Fiscal Convergence, Business Cycle Volatility and Growth *

Transcription

1 Fscal Convergence, Busness Cycle Volatlty and Groth * Davde Furcer European Central Bank and Unversty of Palermo Abstract Ths paper analyzes the effects of fscal convergence on busness cycle volatlty and groth. Usng a panel of EMU and OECD countres and 40 years of annual data, e fnd that countres th smlar government budget postons tend to have smoother busness cycles. That s, fscal convergence (n the form of persstently smlar ratos of government surplus/defct to GDP) s systematcally assocated th smoother busness cycles. We also fnd evdence that reduced busness cycle volatlty through hgher fscal convergence stmulates groth. Our emprcal results are economcally and statstcally sgnfcant, and robust. JEL: E6, H50, O40. Keyords: Fscal Convergence, Busness Cycle Volatlty, Economc Groth. * I am grateful to Antono Afonso, Georgos Karras, Ad van Ret, Jurgen von Hagen and Jurgta Anuprevcute for helpful comments. I alone am responsble for any errors. The opnons expressed heren are those of the authors and do not necessarly reflect those of the ECB or the Eurosystem. Malng Address: European Central Bank, Drectorate General Economs, Kaserstrasse 9, D-603 Frankfurt, Germany. Emal: davde.furcer@ecb.nt / furcer@economa.unpa.t.

2 . Introducton In the last decade, several orks n the lterature have tred to dentfy the effects of fscal polcy on economc performances. Most of these orks have concentrated on to man potental macroeconomc effects: busness cycle volatlty and groth. Whle there s a broad consensus that hgher levels of government expendture are assocated th hgher groth rates, there s lttle consensus on the sgn of the effects of fscal polcy on busness cycle volatlty. On the one hand, governments can smooth out busness cycle fluctuatons by the use of dscretonary changes n fscal polcy (such expansonary spendng and tax cuts n recessons and contractonary polcy n expansons) and by the use of automatc stablzers. On the other hand, fscal polcy tself mght be a source of busness cycle fluctuatons and exacerbate macroeconomc volatlty 3. Moreover, countres are subect to both symmetrc and asymmetrc shocks. If countres are subect to persstent asymmetrc shocks or symmetrc shocks, and those are partally offset by dscretonary fscal polcy or automatc fscal stablzers, then fscal dvergence can, n prncple, be assocated th greater busness cycle synchronzaton and reduced busness cycle volatlty 4. From another pont of ve, f fscal polcy s procyclcal, fscal dvergence ll be assocated th reduced busness cycle synchronzaton and hgher busness cycle volatlty. Thus, the queston s ultmately emprcal. See, for example, Ashauer (989), Barro (990), Evans and Karras (994a,994b), Fölster and Henrekson (00), Munnel (99). See, for example, Sachs and Sala--Martn (99), Bayoum and Masson (995), Asdrubal et al. (996), von Hagen (998), Furcer (005). 3 For example, Fatás and Mhov (003, 004) sho that governments that use fscal polcy aggressvely nduce sgnfcant macroeconomc nstablty, and that dscretonary changes n fscal polcy tend to amplfy output volatlty. 4 In partcular, n monetary unons, hgher busness cycle syncronzaton reduces output volatlty. See, for example, Furcer and Karras (006).

3 Whle the effect of fscal convergence on busness cycle synchronzaton has been recently nvestgated by Darvas et al. (005) 5, no emprcal study to our knoledge has analyzed the effect of fscal convergence on busness cycle volatlty. The frst am of ths paper s to try to fll ths gap. In addton, e also ant to asses the effect that fscal convergence has on groth through reduced (or ncreased) output volatlty. From a theoretcal pont of ve, the relatonshp beteen macroeconomc volatlty and economc groth s a very ntrcate one. On the one hand, rreversble nvestment theores suggest that countres th hgher volatlty ll have loer levels of nvestment and as a consequence loer groth. For example, Bernanke (983), Pndyck and Solmano (993), and Blackburn and Varvargos (005) sho that n models th nvestment rreversblty and fnancal frctons, hgher uncertanty regardng nvestment prces ll determne a loer level of nvestment and groth. On the other hand, hgh groth economes mght be based on rsky technologes and therefore may experence sharp shfts n economc volatlty. These arguments have been tested by Ramey and Ramey (995) n ther emprcal study of the lnk beteen volatlty and groth. The authors found convncng emprcal evdence that busness cycle volatlty s harmful for groth. The robustness of ths relaton has been confrmed by more recent orks 6. Although the queston e approach here s smlar, our man concern s hether fscal dvergence can be responsble for busness cycle volatlty that s harmful to groth. The remander of the paper s organzed as follos. In Secton To, e present a smple theoretcal model shong that the sgn odf the relaton beteen fscal 5 In partcular, the authors found that countres th smlar government budget postons tend to have busness cycles that fluctuate more closely. 6 See, for example, Furcer (007) and Imbs (007). 3

