THE WILLIAM DAVIDSON INSTITUTE AT THE UNIVERSITY OF MICHIGAN BUSINESS SCHOOL

Transcription

1 THE WILLIAM DAVIDSON INSTITUTE AT THE UNIVERSITY OF MICHIGAN BUSINESS SCHOOL Reivested Earigs Bias, The Five Percet Rule ad the Iterpretatio of the Balace of Paymets With a Applicatio to Trasitio Ecoomies By: Josef C. Brada ad Vladimír Tomšík William Davidso Workig Paper Number 543 February 2003

2 Reivested Earigs Bias, The Five Percet Rule ad the Iterpretatio of the Balace of Paymets With a Applicatio to Trasitio Ecoomies* Josef C. Brada** W. P. Carey School of Busiess, Arizoa State Uiversity ad Vladimír Tomšík*** W. P. Carey School of Busiess, Arizoa State Uiversity Prague Uiversity of Ecoomics ad Newto Holdig a.s. * We are idebted to Michael Melvi ad Zdeek Drabek for commets o a previous versio of this paper. All errors are, of course, etirely our resposibility. Brada's research was supported by a Fulbright grat to visit the Czech Republic. Tomšík's research was supported by the W. P. Carey School of Busiess at Arizoa State Uiversity ad by the America Fud for Czechoslovak Relief. ** Josef. C. Brada, Departmet of Ecoomics, W. P. Carey School of Busiess, Arizoa State Uiversity, Tempe, AZ , USA, phoe: (+1) , Josef.Brada@asu.edu *** Vladimír Tomšík, Departmet of Ecoomic Policy, Prague Uiversity of Ecoomics, Prague, 4, W. Churchill Sq., Prague 3, Czech Republic, phoe: (+420) , tomsik@vse.cz

3 ABSTRACT Reivested Earigs Bias, The Five Percet Rule ad the Iterpretatio of the Balace of Paymets With a Applicatio to Trasitio Ecoomies We show that the imputatio of reivested profits of the subsidiaries of foreig firms as a debit item o a host coutry's balace of paymets accout teds to overstate the curret accout deficit. We also show that, because of the workigs of the FDI fiacial life cycle, this pheomeo is most evidet for coutries that have recetly received large iflows of capital. The trasitio ecoomies of East Europe certaily fall amog such coutries, ad we show that, for the Czech Republic ad Hugary, this imputatio has a large effect o their reported curret accout balace. We verify the workig of the FDI fiacial life cycle usig two differet paels of developed, developig ad trasitio ecoomies. Keywords: balace of paymets, fiacial crisis, foreig direct ivestmet, trasitio ecoomies JEL Classificatio Numbers: F21, F23, F34

4 NON-TECHNICAL SUMMARY This paper examies a aomaly i the balace of paymets accoutig framework that teds to overstate the curret accout deficit of coutries that are et recipiets of foreig direct ivestmet (FDI) ad especially of those coutries that are experiecig, or that have recetly experieced, large iflows of FDI. The aomaly is due to the imputatio of the reivestmet of profits by foreig-owed affiliates as a debit item o the curret accout eve though such reivestmet ivolves o trasactios o the foreig exchage market ad, of course, represets a much more stable form of fiacig tha do short-term capital iflows. The paper shows that this bias is of cosiderable quatitative sigificace for a umber of developig coutries ad particularly so for the trasitio ecoomies of Easter Europe. Reivested earigs are reported as a credit item o the capital accout as part of FDI i the capital accout to reflect foreig ivestors' icreased ivestmet i the coutry. To offset this credit item ad maitai the double-etry ature of the balace of paymets, the reivested earigs of a foreig-owed affiliate are also recorded as a liability o the curret accout to reflect the foreig ivestors' ivestmet returs o equity. Coutries that have received large iflows of FDI that geerates large profits that are reivested i the local ecoomy will, paradoxically, appear to have large curret accout deficits eve though the reivested profits may be used to purchase local iputs such as lad, structures, etc. While it is true that propoets of the bechmark for curret accout deficit, which is set at 5% of GDP ( five-percet rule ), emphasize the eed to cosider the way i which the curret accout deficit is fiaced, i commo practice a close aalysis of the fiacig of the deficit is geerally ot udertake, ad, therefore, the bias of reivested earigs which do ot effect floats i exchage market, if quatitatively importat, should be addressed more seriously.

5 This study empirically evaluates a icreasig role of a debit of the balace of icome (icludig reivested earigs) i a debit of the overall curret accout. The paper provides cross-couitry logitdial data that show the statistical relatioship betwee the ratio of the debit i the balace of icome to the debit i the curret accout ad the ratio of a sum of FDI iflow i the period from t through t- to the curret stock of FDI i the period t. The data set covers 32 coutries. The results of pael estimatios with the dummy variables for developig ad trasitio ecoomies establish that the relatioship betwee the ratio of FDI iflow to the stock of FDI ad the ratio of a debit of the balace of icome to a debit of the curret accout balace begis to weake whe the older FDI iflows are icluded i the model. This cofirms our hypothesis that early i the life of foreig ivestmets, whe FDI iflows start to geerate a profit, most of the profit is reivested back to the host ecoomy i the first few years. The results also show that reivestmet of profits is sesitive to coutry characteristics. I the case of the trasitio ecoomies, the high levels of reivestmet of profits fall off more quickly tha they do i developed coutries, a result of the way i which the trasitio ecoomies privatized i the 1990s. We also show that i the trasitio ecoomies reivested profits have caused a sigificat debit of the balace of icome. If the profits are reivested, these reivested earigs cause a bias i a expectatio about the pressure o a foreig exchage market. Thus we coclude that the foreig exchage market should pay much greater attetio to the way i which the curret accout deficit i trasitio ecoomies has bee fiaced.

