Forecasting Macedonian Business Cycle Turning Points Using Qual Var Model

Pe o ska, Magdalena; K s e ska, Ane a; Naumo ski, Nikola A icle Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model Jou nal o Cen al Banking Theo y and P ac ice P o ided in Coope a ion wi h: Cen al Bank o Mon eneg o, Podgo ica Sugges ed Ci a ion: Pe o ska, Magdalena; K s e ska, Ane a; Naumo ski, Nikola (2016) : Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model, Jou nal o Cen al Banking Theo y and P ac ice, ISSN 2336-9205, De G uy e Open, Wa saw, Vol. 5, Iss. 3, pp. 61-78, h ps://doi.o g/10.1515/jcb p-2016-0020 This Ve sion is a ailable a : h ps://hdl.handle.ne /10419/217602 S anda d-Nu zungsbedingungen: Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen Zwecken und zum P i a geb auch gespeiche und kopie we den. Sie dü en die Dokumen e nich ü ö en liche ode komme zielle Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich machen, e eiben ode ande wei ig nu zen. So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen (insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en, gel en abweichend on diesen Nu zungsbedingungen die in de do genann en Lizenz gewäh en Nu zungs ech e. Te ms o use: Documen s in EconS o may be sa ed and copied o you pe sonal and schola ly pu poses. You a e no o copy documen s o public o comme cial pu poses, o exhibi he documen s publicly, o make hem publicly a ailable on he in e ne , o o dis ibu e o o he wise use he documen s in public. I he documen s ha e been made a ailable unde an Open Con en Licence (especially C ea i e Commons Licences), you may exe cise u he usage igh s as speci ied in he indica ed licence. h ps://c ea i ecommons.o g/licenses/by-nc-nd/3.0/ Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model 61 * Na ional Bank o he Republic o Macedonia E-mail: Pe o skaM@nb m.mk ** Na ional Bank o he Republic o Macedonia E-mail: K s e skaA@nb m.mk *** Na ional Bank o he Republic o Macedonia E-mail: Naumo skiN@nb m.mk Jou nal o Cen al Banking Theo y and P ac ice, 2016, 3, pp. 61-78 Recei ed: 6 Ap il 2016; accep ed: 25 July 2016 UDK: 330.1:65.012.511(497.7) DOI: 10.1515/jcb p-2016-0020 Magdalena Pe o ska * Ane a K s e ska ** Nikola Naumo ski *** Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model1 Abs ac : This pape aims a assessing he use ulness o leading in- dica o s in business cycle esea ch and o ecas . Ini ially we es he p edic i e powe o he economic sen imen indica o (ESI) wi hin a s a ic p obi model as a leading indica o , commonly pe cei ed o be able o p o ide a eliable summa y o he cu en economic condi ions. We u he p oceed analyzing how well an ex ended se o indica o s pe o ms in o ecas ing u ning poin s o he Macedo- nian business cycle by employing he Qual VAR app oach o Dueke (2005). In con inua ion, we e alua e he quali y o he selec ed indi- ca o s in pseudo-ou -o -sample con ex . The esul s show ha he use o su ey-based indica o s as a complemen o mac oeconomic da a wo k sa is ac o y well in cap u ing he business cycle de elop- men s in Macedonia. JEL classi ica ions: F42, C25, C22 Keywo ds: Fo ecas ing, Business cycle u ning poin s, Qual VAR, MCMC, La en a iable 1 The esponsibili y o his pape lies solely wi h he indi idual au- ho s. The iews exp essed he ein do no necessa ily ep esen hose o he Na ional Bank o he Republic o Macedonia. Jou nal o Cen al Banking Theo y and P ac ice 62 1. In oduc ion Economic analys s ha e been making g ea e o s o de i e a mix o mac oe- conomic indica o s ha could summa ize and p edic business cycle pa e ns. Mo eo e , he p edic