4 convergence and output volatlty s ambguous. In Secton Three, e explan our methodology to assess the mpact of fscal convergence on busness cycle volatlty and groth. In Secton Four, e present the emprcal analyss and dscuss ts results. Secton Fve summarzes the paper s man fndngs and provdes some polcy mplcatons.. Theoretcal Model The theoretcal frameork follos the models proposed by Kydland and Prescott (977), and Barro and Gordon (983). In partcular, e consder an open economy n hch the polcymakers of each economy mnmze the follong loss functon: { ( y k ) } d L = () here y denotes output (n devatons from the trend), d s defct, s the relatve eght of output devatons, from ts target k. Ths target s assumed to be greater than zero because of dstortons such as mperfect competton or taxes. For each (open) economy e assume that the aggregate demand s gven by: y = A θ d ε ε () here A s the autonomous component of aggregate demand, ε and ε are the domestc and foregn shocks hch e assume to have zero mean and varance equal respectvely to σ and σ, and s the effect of foregn shock on domestc output. ll be: Mnmzng () th respect to (), n equlbrum the optmal level for the defct d ( k A) ε ε = θ (3) 4

5 5 Substtutng ths n () and computng the varance of output, t ll result that output volatlty ll be: y ε ε θ σ θ σ θ, cov var = (4) In order to assess the mpact of fscal convergence on output volatlty for each economy e rearrange (3) and express the country domestc shock as functon of the domestc level of defct and the foregn shock. d a k θ ε ε = ) ( (5) Successvely, gven the symmetrcal equlbrum for each economy, by substtutng for the foregn shock n (5), e can express each domestc shock as functon of domestc and foregn defct: d d A k = ) ( θ θ ε (6) Ths ll lead to the follong varance: d d d d, cov = θ θ σ θ σ θ σ (7) and covarance: d d d d, cov, cov = θ θ σ θ σ θ ε ε (8) Analyzng (4) together th (7) and (8), e can see that the effect of fscal coordnaton on output volatlty s ambguous. From one hand, assumng that the foregn shock s postvely transmtted > 0 and the fscal defct s

6 θ countercyclcal > 0, hgher defcts convergence ( ( d, ) cov ll decrease the varance of country specfc shocks, and thereby ll reduce output volatlty. On the other hand, assumng that the foregn shock s postvely transmtted ( > 0) θ fscal defct s countercyclcal > 0 d and the, hgher defcts convergence ( cov ( d, ) ll ncrease the covarance beteen domestc and foregn shock, and thereby ll ncrease output volatlty. Summarzng, the sgn of the relaton beteen fscal coordnaton and output volatlty s ambguous, and t ll further depend on the transmsson of the foregn shock and on the countercyclcalty of the defcts n each country. Thus, the queston remans ultmately emprcal. d 3. Emprcal Methodology 3. Estmaton Strategy The am of our paper s to nvestgate the effects of cross-country dfference n fscal polcy on busness cycle volatlty and groth. In lne th Darvas et al. (005), e ll use several measure of fscal dvergence. Our prmary measure of fscal dvergence s the dfference beteen countres n the general government budget surplus () or defct (-), measured as a percentage of natonal GDP. Takng the average of ths over a decade of annual data yelds our measure of fscal dvergence beteen countres: FscalDvergence τ.*σ τ ( Budg t - Budg t ) (9) 6

7 Fnally, to get a measure of fscal dvergence for each country, e take the average of ths measure over the set of countres (th = n-): FscalDvergence τ Σ FscalDvergence τ (0) here Budg t s the general government budget surplus () or defct (-) at tme t expressed as a percentage of nomnal GDP for country. A larger value of FscalDvergence corresponds to hgher average dvergence beteen the fscal postons of the to countres over a long perod of tme. In the same ay, a hgher value of FscalDvergence means that the country has a fscal poston very dssmlar to the rest of the other countres. We obtan the output busness cycle measures by detrendng the seres of real GDP. Four dfferent methods are used to detrend the output seres of each country and estmate ts cyclcal component. The frst measure s smple dfferencng (groth rate of the real GDP). The second and the thrd methods use the Hodrck-Prescott (HP) flter. The second method conssts of usng the value recommended by Hodrck and Prescott (997) for annual data for the smoothness parameter ( λ ) equal to 00. The thrd method consders the smoothness parameter ( λ ) equal to 6.5. In ths ay, as ponted out by Ravn and Uhlg (00), the Hodrck-Prescott flter produces cyclcal components comparable to those obtaned by the Band-Pass flter. The fourth method makes use of the recently popular Band-Pass (BP) flter proposed by Baxter and Kng (999), and evaluated by Stock and Watson (999) and Chrstano and Ftzgerald (003) that also compares ts propertes to those of the HP flter. Whle mnor dfferences among the results obtaned by the three flters are not dffcult to detect (for example, dfferencng generally produces the most volatle seres, hle the BP produces the smoothest ones), 7