6 1. INTRODUCTION While currecy crises have a variety of causes (Eichegree et al., 1996; Goldstei et al., 2000; Krugma, 2000; Summers, 2000), both the ecoomic literature ad practical experiece with crises shows that the ability to forecast them remais cotroversial. 1 Despite this, iteratioal leders, orgaizatios like the IMF ad the World Bak, as well as the ecoomic press, have adopted certai rules of thumb that serve, if ot as predictors of crises, the at least as warig sigs that coutries that violate such rules of thumb are i dager of experiecig a currecy crisis or speculative attack o their currecies. Perhaps the best kow of these iformal rules is that a coutry's curret accout deficit should ot exceed five percet of GDP. For example, Milesi-Ferreti ad Razi (1996), observe that [c]ovetioal wisdom is that curret accout deficits above 5% of GDP flash a red light, i particular if the deficit is fiaced with short-term debt ad Summers (1996) wars that close attetio should be paid to ay curret accout deficit i excess of 5% of GDP. Such high ad sustaied curret accout deficits are viewed as precursor to a currecy crisis because they are ofte fiaced by short-term capital iflows ito the coutry, ad such iflows are subject to sudde reversals. I this paper we call attetio to a aomaly i the balace of paymets accoutig framework that teds to overstate the curret accout deficit of coutries that are et recipiets of foreig direct ivestmet (FDI) ad especially of those coutries that are experiecig, or that have recetly experieced, large iflows of FDI. The aomaly is due to the imputatio of the reivestmet of profits by foreig-owed affiliates as a debit item o the host coutry's curret accout eve though such reivestmet ivolves o trasactios o the foreig exchage market ad, of course, represets a much more stable form of fiacig tha do short-term capital 1 See Goldfaj ad Valdés (1998), Berg ad Pattillio (1999), Kamisky ad Reihart (1999), Burkart ad Coudret (2002). 1

7 iflows. We also show that this bias is of cosiderable quatitative sigificace for a umber of developig coutries ad particularly so for the trasitio ecoomies of Easter Europe. The remaider of the paper is orgaized as follows. I the ext sectio we show how reivested profits of foreig firms are imputed as a debit item o the curret accout. I sectio three we discuss the factors that determie the magitude of the distortio i the curret accout balace that this imputatio causes ad we show that, for a umber of coutries, ad particularly for some of the trasitio ecoomies of East Europe ad the former Soviet Uio, such reivested profits may actually represet a very sigificat part of the observed curret accout deficit. I Sectio 5 we show that coutries that have experieced relatively large recet iflows of FDI ted to have a higher ratio of icome accout debits to curret accout debits, a fidig that leds support to our thesis. 2. WHY SOME COUNTIES' CURRENT ACCOUNT DEFICIT IS "OVERSTATED" Normally, a trasactio is recorded o the balace of paymets whe foreig ad the domestic currecies are exchaged betwee the residets of a coutry ad the rest of the world. Such exchages usually have a couterpart trasactio o the foreig exchage market. However, accordig to the IMF Fifth Balace of Paymets Maual (1993), i some cases where o actual currecy flows betwee a coutry's residets ad the rest of the world occur, trasactios are evertheless imputed ad etries are made i the balace of paymets accouts. The reivested earigs of foreig-owed affiliates are a example of such a imputed etry i the balace of paymets because the earigs of the foreig-owed affiliate, whether distributed i the form of divideds paid to the paret firm or reivested i the local affiliate, are icluded i the balace of paymets as a deficit item o the curret accout. I the case of divideds remitted to the paret compay, the ratioale for the etry is obvious because host-coutry currecy has to be coverted 2

8 ito the currecy of the coutry that is the MNC's home. I the case of profits that are reivested i the host coutry affiliate, however, there is o exchage of home-coutry currecy for foreig exchage. Profits eared i the host coutry's currecy remai i the host coutry. This meas that profits reivested i the local affiliate of a foreig-owed firm are treated as a curret accout deficit item, that is, a exchage of local currecy for foreig currecy, eve though there is o such exchage takes place. Such a imputatio is ecessary to preserve the doubleetry ature of the balace of paymets accout, which requires the balace of paymets to accout for a icrease i foreigers' ivestmets i the host coutry. Nevertheless, while reivested profits ad divided remittaces are reported i the icome accout as seemigly similar debit trasactios, the latter must be fiaced i some way o the foreig exchage market while the former do ot. As Box 1 shows, reivested earigs are reported as a credit item o the fiacial accout as part of FDI to reflect foreig ivestors' icreased asset holdigs i the host coutry. To offset this credit item ad to maitai the double-etry ature of the balace of paymets, the reivestmet of earigs by a foreig-owed affiliates are also recorded as a liability o the curret accout. A et iflow of reivested earigs ito the domestic ecoomy has a positive impact o the fiacial accout i the form of et direct ivestmet, but the impact o et icome receipts withi the curret accout is of a equal, but opposite, amout. Thus, coutries that have received large iflows of FDI that geerates large profits that are reivested i the local ecoomy will, paradoxically, appear to have large curret accout deficits eve though the reivested profits purchase local iputs such as lad, structures, etc., ad require o foreig exchage fiacig. While it is true that propoets of the "five-percet rule" emphasize the eed to cosider the way i which the curret accout deficit is fiaced, i commo practice a close aalysis of the 3

