The Austrian Business Cycle A Characterization. Sandra M. Leitner *) Working Paper No October 2007

Transcription

1 DEPARTMENT OF ECONOMICS JOHANNES KEPLER UNIVERSITY OF LINZ The Ausrian Business Cycle A Characerizaion by Sandra M. Leiner *) Working Paper No. 077 Ocober 2007 Johannes Kepler Universiy of Linz Deparmen of Economics Alenberger Srasse 69 A-4040 Linz - Auhof, Ausria *) sandra-marina.leiner@jku.a Phone ++43(0) Fax:++43(0)

2 The Ausrian Business Cycle A Characerizaion Inroducion Economies consanly undergo significan cyclical variaions of disinc paern and origin. In duraion, business cycles vary from more han one o en or welve years, and comprise a boom, or expansionary phase, followed by a recession, or conracionary phase. Such cycles recur a unpredicable inervals and do no las for a fixed lengh of ime. Recessions are characerized by high unemploymen and low produciviy wih highly asymmeric shor bu sharper cycles han expansions. The purpose of his paper is o provide an overall picure of he Ausrian business cycle covering he period 970 unil 2004 by sudying he cycle s sylized facs in erms of volailiy, co-movemens and persisence. The noion of sylized facs sems from he renowned work of Burns and Michell (946) inending o provide and inerpre model-free observaions of macroeconomic variables behavior. Addiionally poenial hypoheses as o he source of observed cycles will be esed, poining a he significan role supply-side shocks like echnology play in riggering observed business cycles. This paper is organized as follows: Secion I describes he mehodology applied o exrac he cyclical componen from observed macroeconomic daa. Secion II gives a brief descripion of business cycle regulariies observed in indusrialized counries, while secion III applies he empirical es o he economy case Ausria and discusses he resuls on he basis of he daa s second momens. Secion IV aemps o idenify poenial sources of observed cycles, focusing on prices and he real wage rae as discriminaory facors. A discussion of he baseline Real Business Cycle Model s basic characerisics as well as resemblances wih he observed daa represens he focal poins of secion V while secion VI concludes. I. Mehodology he Hodrick-Presco Filer The sudy of business cycle sars wih he process of derending key macroeconomic variables since second momens of ime series variables, i.e. volailiy, co-movemen and persisence can only be applied o saionary sochasic processes. The lieraure offers numerous derending or equivalenly, smoohing procedures, including firs differencing, band-pass filers (Baxer/King (999)) and Hodrick-Presco (HP) filers (Hodrick/Presco (997)). The laer, as a wo-sided symmeric moving average filer, is he mos frequenly applied rend-cycle decomposiion echnique due o is simple mechanical and judgmenal-free implemenaion and he resemblances wih acual business cycle movemens. However, is shorcomings are also highlighed in he lieraure. In ha respec, Cogley and Nason (995) and Harvey and Jaeger (993) argue ha he Hodrick-Presco filer migh cause spurious cycles in he case of difference-saionary daa, ha he filer migh creae mos of he cycles

3 in he daa or ha he filer is only opimal in special cases, as poined ou by King and Rebelo (993). However, Ravn and Uhlig (997) hypohesize ha he filer will remain he sandard mehod for derending for a long ime o come since he HP-filer has wihsood he es of ime and he fire of discussion remarkably well. According o Hodrick and Presco (997) observed ime series y are he sum of cyclical c y and growh componens can be represened as follows: g y wih he laer varying smoohly over ime. Mehodologically, his observaion y = y + y for =,..., T c g Assuming ha he smoohness of growh can be measured by he sum of squares of is second-order differences, he growh componens can be calculaed by minimizing he following leas squares problem: = = 0 min g { y } g 2 g g g g 2 {( y y ) + λ [( y + y ) ( y y )] } The above problem consiss of wo componens. The firs componen can be inerpreed as a measure g of finess and becomes minimal for y = y, while he second componen, as a moving-average represenaion, can be viewed as a measure of smoohness and becomes minimal for invariable growh in g y, i.e. linear growh over ime. The parameer λ is a posiive number represening a penaly on growh variabiliy, wih higher values resuling in smooher growh componens and consequenly more erraic cyclical componens. Afer Burns and Michell s influenial work on pre-second world war U.S. business cycle regulariies, he lengh of cycles was widely acceped o vary beween ½ and 8 years. Hence, based on heir observaions filers are specified o cu off componens of higher or lower frequencies in order o capure beer he cyclical componen. For quarerly daa, he convenion is o se λ equal o,600. However, an unrefleced applicaion of a sandardized smoohness parameer is unadvisable, paricularly aking ino accoun ha he sandard is based on oudaed and counry-specific business cycle observaions. To ha end, Pedersen (998) sresses ha he adopion of he Burns/Michell observaion of 8-year cycles should be reconsidered, since he sudy covers he pre-world-war II period only while imporan characerisics of he classical cycles have changed in he US from he pre- o he pos-war period along wih srucural and insiuional changes in he economy. More imporanly, heir measure of he US cycles does no discriminae beween economic growh and cyclical variaions of ime series hence heir measure of he cycle duraion from peak o peak is oversaed. His analysis of OECD counries (including he US) reveals ha poswar cycles are