8 the man characterstcs are remarkably smlar. Ths robustness ll be formally assessed by the estmatons of the emprcal secton. In practce, e measure GDP busyness cycle volatlty for each country and for each decade as the standard devaton of the country s cyclcal component over the relevant decade (σ τ ). Thus, n order to estmate the effect of fscal dvergence on busness cycle volatlty, our benchmark regresson ould be: σ τ =α FscalDvergence τ γ X σ ε τ. () Where, σ τ and FscalDvergence τ denote our measures of busness cycle volatlty and fscal dvergence for country over the decade τ, respectvely. X σ s a set of control varables ncludng: ) openness, measured as the natural log of the GDP s share of total exports and mports (lnopen); ) country sze, measured as the natural log of the total populaton (lnpop). In fact, both varables are found to have an extremely sgnfcant effect on busness cycle volatlty, and ther excluson could cause serous problems of omsson varables 7. The obect of nterest of our frst analyss s the slope coeffcent. A postve estmate of ndcates that a decrease n fscal dvergence (or an ncrease n fscal convergence) s assocated th reduced busness cycle volatlty. That s, fscal polcy convergence s lnked to smoother busness cycles. As e dscussed n the prevous secton, busness cycle volatlty has a sgnfcant and negatve effect on groth. Thus, n prncple, a hgher fscal convergence could be helpful for groth through reducton of busness cycle volatlty. The second obect of 7 See Furcer and Karras (007a, 007b). 8

9 our analyss s to test f ths effect exsts and ts relatve magntude. For ths purpose, e ll consder the follong groth equaton: g τ =θ δσ τ γ X g ζ τ () Where, g τ and σ τ denote our measures of real GDP groth and busness cycle volatlty for country over the decade τ, respectvely. X g s a set of controls varables ncludng those varables that Levne and Renelt (99), applyng the Extreme Bound Analyss ntally proposed by Leamer (983), found to be robust cross-country groth correlates: () the average nvestment share of GDP (nvestment); () the ntal log of GDP per capta(y o ); () ntal human captal (human captal) ; and (v) the average groth rate of the populaton (n). Thus, n order to estmate the effect of fscal convergence on groth, frst e ll estmate the effect of fscal convergence on busness cycle volatlty and then the effect of (nstrumented) busness cycle volatlty on groth. 3. Data Set In ths paper, e focus on tenty-one OECD countres. The countres ncluded n the analyss are eleven of the ffteen Euro Area members (excludng Luxembourg), thereafter ndcated as EMU (Austra, Belgum, France, Fnland, Germany, Greece, Italy, Ireland, the Netherlands, Portugal, Span), and the other ten OECD countres: Australa, Canada, Denmark, Japan, Ne Zealand, Noray, Seden, Stzerland, the UK and the Unted States. The analyss of ust these to samples of countres offers several advantages. Frstly, a longer span of data s avalable than for a broader set of countres, ncludng, for 9

10 nstance, developng countres. Secondly, data qualty and cross-country comparablty are also lkely to be of a hgher standard for the OECD, and ths s extremely mportant hen e consder the measure of busness cycle volatlty, snce volatlty ould ncrease n presence of measurement errors. Thrdly, as argued by Grer and Tullock (989), data from the OECD and the rest of the orld do not share a common set of coeffcents n cross-country groth regressons and thus should not be pooled. Fscal data, n terms of budget surplus () or defct (-), measured as a percentage of natonal GDP, are obtaned from the OECD Economc Outlook database and from Andre Rose s ebste. Real GDP, real GDP per capta groth, average nvestment, openness and populaton are taken from the Heston-Summers-Aten (006) dataset (Penn World Table 6.). Human captal s taken from the Barro-Lee (00) data set Results 4. Fscal Convergence and Busness Cycle Volatlty We start our nvestgaton on the effects of fscal convergence on busness cycle volatlty, analyzng the smple correlaton beteen budget dvergence and busness cycle volatlty. Fgure provdes a scatter plot of busness cycle output volatlty (measured by the natural logarthm of the standard devaton of the GDP s annual groth rate) aganst budget dvergence (measured by natural logarthm of the budget balance devaton) relatve to the full set of countres, for the four decades: , , , and Fgure exhbts the relaton beteen these to varables. In partcular, the estmate of ths smple bvarate relaton s: 8 See Appendx for a detaled descrpton of all the varables and ther avalablty. 0

