DOCUMENT DE TRAVAIL N 468

Transcription

1 DOCUMENT DE TRAVAIL N 468 ASSESSING THE LOSSES IN EURO AREA POTENTIAL PRODUCTIVITY DUE TO THE FINANCIAL CRISIS Valérie Chouard, Daniel Fuenes Casro, Delphine Irac and Mahieu Lemoine December 2013 DIRECTION GÉNÉRALE DES ÉTUDES ET DES RELATIONS INTERNATIONALES

2 DIRECTION GÉNÉRALE DES ÉTUDES ET DES RELATIONS INTERNATIONALES ASSESSING THE LOSSES IN EURO AREA POTENTIAL PRODUCTIVITY DUE TO THE FINANCIAL CRISIS Valérie Chouard, Daniel Fuenes Casro, Delphine Irac and Mahieu Lemoine December 2013 Les Documens de ravail reflèen les idées personnelles de leurs aueurs e n'exprimen pas nécessairemen la posiion de la Banque de France. Ce documen es disponible sur le sie inerne de la Banque de France « Working Papers reflec he opinions of he auhors and do no necessarily express he views of he Banque de France. This documen is available on he Banque de France Websie

3 Assessing he losses in euro area poenial produciviy due o he financial crisis * Valérie Chouard, Daniel Fuenes Casro, Delphine Irac and Mahieu Lemoine * We are graeful o Annabelle Mourougane and seminar paricipans a he AFSE meeing 2013 and he Banque de France seminar for useful commens and suggesions. The views expressed in his paper do no necessarily reflec hose of Banque de France. Banque de France, 39 rue Croix des Peis Champs, Paris Cedex 01 France. Corresponding auhor, mahieu.lemoine@banque-france.fr. 1

4 Résumé Nous monrons dans ce aricle que la récene crise financière a rédui significaivemen la producivié globale des faceurs (PGF) poenielle des quare plus grands pays de la zone euro, e égalemen celle du rese de la zone euro. Nous avons uilisé une équaion en forme réduie de la PGF, fondée sur une approche développée dans Cahn e Sain-Guilhem (2010). Nos résulas empiriques monren que l impac permanen sur la PGF poenielle varie selon les pays enre -3,9 e -1,3 poins de pourcenage au T Lorsque l on prend en compe ces peres, l écar de PGF, défini comme l écar enre la PGF e son poeniel, évolue pour chaque pays de façon rès similaire au aux d uilisaion des capaciés de producion (TUC). De plus, dans le cas de la France, la prise en compe du TUC dans le modèle de la PGF perme de diminuer les révisions affecan l écar de PGF en emps quasi-réel. Mos clés : foncion de producion, producivié globale des faceurs, crise financière, uilisaion des capaciés de producion. Codes JEL : E22, E23, E32, O47. Absrac In his paper, we show ha he recen financial crisis has significanly affeced he poenial oal facor produciviy (TFP) of he four larges euro area economies, as well as ha of he res of he euro area. We used a reduced-form equaion of TFP, based on an approach recenly developed by Cahn and Sain-Guilhem (2010). Our empirical findings show ha he permanen impac on poenial TFP varies across counries from -3.9 poins o -1.3 poins in Q When hese losses are incorporaed, TFP gaps develop closely in line wih capaciy uilisaion raes (CUR). Moreover, in he case of France, including CUR in our TFP model improves he quasi real-ime reliabiliy of TFP gap esimaes. Key words: producion funcion, oal facor produciviy, financial crisis, capaciy uilisaion. JEL codes: E22, E23, E32, O47. 2