4 beer described as covering six- year cycles a mos, calling for a modificaion of he smoohness parameer λ o range beween 30 and 340 for quarerly daa. Table Augmened Dickey-Fuller Tess for differen Lambdas Variables -value p-value λ = 00 p-value λ = 25 p-value λ = p-value -value -value -value Real GDP * * * Real Governmen Expendiures * * Real Privae Consumpion * * * Real Invesmens * * * Real Expors * * * Real Impors * * * Real Ineres Rae * * * CPI * 0.00 Inflaion Rae * * * Real Wage Rae * * * * Capial Sock * * Money Supply * * * Dependen Employmen * * Toal Employmen * Labor Produciviy * * * Source: Own Calculaions All regressors include a consan. The criical value a he 5 % significance level is from Fuller (976), indicaed by an aserix. As for annual daa frequencies, differen smoohness parameers are suggesed by he lieraure. On he one hand, Backus and Kehoe (992) in heir sudy on business cycle properies of a cross-secion of counries apply a smoohness parameer of 00, while, on he oher, Correira e al (992) and Cooley and Ohanian (99) use a lambda of 400. Relaed o ha, Ravn and Uhlig (997) - in heir sudy on he proper adjusmen of filers o diverse daa frequencies - sugges alernaive smoohness parameers. Based on a graphical represenaion of differen parameers for annual daa o bes replicae he convenionally applied parameer of λ of.600 for quarerly daa, hey recommend a smoohness parameer of 6.25 as he closes mach. Wih respec o he opimal choice of he smoohness parameer, since he filer eliminaes he secular rend componen, he cyclical componen of he observed ime series should be esed for saionariy o ensure any long erm rend o be eliminaed by he filer. To do his, he augmened Dickey-Fuller uni roo es was firs applied o all original daa variables as shown in he firs column of Table. Wih he excepion of real wages, all variables appear non-saionary and call for derending for furher analysis. Columns 2, 4, 6 and 8 show he resuls of he augmened Dickey-Fuller ess for differen λ s like 00, 25 and 6.25 and he respecive p-value of significance and hence represen he raionale for choosing 6.25 for he analysis.

5 Afer derending he series, basic characerisics can be inferred; Volailiy assesses he ampliude of flucuaions and indicaes he magniude of he variable s conribuion, and is sensiiviy, o aggregae flucuaions. This is measured by he sandard deviaion, where a low sandard deviaion implies he variable does no conribue much o aggregae flucuaions. Co-movemens wih conemporaneous oupu series indicae he cyclicaliy of key macroeconomic variables like privae consumpion, invesmen, governmen expendiures, and money supply. These are measured by he correlaion coefficiens where posiive, negaive or near-zero coefficiens poin o pro-, couner-, and acyclicaliy, respecively. Finally, persisence indicaes he ineria in business cycles, paricularly he cyclical componen, and capures he lengh of observed flucuaions. This is measured by he firs-order auocorrelaion coefficien where a high coefficien implies a very persisen, i.e., long, economic flucuaion. Posiive coefficiens indicae ha high values follow high values or low values follow low ones, whereas negaive coefficiens indicae reversals from high o low values or he reverse. II Business Cycles in Indusrialized Counries: A Brief Descripion VOLATILITY. Business cycles in indusrialized counries are found o cover a period of approximaely six o eigh years wih high volailiy in invesmen in consumer and producer durables and low one in privae consumpion. The mos volaile expendiure caegory however is invenory invesmen. Employmen also varies considerably over observed cycles bu less han real GDP, while produciviy as well as wages show comparably lower variaion. The capial sock is found o show lile volailiy. Given differen preferences, insiuional se-ups, and war crisis, governmen expendiures reveal diverse behaviors across counries and are found be eiher more or less variable han real naional GDP (Fiorio/Kollinzas, 994). In he same vein, money does no have a clear cu paern bu varies differenly across counries and money sock definiions. Expors and impors are found o be more variable han consumpion bu less variable han invesmens. Finally, inflaion exhibis high volailiy. CO-MOVEMENTS. For indusrialized counries, King and Rebelo (999) sress ha mos macroeconomic variables are procyclical wih a paricularly high degree of co-movemen beween aggregae oupu and oal hours worked. Addiionally, wages, governmen expendiures, and he capial sock seem o display no sysemaic cyclicaliy wih aggregae oupu. PERSISTENCE. Macroeconomic variables in indusrialized counries display non-negligible persisence.