9 fiacig of the deficit is geerally ot udertake, ad, therefore, the bias described here, if quatitatively importat, eeds to be addressed more seriously i evaluatig whether a coutry s curret-accout deficit is sustaiable or ot. Moreover, there is some iroy to the fact that FDI, the most stable source of exteral fiace, ad oe that Ferádez-Arias ad Hausma (2001) foud to actually reduce the risk of currecy crises ad speculative attacks, is also the oe form of fiacig that is icluded i the balace of paymets i a way that makes the coutry appear more vulerable to such crises ad attacks. 3. IS THE REINVESTED EARNINGS BIAS A LARGE PART OF THE CURRENT ACCOUNT DEFICIT? A. The Magitude of Reivested Earigs from FDI Whether the imputatio of reivested earigs i the curret accout is sufficiet to materially affect the magitude of a coutry's exteral deficit depeds o three broad sets of factors. The first of these is quatitative. The larger the stock of FDI relative to the size of the ecoomy ad the more profitable are foreig firms, the greater is the pool of moey that ca be reivested ito the local affiliates of foreig firms. The secod set of factors cosists of coutryspecific characteristics of the host ad home coutries that ifluece the distributio of total affiliate profits ito divideds that are remitted to the paret compay ad ito fuds that are reivested back ito the local affiliate. This decisio depeds o a variety of factors icludig perceptios of host coutry risk; tax treatmet of divideds by the home ad host coutries; opportuities for extractig fuds from the affiliate through trasfer pricig, maagemet fees, etc.; ad the attractiveess of alterative ways of fiacig the affiliate's ivestmet eeds 4

10 (Robbis ad Stobaugh, Ch. 5, 1973). 2 The third set of factors cosists of the time path of FDI ito the host coutry, which affects both the volume of profits ad their distributio betwee reivestmet ad divideds. This third set of factors we characterize as the FDI fiacial life cycle. The FDI fiacial life cycle model is described i Figure 1, which presets a stylized relatioship betwee profits, divideds ad reivested profits over the life of a foreig direct ivestmet project. At the outset, i what we call Stage 1 i the diagram, the MNC makes a ivestmet i the foreig coutry to foud a affiliate. At first, the affiliate will operate at a loss. I the case of a acquisitio, this period may be short if the acquired firm is, or ca be easily reorgaized to become, profitable. I the case of a greefield ivestmet, durig the time take to acquire a site, build ad equip a productio facility, trai workers ad begi productio, the iterest o the capital ivested may result i sizable ad loger lastig start-up losses. Thus, i Stage 1, the affiliate operates at a loss ad pays o divideds. I Stage 2, the affiliate begis to operate at a profit as productio starts or as the firm becomes more competitive as the result of the restructurig or other competitive advatages provided by the paret firm. However, as the affiliate becomes more successful o the market, it is likely to have sigificat eeds for additioal ivestmet, both for workig capital as well for icreased plat ad equipmet. Thus, at first, all profits may be reivested to meet these eeds. As time passes ad profits cotiue to grow, the paret firm may begi to require that the affiliate remit some of the profits i the form of divideds, although the moetary value of reivested profits may cotiue to icrease. The legth of the secod stage will i part deped o the size of the domestic market, which will determie for how log the affiliate ca cotiue to expad its 2 Despite the existece of these other optios, divided remittaces have accouted for over 50% of the fuds flowig from foreig affiliates to US MNCs i the post-world War II period, ad this proportio has show little chage over time. 5

11 capacity, o the availability of export markets to the affiliate ad o the attractiveess of alterative ways of fiacig the affiliate's expasio. I Stage 3, the affiliate has reached a "mature" stage where its market share ad profit margis i the host coutry have stabilized. At this poit, the paret firm will choose to repatriate a larger share of the profits i the form of divideds so that these fuds ca be used to fiace ivestmet opportuities that offer more dyamic prospects elsewhere, ad reivested earigs will declie both as a share of profits ad absolutely. 3 B. FDI Reivestmet I the Balace of Icome - Some Evidece Table 1 shows how the three determiats of FDI reivestmet discussed above ifluece the size of the bias i the curret accout of four coutries, Brazil, the Czech Republic, Irelad ad Portugal. These four coutries provide good data o reivested profits ad also illustrate the sigificace of the three factors discussed above. The experiece of at least three of these coutries also provides strikig evidece that, i some cases, reivested profits are a very sigificat compoet of the curret accout deficit. The first factor, the amout of FDI ad its profitability, is most evidet i a compariso of Irelad with the other three coutries. For Irelad, the differece betwee the curret accout with reivested divideds reported as a debit item ad without the iclusio of reivested profits is aroud 10 percet of GDP. That is, without the imputatio of reivested profits by foreig MNCs located i Irelad as a debit item, Irelad's curret accout surplus would be higher by about 10 percet of GDP. This differece betwee the two measures of the curret accout surplus is much greater tha it is i the other three coutries. I part, the greater gap betwee the 3 A iterestig example of the workigs of the FDI fiacial life cycle is provided by Koretz (2002), who writes: "The U.S. became a debtor atio durig the 1990s. Yet util this year it actually received more icome from its direct ivestmets overseas ad holdigs of foreig fiacial assets tha foreigers received from their U.S. ivestmets." He suggests that this is because 6