ion o business cycle phases is e y challenging gi en ha luc ua ions in mac oeconomic condi ions a e ne e di ec ly obse able. A good comp ehension o business cycle ac s is essen ial o mac oeconomic decision- making, especially in mone a y and iscal policies. The GDP cycle was ex ac ed using a double Hod ick-P esco (HP) il e . This pa icula app oach wo ks o- wa ds achie ing g ea e u ning poin s signal s abili y. Fu he mo e, gi en ha business cycle o ecas ing is he c ea ion o pos ula es abou how a business cycle will e ol e in he u u e, analys s use a a ie y o economic indica o s o o m he p edic ions. Fo ins ance, he so called leading indica o s a e a iables ha ha e a endency o change ahead o he economy as a whole, so hey a e equen ly used in business cycle o ecas ing. Following his p ac ice, we conside ed employing he Eu opean Commission’s Economic sen imen indica o (ESI) o Macedonia in ou analysis. Namely, he ESI as a single a iable e lec ing economic agen s’ sen imen in eg a es mos o he in o ma ion embedded in mul iple indica o s. We examined he leading p ope y o he ESI ela i e o GDP i s ly by using a s a ic p obi model. In his ega d, we ind ha he cyclical mo emen o he ESI is help ul in moni o ing and p edic ing he u ning poin s o GDP. A la ge and g owing body o esea ch (like o ins ance, on he case o U.S. eces- sions, he wo k by Kauppi and Saikkonen (2008), Nybe g (2010), and Ng (2012), among o he s), has demons a ed ha including dynamic elemen s (e.g., lags o he bina y esponse a iable) in he p obi models can p oduce mo e p ecise o e- cas s o ecessions han s anda d p obi eg essions. Consequen ly, we ocus on he Qual VAR me hod o Dueke (2005). One imp o emen ha Dueke con ib- u es o business cycle u ning poin s o ecas ing li e a u e is ha his au ho ’s me hod includes a la en a iable unde lying he ecession pe iods along wi h o he indica o s in a VAR se ing. To his end, good o ecas ing indica o s should closely ma ch he ola ile economic de elopmen . Ou Qual VAR model showed ha bo h g ow h a es o eal GDP, as mos na u al choice, and he ESI alone a e good candida e indica o s, howe e , hey do no qui e well e lec he ecen Macedonian business cycle his o y. When i comes o he ESI o he Macedonian economy, we would like o poin ou ha i is qui e newly publicly a ailable indica o and, o his end, we belie e ha his pa- pe could con ibu e owa ds enligh ening i s o ecas ing powe in business cycle analysis. The e o e, we p oceeded wi h addi ional expe imen s ha combine di - e en inancial and mac oeconomic ac i i y a iables wi h he abo e selec ion Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model 63 o da a. A e g ea many such simula ions, we concluded ha ESI along wi h he eal GDP and he capaci y u iliza ion a e in manu ac u ing (which ypically climbs when he economy is ib an and alls when demand so ens), can ac ually smoo hly p edic he Macedonian business cycle luc ua ions. In addi ion, o ou knowledge, his wo k is be ween he i s pape s on his subjec o he Macedo- nian economy (aside om business cycle synch oniza ion analysis). This pape is o ganized as ollows. The nex chap e p o ides li e a u e o e iew, while chap e 3 e alua es he ESI use ulness in assessing he cu en s a e o he Macedonian economy. Chap e 4 explains he Qual VAR app oach ha is used and chap e 5 p o ides he esul s, ollowed by he concluding pa . 