6 III Empirical Tes: The Ausrian Case Figure uses daa from he Toal Economy Daabase provided by he Conference Board and Groningen Growh and Developmen Cenre (2006) and suggess ha wih he excepion of he pos- World-War II period of he 950s, he Ausrian business cycles move wihin a bandwidh of +/-2 % For he res of he paper, he analysis avails of annual macroeconomic variables covering he period 970 o 2004 from he Economic Oulook (ECO) daabase available on he WIFO daabase (he Ausrian Economic Research Insiue) and covers real gross domesic produc (GDP), privae consumpion, governmen expendiures, invesmens, expors, impors, capial sock in he business secor, real money supply (M2), he consumer price index (CPI), he inflaion rae calculaed from he GDP-deflaor, he real long-erm ineres rae deflaed wih he GDP-based inflaion rae, oal employmen, oal dependen employmen, he real wage rae for he business secor deflaed wih CPI and labor produciviy. Only real money supply, he real ineres rae and he inflaion rae cover shorer periods: 970 o 998 for real money supply and 97 o 2004 for he real ineres rae and he inflaion rae. All real componens are expressed in erms of 2000 prices. Figure Ausrian Business Cycles, GrowhRaes Years Source: GGCD-Daa In correspondence wih he suggesions of Uhlig and Ravn (997) paired wih he Dickey-Fuller ess of saionariy for differen smoohness parameers moivae λ o be se equal o 6.25 and he applicaion of he Hodrick-Presco filer procedure o decompose available ime series. The applicaion of alernaive smoohness parameers and associaed implicaions are discussed in he daa appendix A. Figure 2 gives a represenaion of he Ausrian business cycles covering he period 970 o While he 970s appear o be dominaed by he effecs of he wo oil price shocks in 974 and 978, saring in 983a peak-o-peak duraion analysis of observed cycles reveals wo very similar cycles: One cycle covers he eigh-year period from 983 o 99 wih a recession in 987. The oher cycle

7 covers a nine-year period from 99 o 2000, wih a recession in 997. Ineresingly, he economy was on is way o economic recovery already when he raher severe recession of 987 hi Ausria. In correspondence wih previous observaions, recessions seem o be sharper bu less deep. Figure 2 Ausrian Business Cycles Growh Raes Years Source: Own calculaions Volailiy. Table 2 repors sandard deviaions of HP-filered naional expendiure componens, as well as he real long-erm ineres rae, he consumer price index, he inflaion rae, he capial sock, M2, dependen employmen, oal employmen, he real wage rae and labor produciviy. Among he naional expendiure componens, invesmen is he mos volaile variable while governmen expendiures is he leas volaile one. Relaive volailiy, defined as he raio of a componen s volailiy o volailiy of real GDP furhermore shows ha wih he excepion of governmen expendiures, he capial sock, oal as well as dependen employmen and he real wage rae, all componens have higher volailiy han real GDP, wih invesmens four imes more volaile han real GDP. The relaively high volailiy of invesmens is ypically associaed wih Keynes famous asserion of animal spiris of invesors exogenous and perhaps self-fulfilling waves of opimism and pessimism. This also forms par of he widely acceped belief ha invesmen burss se off business cycles. Among he moneary and price variables, money supply M2 is wo imes more volaile han real GDP and CPI is equally volaile. Addiionally, he inflaion rae and he long-erm ineres rae represen he wo mos volaile moneary componens wih a 7-imes and a 48-imes higher volailiy as compared o real GDP. As expeced, he capial sock is he leas volaile variable. Labor marke specific variables all exhibi relaively low volailiy wih all bu labor produciviy revealing lower han GDP volailiy. As o capial sock and employmen, expensive adjusmens in he major producion inpus like capial and labor migh be held responsible for he observed sluggish changes. Afer a decomposiion of he whole period under invesigaion ino wo subperiods, one covering 970 o 983 and he oher covering 983 o 2004, i becomes apparen ha for he firs subperiod, wih he