5 I. Inroducion The financial crisis coninues o generae significan uncerainy regarding he measuremen of he oupu gap (he gap beween acual and esimaed poenial oupu). Reliable measures of his gap are especially valuable o policy makers as a vial ool for moneary and fiscal policy. The breakdown of TFP beween a poenial componen and a cyclical one (he TFP gap) is crucial since he TFP gap conribues o he oupu gap proporionaely in mos sandard producion funcions. In his paper we invesigae he size and duraion of he impac of he crisis on poenial TFP. The heoreical lieraure and firm-level empirical lieraure show ha he effecs of recessions on oal facor produciviy are ambiguous. Indeed, poenial cleansing effecs, where crises eliminae less producive firms, may increase TFP levels (Caballero and Hammour, 1994). Bu oher facors may have a negaive impac on TFP, such as scarring effecs where young, innovaive firms have fewer incenives o coninue heir aciviies in an uncerain environmen (Ouyang, 2009), or sullying effecs, where he mos producive firms face igher financial consrains (Barlevy, 2002 and Barlevy, 2003). Long-erm invesmen and R&D spending may also be reduced in imes of ighened financial consrains, lowering oal facor produciviy (Aghion e al., 2012). According o he recen empirical lieraure on crises, he long-run impac of pas financial crises 1 on poenial oupu is a loss of around 1.5 o 2.4% on average, and of up o 4% for he mos severe crises (Furceri and Mourougane, 2012). Moreover, Oulon and Sebasia-Barriel (2013) show ha he long-run impac of pas recessions on labour produciviy is beween 0.8% and 1.1% on average. Noe ha hese assessmens mainly rely on narraive approaches à la Romer and Romer (1989), where he auhors consider a panel of counries ha suffered a financial crisis beween he 1960s and he 2000s, and where ha crisis was generally confined o a single counry or group of counries. The crises are proxied using dummy variables and he esimaed impac is more or less he average of all he counry-specific impacs. Since he las financial crisis was a global one, however, affecing all counries in he world simulaneously, i is likely ha is impac on growh will be more pronounced. The aim of his paper is o invesigae how he recen recession has affeced poenial oal facor produciviy in he euro area, ha is in France, Germany, Ialy, Spain and in he res of he euro area. Admiedly, poenial TFP and he TFP gap are unobservable variables which can only be 1 Excluding he financial crisis of

6 derived from heoreical and/or saisical models; for insance, he OECD uses an HP filer o idenify poenial TFP. However, we argue ha economiss have a heir disposal a key survey variable, he capaciy uilisaion rae (CUR), which conveys crucial informaion abou he TFP gap. Since any underuilisaion of capial sock leads o a decline in apparen TFP, i is economically inuiive o view he CUR as a srong deerminan of he TFP gap. Saisically, his variable also has a number of feaures ha make i exremely valuable from a pracical poin of view: 1) he CUR is observable and available for many counries, which faciliaes inernaional comparisons (see he EU harmonised Business Tendency Surveys); 2) CUR surveys are rapidly available (generally a he beginning of he following monh); 3) he CUR is no revised. By focusing on capaciy uilisaion, our approach builds heavily on Planas e al. (2013) bu in a rend-saionary framework à la Perron (1989). Our conribuion o he lieraure is herefore hreefold. We propose a new measure of TFP gaps closely correlaed o capaciy uilisaion raes. In he case of France, we also show ha including he CUR in our TFP model improves he quasi real-ime reliabiliy of TFP gap esimaes. Our second conribuion is ha we consider he precise iming of he crisis and quanify he size of he losses in poenial TFP. Lasly, hese losses are broken down by source (permanen losses, capial ageing impac and consrucion secor impac). II. TFP model In his secion, we exend a reduced-form equaion developed by Cahn and Sain-Guilhem (2010) o measure he cos of he recen financial crisis in erms of poenial euro area TFP. Economywide producion echnology is represened by a Cobb-Douglas-like producion funcion wih a consan reurn o scale on labour and capial. As described in Cahn and Sain-Guilhem (2010), he producion funcion can be expressed analyically as: Y = σe γ ~ 1 K α ( ) α N H, 0 < α < 1, where Y sands for oal GDP, K ~ for he sock of available producive capial, N for oal employmen, and H for per capia hours worked. The parameer α is he elasiciy of oupu wih respec o employmen, γ is he growh rae of a pure exogenous deerminisic echnical change and σ a scale facor. Here, available producive capial is ied up wih measured capial sock K and he age of capial according o: 4