12 two measures of the curret accout reflects the fact that the stock of FDI i Irelad is equivalet to over 40 percet of GDP while for the other three coutries, it rages form 11 to 15 percet of GDP. Moreover, as colums 1 ad 2 of Table 1 show, FDI i Irelad appears to be more profitable tha it is i the other three coutries. 4 Thus the much larger volume of MNC profits i Irelad relative to aggregate ecoomic activity does much to explai why the bias i the measuremet of the curret accout is so large. Coutry-specific factors also play a role i the magitude of the curret accout bias caused by FDI. Although Brazil, the Czech Republic ad Portugal have similar ratios of FDI to GDP, a examiatio of colum 4 of Table 1, the percetage of FDI profits that is reivested i the coutry, reveals that Brazil is somethig of a outlier. I the other two coutries, as well as i Irelad, about half of FDI profits are reivested. I Brazil, the rate of reivestmet is quite low, ad i some years egligible. 5 As expected, the differece betwee the Brazilia curret accout deficit measured with ad without reivested profits is virtually oexistet. O the other had, for the Czech Republic ad Portugal, the differece is appreciable, usually over oe percet of GDP for the Czech Republic ad early oe percet for Portugal. These are sigificat biases whe cosidered i the cotext of the "five percet rule". Of particular relevace to the trasitio ecoomies is the FDI fiacial life cycle effect o the volume of reivested earigs i the curret accout balace. This is so because the trasitio ecoomies have goe from a state where they had virtually o FDI at the start of the 1990s to a situatio where some, such as Hugary, Polad ad the Czech Republic, have FDI stocks of a magitude, whether measured relative to GDP or populatio, that compares with may other " a lot of recet direct ivestmet i the U.S. has faced big startup costs. Ivestmet by U.S. compaies overseas is older, so it ears higher returs." 4 It is importat to bear i mid that FDI i Irelad may appear more profitable because Irelad's accoutig stadards may make it more difficult for MNCs to uderstate profits through trasfer pricig, royalties, maagemet fees, etc. 7

13 middle-level icome coutries that have received FDI iflows for much loger periods of time. The major differece betwee the trasitio ecoomies ad other coutries the is ot i the stock of FDI but rather i its vitage. If the FDI fiacial life cycle model is correct, the, curretly, the amout of reivestmet of MNC earigs i the trasitio coutries is abormally high ad is likely to be icreasig because most of the foreig affiliates are eterig or operatig i Stage 2 of the FDI fiacial life cycle. Oly later, as they eter Stage 3, will the bias i their curret accout steadily dimiish as the reivestmet of earigs drops off ad is replaced by divided repatriatio, which, ulike reivested earigs, does create claims o the foreig exchage market. The workigs of the FDI fiacial life cycle are illustrated i Figure 2, which shows the role that reivested earigs play i the FDI positio ad i the balace of icome of two trasitio ecoomies, the Czech Republic ad Hugary. Hugary attracted a large stock of FDI early o i the trasitio. This was due to the fact that Hugaria privatizatio was cosciously desiged to attract foreig "strategic" ivestors for Hugary's state-owed firms ad, later, fiacial istitutios. For the first half of the 1990's Hugary was by far the leader i both the stock of FDI ad aual FDI iflows amog the East Europea trasitio ecoomies. The Czech Republic, o the other had, chose to privatize the bulk of its state-owed firms by meas of the "voucher privatizatio" that put firms i the hads of domestic rather tha foreig owers. While some Czech firms, SPT Telecom, the telephoe moopoly, ad the carmaker Škoda beig prime examples, were sold to foreigers, much of the ivestmet i the Czech Republic through mergers ad acquisitios had to wait util the ew domestic owers could take cotrol of their firms ad the decide to sell them to foreigers. As a result, much more of the FDI ito the Czech 5 The low rate of reivestmet i Brazil may reflect the coutry's poor ecoomic performace i the late 1990s. 8

14 Republic had to take the form of greefield ivestmets, which aturally took loger to pla ad implemet. Cosequetly, while the two coutries had similar levels of stocks of FDI by the ed of the 1990s, the vitage of Czech ivestmets was much ewer tha Hugary's. This timig of FDI i the two coutries is reflected i Figure 2, which shows the cotributio of reivested earigs to both et FDI o the fiacial accout (bars above the zero lie) ad to the balace of icome o the curret accout (bars below the zero lie). I the case of the Czech Republic, reivested earigs are a small part of total FDI iflows, but they are a large part of the deficit o the icome accout. The former is due to the fact that most of the stock of FDI i the Czech Republic has etered the coutry i the secod half of the 1990s so that FDI iflows from abroad still costitute the mai aveue for foreigers to acquire or icrease ivestmets i the Czech Republic. The FDI fiacial life cycle suggests that this recet ivestmet should yield o or low profits or, to the exert that it does yield profits, these should maily be reivested i the Czech affiliates that geerate them. Thus, whe we examie the Czech balace of icome, these reivested profits form a large share of the deficit o this accout because few of the foreig ivestmets i the Czech Republic are sufficietly mature to be i Stage 3 of the FDI fiacial life cycle where their profits would be repatriated to the paret compay i the form of divideds. I Hugary, total profits o FDI are higher tha they are i the Czech Republic, as the FDI fiacial life cycle model would predict. Also, reivested profits accout for a larger share of total FDI flows i Hugary tha they do i the Czech Republic, both because the iflows of ew FDI are lower i Hugary tha they are i the Czech Republic ad because the earigs of the more mature foreig ivestmets i Hugary are greater tha those of relatively ewer ivestmets i the Czech Republic. Moreover, because profits o FDI i Hugary are higher, as 9

15 suggested by the FDI fiacial life cycle model, eve if higher divideds are paid out, there is evertheless more moey to reivest as well. I Hugary a larger proportio of FDI occurred i the early 1990s, ad thus Hugaria FDI projects are more mature ad some ivestmet projects may be approachig Stage 3 of the FDI fiacial life cycle. As more FDI projects eter Stage 3, Hugary is experiecig larger divided outflows tha are evidet i the Czech Republic. Of course, as other ivestmet projects i Hugary eter Stage 2, reivested profits will cotiue to grow as well, eve if they do accout for a smaller share of the deficit o the balace of icome. While the data we have preseted show that the imputatio of reivested earigs has a sigificat effect o the reported curret accout deficits of the two trasitio ecoomies, it is also worthwhile to examie the dyamics of this bias i order to see how it has evolved ad what its likely effect may be i the future. To this ed we examie more carefully the case of the Czech Republic. At the ed of the 1990s ad i 2000 ad 2001, the most importat item of the Czech curret accout balace, the trade deficit, declied due to favorable developmets i the terms of trade. The services surplus remaied stable from 1995 o. Nevertheless, there was a steady icrease i the curret accout deficit due to the icreasig deficit i the balace of icome. I Figure 3, we show the et balaces o the icome accout of the Czech Republic. Figure 3 shows that the growth of the icome accout deficit was almost etirely due to the imputatio of et reivested earigs as a debit item o this accout. I 1995, reivested earigs played virtually o role i the icome accout ad perhaps were ot eve measured or reported as a separate item i the balace of paymets. There was a small surplus i the compesatio of oresidet employees ad small deficits i the iterest balace ad divideds ad redistributed earigs. It was oly i 1998 that a deficit i reivested earigs appeared, ad its magitude was the about equal to those of the deficits i the iterest balace ad oresidet employee 10