8 excepion of he inflaion rae, all variables depic higher volailiy, while for he second subperiod, wih he excepion of expors, he inflaion rae, oal employmen and labor produciviy, all variables depic lower volailiy han observed for he period as a whole. The relaively higher volailiy of all variables during he firs subperiods migh be pu down o he effecs of he wo oil-price shocks in he 970s. Table 2 Sandard Deviaion Variables No. of Obs. Overall Volailiy relaive o ha of oupu Subperiod Volailiy relaive o ha of oupu Subperiod Volailiy relaive o ha of oupu Real GDP Real Governmen Expendiures Real Privae Consumpion Real Invesmens Real Expors Real Impors Money Supply Real Ineres Rae CPI Inflaion Rae Capial Sock Toal Employmen Dependen Employmen Real Wage Rae Labor Produciviy Source: Own calculaions Co-Movemens. Table 3 repors cross-correlaions of naional expendiure componens, he real ineres rae, CPI, he inflaion rae, he capial sock, money supply, dependen employmen, oal employmen, he real wage rae as well as labor produciviy wih oupu. Wih he excepion of governmen expendiures, he inflaion rae and money supply, all variables show srong and posiive correlaion wih real GDP while he real ineres rae and CPI depic srong and negaive correlaion wih real GDP. Governmen expendiures, he inflaion rae and money supply do no seem o reveal any significan cyclicaliy wih real GDP. A closer look a he periodical decomposiion furher reveals ha wih he excepion of invesmens, impors, he real ineres rae and CPI, money supply and labor produciviy, all variables in he firs subperiod show lower correlaion wih real GDP, while for he second subperiod governmen expendiures, privae consumpion, he inflaion rae, he capial sock, dependen employmen and oal employmen show higher correlaion wih real GDP han observed for he period as a whole. In he firs period, only privae consumpion, invesmens, impors and labor produciviy are

9 significanly and posiively correlaed wih oupu while for he second period, privae consumpion, expors and impors, he capial sock, dependen as well as oal employmen, he real wage rae and labor produciviy reveal posiive significan correlaion wih real GDP. Table 3 Correlaion wih real GDP Variable Overall -value Subperiod Subperiod Real Governmen Expendiures Real Privae Consumpion * * 0.757* Real Invesmens * * Real Expors 0.535* * Real Impors * * * Nominal Ineres Rae * Real Ineres Rae * CPI * Inflaion Rae Real Wage Rae * * Capial Sock * * Money Supply Dependen Employmen 0.60* * Toal Employmen * * Labor Produciviy * * * Source: Own calculaions Aserix denoes significan a 5 % level Persisence. Table 4 shows he persisence of variables measured by he firs order auocorrelaion. The daa illusrae fairly persisen variables wih expors, CPI, he real wage rae, he capial sock and oal employmen as he mos persisen macroeconomic variables for he overall period. Wih respec Table 4 Firs-Order Auocorrelaion Variable Overall p-values Subperiod p-values Subperiod p-values Real GDP Real Governmen Expendiures Real Privae Consumpion Real Invesmens Real Expors Real Impors Real Ineres Rae CPI Inflaion Rae Real Wage Rae Capial Sock Money Supply Dependen Employmen Toal Employmen Labor Produciviy Source: Own calculaions