11 σ τ = FscalDvergence τ (-9.79) (3.5) th R =0.3, and t statstcs are n parenthess. The relatonshp s clearly postve and statstcally sgnfcant. Moreover, t does not seem to be affected by outlers. In partcular, ths estmate suggests that a one percent ncrease of fscal dvergence translates nto 0.6 percent ncrease of busness cycle volatlty. Or that, n other ords, a one percent ncrease of fscal convergence determne a reducton of 0.6 percent n busness cycle volatlty. In Fgure, e repeat ths smple exercse, ncludng (the natural logarthm of) openness and populaton as control varables. In partcular, the assocated estmate s: σ τ = FscalDvergence τ (-4.3) (4.35) th R =0.9, and t statstcs are n parenthess. The relatonshp s clearly stll postve and statstcally sgnfcant. Moreover, also n ths case t does not seem to be affected by outlers, and the effect of fscal convergence on busness cycle volatlty s almost unchanged. No e proceed th more formal statstcal evdence. Table reports the estmated slope coeffcent () of Fscal Dvergence, along th the assocated t-statstcs n parentheses for several specfcatons of equaton (). The Table s four columns correspond to the OLS and country fxed effect estmates for both the EMU and the OECD countres. The detrendng method employed for the results n the table s dfferencng (groth rate of GDP), addtonal results obtaned usng the other detrendng methods dscussed n the prevous secton are presented n Appendx.

12 The fve ros of Table correspond to seven dfferent measure of budget dvergence: () Average of absolute value of budget devatons form the Maastrcht reference value; () Absolute value of budget devatons form the Maastrcht reference value; () Average absolute value of budget balance dfferental; (v) Absolute value of average budget balance dfferental; (v) Standard devaton of budget dfferental. Focusng on the frst column, t can be readly seen that the relaton beteen fscal dvergence and busness cycle volatlty s postve and statcally sgnfcant: the hgher s the level of fscal convergence, the less volatle s the busness cycle. Ths result holds for each of the fve measures of fscal dvergence. A smple average of the seven estmates s Ths mples that a one percent ncrease of fscal convergence ould decrease busness cycle volatlty by percent. The results are stll sgnfcant hen e use country fxed effects, and actually the magntude of the effects of fscal convergence on busness cycle volatlty s larger. Analyzng the overall set of OECD countres, e can see that the results are stll postve and extremely sgnfcant. Hoever, comparng the results for the to sets of countres, t emerges that the effect of fscal convergence on busness cycle volatlty s larger for the EMU countres than for the rest of OECD countres. Ths s partcularly mportant for to reasons. Frst, the EMU countres are those that manfested a hgher convergence of fscal poston. Second, t seems that the process of fscal convergence for these countres has sensbly reduced the possble stablzaton cost assocated th the creaton of the EMU and the loss of ndependent domestc monetary polces. The analyss dscussed so far s extremely robust th respect to the dfferent detrendng methods (HP00, HP6.5 and Band Pass). In fact, repeatng the same

13 analyss th dfferent detrendng methods, e can see (Table A, A and A3 of Appendx ) that the effect of fscal dvergence (convergence) on busness cycle volatlty s stll postve (negatve) and extremely sgnfcant. We also examned the robustness of the relaton beteen fscal convergence and busness cycle volatlty th respect to dfferent tme perods. In partcular e consdered to tme sample: the frst, ncludng the frst to decades ( , ); the second ncludng the second to decades ( , ). Table presents, across the above mentoned tme perods, the coeffcent on fscal dvergence obtaned usng the OLS for both EMU and OECD countres. Our results suggest that hle the statstcal sgnfcance of the relaton s broadly constant over the to tme samples, the magntude of the coeffcent for all measure of fscal convergence and for all groups of countres s remarkably ncreased. From an economc pont of ve, a possble explanaton s that n the last perod the economes under nvestgaton became more nterdependent, amplfyng spllover effects of stablzaton polces from one country to others. Another robustness check that e provde s to control for possble heterogenety over tme. Indeed, our dependent varable has a trend and decreased remarkably over the four decades. 9 In order, to consder ths ssue e replcate estmaton of equaton () usng both fxed and random tme effects. 0 Table 3 reports the coeffcent on fscal dvergence th respect to all measures consdered and th respect to EMU and OECD countres. Analyzng the table e can mmedately see the effect of fscal dvergence on 9 In contrast, the data do not sho a partcular pattern overtme for our measures of fscal convergence. 0 Smlar results are obtaned hen e nclude tme trend as expalnatory varable nstead of usng tme effects. 3