7 ~ K = CUR. K. e ε ( τ ) where CUR is he capaciy uilisaion rae of he capial sock and τ, he age of capial assuming ha, hanks o capial-embodied echnological progress, one uni of invesmen shows a produciviy gain a each period amouning o (1 + ε), wih ε> 0. We compue he age of capial as a weighed average τ = l 1 j= 0 ( 1 δ ) I ~ K j j j, a each period. This formula is derived from he accumulaion of invesmen flows, where δ denoes he depreciaion rae of capial, 0 <δ < 1, and I denoes invesmen. 2 where G Therefore, he producion funcion can be expressed as: ( N H ) α 1 Y α = G K γ 1 α (1 α ) ε ( τ ) = e. CUR. e sands for he apparen oal facor produciviy (TFP). Thus, we can measure TFP as he Solow residual of he neoclassical model: G 1 α α = Y K ( N H ) (1) TFP measures he effeciveness wih which accumulaed facors of producion (capial K and labour NH) are used o produce oupu Y. Cahn and Sain-Guilhem (2010) deduce from he previous heoreical framework a reduced-form equaion of he TFP logarihm. Here, we modify heir TFP equaion in wo ways in order o ake ino accoun he impac of he financial crisis: we add an exogenous rend break break I+1 in T B,I+1 = 2008Q3 for all counries, in order o es he possibiliy of a long-run effec of he crisis on TFP growh; we add a permanen shock in he AR(1) residuals, in order o allow for a long-run impac of he financial crisis on he level of TFP. These modificaions lead o he following equaions, expressed in logarihmic form: I g ecur ecage + 1 = α + α + α + α + α u i i break i = 1 5,, (2a) u = u 1 + φ 1 ρ ) st C, ρ ( + ε (2b) j ~ j 2 Inuiively ( 1 δ ) I / K represens he remaining share in he available capial sock of capial of vinage j. 5

8 wih: ecur = ( cur cur), ecage = ( age age), break = ( T i B, i ) 1, T. CUR and he age of B, i capial sock are adjused by heir long-erm averages cur and age. 3 The parameer α 2 should be posiive, reflecing he fac ha TFP grows as domesic producion capaciies are used more inensively han usual. Moreover, a higher-han-average age of capial sock could have a negaive impac on TFP, so he parameer α 3 should be negaive. In his TFP model, g is a funcion of he CUR, he age of he sock of capial equipmen and a deerminisic rend wih I+1 breaks a daest B, i, i = 1 o I. The residual u capures he omied variables and is supposed o follow an AR(1) process, wih dampening facor ρ, driven by a Gaussian whie noise ε. A firs impac of he onse of he crisis in our model is capured hrough a long-lasing disurbance in he dynamics of u, which formally maerialises hrough 1 hereafer. s T C, a qualiaive indicaor of he crisis which is equal o 0 up o ime T C and o We reformulae equaions (2a) and (2b) as a single non-linear equaion, which we esimae wih non-linear leas squares (NLS): g ( 1 ρ) = α + ρ g + α ( ecur 1 ) + ( 1 ) + ( ( 1)) ρ ecur α ecage ecage 3 ρ α ρ 4 I α 5, i ( break ρ break ) + φ( 1 ρ) S + ε T (2 ) i= 1 i, i, 1, C According o his specificaion wih he qualiaive indicaor s T C, he parameer φ measures he impac of he crisis on he level of poenial TFP. The parameer 5, I + 1 associaed wih T C deecs a possible impac of he crisis on he growh of TFP. Le us noe ha he occurrence of α s T can be C inerpreed as progressively impacing he consan erm α 1 via he impulse response funcion of an AR(1) o a shock ha akes he form of a sep of ampliude φ. For he sake of simpliciy, we make he assumpion ha he financial crisis diffuses a he same rae ρ as he oher shocks affecing u dampen. 3 This ransformaion does no change heir elasiciy (i only changes he esimaed inercep). For he CUR, i allows us o disenangle he cyclical par relaed o he TFP gap from he long-run reference value of he CUR (measured by is mean) relaed o he poenial TFP. For he age of capial, his ransformaion would be useful, if we were ineresed in a long-run poenial TFP which would also exclude flucuaions in he age of capial. 6