16 compesatio. Sice the, the deficit from reivested earigs has made up the largest share of the deficit o the icome accout, ad it is almost etirely the source of the growth of the icome accout deficit. Moreover, ot oly has the growth of et reivested earigs drive the icome accout deficit, ad, by extesio, the curret accout deficit, but the importace of reivested earigs to the curret accout deficit has also created additioal ucertaity about the size of curret accout deficit itself. I Figure 3 we preset the prelimiary ad revised icome accouts for 2000 ad for I 2000, the prelimiary figures cosiderably uderestimated the volume of et reivested earigs. The Czech Natioal Bak estimates reivested earigs i the curret period through extrapolatio, surveys ad forecasts, ad estimates based o such methods require larger revisios tha do some other etries i the balace of paymets accouts. For example, the Czech Natioal Bak revised the et reivestmet of profits figure for 2000 upward by 84 percet from the formerly published figure of CZK 20,000 millio to CZK 36,871 millio. Figure 3 shows that this revisio accouted for the bulk of the revisio i the icome accout. More importat, as a result of this revisio, the curret accout deficit jumped to above 5 percet of GDP i 2000, raisig cocers about the log-term viability of the exteral balace. If a revisio of similar proportios were to be required for the estimated earigs reivestmet of foreig MNCs reported for 2001, which was published as CZK 32,000 millio, the revised figure would be i the eighborhood of CZK 57,600 millio. This potetial revisio is labeled Forecast by Authors i Figure 3, ad it shows the serious cosequeces that such a revisio would have for the deficit o the icome accout. 11

17 4. THE FDI FINANCIAL LIFE CYCLE IN INTERNATIONAL PERSPECTIVE I the foregoig parts of this paper, we have show that the imputatio i the balace of paymets of reivested earigs o FDI ca have a potetially large impact o a coutry s curret accout deficit ad that there are systematic factors that explai the magitude of the volume of reivested earigs over time ad across coutries. While it is quite true that a sophisticated iterpretatio of a coutry s curret accout deficit would take ito accout the sources of fiacig of the deficit ad that such a aalysis would brig to light the cotributio of reivested earigs, very few coutries actually report the magitude of reivested earigs o FDI i their balace of paymets accouts, makig such a careful aalysis difficult. I our ow research, we have had to obtai the iformatio o reivested earigs for may coutries from specialized sources rather tha from their balace of paymets accouts, ad eve specialized sources do ot always provide this iformatio o a cosistet basis. Due to the paucity of data o reivested earigs from FDI, i this sectio we tur our attetio to the icome balace. As is evidet from Table 1 ad Figures 2 ad 3, for may coutries, reivested profits o FDI make up a large part of the icome accout deficit. Moreover, as Figure 3 shows, due to the workigs of the FDI fiacial life cycle, reivested earigs ca chage quite rapidly over time, much more so tha do other items icluded i the icome accout balace, ad, as a result, chages i the etire icome balace ted to reflect chages i the this compoet of the balace. Cosequetly, for may coutries, chages i the debit items i the icome accout ca be a good proxy for movemets i the reivestmet of earigs o FDI, ad it is this variable whose behavior we examie i this sectio. Because the icome accout debits iclude both reivested profits ad divided repatriatio, the importace of the former ca be tested by examiig the relatioship betwee the 12

18 vitage of the stock of FDI ad movemets i the debit items o the icome accout. If the FDI fiacial life cycle model is correct, the we should see a icrease i the icome accout debit soo after the iflow of FDI because of the icrease i reivested earigs o FDI, which will occur sooer tha does profit repatriatio by the legth of time eeded for firms to move from Stage 2 to Stage 3. O the other had, if there is oly a weak lik betwee the icome accout debit items ad recet FDI iflows, the either reivested FDI does ot play a importat role i movemets of the icome balace debit items or the FDI fiacial life cycle model represets a uimportat pheomeo because it is profit repatriatio, which occurs somewhat later after the iflow of FDI, that would be the major drivig force behid movemets i the icome accout. I this sectio, we test for the existece of a relatioship betwee the vitage of FDI iflows ad profit reivestmet, the latter proxied by the debit items i the icome balace. We also examie differeces i this relatioship betwee trasitio ecoomies ad developed ad developig coutries. We hypothesize that FDI reivestmet i developed, developig ad trasitio ecoomies coutries follows a path over time that is similar i shape to that described i Figure 1 but that the legth of time eeded to reach the various stages of the FDI fiacial life cycle or the propesity to reivest profits at ay stage of the FDI fiacial life cycle may differ betwee these three categories of host coutries. Most coutries experiece a cotiuous stream of FDI, some ew ad some i the form of reivested profits. We have argued that the age structure of FDI plays a major role i determiig the magitude of reivested profits alog with coutry-specific factors such as the magitude of FDI, its profitability, etc. I this sectio we employ a sample of thirty two coutries with data extedig from 1993 to 2000 to test whether the vitage of FDI plays a sigificat role i the magitude of FDI. The sample cosists of developed, developig ad trasitio ecoomies for 13