10 o real GDP, he periodical decomposiion poins a higher persisence in he second subperiod as opposed o he firs and he overall period. During he firs subperiod, only CPI reveals srong persisence, while for he second subperiod, CPI, he capial sock, oal dependen and oal employmen show srong persisence. IV Poenial Sources of he Ausrian Business Cycles Over he pas decade, considerable effor has been devoed o shedding ligh on he poenial sources of business cycles by sudying he price-oupu and inflaion-oupu correlaions, i.e. he cyclicaliy of prices and inflaion. To ha end, if supply shocks like erms of rade shocks, echnology shocks or shocks due o exreme weaher condiions are idenified as he basic source, a negaive price-oupu correlaion should be observed. Conversely, if demand shocks like shocks o privae or public consumpion or moneary shocks represen he source of he business cycle, a posiive price-oupu correlaion should be idenified. This can bes be analyzed wihin a radiional AS-AD framework, where a shif of he supply curve (AS) along he negaively-sloped demand curve (AD) leads o counercyclical prices while a shif of he demand curve along he posiively-sloped supply curve resuls in procyclical prices/inflaion. The real business cycle heory, as a purely supply side approach, and he radiional Keynesian heory, as a purely demand side approach, lead o differen price-oupu correlaions and herefore help discriminae beween differen economic schools. Tradiionally, he Keynesian demand driven models focus on he inflaion rae raher han he price level as he variable of ineres so ha a posiive inflaion oupu correlaion emerges. In ha respec, he real-business-cycle heory leads o counercyclical prices while he radiional Keynesian heory predics procyclical inflaion raes. Empirical resuls on developed and developing counries however remain inconclusive as o he exac source of observed cycles. Chadha/Prasad (994), using quarerly daa from he IMF s Inernaional Financial Saisics poin a he counercyclicaliy of he price level for he G7 counries. However, his resul does no carry over o he inflaion-oupu correlaion and no clear-cu inference abou he source of he shock can be made. Addiionally, Kim e al. (2003) repor fairly srong counercyclicaliy for prices for he G7 for he period covering 960 o 996 for daa aken from IFS while Fiorio and Kollinzas (994) show srong counercyclical prices for he G7 counries for quarerly daa beween 960 and 989. Furhermore, describing hisorical properies of business cycles of a cross-secion of counries for he prewar, inerwar and poswar period, Backus and Kehoe (992) repor negaive correlaions of naional inflaion raes wih real GDP for he poswar period for Ausralia, Canada, Denmark,

11 Germany, Ialy, Japan, Norway, Sweden, he UK and he US and posiive or mixed correlaions wih oupu for he prewar and inerwar period, respecively. Wih respec o price levels, all counries excep Canada and Germany reveal negaive correlaions during he poswar period. For he prewar period, only Germany, Ialy and Japan show negaive correlaions wih oupu, while for he inerwar period, only Denmark displays a negaive oupu-price correlaion. Hence, supply-led business cycles appear o be more prominen in he aferwar period while demand-side driven business cycles seem o be prevalen for he prewar period. Addiionally, Mendoza (995) emphasizes ha he erm supply shock enails more han jus echnology shocks by showing ha abou 50 percen of overall oupu flucuaions for G7, as well as for developing counries, are due o shocks o erms of rade. Table 5 Correlaion of real GDP wih derended and conemporaneous Inflaion and CPI Panel A. Filered Inflaion and Oupu Panel B. Filered CPI and Oupu Lag Overall Subperiod Subperiod Overall Subperiod Subperiod * * * * Source: Own calculaions Aserix denoes significan a 5 % level Table 5 repors correlaions beween HP-filered inflaion and real oupu in panel A and correlaions beween HP-filered prices wih real oupu in panel B, all for conemporaneous real GDP as well as periods of one lead and lag each. Again, a periodical decomposiion for boh correlaion analyses is conduced. Panel A shows insignifican posiive correlaion of inflaion wih conemporaneous oupu for he overall period and second subperiod and insignifican negaive correlaion for he firs subperiod. All in all, only he second subperiod shows a srong and significan posiive correlaion beween inflaion and oupu wih inflaion, leading oupu by one period. For he second subperiod, posiive however insignifican inflaion-oupu correlaions dominae he picure, poining a demand-side driven business cycle heories o play a raher insignifican role in riggering he Ausrian business cycles. For he period as a whole as well as boh subperiods, panel B repors negaive correlaions beween conemporary CPI and GDP wih a significan negaive correlaion only for he overall period. This ogeher wih he counercyclical price-oupu relaions for Ausria beween 964 and 990 by Brandner and Neusser (99?) suppors predicions of supply-side led heories of economic flucuaions for Ausria, like he real business cycle approach. Addiionally, he lead-lag analysis poins a prices counercyclically lagging real oupu by one year.