9 Applying he Bai-Perron es o a linearised version of his equaion 4 for he rend coefficien wih a rimming facor of 10% and over he pre-crisis period (from 1960 o 2007), we find he following break daes: Q and Q for France; Q and Q for Germany; Q for Ialy; Q for Spain; Q for he res of he euro area. For all counries, we also es an addiional break a he dae of he Lehman bankrupcy (Q3 2008). 5 Finally, poenial TFP is derived from (2 ) as follows: 6 g I = α + α ecage + α rend * i= 1 α 5, i Break 1, i 1 ρ + φ S 1 ρl Now, in equaion (3) below, he TFP gap (gap which is equal o g-g*) is direcly linked o he CUR. An addiional erm depics AR(1) residuals ha remain afer fiing he gap wih he CUR: TC gap ε = α 2 ecur + (3) 1 ρl By over-differeniaing he variables, we can wrie his equaion in he spiri of an error correcion represenaion: gap = α ρ α + ε 2 ecur ( 1 )( gap 1 2ecur 1) (4) gap = gap gap where 1. Thus, we can inerpre he parameer 1- ρ as an adjusmen speed of he TFP gap o he CUR arge. III. Daa The daa cover he period Q o Q for France, Germany, Ialy, Spain and he res of he euro area. Daa on labour markes, oupu and invesmen are from naional saisical insiues and Eurosa daabases. Capial sock K is compued as he sum of wo ypes of capial: KM, capial for machinery and equipmen and KB, capial for consrucion. These socks of capial are derived from he sandard permanen invenory equaion. 7 Capial for machinery and equipmen is deermined using invesmens in machinery and equipmen wih an obsolescence rae of 0.024, 4 If we relax he consrains relaing o he coefficiens of equaion (2 ) by regressing he TFP on an inercep, is lag and lags of all regressors (ecur, ecage, he ime rend, breaks and he qualiaive indicaor), we ge a linear equaion and we can apply he Bai-Perron es. Anoher possible approach would have been o es break daes wih he non-linear equaion, by using he es of Boldea and Hall (2013). 5 Since his break dae is exogenous, esing is significance does no require ess of endogenous break daes (Bai-Perron or Boldea-Hall ess) and we simply use he sandard disribuion of he Suden es. 6 L is he lag operaor. 7 K = KM +KB wih KM = ( )*KM -1+IM and KB = ( )*KB -1+IB, a each dae. 7

10 equivalen o a lifeime of nearly 10 years, and capial for consrucion is deermined using invesmens in consrucion wih an obsolescence rae of 0.004, equivalen o nearly 60 years (see Cahn and Sain-Guilhem, 2010). Age of capial is calculaed using hisorical ime-series on invesmens in machinery, vehicles and equipmen. 8 Figure 1 shows he evoluion of he age of machinery and equipmen capial for France, Germany, Ialy, Spain and he res of he euro area. CUR series are colleced from regular naional surveys on indusrial secors. The oupu elasiciy of labour α is calibraed as he average shares of wages in GDP. IV. Impac of he financial crisis on poenial TFP The esimaion resuls obained using he previous model for TFP are shown in Table 1 and summarised in Table 2. Table 1 presens he coefficien esimaes of equaion (2 ) for each of he economies under examinaion. 9 We find ha inroducing an indicaor of he crisis maers: is parameer φ is significan for all he economies in quesion. This resul shows ha, for all counries, permanen losses in poenial TFP are subsanial. The parameer α 3 indicaing he age of capial is significan and negaive for France, Germany and Spain. This represens a loss resuling from capial ageing. The TFP rend breaks assessed via he parameers α 5, i show ha he dynamics of TFP rends differ across counries. The negaive signs for α 5, 1 and α 5, 2 reflec decreases in TFP rends afer he breaks. For France, Germany and Ialy in paricular, TFP rends reached 1.0%, 0.7% and -0.1% respecively before he crisis. For Spain, however, he parameer α 5, 3 capures a significan posiive impac from he crisis on he growh of TFP in Q In Spain, he TFP rend reached 0.7% over he period , afer 0.2% over and 1% over These hree sub-periods correspond o srucural breaks in he Spanish consrucion secor (he percenage share of his secor in value added follows he same paerns as hese TFP rends). An explanaion for he increasing produciviy rend over could be he disappearance of he low-produciviy firms ha were creaed in he consrucion secor during he real esae bubble. For oher counries, we wondered 8 Hisorical daa on invesmen are derived from Maddison s daabases for France and Germany, from Prados (2003) for Spain and from Baffigi (2011) for Ialy. 9 We have performed robusness ess on residuals which sugges homoscedasiciy and non-auocorrelaion. These addiional resuls are available upon reques from he auhors. 8