19 which we were able to fid cosistet data for the etire sample period. Some coutries are small, others, like the Uited States ad the Peoples Republic of Chia, are both large ad major recipiets of FDI. 6 We also use a subset of these coutries that cosists of oly developed ad developig coutries to test the same relatioship over a sigificatly loger time period from 1981 to The magitude of the debit items o the balace of icome, reflectig chages i reivested FDI, should be a greater or smaller share of the debits o the curret accout, ceteris paribus, depedig o the vitage of the FDI iflows. To accout for coutry size effects ad the differeces i opeess to iteratioal commerce amog coutries, we ormalize the icome debit items by the coutry's curret accout debits. Moreover, because we lack data o the vitage of FDI i a coutry, we use as a idicator of vitage the ratio of the sum of FDI for the most recet years to the total stock of FDI. This relatioship ca be stated as: FDIt m BI _ debit = m= 0 t f Eq. 1 CA_ debit Stock _ FDI t where t deotes the year; BI_debit is the value of the debit items i the balace of icome, CA_debit is the value of the debit items i the curret accout; FDI t m= 0 m is a sum of FDI iflows i the period from t through t-, where {1,2,3, }; ad Stock_FDI t is the stock of 6 Amog the developed coutries are Australia, Frace, Germay, Great Britai, Irelad, Italy, Japa, New Zealad, Portugal, Spai, ad the Uited States of America; amog the developig coutries, Argetia, Brazil, Chile, Chia, Idia, Idoesia, Korea, Mexico, Peru, Sigapore, South Africa, Turkey, ad Veezuela. The trasitio coutries are Bulgaria, the Czech Republic, Estoia, Hugary, Romaia, Russia, Slovakia, ad Sloveia. Polad was ot icluded amog the trasitio coutries due to a chage i its balace of paymets methodology i 1995, which redered the Polish balace of paymets data icosistet with that of other coutries. The sample period begis i 1993 to allow us to accout for the separatio of Czechoslovakia ito two coutries. 14

20 FDI i period t. The higher the ratio of recet FDI iflows, FDI t m= 0 m, to the total stock of FDI i the coutry, the ewer the vitage of FDI i the coutry, ad the more we expect to see the reivestmet of earigs characteristic of Stage 2 rather tha the capital repatriatio of Stage 3. We use cumulated ivestmet for the past oe, two ad three years for the pael that icludes trasitio ecoomies ad FDI cumulated for from oe to five years for the loger pael that does ot iclude the trasitio ecoomies. Because there are coutry-specific factors such as the actual stock of FDI i the coutry, domestic market size, access to export markets, political stability, etc., at work, we employ a pael estimatio with fixed effects to take these coutry-specific factors ito accout. To test whether the timig of the FDI fiacial life cycle differs betwee developed, developig ad trasitio ecoomies, we specify Equatio 1 for estimatio purposes as: Y i, t c, i +, i, t X, i, t + β, i, tdum _ 1, i, t X, i, t + γ, i, tdum _ 2, i, t = α X + ε Eq. 2 where Y i, t = (BI_debit/CA_debit) i, t ; X, i, t = FDI t m= 0 m, i, t, i, t /Stock_FDI i, t ; ad two slope dummy variables are itroduced: dum_1 ad dum_2. These dummy variables are defied as follows: dum_1 = 1 if coutry i is a developig coutry = 0 otherwise, dum_2 = 1 if coutry i is a trasitio coutry = 0 otherwise. From this pael estimatio we ca calculate the slope for the developig coutries as α + β, ad for the trasitio coutries as α + γ. For the developed coutries, the slope coefficiet is equal to α. The results of the estimatio for the pael that icludes the trasitio ecoomies are reported i Table 2. The coefficiet for the developed coutries, α, is sigificat for all values of, the 15

21 umber of years over which we cumulate the most recet flows of FDI, ad it decreases steadily with. The effect of a ewer vitage of FDI capital stock i the ratio of icome balace debits to curret accout debits is eve greater for developig ad trasitio ecoomies because the coefficiets β ad γ are also positive ad sigificat. This implies that the propesity to reivest earigs o FDI is be greater i these coutries tha it is i developed coutries because recet acquisitios i these coutries may ivolve firms that are i greater eed of restructurig ad upgradig of techology ad capacity tha is the case i acquisitios made i developed coutries. For example, Geeral Electric, which acquired Hugary s flagship maufacturig firm, Tugsram, had to udertake a aggressive program of ivestmet i its acquisitio i order to make it fully competitive o the world market for light bulbs (Marer ad Mabert, 1999). Similarly, Volkswage had to make large ivestmets i its Czech acquisitio, Škoda, i order to brig the firm's products, techology ad quality stadards up to par (Bohatá, 2000). Alteratively, the legth of Stage 1 may differ betwee developed ad developig ad trasitio coutries. The differece betwee developed ad trasitio ecoomies, however, does ot exted to beyod =2, sice thereafter γ is o loger statistically sigificat. A obvious problem with icludig trasitio ecoomies i our sample is that it is ot possible to costruct a balaced pael that goes back beyod the start of the 1990s. Therefore, i order to explore larger values of, we use a subset of our coutry pael that icludes oly developed ad developig coutries, ad exted the data back to 1981, thus allowig us to icrease the value of beyod the value of 3 used with the larger sample of coutries. The results for this regressio are reported i Table 3. The value of α is, except for =1, quite similar to that obtaied from the sample used to geerate the estimates reported i Table 2. It does, however, decrease with icreasig, agai reflectig the decreasig ifluece of reivested profits as we iclude older 16