2. Li e a u e e iew The complica ed na u e o business cycle modelling has caused many di icul- ies in mode n mac oeconomics. Du ing he las decades, esea che s ha e kep on sea ching o he mos accu a e and eliable economic models o e lec in a bes way he u u e economic beha iou s. Mos ecession o ecas ing models use ma hema ical echniques o igina ed om he ec o au o eg ession me hod. Howe e , a numbe o ecen s udies combine adi ional p obi models and ec- o au o eg ession models o de elop mo e comp ehensi e dynamic p obi VAR app oaches. Dueke (2005) p esen ed a new Qual VAR model by inco po a ing in o ma ion om quali a i e and/o disc e e a iables in ec o au o eg essions. Wi h a Qual VAR, i is possible o c ea e dynamic o ecas s o he quali a i e a iable using s anda d VAR o ecas s. P e ious o ecas ing me hods o quali a i e a iables, in con as , only p oduce s a ic o ecas s. Dueke applied he Qual VAR o o e- cas ing he 2001 US business ecession. Ou o sample, he model p edic ed he iming o he 2001 ecession qui e well ela i e o he ecession p obabili ies pu o h a he ime by p o essional o ecas e s. Mos impo an ly, unlike adi ional app oaches in ecession modelling, Dueke included a bina y indica o o each ecession as a dynamic componen in he Qual VAR model. As an augmen ed VAR, he Qual VAR, which includes in o ma ion abou he quali a i e a iable in he o m o a unca ed no mal la en a iable, can also enhance he quali y o densi y o ecas s o o he a iables in he sys em, such as ou pu g ow h, ela i e o he VAR wi hou he quali a i e a iable. Chen (2014) applied he Qual VAR me hod de eloped by Dueke (2005) o sea ch o he mac oeconomic indica o s ha ully i he in-sample mo emen s o busi- Jou nal o Cen al Banking Theo y and P ac ice 64 ness cycle luc ua ions and accu a ely p edic he ou -o -sample ecession p ob- abili y in Canada. This s udy examined he pe o mance o se e al leading indi- ca o s in p edic ing u u e Canadian ecessions using he sample om 1959 o 2012. Consis en wi h he indings om Es ella and Mishkin (1998), his pape inds ha he slope o he yield cu e p o ides he mos eliable in-sample i ness o economic u ning poin s in ecen Canadian his o y. Besides, he US-Canada noon-spo ed exchange a e also p esen s some in-sample o ecas ing powe . Howe e , none o he single a iables included in his s udy p o ided signi i- can ou -o -sample o ecas ing powe . Combining he exchange a e and he e m sp ead p o ided he mos accu a e sho - e m signal o upcoming ecessions in Canada. The addi ion o he g ow h a e o eal chain-weigh ed GDP o he ex- change a e and he e m sp ead p oduced he mos us able o ecas ing model in he long un. Besides he ad an ages, one canno igno e he ac ha Qual VAR model may s ill su e om he po en ial o e i ing p oblem, hus ca e ul choice o he app op ia e o ecas ing indica o s is c ucial. Moon and Lee (2012) e alua ed he o ecas ing pe o mance o he Ko ean ESI wi h espec o GDP g ow h and cycle. They ound ou ha hecons uc ed ESI (based on business su ey index and consume su ey index) had a good ack- ing pe o mance as a leading indica o o GDP. Using he G ange causali y es s, hey showed ha he ESI p ecedes GDP g ow h, implying he o me con ains use ul in o ma ion in p edic ing GDP g ow h. Also, using a p obi model, he au ho s con i med ha he ESI is help ul in moni o ing and p edic ing he u n- ing poin s o GDP. 3. T acking he leading ea u es o ESI wi h espec o GDP cycle This Chap e aims o assess he ex en o which he Macedonian cyclical de elop- men is a ec ed by changes in consume and p oduce business cycle e alua ions, as summa ized by he Economic sen imen indica o o he Eu opean Commis- sion. Sen imen su ey da a a e usually assumed o be leading a iables ela i e o business cycles. The span o lead ime, howe e , is less ce ain and subjec o empi ical e i ica