12 Moreover, apar from he price-oupu correlaion as indicaive facor o discriminae beween radiional Keynesian versus real business cycle models of economic flucuaions, correlaion of real wages wih real GDP represen an addiional discriminaory facor. In ha respec, radiional Keynesian models of business cycles wih sicky nominal wages predic counercyclical real wages while he real business cycle radiion predics procyclical wages wih procyclical labor produciviy deermining real wages. Table 6 Correlaion of real GDP wih derended and conemporaneous real wages Filered Real Wage and Oupu Lag Overall Subperiod Subperiod * 0.652* * * Source: Own calculaions Aserix denoes significan a 5 % level The analysis furher reveals ha for he Ausrian economy for he overall period, real wages exhibi significan procyclical correlaion wih conemporaneous real GDP as well as lagged real GDP, furher emphasizing supply-side led shocks as a source of business cycles in Ausria. A periodical breakdown of correlaions shows a significan posiive relaionship beween lagged real GDP and real wages for he period 970 o 983 and a significanly posiive conemporaneous correlaion beween said variables for he period 983 o V. Resemblances of he Findings wih he Baseline Real Business Cycle Model The baseline Real Business Cycle Model is concepualized as a one secor, closed economy case wih exernal, secor-neural echnological shocks iniiaing economic flucuaions and raional, maximizing agens opimally responding o said shocks and hence, no significan role aribued o he sae wihin he realm of (fiscal or moneary) sabilizaion policy. In ha respec, represenaive consumers are characerized as maximizing expeced lifeime uiliy from consumpion of goods and services c (as consumpion per household member) as well as leisure l N E U = E ρ e u( c, l ) = 0 H, wih u ( ) as he insananeous uiliy funcion of he represenaive household member, ofen characerized as log-linear in wo argumens: u = ln c + b ln( l ), ρ as he discoun rae and N and H as populaion and he number of households, respecively. A he same ime consumers are expeced o supply labor and ren ou capial on compeiive markes, wih real wages and real ineres

13 rae deermined by heir marginal producs. Addiionally, populaion is expeced o exhibi consan growh wih N η = N 0e wih he annual growh rae given by η. Represenaive producers are expeced o maximize profis given he neoclassical producion echnology Y = K A L, α α ( ) wih capial (K), labor-augmening echnology (A) and labor (L) as inpus o producion. Technology is assumed o evolve according o he following sochasic process: ~ ln A = A + g + A, where A is he average level of echnology, g is he rae of echnological progress and A ~ capures he effec of a random echnology shock. Said shocks are modeled as firs-order auoregressive ~ ~ processes: = ρ A + ε, wih ρ A as he ineremporal ransmission facor wih < ρ A < and A A A, he uncorrelaed whie noise shock ε A, wih zero mean ( E A, ε ). Wih < ρ < he effec of he A shock evenually dissipaes. Consumers have wo decision variables o ake ino accoun when maximizing heir lifeime uiliy. Wih respec o labor l, an ineremporal decision rule emerges wih consumers coninuously deciding over he exac amoun of labor o be supplied depending on he relaive real wage rae in period 2 relaive o he one in period : l l 2 = e w 2. ρ ( + r) w Hence, wih an increase in he real wage rae in period, supply of labor also increases in said period. And wih higher income (approximaed by a higher real wage rae and higher labor supply) consumpion also increases in period. Addiionally, wih an increase in he real ineres rae, firs period labor supply also rises relaive o second period labor supply, wih augmened income ransferred o bank accouns for reasons of savings o be len ou o producers for invesmens. Wih respec o consumpion under uncerainy, he problem of ineremporal redisribuion of foregone consumpion arises, capured by he following equaion c ρ = e E ( + r + c+ ).

14 And wih he produc of wo expecaion variables given by he produc of heir expecaions plus heir covariance, ineremporal consumpion is deermined by c e E E c+. [ + r ] Cov, r c+ ρ = + Hence, he covariance beween + consumpion and + ineres raes deermines consumpion in period. In case of posiive correlaion beween said variables, covariance is negaive, rendering curren consumpion more aracive. Addiionally, in case of negaive correlaion beween said variables, covariance is posiive, rendering curren consumpion less aracive. Finally, a radeoff beween curren consumpion and curren leisure can be inferred by equaing disuiliy of addiional work wih uiliy of addiional consumpion, given by c l = w b. Wih high curren real wages, curren consumpion is also high while curren leisure is comparably lower. So, wih an exernal, secor-neural echnology shock hiing he economy, oupu increases wih only slowly dying-off effecs. A he same ime, boh real wages and real ineres raes increase, iniiaing a rise in boh, labor supply and consumpion on he one hand, and savings on he oher. Firms avail of he capial rened ou by consumers o inves in new machineries and equipmen, for example, increasing capial sock. By solving he model he following characerisics can be idenified: As noed above, here is no role for sabilizaion policy money supply and governmen expendiures are expeced o exhibi no sysemaic correlaion wih real GDP. Since he baseline model is a closed-economy case, rade plays no role. Exensions of he baseline model can however capure observed regulariies of procyclical and coincidenal expors and impors (and counercyclical ne expors). The real wages and he real ineres rae are deermined by heir marginal producs and hence are expeced o vary procyclically due o random produciviy shocks. Employmen as deermined by he ineremporal subsiuion decisions of households is expeced o vary procyclically since households work more during periods of high produciviy of labor and work less during periods of low produciviy of labor. Invesmens vary procyclically since households save more during imes of high produciviy, accumulaing capial o be len ou o firms for invesmens.