22 vitages of FDI i the explaatory variable. However, i cotrast to the results i Table 2, β, the slope dummy for developig coutries, is ot sigificat for ay value of, ad thus the effect of the FDI fiacial life cycle is the same for both sets of coutries. Give differeces i time ad coutry coverage betwee the two samples, these mior differeces are ot surprisig. What is importat is that, for both samples, there exists a positive relatioship betwee the vitage of a coutry's FDI stock ad the share of icome balace deficits items i the curret accout deficit. Give the rather low values of at which this pheomeo is observable, it is reasoable to coclude that the debit item o the icome balace that is icreasig as the vitage of the capital stock becomes ewer is the imputatio of reivested profits rather tha divided repatriatio. 5. CONCLUSIONS We have show that the imputatio of reivested earigs as a debit item i the balace of paymets of host coutries creates a situatio where the curret accout deficit ca appear to be i deficit eve though there is o eed to fiace some or a large part of this deficit o the foreig exchage market. We have also show that, because of the workigs of the FDI fiacial life cycle, such a bias is most evidet for coutries that have recetly received large iflows of capital. Our aalysis also shows that two of the trasitio ecoomies of East Europe, the Czech Republic ad Hugary, have received large iflows of FDI over a short spa of years, ad this imputatio has had a large effect o their curret accout balaces. We verified the workig of the FDI fiacial life cycle usig two differet paels of developed, developig ad trasitio ecoomies. Coutries that do ecouter large iflows of FDI, especially if existig stocks of FDI are relatively small, should make a effort to call attetio to this pheomeo so that foreig 17

23 ivestors ca evaluate their ecoomic performace more accurately. Moreover, trasitio ad developig coutries that have ot take care to report reivested MNC earigs i their balace of paymets accout should take care to do so i order to clarify the fiacig eeds implied by their reported curret accout deficits. 18

24 6. REFERENCES Berg, Adrew, ad Pattillio Catherie, Are currecy crises predictable? IMF Staff Papers, Vol. 46, No.2, (1999), pp Bohatá, Marie, "Škoda automobilová a.s.", i Saul Estri, Xavier Richet ad Josef C. Brada (eds.), Foreig Direct Ivestmet i Cetral Easter Europe: Case Studies of Firms i Trasitio. Armok ad Lodo: M. E. Sharpe, Burkhart, Oliver, ad Coudret, Virgiie, Leadig idicators of currecy crises for emergig coutries. Emergig Markets Review, Vol. 3, (2002), pp CNB: Balace of Paymets Aual Report Prague, Czech Natioal Bak, CNB: Foreig Direct Ivestmet Prague, Czech Natioal Bak, March CNB: Balace of Paymets Statistics, Czech Natioal Bak, November CZSO: Natioal Accouts Statistics, Czech Statistical Office, November Eichegree, Barry, Rose, Adrew, ad Wyplosz, Charles, Cotagious currecy crises. Natioal Bureau of Ecoomic Research Workig Paper No. 5681, Feradez-Arias, Eduardo, ad Hausma, Ricardo, Is foreig direct ivestmet a safer form of ivestmet? Emergig Markets Review, Vol. 2, No. 1, (March 2002), pp Goldfaj, Ila, ad Valdéz, Rodrigo, Are currecy crises predictable? Europea Ecoomic Review, Vol. 42, No. 3-5, (1998) pp Goldstei, Morris, Kamisky, Gracilela L., ad Reihart, Carme R., Assessig fiacial vulerability; a early warig system for emergig markets. (mimeo). Washigto DC: Istitute for Iteratioal Ecoomics, GUS: Natioal Accouts Statistics, Statistical Office of Polad, November IMF: Balace of Paymets Maual. Washigto, Iteratioal Moetary Fud, IMF: Balace of Paymets Statistics Yearbook. Washigto, Iteratioal Moetary Fud, IMF: Iteratioal Fiacial Statistics Yearbook Washigto, Iteratioal Moetary Fud, INEP: Natioal ad Regioal Accouts, Natioal Istitute of Statistics i Portugal, November INGE: Natioal Accouts of Brazil, Brazil Istitute of Geography ad Statistics, November Kamisky, Graciela L., ad Reihart, Carme M., The twi crises: the causes of bakig ad balace of paymets problems. America Ecoomic Review, Vol. 89, No. 3 (Jue, 1999) pp Koretz, Gee, "Ecoomic treds: the U.S. fially pays the piper." Busiess Week, November 11,

25 Krugma, Paul (ed.), Currecy Crises. Chicago: Uiversity of Chicago Press, KSH: Natioal Accouts Statistics, Hugaria Cetral Statistical Office, November Marer, Paul ad Mabert, Vicet, "Tugsram", i Josef C. Brada ad Iderjit Sigh (eds.), Corporate Goverace i Cetral Easter Europe: Case Studies of Firms i Trasitio: Armok ad Lodo: M. E. Sharpe, Milesi-Ferrretti, Giamaria, ad Razi, Assaf, Sustaiability of persistet curret accout deficits. Natioal Bureau of Ecoomic Research Workig Paper No. 5467, NBB: Time Series: Balace of Paymets Statistics, Natioal Bak of Brazil, November NBH: Balace of Paymets Statistics, Natioal Bak of Hugary, November NBP: Balace of Paymets Statistics, Natioal Bak of Polad, November NBP: Balace of Paymets Statistics, Natioal Bak of Portugal, November NBS: Balace of Paymets Statistics, Natioal Bak of Slovakia, November OECD: OECD Ecoomic Surveys: Hugary, Paris, Vol. 2002/10, Jue SOSR: Natioal Accouts Statistics, Statistical Office of the Slovak Republic, November Summers, Laurece H., Commetary. I Ricardo Hausma ad Liliaa Rojas-Suarez (eds.) Volatile capital flows. Washigto, DC: Iter-America developmet Bak, Summers, Laurece H. Iteratioal fiacial crises: causes, prevetio ad cures. America Ecoomic Review, Vol. 90, No. 2 (May, 2000) pp