ion. The amoun o lead ime may change o e ime as well. In his sec ion, pa e ns o bo h he Macedonian business cycle and he economic sen imen cycle a e analysed. We ini ia ed he p ocedu e o mapping he peaks and oughs in he Macedo- nian economic down u ns by i s adop ing a mul i a ia e case o Chow and Lin (1971) as a empo al disagg ega ion me hod o es ima e mon hly om qua e ly igu es o he o icial GDP. Fo he economic cycle indica o , he indus ial p o- Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model 65 duc ion and e ail ade we e embedded as high equency ex e nal in o ma ion ha is belie ed o sa is ac o ily ep esen he dynamics o he a ge a iable2. The GDP cycle was ex ac ed by applying a double Hod ick-P esco (HP) il e o he p e iously ob ained mon hly igu es. The HP il e is un wice o achie e a smoo hed de- ended cycle ( emo ing a long- e m end om he season- ally adjus ed GDP and hen smoo hing his de- ended se ies). Namely, se ing λ = 133,107.94 in he i s s ep and hen imposing λ = 13.93 in he second s ep o he il e ing p ocedu e allows us o emo e he cyclical componen s ha ha e a cycle leng h longe han 120 mon hs and hose ha ha e a cycle leng h sho e han 12 mon hs.3 The ESI is composed o he indus ial, se ices, consume , cons uc ion and e- ail ade con idence indica o s. The economic sen imen indica o is compiled as an index wi h mean alue o 100 and s anda d de ia ion o en o e a ixed sam- ple pe iod. Values o he economic sen imen indica o abo e (below) 100 indi- ca e abo e-a e age (below-a e age) economic sen imen . Fo he pu poses o he econome ic analysis we gene a ed a smoo hed-cu e ep esen a ion o ESI. The adjus men o he sensi i i y o he end o sho - e m luc ua ions is achie ed by se ing he alue o he mul iplie λ o he HP il e o 13.93. Eu opean Commission publishes ESI o Macedonian economy (and se e al o h- e candida e coun ies) on mon hly basis, seasonally adjus ed. The su eys used o cons uc his indica o o Macedonia we e launched o he i s ime in May 20084. The da a se managed in his pape spans o e he pe iod om May 2008 o Decembe 2014, e lec ing his pa icula limi a ion. Figu e 1 displays he ajec o y o bo h ESI cycle and GDP cycle, wi h mon hly equency. ESI cycle se es as a p oxy o consume s’ and p oduce s’ se o opin- ions abou he gene al s a e o he economy. On he o he hand, GDP cycle de- 2 We also conside ed mon hly da a o cons uc ion wo ks, bu due o high ola ili y hey we e no included in he in e pola ion p ocedu e. 3 Going om equencies o λ pa ame e is achie ed by subs i u ing in o he o mula: λ=[4(1=cos(ω0))2]-1 whe eas ω0 is he equency exp essed in adians, and τ deno es he numbe o pe iods i akes o comple e a ull cycle. The wo pa ame e s a e ela ed h ough ω0 = 2π/τ . So he λ alues abo e co espond o τ =120 mon hs and τ =12 mon hs (Sou ce: OECD, h p://www.oecd.o g/s d/composi eleadingindica o scli equen lyaskedques ions aqs.h m#12). 4 Howe e , he da a on con idence indica o o consume s a e published since May 2012. Ac- co ding o he Use guide by EC, (Ma ch 2014), he ESI summa izes he con idence indica o s in i e a eas o he economy, acco ding o he ollowing weigh s: indus y (40%), se ices (30%), consume s (20%), cons uc ion (5%) and e ail ade (5%). Jou nal o Cen al Banking Theo y and P ac ice 66 sc ibes he economic condi ions om he pe spec i e o p oduc ion and supply, as well om consump ion and demand s anding poin . The leading ea u e o he cyclical mo emen o ESI wi h espec o ha o GDP, obse ed in Figu e 1, is also ully-con i med by he c oss-co ela ion analysis o he cycles. To his end, Table 1 sugges s pa e ns o small leads and lags be ween hese a iables, a ac uali y which will be picked up la e by he Qual VAR model, gi en i s dynamic s uc u e. The maximum c oss-co - ela ion is 0.71 and 0.68, when he cyclical com- ponen o ESI is 0 and 1 mon h ahead o GDP, espec i ely, as shown in Table 1. Table 1: C oss-co ela ion o he cycles o he ESI and GDP Leading (-1) o Lagging (+1) mon hs -6 -5 -4 -3 -2 -1 012345 6 0.06 0.22 0.38 0.51 0.61 0.68 0. 