15 Hence, he capial sock is also expeced o vary procyclically due o procyclical invesmen decisions of firms. Consumpion is procyclical since marginal uiliy o consumpion is higher in imes of higher real wages. Addiionally, consumpion exhibis lower variabiliy han invesmens given he consumers preferences for consumpion smoohing. Prices are counercyclical due o he shif of he supply funcion along a negaively sloped demand funcion. A closer look a he Ausrian business cycles covering he period 970 o 2004 reveals ha all bu one predicion of he baseline real business cycle model is me. Real ineres raes are expeced o vary procyclically, however, he analysis poins a a saisically significan negaive correlaion wih real GDP. Addiionally, he wo discriminaory facors real wages and prices (CPI) ha help o discriminae beween he radiional Keynesian demand-side led approach and he RBC supply-side led approach reveal procyclical real wages and counercyclical prices. Hence, he model-free analysis of he basic sylized facs poins a he Ausrian economy following he real business cycle heory, aribuing a leading role o supply-side led facors like echnological progress as riggering impulses.

16 VI. Conclusion/Overall Findings The Ausrian economy underwen wo business cycles from he period broken down ino he following and The span of he wo cycles is eigh and nine years, respecively wih he hird cycle saring in 2000 sill ongoing. The basic characerisics of he cycles are as follows:. Volailiy, as a preliminary indicaion of significance for macroeconomic flucuaions, poins o invesmens as he mos volaile expendiure componen and governmen expendiures as he leas volaile one. This is suggesive of invesors animal spiris affecing economic flucuaions. Addiionally, employmen and he capial sock appear fairly sable over observed business cycles poining a significan adjusmen coss hampering coninuous adjusmen of exising capial or labor socks o economically opimal ones. 2. The co-movemens of GDP wih all variables, wih he excepion of governmen expendiures and money supply, exhibi srong posiive correlaions while he long-erm real ineres raes and CPI exhibi srong negaive correlaions. 3. All key economic variables reveal fairly weak persisence, wih oal employmen, capial sock and CPI as he mos persisen and predicable ones. All bu one variable follow he predicions of he baseline real business cycle model (and for expors and impors for an exended model). The poenial source of economic flucuaions for he enire period leads o one dominan culpri supply-side shocks. This is derived from he significan and negaive conemporaneous correlaion of prices wih oupu. Addiionally, prices seem o significanly lag oupu by one year poining a sluggishly adjusing prices. Furhermore, he weak posiive, however insignifican correlaion of conemporaneous inflaion wih oupu suppors he finding ha flucuaions are more supply-side driven han demand-side driven in origin. Procyclical real wages furhermore poin a he significan role of he real business cycle approach for he Ausrian economy beween 970 and 2004.