26 Box 1 Reivested Earigs i the IMF Balace of Paymets Methodology A. Curret accout 1. Trade balace 2. Balace of services 3. Icome balace 3.1. Credit Iterest accepted, icome from CB reserves Icome from work abroad Divideds ad distributed earigs Reivested earigs abroad 3.2. Debit Iterest paid Paymets to foreig workers Divideds ad distributed earigs Reivested earigs i the reportig coutry B. Capital accout C. Fiacial accout 1. Direct ivestmet 1.1. Abroad (debit) Equity capital Other capital Reivested earigs abroad 1.2. I the reportig ecoomy (credit) Equity capital Other capital Reivested earigs i the reportig coutry 2. Portfolio ivestmet 3. Fiacial derivatives 4. Other ivestmet D. Net errors ad omissios, valuatio chages E. Chage i reserves (-icrease) Source: Compiled from IMF, Fifth Balace of Paymets Maual, 1993, p

27 Figure 1. The FDI Fiacial Life Cycle Stage 1: Etry Stage 2: Growth Stage 3: Ivestmet Repatriatio $ Profit Reivested Profit Divideds 0 TIME 22

28 Figure 2: Reivested Earigs i the Balace of Paymets of the Czech Republic ad Hugary (mil. EUR) Note: Oly prelimiary data are available for Reivested Earigs Net FDI Divideds + Iterest Remittaces Balace of Icome Czech Republic Hugary of w hich Net Reivested Earigs (OECD estimates i Hugary) Net Icome Receipts Net Foreig Direct Ivestmet Iflow Balace of Icome of w hich Net Reivested Earigs (OECD estimates i Hugary) Sources: Czech Natioal Bak, Natioal Bak of Hugary, ad OECD Ecoomic Surveys: Hugary, Paris, Vol. 2002/10, Jue 2002, p

29 Figure 3: Structure of the Icome Accout of the Czech Republic ( all items are et i mil. CZK) Forecast by Authors Revised Figures Revised Figures Revised Figures Revised Figures Revised Figures Prelimiary Figures Revised Figures Prelimiary Figures Revised Figures Reivested Earigs Iterest Balace Divideds ad Distributed Earigs Compesatio of Employees Balace Source: Balace of Paymets Statistics at the website of the Czech Natioal Bak ( Data o reivested earigs available i Foreig Direct Ivestmet Prague, Czech Natioal Bak, March 2002, p

30 Table 1. Importace of Reivested Profits I the Curret Accouts of Four Coutries Pael A. Brazil Stock of FDI mil. USD (1) Profits o FDI i Brazil mil. USD (2) Of which: Reivested Earigs mil. USD (3) Ratio of Reivested Earigs to FDI Profits (4 = 3/2) % % % % % % % NA NA NA NA NA NA Curret Accout Balace (Icludig Reivested Earigs) Curret Accout Balace (Excludig Reivested Earigs) mil. USD % of GDP mil. USD % of GDP CA Balace Differece as Percet of GDP NA NA NA NA NA NA NA NA NA Source: IMF Database; Cetral Bak of Brazil 25

31 Stock of FDI mil. USD (1) Pael B. Czech Republic Profits o FDI i Czech Rep. mil. USD (2) Of which: Reivested Earigs mil. USD (3) Ratio of Reivested Earigs to FDI Profits (4=3/2) NA NA % % % % Curret Accout Balace (Icludig Reivested Earigs) Curret Accout Balace (Excludig Reivested Earigs) mil. USD % of GDP mil. USD % of GDP CA Balace Differece as Percet of GDP Source: WIIW Database, IMF Database, Czech Natioal Bak Stock of FDI mil. USD (1) Pael C. Irelad Profits o FDI i Irelad mil. USD (2) Of which: Reivested Earigs mil. USD (3) Ratio of Reivested Earigs to FDI Profits (4=3/2) % % % % 26

32 Curret Accout Balace (Icludig Reivested Earigs) Curret Accout Balace (Excludig Reivested Earigs) mil. USD % of GDP mil. USD % of GDP CA Balace Differece as Percet of GDP Source: IMF Database; Cetral Bak of Irelad Stock of FDI mil. USD (1) Pael D. Portugal Profit o FDI i Portugal mil. USD (2) Of which: Reivested Earigs mil. USD (3) Ratio of Reivested Earigs to FDI Profits % % % % % % Curret Accout Balace (Icludig Reivested Earigs) Curret Accout Balace (Excludig Reivested Earigs) mil. USD % of GDP mil. USD % of GDP CA Balace Differece as Percet of GDP Source: IMF Database; Cetral Bak of Portugal 27

33 Table 2: Parameter Estimates from Pael Estimatio with Fixed Coutry Effects for 32 Coutries ( )* No. of Years Over Which FDI Iflows Are Cumulated Parameter Estimate t-statistic Probability value No. Obs. Adjusted F- Statistic R 2 α =1 β dum _ γ dum _ α =2 β dum _ γ dum _ α =3 β dum _ γ dum _ * Coutry dummies available from the authors upo request. 28

34 Table 3: Parameter Estimates from Pael Estimatio with Fixed Coutry Effects for 18 Developed ad Developig Coutries* ( )** No. of Years Over Which FDI Iflows Are Cumulated Parameter Estimate t-statistic Probability value No. Obs. Adj. R 2 F-Statistic =1 =2 =3 =4 =5 α β dum _ α β dum _ α β dum _ α β dum _ α β dum _ * I order to exted the series back to 1981, we had to exclude Chile, Idia, Peru, South Africa, Turkey, ad Veezuela as well as all the trasitio ecoomies. ** Coutry dummies available from the authors upo request. 29