71 0.61 0.48 0.34 0.21 0.08 -0.02 The de ec ed ela ion be ween he c oss- abula ed a iables leads us o a s a ic p obi model aiming a u he examining he leading p ope y o he cyclical componen o ESI wi h espec o ha o GDP. To his end, ou app oach is ex- ac ly analogous o ha o Moon and Lee (2012). Addi ionally, in con inua ion in his Chap e we p o ide an e alua ion o he o ecas pe o mance o he ESI cy- cle in iden i ying he business cycle u ning poin s in bo h in-sample and pseudo ou -o -sample con ex . Following Moon and Lee (2012), we de ine a bina y dependen a iable, Y which akes on only alues o one and ze o as ollows: 1, i he economy is in decele a ion pe iod Y = 0, o he wise Figu e 1: Cyclical componen s o ESI and GDP Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model 67 Then he es ima ed p obabili y o being in he decele a ion pe iod is o he o m whe e x1 ... , xk a e k explana o y a iables, β1 ... , βk, a e he co esponding eg es- sion coe icien s and F is he cumula i e dis ibu ion unc ion (cd ) o a s anda d no mal dis ibu ion, i.e., He e he decele a ion pe iod is de ined based on he peak and ough poin s o he GDP cycle. The con empo aneous and lagged alues o he cyclical compo- nen o he ESI a e in oduced as explana o y a iables in he p obi model om abo e. Mo eo e , Table 2 shows ha bo h o hem a e s a is ically signi ican . No e ha he sign o es ima ed coe icien o ESI_CYC(-1) is posi i e, implying ha ESI_CYC(-1) and he con empo aneous GDP cycle end o co-mo e. We also ca ied ou he LM es o he e oscedas ici y using he a i icial eg es- sion me hod desc ibed by Da idson and MacKinnon (1993, sec ion 15.4). We es ed he null hypo hesis o homoscedas ici y agains he al e na i e o he e o- scedas ici y. So o his end, he null was accep ed. We also pe o med he Hos- me –Lemeshow goodness-o - i es . The es esul s con i med ha he model is co ec ly speci ied. Table 2: P obi model es ima ion Va iable Coe icien S d. E o z-S a is ic P obabili y C9.1340 3.0781 2.9674 0.0030 ESI_CYC -0.8864 0.1829 -4.8453 0.0000 ESI_CYC(-1) 0.7965 0.1650 4.8263 0.0000 McFaddenR-squa ed 0.3953 Hosme -Lemeshow chi2(8) = 3.5217 P ob > chi2 = 0.8975 He e oscedas ici y L-M es = 2.8737 P obabili y = 0.2377 F om he es ima ed p obi eg ession, he p obabili y o being in he decele a ion pe iod can be compu ed o each da a poin , which is called in-sample o ecas . The p obabili ies om he in-sample o ecas a e g aphed oge he wi h he GDP cycle in Figu e 2. Jou nal o Cen al Banking Theo y and P ac ice 68 The es ima ed p ob- abili ies a e shown o be high in he shaded a eas which ep esen Macedo- nian business cycle eces- sions. In pa icula , hey a e close o 1 du ing he decele a ion phase span- ning om mid-2008 o mid-2009, e lec ing he new global eali y i.e. he global down u n and mid-2011 o end-2012 e- lec ing he second wa e o he c isis (G eek deb c isis). In addi ion, ecu si e 1-s ep ahead ou -o -sample o ecas s we e un o e he pe- iod om May 2011 o Decembe 2014. The p edic i e accu acy o hese o ecas s is e alua ed by compa ing a pe cen age o co ec classi ica ion based on he p e- speci ied cu o alue, which is p ocedu e analogous o ha desc ibed