14 busness cycle volatlty, s stll postve and sgnfcant, even though the magntude s loer for some measures of fscal dvergence for the fxed effect specfcaton. 4. Busness Cycle Volatlty and Groth In Table 4, e report the estmate of the effect of busness cycle volatlty on groth, controllng for the ntal level of GDP, average nvestment share, human captal and populaton groth. The results n the frst and fourth columns are relatve to the OLS estmates for the EMU and the OECD countres, respectvely. Analyzng these results, e can see that the effect of busness cycle volatlty on groth s negatve and extremely sgnfcant. Successvely, n order to evaluate the mportance of fscal convergence for groth, e use our measure of fscal convergence as an nstrument for busness cycle volatlty. Columns and 5 report the coeffcent on output volatlty hen output volatlty s nstrumented by fscal dvergence (measured as the absolute value of budget balance dfferentals). The estmate s hghly sgnfcant and negatve and mples that the volatlty n output nduced by fscal dvergence has negatve consequences for groth. Or, conversely, that reduced output volatlty determned by fscal convergence has postve effects on groth. In addton to shong that fscal dvergence s harmful for economc groth, the results provde addtonal evdence of the sgnfcant effect of fscal convergence on busness cycle volatlty. In fact, f the relaton beteen fscal dvergence and output volatlty dentfed n the prevous secton as mspecfed, then the use of our measure as an nstrument n the groth equaton ll smply replcate the varaton n busness cycle volatlty and ll not lead to changes n the coeffcent. Instead, the 4

15 coeffcent on output volatlty changes sharply and sgnfcantly n the expected drecton. In partcular, a one percent ncrease n the nducted (by fscal dvergence) busness cycle volatlty ll determne a decrease of.4 (.5) percentage ponts n the average groth rate for the EMU (OECD) countres. The results are also sgnfcant hen e nclude openness and country sze as nstrument for busness cycle volatlty, and t does not seem that t sgnfcantly dffers beteen EMU and OECD countres (Columns 3 and 6). Also, ths analyss s extremely robust th respect to dfferent detrendng methods (HP00, HP6.5 and Band Pass). In fact, repeatng the same analyss th dfferent detrendng methods, e can see (Table A4, of Appendx ) that the effect of nducted busness cycle volatlty on groth s stll negatve and extremely sgnfcant. Fnally, repeatng the analyss for the fve measures of fscal dvergence analyzed n the prevous secton, e can see that the results are almost unchanged, both n terms of magntude and sgnfcance level (Table A5 of Appendx ). 5. Conclusons The am of ths paper s to analyze the effects that fscal convergence has on busness cycle volatlty and groth. From a theoretcal pont of ve, there s no clear consensus about the sgn of these to effects. In fact, f countres are subect to persstent asymmetrc shocks or symmetrc shocks, and those are partally offset by dscretonary fscal polcy or automatc fscal stablzers, then fscal dvergence can, n prncple, be assocated th greater busness cycle synchronzaton and reduced busness cycle volatlty. On the other hand, f fscal polcy s pro-cyclcal fscal dvergence ll be assocated th reduced busness cycle synchronzaton and hgher busness cycle 5

16 volatlty. At the same ay, hle rreversble nvestment theores suggest that countres th hgher volatlty ll have loer levels of nvestment and as a consequence loer groth, hgh groth economes mght be based on rsky technologes and therefore may experence sharp shfts n economc volatlty. Thus, both questons are ultmately emprcal. To help anserng these questons e nvestgate the relaton beteen fscal convergence and busness cycle volatlty, and the relaton beteen busness volatlty (nducted by fscal convergence) and groth, for a panel of OECD countres over a perod of 40 annual years. Usng several measures of fscal convergence and busness cycle volatlty our results suggest that (on average) a one percent ncrease of fscal convergences reduces busness cycle volatlty by percent. Moreover, ths effect s larger for the EMU countres than for ten other OECD countres. At the same ay, e found that a percent ncrease of nducted (by fscal dvergence) busness cycle volatlty determnes a decrease of.4 percentage ponts n the average groth rate. These results are robust to the dfferent detrendng methods and measures of fscal convergence consdered. The polcy mplcatons of these results are qute straghtforard. Fscal convergence, especally n ntegrated economes as the EMU, plays an mportant role. In fact, not only t could have helped to dramatcally reduce the possble stablzaton costs assocated th the creaton of a common currency but can also contrbute, through reducton of macroeconomc uncertanty, to hgher long-term groth. 6

