs spans) (IMEMO), diffusion index Anicipaed growh (in a /92 n.a. 5 X X monhs span) (IET) REB-28: Anicipaed 02/93 n.a. 4 X X improvemen in financial siuaion, indusry (percen improving over 3-monhs spans), diffusion index (IMEMO) Anicipaed growh in 05/95 n.a. n.a. n.a. n.a inernal financial resources (CEA) Anicipaed change in demand (IET) 10/95 n.a. n.a. n.a. n.a Sock price index («Moscow 09/94 n.a. n.a. X X Times» dollar index) Foreign exchange reserves 03/96 n.a. n.a. n.a. n.a a he Bank of Russia (gold excluded), $ billion. Claims of credi insiuions 06/95 n.a. n.a. n.a. n.a on enerprises and individuals, a 1996 prices Real ineres rae (MIACRovernigh), % per annum 6 10/94 n.a. n.a. n.a «T» - rough, «P» peak of he indusrial producion index; n.a. non-available; X «false» or «exra» peak or rough; «-» - a urning poin missing. 1. Agains peaks and roughs of indusrial producion index. 2. Whole names of series and sources are given in he noes o he Appendix A1. 3. The momen when regular monhly daa became available. 4. Inermediae urning poins of indusrial producion index for which some oher indicaors have corresponding urning poins. 5. In he second half of 1994 firs half of 1995, a range of indicaors had exra urning poins (a rough and a peak) which had no corresponding urning poins of he indusrial producion index. 6.Before daing of urning poins, he indicaor is aken wih a reversed sign. Source: Developmen Cener

23 26 h CIRET Conference, Taipei 23 A3 Calculaion of he Composie Leading Index 1. Symmerical monhly incremens x i ( presen insan of ime) are i calculaed for all seleced leading indicaors X (i - he series number): x i i ( X - X = 200 i ( X + X ) ) i -1 i 1 Symmerical incremens (as disinc from regular incremens) have a characerisic qualiy ha equal percen changes in opposie direcions, observed in wo consequen insans of ime, in sum, bring abou he iniial level of he i indicaor X. i 2. Average values x av and sandard deviaions s i of he obained incremen series are calculaed (n - he number of monhs in he basic period): x i av = å x n i s i å i ( x x = ( n 1) i av ) 3. For each, he averaged incremen g is calculaed, as well as i average and sandard deviaion (m - he number of iniial indicaors) 16: g g av = = å i i ( x / s ) m å n g s g å ( g g = n 1 av ) 4. Seps 1-2 are performed for indusrial producion indexes Y. The resuls y are he average ( y av ) and he sandard deviaion ( s ) of he incremen series. 16 In combining sandardised series, differences in heir lag srucure are no aken ino accoun.

24 26 h CIRET Conference, Taipei 24 5.The g indicaor is adjused so ha is volailiy should be equal o he volailiy of incremens in he indusrial producion index. G = g s s y g 6. Values of he composie leading index Z are calculaed by a recursion formula (going back from incremens o he aggregae): Z Z 1 (200 + G1) = (200 G ) = Z 1 1 (200 + G ) (200 G ) 7. The obained index Z is adjused o he base of indusrial producion index (1996 = 100). To do so, we divide all values of Z by he average monhly level of 1996, and muliplied by 100. As a resul, he composie leading index has he common base wih he indusrial producion index, and symmerical incremens of he wo indicaors have he same volailiy.