in Moon and Lee (2012). So, he e a e wo ypes o co ec classi ica ions. One is ha he p edic ed p obabili y is g ea e han he cu o and he obse ed Y =1, and an- o he is ha he p edic ed p obabili y is less han o equal o he cu o and he obse ed Y =0.5 The po ion o Y =1 obse a ions ha a e co ec ly p edic ed is called sensi i i y, while he ac ion o Y =0 obse a ions ha a e co ec ly classi- ied is called speci ici y. In his pa icula empi ical implemen a ion, he sensi i - i y is calcula ed o he decele a ion phase and he speci ici y o he accele a ion phase. Mo eo e , a pe cen age o co ec classi ica ion among o al obse a ions is calcula ed. 5 Recessions / expansions a e de ined by he bina y code 1 / 0. Fo example, wi h cu o alue o 0.6, all esul s o he p obi model o ecession wi h p obabili y equal o highe han 0.6 will be plausible. Figu e 2: Es ima ed p obabili ies and he GDP cycle No e: P-peak, T- ough. Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model 75 in pu e eal ime bu om he la es a ailable da a se – a p ac ice pe cei ed as a na u al amewo k o e alua e he alue added in o ecas ing. The Figu e 3 plo s he pseudo eal- ime p edic- ion o he di ec ion o he business cycle in 2013 om in o ma ion se end- ing in Decembe 2012. The ex pos knowledge on he eal di ec ion o he GDP cycle in he co - esponding pe iod poin s ou ha he accu acy o he index is limi ed o only nea - e m business cycle de elopmen s ( he i s ew mon hs). In o de o achie e g ea e quali y assu ance, we pe o med an addi ional obus - ness check by epea ing he abo e p ocedu e on a da a se ending in Decembe 2010. This ime, esul s comply wi h subsequen knowledge o he cyclical ac s o he en i e 2011. Howe e , like in he p e ious case, he empi ical indings a e s a is ically signi ican only o a e y sho ho izon. 5.2. Pu e ou -o -sample esul s Figu e 4 displays he model simula ed ajec- o y o he business cycle di ec ion in 2015 om an in o ma ion se ending in Decembe 2014. The p edic ed aise in he index indica es ha he business cycle condi ions a e expec ed o imp o e in 2015, i.e. a ecession is no o eseen ha a Figu e 3: Pos e io mean o la en a iable and i s 12-mon hs-ahead o ecas (shaded a eas indica e ecession da es) Figu e 4: Pos e io mean o la en a iable and i s 12-mon hs-ahead o ecas (shaded a eas indica e ecession da es) Jou nal o Cen al Banking Theo y and P ac ice 76 ahead. Bu again, he s a is ically signi ican pe iod spans only o e he i s ew mon hs o 2015. All in all, he esul s sugges ha ou Qual VAR p edic s he business cycle de- elopmen s easonably well on a nea - e m ho izon, so ha o ecas s o i a e pe cei ed o be a use ul addi ion o NBRM’s exis ing amewo k aimed a GDP nowcas ing. 6. Conclusion The de i a ion o a mix o mac oeconomic indica o s ha could summa ize and p edic business cycle pa e ns is usually a challenging ask, gi en he luc ua- ions in mac oeconomic condi ions. A good unde s anding o business cycle ac s is c ucial o mac oeconomic decision-making, especially in mone a y and iscal policies. This pape desc ibes he p ocedu e o mapping he peaks and oughs in he Macedonian economic cycle. A numbe o ecen s udies in his a ea com- bine adi ional p obi models and ec o au o eg ession models o de elop mo e comp ehensi e dynamic p obi VAR app oaches and ha is also done in his pa- pe in he case o Macedonian economy. Sen imen su ey da a a e commonly assumed o be leading a iables ela i e o business cycles, he e o e, we use he Economic sen imen indica o (ESI) pub- lished by he Eu opean Commission o he Macedonian economy. The de ec ed ela ion be ween hese wo a