17 References Alt, J. and Lory, R. (994). Dvded Governments, Fscal Insttutons and Budget Defcts: Evdence for the States, Amercan Poltcal Scence Reve, Vol. 88, No. 4, pp Asdrubal, P. Sorensen, B. E. and Yosha, O. (996). Channels of Interstate Rsk-sharng: US , Quarterly Journal of Economcs, Vol., No.4, Ashauer, D.A.(989). Is Publc Expendture Productve?, Journal of Monetary Economcs, Vol. 3, pp Barro, R. J. (990). Government Spendng n a Smple Model of Endogenous Groth, Journal of Poltcal Economy, Vol. 98, no. 5, pp Barro, R. and Gordon. D. (983). Rules, Dscreton and Reputaton n a Model of Monetary Polcy. Journal of Monetary Economcs,, 0-. Barro, R.J. and Lee, J.W. (00). Internatonal Data on Educatonal Attanment: Updates and Implcatons, Oxford Economc Papers, Vol. 53, No. 3, pp Bayoum, T. and Masson, P.R. (995). Fscal Flos n the Unted States and n Canada: Lessons from Monetary Unon n Europe, European Economc Reve, 39(), Baxter, M. and Kng, R. (999). Measurng Busness Cycles: Approxmate Band-Pass Flters for Economc Tme Seres, Reve of Economcs and Statstcs, Vol. 8, No. 4, pp Bernanke, B. S. (983). Irreversblty, Uncertanty and Cyclcal Investments, Quarterly Journal of Economcs, Vol. 97, No., pp Blackburn, K and Varvagos, D. (005). Groth, Uncertanty and Fnance, Centre for Groth and Busness Cycle Research Dscusson Papers, 48. Chrstano, L. and Ftzgerald, T. (003). The Band Pass Flter, Internatonal Economc Reve, Vol. 4, No., pp Darvas, Z., Rose, A. and Szapáry, G. (005). Fscal Dvergence and Busness Cycle Synchronzaton : Irresponsblty s Idosyncratc, NBER Workng Paper, 580. Evans, P. and G. Karras (994a). Are Government Actvtes Productve? Evdence from a Panel of U.S. States, Reve of Economcs and Statstcs, Vol.76, No., pp. -. Evans, P. and G. Karras (994b). Is Government Captal Productve? Evdence from a Panel of Seven Countres, Journal of Macroeconomcs, Vol. 6, No., pp

18 Fatás, A. and Mhov, I. (003). The Case for Restrctng Fscal Polcy Dscreton, Quarterly Journal of Economcs, Vol. 8, No. 4, pp Fatás, A. and Mhov, I. (006). The Macroeconomcs Effects of Fscal Rules n the US States, Journal of Publc Economcs, Vol. 90, No., pp Fölster, S., Henrekson, M., (00). Groth Effects of Government Expendture and Taxaton n Rch Countres. European Economc Reve, Vol. 45, No. 8, pp Furcer, D. (005). Is an Increase of the Fscal Budget at EMU level Desrable?, n Arests P., Ferrero J. and F. Serrano (Eds.): Fnancal Development n Natonal and Internatonal Markets, Palgrave Macmllan. Furcer, D. (007a). Long-Run Groth and Volatlty: Whch Source Really Matters?, Manuscrpt, Unversty of Illnos at Chcago. Furcer, D. (007b), Is Government Expendture Volatlty Harmful for Groth? A Cross-Country Analyss, Fscal Studes, Vol. 8, No., pp Furcer, D. and Karras, G. (006). Are the Ne EU Countres Ready for the EURO? A Comparson of Costs and Benefts, Journal of Polcy Modelng,Vol. 8, No., pp Furcer, D. and Karras, G. (007a). Busness Cycle Volatlty and Country-Sze: Scale Really Matters, Journal of Japanese and Internatonal Economes (forthcomng). Furcer, D. and Karras, G. (007b). Busness Cycle Volatlty and Country Sze: Evdence for a Sample of OECD countres, Manuscrpt, Unversty of Illnos at Chcago. Grer, K., and Tullock, G. (989). An Emprcal Analyss of Cross-Natonal Economc Groth: , Journal of Monetary Economcs, Vol. 4, No., pp Heston, A., Summers, R. and Aten, B. (006). Penn World Table Verson 6., Center for Internatonal Comparsons of Producton, Income and Prces at the Unversty of Pennsylvana, September. Hodrck, R. and Prescott, E. (997). Postar U.S. Busness Cycles: An Emprcal Investgatons, Journal of Money, Credt and Bankng, Vol. 9, No., pp. -6. Imbs, J. (007). Volatlty and Groth, Journal of Monetary Economcs, forthcomng. Kydland, F. and Prescott. E. (977). Rules Rather Than Dscreton: The Inconsstency of Optmal Plans, Journal of Poltcal Economy, 85,

19 Lane, P. (003). The Cyclcal Behavor of Fscal Polcy: Evdence from the OECD, Journal of Publc Economcs, Vol. 87, No., pp Leamer, E. (983). Let s Take the Con Out of Econometrcs, Amercan Economc Reve, 73, Levne, R. and Renelt, D. (99). A Senstvty Analyss of Cross-Country Groth Regressons, Amercan Economc Reve, Vol. 8, No. 4, pp Levnson, A. (998). Balanced Budget and Busness Cycles: Evdence from the States. Natonal Tax Journal, Vol. 5, pp Munnel, A. (99). Why Has Productvty Groth Declned? Productvty and Publc Investment, Ne England Economc Reve, Jan.-Feb, pp. 3-. Pndyck, R.S and Solmano, A (993). Economc Instablty and Aggregate Investments, NBER Macroeconomcs Annual, eds. O.J. Blanchard and S. Fsher, The MIT Press, Cambrdge. Poterba, J. (995). State responses to Fscal Crses: the Effects of Budgetary Insttutons, Journal of Poltcal Economy, Vol. 0, No.4, pp Ramey G. and Ramey V.A. (995). Cross-Country Evdence on the Lnk beteen Volatlty and Groth, Amercan Economc Reve, Vol. 85, No.5, pp Ravn, M. O. and Uhlg, H. (00). On Adustng the Hodrck-Prescott Flter for the Frequency of Observatons, Reve of Economcs and Statstcs, Vol. 84, No., pp Roubn, N. and Sachs, J. (989). Government Spendng and Budget Defcts n the Industralzed Countres, Economc Polcy, 8. Sachs, J. and Sala--Martn, X. (99). Fscal Federalsm and Optmum Currency Areas: Evdence for Europe from the US, NBER Workng Paper, Stock, J. and Watson, M. (999). Busness Cycle Fluctuatons n U.S. Macroeconomc Tme Seres, n Taylor, J. and Woodford, M. (eds.), Handbook of Macroeconomcs, vol., 3-64, Elsever. Von Hagen, J. (998). Fscal Polcy and Intranatonal Rsk-Sharng, CEPR Dscusson Paper, 6. 9