11 wheher bubbles had also occurred before he crisis, bu we do no find any significan break in TFP growh over his period. This issue is discussed in Secion IV. The coefficien ρ represens he persisence of TFP shocks. The smaller his parameer, he faser TFP reurns o is mean value. We find ha losses are more persisen in France and in Ialy han in oher counries such as Germany and Spain. Impulse responses of (Figure 2). s T also illusrae his finding C Table 2 shows ha he oal impac on poenial TFP sands a 4 percenage poins for he euro area in Q Ineresingly, his impac sems from a permanen impac (-2.8pp), a capial ageing impac (-1.4pp) and a consrucion secor impac (+0.2pp). Permanen losses reflec he longrun impac of he financial crisis. They appear o be greaer in France (-3.3pp), in Ialy (-3.7pp) and in he res of he euro area (-3.9pp). The losses caused by capial ageing resul from depressed invesmen in capial goods in France, Germany and Spain. There is no impac from he age of capial in Ialy and he res of he euro area (REA) since his variable is no saisically significan. The posiive impac from he consrucion secor comes from Spain (wih an impac of +2.1pp on Spanish TFP). Figures 3 and 4 depic he TFP gap and poenial TFP growh for he economies under examinaion. In Figures 3a o 3e, he magniude of he permanen impac is observed in he TFP gap curve. Afer being adjused for he permanen impac, he TFP gap is more closely correlaed o he CUR, which now has a significan impac on TFP, excep in he case of he REA (according o esimaes of α3 in Table 1). Figures 3a o 3e also compare he TFP gap for our TFP model ( TFP gap wih producion funcion ) wih he TFP gap obained using he Hodrick-Presco (HP) filer. We can see ha, for boh mehods, he gaps obained and herefore he loss of poenial TFP, are comparable, paricularly over he end of he period. Figures 4a o 4e show he growh of HP TFP rends and ha of our poenial TFP. The resuls reveal ha HP esimaes of TFP rend had already begun o decrease before he crisis in each euro-area counry (e.g. in Q for Germany), while our esimaes of poenial TFP growh sar o decrease a he dae of he crisis (Q3 2008). In Figures 5a o 5f, he loss in poenial TFP is broken down according o he conribuion of each of he hree impacs lised in Table 2: he impac of he crisis, of capial ageing and of he consrucion secor. We can see he sharp decrease in TFP and poenial TFP afer he end of

12 This loss resuls mainly from he permanen impac caused by he crisis, followed by he impac of capial ageing. V. Robusness and quasi real-ime reliabiliy How can we ensure ha he analysis is no affeced by a diagnosis error concerning he precrisis period? Our esimae of he impac of he crisis on poenial TFP hinges crucially on he breakdown of effecive TFP beween is cyclical componen and is rend during he pre-crisis period. Any overesimaion of poenial growh before he crisis may resul in an underesimaion of he precrisis (posiive) oupu gap and an overesimaion of he pos-crisis loss of poenial oupu. The quesion of a possible negaive break in TFP growh in he decade preceding he crisis has been exensively discussed. Bénassy-Quéré e al. (2009) argue ha beween 1999 and 2007 Cheap credifaciliaed deb financed invesmen in real esae and financial asses, and conribued o excessive risk aking. From a macroeconomic sandpoin also, his crisis has been a crisis of leverage. Many oher expers develop he same argumen (Turner Review, Financial Services Auhoriy, 2009; De Larosière (2009) repor prepared a he reques of he European Commission). Indeed, he main sylised fac of he 2000s was he generous availabiliy of credi. Table 3 shows he significan rise in euro area deb levels beween 1999 and However, we argue ha he hypohesis of credi-fuelled economic growh is rejeced by our model. Firs, we noe ha Suden ess confirm he sabiliy of he TFP rend, since here was no negaive break in TFP rend for he economies under examinaion in he period preceding he crisis, excep in he case of Spain where he es deecs a posiive break. Secondly, a more convincing argumen is ha no prolonged increase of CUR can be observed during he pre-crisis period for hese economies (excep in he case of Spain, increases in CUR lased less han wo years on average during he pre-crisis period) and no insabiliy in he relaionship beween he CUR and he oupu gap is observable. The CUR increases observed during bubble episodes in oher counries appear o have been a sympom of excess demand: for example, in Japan, before he speculaive bubble a he beginning of he 1990s, or in he Unied Kingdom, before he crisis a he end of he 1990s, he CUR increased coninuously over he four years preceding he even (OECD, 1996; OECD, 1997). Consequenly, for he economies sudied here, we can consider ha excess demand is no maerialised. Figure 6 shows he residuals of he TFP model for he differen counries in quesion, which deec 10