iables poin s us o a s a ic p obi model aiming a u he examining he leading p ope y o he ESI wi h espec o GDP. Ad- di ionally, we p o ide an e alua ion o he o ecas pe o mance o he ESI cycle in iden i ying he business cycle u ning poin s in bo h in-sample and pseudo ou -o -sample con ex . The es ima ed p obabili ies wi hin in-sample o ecas a e shown o be high du ing he decele a ion phase o he Macedonian business cycle spanning om mid-2008 o mid-2009, e lec ing he global economic down u n and om mid-2011 o end-2012 e lec ing he second wa e o he c isis (G eek deb c isis). The p edic i e accu acy o ou -o -sample o ecas s is e alua ed by compa ing a pe cen age o co ec classi ica ion based on he p e-speci ied cu o alue, which is p ocedu e analogous o ha desc ibed in Moon and Lee (2012). The es ima ed p obi model seems o iden i y he business cycle u ning poin s sa is ac o ily well. Acco ding o a la ge and g owing body o li e a u e, he inclusion o dynamic elemen s in he p obi models can yield mo e accu a e o ecas s o ecessions han s anda d p obi eg essions. The e o e, we adap he Qual VAR model om Fo ecas ing Macedonian Business Cycle Tu ning Poin s Using Qual Va Model 77 Dueke (2005). Ou Qual VAR model pu oge he he ollowing indica o s: eal GDP (in e pola ed GDP, wi h mon hly equency), ESI and he capaci y u iliza- ion a e in manu ac u ing. Conside ing i s easonably good ack eco d in ap- p oxima ing he business cycle ac s in pseudo ou -o -sample con ex , we ound ha ou Qual VAR is sa is ac o ily success ul in he accu a e and imely assess- men o nea - e m business cycle de elopmen s. To his end, he o ecas s o i a e pe cei ed o be a use ul addi ion o he NBRM’s exis ing amewo k aimed a GDP nowcas ing. Jou nal o Cen al Banking Theo y and P ac ice 78 7. Re e ences 1. Chau e , M. and Hamil on, J. D. (2005). Da ing Business Cycle Tu ning Poin s, NBER Wo king Pape No. 11422. 2. Chen, L. (2014). Fo ecas ing Canadian Recession using Qual VAR Model, Mas e hesis, Uni e si y o O awa, Depa men o Economics, ECO6999, Augus 2014. 3. Chow, G.C. and Lin, A. (1971). Bes linea unbiased in e pola ion, dis ibu ion and ex apola ion o ime se ies by ela ed se ies. The Re iew o Economics and S a is ics, 53, 372-375. 4. Da idson, R. and MacKinnon, J.G. (1993). Es ima ion and In e ence in Econome ics, New Yo k: Ox o d Uni e si y P ess. 5. Dueke , J. M. (2005). Dynamic Fo ecas s o Quali a i e Va iables: A Qual VAR Model o U.S. Recession, Jou nal o Business & Economic S a is ics, 23, 96-104. 6. Es ella, A. and Mishkin F.S. (1998). P edic ing U.S. Recessions: Financial Va iables as Leading Indica o s, The Re iew o Economics and S a is ics, 80, 45-61. 7. Eu opean Commission (2014). The join ha monized EU p og amme o he business and consume su eys, Use guide. B ussels: Eu opean Commission. 8. Kauppi, H. and Saikkonen, P. (2008). P edic ing US ecessions wi h dynamic bina y esponse models, The Re iew o Economics and S a is ics, 90, 777-791. 9. Meinusch, A. and Tillmann, P.(2014). The Mac oeconomic Impac o Uncon en ional Mone a y Policy Shocks, MAGKS Pape s on Economics 201426. Philipps-Uni e si ä Ma bu g, Facul y o Business Adminis a ion and Economics, Depa men o Economics. 10. Moon, H. and Lee, J. (2012). Fo ecas E alua ion o Economic Sen imen Indica o o he Ko ean Economy. Basel, IFC Con e ence, BIS. 11. Ng, E. (2012). Fo ecas ing US ecessions wi h a ious isk ac o s and dynamic p obi models, Jou nal o Mac oeconomics, 34, 112-125. 12. Nybe g, H. (2010). Dynamic p obi models and inancial a iables in ecession o ecas ing, Jou nal o Fo ecas ing, 29, 215-230.