17 References Backus, D.K., and P.J. Kehoe, 992, Inernaional Evidence on he Hisorical Properies of Business Cycles, American Economic Review 82(4), Brandner, Peer, and Klaus Neusser, 99?, Business Cycles in Open Economies: Sylized Facs for Ausria and Germany, Welwirschafliches Archiv, Burns, Arhur F., Wesley C. Michell, 946, Measuring Business Cycles, New York Chadha, Bankim, and Eswar Prasad, 994, Are Prices Counercyclical? Evidence from he G-7, Journal of Moneary Economics 34, Cooley, Thomas J. and Lee E. Ohanian, 99, The Cyclical Behaviour of Prices, Journal of Moneary Economics 28, Correia, Isabel H., Joao L. Neves, and Sergio T. Rebelo, 992, Business Cycles from New Facs abou Old Daa, European Economic Review 36(2/3), Fiorio, Riccardo, and Tryphon Kollinzas, 994, Sylized Facs of Business Cycles in he G7 from a Real Business Cycles Perspecive, European Economic Review 38, King, Rober G., and Serbio T. Rebelo, 993, Low Frequency Filering and Real Business Cycles, Journal of Economic Dynamics and Conrol 7, Kim, Sunghyun Henry, Kose, M. Ayhan, and Michael G. Plummer, 2003, Dynamics of Business Cycles in Asia: Differences and Similariies, Review of Economic Dynamics 7(3), Mendoza, Enrique G., 995, The Terms of Trade, he Real Exchange Rae, and Economic Flucuaions, Inernaional Economic Review 36(), 0-37 Pedersen, Torben Mark, 998, How Long are Business Cycles? Reconsidering Flucuaions and Growh, Insiue of Economics, Universiy of Copenhagen, Discussion Paper No Ravn, Moren O., and Harald Uhlig, 997, On Adjusing he HP-Filer for he Frequency of Observaions, CEPR Working Paper The Conference Board and Groningen Growh and Developmen Cenre, May 2006, Toal Economy Daabase, hp://

18 A. Daa Appendix Table A repors sandard deviaions for differen smoohing parameers suggesed by he lieraure and discussed in secion I. Wih lile surprise, higher lambdas resul in higher sandard deviaions, hence, more erraic cyclical variaions wih he mos dramaic changes observable in ime series of real GDP, governmen expendiures, CPI, capial sock, dependen employmen and real wages. Table A Sandard Deviaion Variable Lambda=6.25 Lambda=25 Lambda=00 Real GDP Governmen Expendiures Privae Consumpion Invesmens Expors Impors Money Supply Real Ineres Rae CPI Inflaion Rae Capial Sock Toal Employmen Dependen Employmen Real Wage Rae Labor Produciviy Source: Own calculaions Table 2A describes correlaions of all relevan variables wih real GDP, deermining he variables cyclicaliy. Wih he excepion of he real ineres rae and CPI which appear non-significan wih a lambda of 00, no variable changes eiher significance or sign. Hence, a leas for he Ausrian economy, basic resuls and inferences are robus o he choice of differen smoohing parameers. Table 2A Correlaion wih Real GDP Variable Lambda=6.25 Lambda=25 Lambda=00 Governmen Expendiures Privae Consumpion * 0.748* * Invesmens * 0.732* * Expors 0.535* * * Impors * * * Money Supply Real Ineres Rae * * CPI * * Inflaion Rae Capial Sock * * * Toal Employmen * * 0.806* Dependen Employmen 0.60* 0.744* * Real Wage Rae * * 0.706* Labor Produciviy * * 0.76* Source: Own calculaions Aserix denoes significan a 5 % level

19 Resuls however significanly change once he effec of differen smoohing parameers on persisence is analyzed (able 3A). Signs as well as exen of persisence parameers change in almos all cases. Wih he excepion of he real ineres rae and he inflaion rae, all variables become more persisen wih higher lambdas. Table 3A Persisence Variable Lambda=6.25 Lambda=25 Lambda=00 Real GDP Governmen Expendiures Privae Consumpion Invesmens Expors Impors Money Supply Real Ineres Rae CPI Inflaion Rae Capial Sock Toal Employmen Dependen Employmen Real Wage Rae Labor Produciviy Source: Own calculaions As o he relevance of supply-side shocks for he Ausrian economy capured by correlaion coefficiens of real GDP wih eiher inflaion or CPI, able 4A shows ha he correlaion of inflaion wih real GDP is unaffeced by he applicaion of differen smoohing parameers. Wih he excepion of he very high lambda of 00, he correlaion of real GDP wih CPI remains unmodified wih differen smoohing parameers. Hence, irrespecive of he smoohness parameers applied, supply-side shocks seem o be of imporance for he Ausrian business cycles beween 970 and Table 4A Correlaion of real GDP wih derended and conemporaneous Inflaion or CPI Panel A. Filered Inflaion and Oupu Panel B. Filered CPI and Oupu λ = 6.25 λ = 25 λ = 00 λ = λ = 25 λ = 00 Lag Overall Overall Overall Overall Overall Overall * * * * * Source: Own calculaions Aserix denoes significan a 5 % level