13 poenial insabiliies in he relaionship (4) beween he CUR and he TFP gap. Excep for he res of he euro area, he residuals of he TFP model for he differen counries remain small (heir sandard deviaion is small relaive o ha of he TFP gap). Moreover, Suden mean ess indicae ha heir mean is zero over he esimaion period, confirming he sabiliy in he relaionship beween he CUR and he TFP gap. Like Planas e al. (2013), bu in a differen framework, we find ha including he CUR in our TFP model improves he quasi real-ime reliabiliy 10 of TFP gap esimaes. We ieraively esimae he TFP equaion over samples going from a fixed sar dae o a rolling end dae. The end dae rolls over he Q Q period, i.e. afer he financial crisis. The resuls show ha sandard errors of quasi real-ime revisions over he pos-crisis period decrease when he CUR is included, as in our TFP model. For example for France, he sandard errors decrease from 0.64 (when he CUR is no aken ino accoun) o 0.23 (when he CUR is included, as in our TFP model). Anoher issue is ha errors in he measuremen of TFP migh affec our esimaes of TFP losses. Working on he Unied Kingdom, O'Mahony and Oulon (1994) sress he imporance of he depreciaion rae for measuring produciviy. In Table 4, as a firs aemp o invesigae he sensiiviy of our resuls o depreciaion raes, we examine he impac of using higher and lower values for hese raes. We use he ranges proposed in O Mahony and Oulon (1994) as he lower and upper bounds for our raes (from 4 o 14.27% for equipmen and from 1.67% o 3.68% for buildings). The figures in Table 4 show he variaion in he permanen losses beween he baseline scenario and hese wo alernaive esimaes. We show ha permanen losses caused by he crisis change by less han 0.4 a he euro area level wih hese alernaive assumpions. VI. Conclusion In his aricle we exend a reduced-form equaion developed by Cahn and Sain-Guilhem (2010) o measure he cos of he recen financial crisis in erms of poenial TFP in he euro area. We find ha euro area economies have seen significan losses in heir poenial TFP. In our model, TFP gaps closely follow he CUR. Moreover, in he case of France, including he CUR in our TFP model improves he quasi real-ime reliabiliy of TFP gap esimaes. 10 We use he erminology of Orphanides and Van Norden (2002), who disinguish he quasi real-ime reliabiliy from he realime one: he quasi real-ime reliabiliy is only based on revision relaed o he sample availabiliy, while he real-ime reliabiliy also akes ino accoun daa revisions. 11

14 In fuure research, several facors migh revise our diagnosis. Firs, we would like o find economic facors ha could explain he heerogeneiy of losses wihin he euro area. Second, wih more hindsigh, we should improve our ess of rend breaks (coefficien esimaes of a break in TFP growh for France, Germany, Ialy and he res of euro area were no significan during he financial crisis). This resul can be inerpreed as he crisis having no impac on poenial TFP growh. Perhaps a larger sample, in he years o come, may improve heir significance. 12