20 Fgure. Fscal Dvergence and Volatlty Busness Cycle Volatlty Budget Dvergence Fgure. Fscal Dvergence and Volatlty (controllng for openness and sze) Busness Cycle Volatlty Budget Dvergence 0

21 Table. Effect of Fscal Dvergence on Busness Cycle Volatlty (Dfferencng) EMU OECD OLS FE OLS FE Maastrcht Budget Devatons 0.38*** 0.533*** 0.307*** 0.37*** (4.8) (6.48) (4.3) (5.67) Maastrcht Budget Devaton Varant 0.343*** 0.504*** 0.87*** 0.353*** (4.00) (6.34) (4.) (5.65) Budget Balance Devaton 0.699***.005*** 0.67*** 0.744*** (4.07) (5.58) (4.4) (5.5) Budget Balance Devaton Varant 0.653*** 0.93*** 0.600*** 0.670*** (4.37) (6.37) (5.4) (5.96) Standard Devaton Absolute Value Budget Devatons 0.7***.433*** *** (.88) (5.88) (0.98) (4.55) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). Table. Effect of Fscal Dvergence on Busness Cycle Volatlty (Dfferencng) EMU OECD Maastrcht Budget Devatons 0.73*** 0.766*** 0.90*** 0.86*** (3.63) (3.36) (3.56) (5.4) Maastrcht Budget Devaton Varant 0.6*** 0.699*** 0.79*** 0.760*** (3.84) (3.05) (3.66) (5. 65) Budget Balance Devaton 0.479***.09*** 0.386***.3*** (.79) (.88) (3.47) (4.33) Budget Balance Devaton Varant 0.498*** 0.984** 0.36***.076*** (3.68) (.8) (4.09) (5.59) Standard Devaton Absolute Value Budget Devatons.050***.4*** 0.54**.50*** (3.9) (4.4) (.39) (.98) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). = refers to the frst to decades, =refers to the second to decades.

22 Table 3. Effect of Fscal Dvergence on Busness Cycle Volatlty (Dfferencng) EMU OECD FE RE FE RE Maastrcht Budget Devatons 0.46*** 0.38*** 0.0** 0.308*** (.79) (4.64) (.3) (4.83) Maastrcht Budget Devaton Varant 0.*** 0.504*** 0.09** 0.85*** (.74) (6.34) (.9) (5.00) Budget Balance Devaton 0.30** 0.653*** 0.84** 0.65*** (.05) (4.60) (.9) (4.36) Budget Balance Devaton Varant 0.394** 0.93*** 0.408*** 0.60*** (.95) (6.37) (.53) (6.) Standard Devaton Absolute Value Budget Devatons.*** 0.7*** (4.66) (3.59) (0.68) (0.90) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ).

23 Table 4. Effect of Busness Cycle Volatlty on Groth (Dfferencng) EMU OECD OLS IV IV OLS IV IV Intal GDP (Y o ) *** *** *** -0.00*** *** *** (-5.97) (-5.59) (-5.86) (-4.65) (-5.00) (-5.6) Investment ** 0.085* 0.095** (.3) (.0) (.3) (.38) (.83) (.56) Human Captal (0.98) (0.75) (.4) (.7) (.7) (.3) Populaton Groth -.00* -.595** -.474** (-.73) (-.47) (-.48) (-0.97) (-.3) (-.) Busness Cycle Volatlty (σ τ ) -.44*** -.384*** -.09*** -0.99*** -.56*** -.637*** (-3.9) (-4.37) (-4.6) (-3.9) (-4.0) (-4.5) R F T-statstcs n parentheses. Whte Robust Standard Errors. Constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). 3