15 References Aghion, P. Askenazy, P. Berman, N. Cee, G. and Eymard, L. (2012) Credi Consrains and he Cyclicaliy of R&D Invesmen: Evidence from France, Journal of he European Economic Associaion 10(5), Andrews, D. W. K. (1993) Tess for Parameer Insabiliy and Srucural Change wih Unknown Change Poin, Economerica, 61, Baffigi, A. (2011) Quaderni di Soria Economica, Ialian Naional Accouns, , Ocober. Bai, J. and Perron, P. (1998) Esimaing and Tesing Linear Models wih Muliple Srucural Changes, Economerica, 66, Barlevy, G. (2002) The Sullying Effec of Recessions, Review of Economic Sudies 69, Barlevy, G. (2003) Credi Marke Fricions and he Allocaion of Resources Over he Business Cycle, Journal of Moneary Economics 50, Bénassy-Quéré, A. Coeuré, B. Jacque, P. and Pisani-Ferry, J. (2009) The Crisis: Policy Challenges and Policy Responses, Bruegel Working Paper, December. Boldea, O. and Hall, A.R. (2013). Esimaion and Inference in Unsable NLS Models. Journal of Economerics, 172(1), pp Bonleu, A. Cee, G. and Horny, G. (2013) Capial Uilisaion and Reiremen, Applied Economics, 45, pp Caballero, R. and Hammour, M. (1994) The Cleansing Effec of Recessions, American Economic Review, 84, Cahn, C. and Sain-Guilhem, A. (2010) Poenial Oupu Growh in Several Indusrialized Counries: A Comparison, Empirical Economics, 39(1), De Larosière, J. (2009) Repor of he High-Level Group on Financial Supervision in he EU, February. Financial Services Auhoriy (2009) The Turner Review: A regulaory response o he global banking crisis. Furceri, D. and Mourougane, A. (2012) The Effec of Financial Crises on Poenial Oupu: New Empirical Evidence from OECD Counries, Journal of Macroeconomics 34,

16 Maddison, A. (1993) Sandardized Esimaes of Fixed Capial Sock: a Six Counry Comparison, in Essays on Innovaion Naural Resources and he Inernaional Economy, Innovazione E Maerie Prime, OECD (1996) Economic Surveys: Unied Kingdom 1995, May. OECD (1997) Economic Surveys: Japan 1997, November. Orphanides, A. and Van Norden, S. (2002) The Unreliabiliy of Oupu-Gap Esimaes in Real Time. The Review of Economics and Saisics, vol. 84(4), pages O'Mahony, M.and Oulon, N. (1994) Produciviy and Growh, Cambridge Books, Cambridge Universiy Press. Oulon, N. and Sebasia-Barriel, M. (2013) Long and Shor Term Effecs of he Financial Crisis on Labour Produciviy, Capial and Oupu, Working Paper, Bank of England. Ouyang, M. (2009) The Scarring Effec of Recessions, Journal of Moneary Economics 56, Perron, P. (1999) The Grea Crash, he Oil Price Shock and he Uni Roo Hypohesis, Economerica 57, Planas, C. Roeger, W. and Rossi, A., M. (2013) The Informaion Conen of Capaciy Uilizaion for Derending Toal Facor Produciviy, Journal of Economic Dynamics and Conrol 37, Prados de la Escosura, L. (2003) El Progreso Economico de España , Fundacion BBVA. Romer, C. and Romer, D. (1989) Does Moneary Policy Maer? A New Tes in he Spiri of Friedman and Schwarz, NBER Macroeconomics Annual 1989, 4,

17 Figure 1: Age of capial for France, Germany, Ialy, Spain and REA Figure 2: Impulse response funcion of TC 0 s Q2 2008Q4 2009Q2 2009Q4 2010Q2 2010Q4 2011Q2 2011Q4 2012Q2 2012Q Germany France Ialy Spain Res of he zone Euro zone Noe: Each curve represens he impac of he crisis on poenial TFP hrough he variable s TC in each economy. In Q3 2008, he impac is equal oφ S2008Q3. In he long-run, i is equal o φ /(1-ρ). For example, France has experienced a loss in is poenial TFP of 3.3 poins in he long run. 15

18 Table 1: Resuls of TFP esimaions beween 1960 and 2012 REA = Res of he Euro Area. Figures in ialics are sandard errors in coefficien esimaes. * Non-saisically significan. Excep for he CUR of REA, all coefficiens are significan a he 5% level. Esimaions are robus o auocorrelaion and heeroscedasiciy ess. ** Coefficiens (and sandard errors) are muliplied by 400 in order o obain he percenage annualised growh rae for TFP rend. For example, for France, he TFP rend growh reached 2.7% before 1973, 2.0% over and 1.0% over Table 2: Impac of he financial crisis on euro-area poenial TFP in Q

19 Figure 3a: TFP gap for France Figure 3b: TFP gap for Germany Figure 3c: TFP gap for Ialy Figure 3d: TFP gap for Spain Figure 3e: TFP gap for he res of he area 17

20 Figure4a: Growh of Poenial TFP-France Figure4b: Growh of Poenial TFP-Germany Figure4c: Growh of Poenial TFP-Ialy Figure4d: Growh of Poenial TFP-Spain Figure4e: Growh of Poenial TFP-Res of he area 18

21 Figure 5a: TFP and poenial TFP for France Figure 5b: TFP and poenial TFP for Germany Figure 5c: TFP and poenial TFP for Ialy 19

22 Figure 5d: TFP and poenial TFP for Spain Figure 5e: TFP and poenial TFP for he res of he area Figure 5f: TFP and poenial TFP for he euro area 20

23 Figure 6: Residuals of TFP model and basic saisics Q1 1993Q1 1994Q1 1995Q1 1996Q1 1997Q1 1998Q1 1999Q1 2000Q1 2001Q1 2002Q1 2003Q1 2004Q1 2005Q1 2006Q1 2007Q1 2008Q1 2009Q1 2010Q1 2011Q1 2012Q1 Germany France Ialy Res of he zone Spain Euro zone In % France Germany Ialy Spain REA Euro area Mean Sd. dev Min Max Noe: The TFP model residuals for he differen counries remain small (excep for REA, he sandard deviaions are small compared o he TFP gap sandard deviaions which are 0.98 for France; 1.09 for Germany; 1.11 for Ialy; 0.38 for Spain and 1.14 for REA. Moreover, Suden mean ess indicae ha he mean of hese residuals is zero over he period of esimaion), confirming he sabiliy in he relaionship (equaion 4) beween he CUR and he TFP gap. Table 3: The rise in privae deb in he euro area beween 1999 and 2007 Ineres-rae liabiliies (loans and non-equiy securiies) as % of GDP Nonfinancial Financial Governmen Households TOTAL Source: ECB, auhor calculaions. Table 4: Changes in permanen losses wih lower and higher depreciaion raes In percenage poins Low depreciaion High depreciaion France Germany Ialy Spain REA Euro area Noe 1: we use he esimaes of O'Mahony and Oulon (1994) as he upper and lower bounds for our depreciaion raes. These range from 4 o 14.27% for equipmen and from 1.67% o 3.68% for buildings. Noe 2: we calculae he changes in permanen losses wih respec o our baseline esimaes based on depreciaion raes of 9.99% for equipmen and 1.64% for buildings. 21

24 Documens de Travail 460. J. Carluccio and M. Bas, The Impac of Worker Bargaining Power on he Organizaion of Global Firms, November F. Canova, F. Ferroni and C. Mahes, Choosing he variables o esimae singular DSGE models, November S. Avouyi-Dovi, G. Horny and P. Sevesre, The dynamics of bank loans shor-erm ineres raes in he Euro area: wha lessons can we draw from he curren crisis?, November D. Fougère, C. Golfier, G. Horny e E. Kremp, Quel a éé l impac de la crise de 2008 sur la défaillance des enreprises?, Novembre T. Duprey, Heerogeneous Banking Efficiency: Allocaive Disorions and Lending Flucuaions, November L. Arrondel, M. Debbich and F. Savignac, Financial Lieracy and Financial Planning in France, December P. Anipa, Fiscal Susainabiliy and he Value of Money: Lessons from he Briish Paper Pound, , December D. Bellas and Vincen Vicard, French firms expors during downurns: evidence from pas crises, December V. Chouard, D. Fuenes Casro, D. Irac and M. Lemoine, Assessing he losses in euro area poenial produciviy due o he financial crisis, December 2013 Pour accéder à la lise complèe des Documens de Travail publiés par la Banque de France veuillez consuler le sie : For a complee lis of Working Papers published by he Banque de France, please visi he websie: Pour ous commenaires ou demandes sur les Documens de Travail, conacer la bibliohèque de la Direcion Générale des Éudes e des Relaions Inernaionales à l'adresse suivane : For any commen or enquiries on he Working Papers, conac he library of he Direcorae General Economics and Inernaional Relaions a he following address : BANQUE DE FRANCE Labolog Paris Cedex 01 él : 0033 (0) ou ou ou HU1404-u@banque-france.fr