24 Appendx Cyclcal Volatlty Table A. Effect of Fscal Dvergence on Busness Cycle Volatlty (HP00) EMU OECD OLS FE OLS FE Maastrcht Budget Devatons 0.3** 0.4*** 0.304*** 0.388*** (.50) (3.) (3.58) (4.4) Maastrcht Budget Devaton Varant 0.8** 0.400*** 0.80*** 0.366*** (.3) (3.9) (3.5) (4.9) Budget Balance Devaton 0.660*** 0.888*** 0.646*** 0.86*** (.75) (3.4) (3.86) (5.8) Budget balance Devaton Varant 0.58** 0.738*** 0.56*** 0.708*** (.43) (3.8) (4.0) (5.7) Standard Devaton Absolute Value Budget Devatons 0.678**.00*** *** (.04) (.9) (.35) (3.93) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). 4

25 Table A. Effect of Fscal Dvergence on Busness Cycle Volatlty (HP6.5) EMU OECD OLS FE OLS FE Maastrcht Budget Devatons 0.3** 0.508*** 0.96*** 0.399*** (.58) (4.7) (3.04) (4.87) Maastrcht Budget Devaton Varant 0.90** 0.498*** 0.76*** 0.380*** (.45) (4.94) (3.0) (4.89) Budget Balance Devaton 0.63** 0.98*** 0.66*** 0.8*** (.64) (4.39) (3.44) (4.95) Budget balance Devaton Varant 0.600*** 0.99*** 0.595*** 0.76*** (.90) (5.5) (3.8) (5.06) Standard Devaton Absolute Value Budget Devatons 0.58**.460*** *** (.5) (4.93) (0.67) (4.37) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). Table A3. Effect of Fscal Dvergence on Busness Cycle Volatlty (Band-Pass) EU OECD OLS FE OLS FE Maastrcht Budget Devatons 0.4** 0.57** 0.5*** 0.3*** (.46) (.3) (.70) (.74) Maastrcht Budget Devaton Varant 0.6** 0.50** 0.09*** 0.07*** (.30) (.7) (.7) (.75) Budget Balance Devaton 0.33** 0.395** 0.5*** 0.47*** (.5) (.48) (.67) (4.7) Budget Balance Devaton Varant 0.306** 0.38*** 0.38*** 0.440* (.) (.89) (3.6) (4.77) Standard Devaton Absolute Value Budget Devatons 0.43* 0.546** *** (.77) (.46) (0.43) (3.80) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). 5

26 Table A4 Effect of Volatlty on Groth, Robustness Check (EMU) Dfferencng HP00 HP6.5 Band Pass σ τ -.44*** *** -.9*** -.64*** (-3.9) (-.98) (-4.63) (-4.0) Table A5. Effect of Busness Cycle Volatlty on Groth (Dfferencng) EMU OECD IV IV IV IV Maastrcht Budget Devatons -.8*** -.070*** -.693*** -.663*** (-4.5) (-4.78) (-4.04) (-4.53) Maastrcht Budget Devaton Varant -.43*** -.69*** -.868*** -.698*** (-4.44) (-4.65) (-3.9) (-4.54) Budget Balance Devaton -.384*** -.070*** -.56*** -.637*** (-4.37) (-4.78) (-4.0) (-4.5) Budget Balance Devaton Varant -.8*** -.070*** -.643*** -.663*** (-4.5) (-4.78) (-4.30) (-4.53) Standard Devaton Absolute Value Budget Devatons -.97*** -.999*** -4.46* -.49*** (-3.65) (-3.90) (-.78) (-.68) Coeffcents represent elastctes. T-statstcs n parentheses. Controls varables and constant term ncluded but not recorded. Coeffcents sgnfcantly dfferent from zero at 0.0, 0.05, and 0.0 level marked th one, to and three astersks, respectvely. Data set has observatons for four decades (964-73, , , ). 6

27 Appendx Data Sources The follong varables are avalable for all countres from 964 to 003: Intal Output (Yo) - The log of Heston-Summers-Aten (006) varable Real GDP per Capta at the Begnnng of Each Tme Perod, 000 Internatonal Prces; Laspeyres Index. Groth-rate (g) - The 0-year average of the Heston-Summers-Aten (006) annual groth rate varable. Populaton Groth (n) - The average of the annual log dfference of Heston-Summers- Aten (006) populaton varable. Sze (pop) - The 0-year average of the Heston-Summers-Aten (006) annual populaton varable. Investment Share of GDP (nvestment) The 0-year average of the Heston-Summers- Aten (006) varable Real Gross Domestc Investment, Prvate and Publc, % Real GDP per Capta, 000 Internatonal Prces. Human Captal (human captal) The ntal value of the Barro-Lee (00) varable Average Schoolng Years n the Total Populaton over Age 5 at the Begnnng of Each Tme Perod. Openness (open) - The 0-year average of the Heston-Summers-Aten (006) varable Exports plus Imports Dvded by Real GDP. Budget, Nomnal GDP - Source: OECD Economc Outlook, Andre Rose ebste. 7