Is Singapore the Model for China s New Exchange Rate Policy? Bennett T. McCallum. Carnegie Mellon University. and

Transcription

1 Is Singapore he Model for China s New Exchange Rae Policy? Benne T. McCallum Carnegie Mellon Universiy and Naional Bureau of Economic Research Augus 19, 2005 Revised: November 30, 2005 Prepared for Shadow Open Marke Commiee meeing of November 6-7, The auhor is indebed o Khor Hoe Ee, Marvin Goodfriend, Edward Robinson, Celine Sia, and Lin Tian for helpful discussions and commens.

2 1. Inroducion Few economic evens of recen years have araced as much aenion as he announcemen by he People s Bank of China on July 21, 2005, ha henceforh he exchange rae of he Renminbi (RMB) would no be kep fixed in relaion o he U.S. dollar, bu would be managed relaive o a baske of currencies and adjused more frequenly via a managed floa arrangemen. Various commenaors picked up on saemens by informed observers o he effec ha he new sysem would be somewha like ha of Singapore. Buron (2005), for example, saed ha China and Malaysia... adoped a version of is [i.e., Singapore s] managed floaing exchange rae sysem afer abandoning heir fixed currency pegs agains he US dollar. (Also see Areddy, e. al. (2005).) The cenral aims of he presen paper are o explain he Moneary Auhoriy of Singapore s exchange rae/moneary policy sysem, which has been unique among he world s cenral banks, and o consider he likelihood ha Chinese exchange rae policy will in fac closely resemble ha of Singapore. I will be argued ha in pracice he degree of similariy is likely o be very small, a leas for several years o come. Before beginning, i migh be of some ineres o explain why some newspapers (such as he Financial Times) refer o he Chinese currency as he Renminbi while ohers (including he Wall Sree Journal and The Economis) usually refer o he Yuan. To he Chinese, renminbi is he name of he currency, ren min bi meaning somehing like he peoples money, whereas he yuan is he uni of accoun. Thus he Renminbi (RMB) exchange rae was (prior o July 21) abou 8.28 yuan/dollar. Evidenly, some newspapers find his usage edious or confusing. From a Chinese perspecive, however, o use yuan as he name of he currency would seem o be grammaically incorrec. 1

3 2. The Singapore Sysem Le us begin wih an informal descripion of he Singapore sysem, before presening i in erms of an analyical model. I will quoe from a one-page summary by he Moneary Auhoriy of Singapore (MAS) iself, as follows. Since 1981, moneary policy in Singapore has been cenered on he managemen of he exchange rae. The primary objecive has been o promoe price sabiliy as a sound basis for susainable economic growh. The exchange rae represens an ideal inermediae arge of moneary policy in he conex of he small and open Singapore economy... Firs, he Singapore dollar is managed agains a baske of currencies of our major rading parners and compeiors... Second, MAS operaes a managed floa regime... The rade-weighed exchange rae is allowed o flucuae wihin a policy band, he level and direcion of which is announced semi-annually o he marke... Third, he exchange rae policy band is periodically reviewed o ensure ha i remains consisen wih he underlying fundamenals of he economy. (MAS, undaed) A careful reading of he foregoing, plus addiional descripions by MAS, reveals a crucial aspec of his procedure. I is ha he band, wihin which he Singapore dollar (S$) exchange rae is kep, is no consan hrough ime. Insead, he band may move upward or downward auomaically as ime passes (o allow for expeced ongoing appreciaion or depreciaion) and, more imporanly, boh he level and slope of he band and even is widh may be discreely adjused each decision period. These adjusmens are made so as o keep inflaion low i.e., o promoe price sabiliy. Some adjusmens of he band may, as well, be made in response o prevailing (or forecased) behavior of real variables such as aggregae oupu or 2

4 employmen. Thus he ype of exchange rae managemen employed by MAS is very differen from a radiional fixed exchange rae. In fac, i would appear ha he MAS operaes wih policy objecives quie similar o hose of he Federal Reserve or he European Cenral Bank or he Bank of England, i.e., o mainain low inflaion as a prioriy, wih some aenion also paid o oupu/employmen consideraions. Furhermore, he MAS procedures seem very much like hose of hese oher cenral banks excep ha is policy managemen involves periodic adjusmens in he exchange rae, raher han a shor erm nominal ineres rae. The reason for his difference in policy behavior is, moreover, quie sraighforward and simple: he Singapore economy is much more open o foreign rade han hose of he Unied Saes or he euro area, or he Unied Kingdom. Insead of an expor/gdp raio of abou 0.15 (or abou 0.25 for he UK), for Singapore he value is currenly abou ! Thus he exchange rae channel of moneary policy ransmission is much more imporan, relaive o he familiar ineres-rae channel, han in larger economies ha are less open o inernaional rade. Accordingly, making open-marke purchases or sales in he foreign exchange marke, raher han he domesic money marke, may be a relaively more effecive way of managing aggregae demand. 1 The policy comparison will be illusraed below. The foregoing suggesion ha he MAS policy framework is basically one in which inflaion is he main arge variable, wih he exchange rae being used primarily as an insrumen or indicaor for specifying policy changes ha are designed o keep inflaion close o arge, is suppored by he behavior of he exchange rae over he years The period discussed begins wih 1981 because ha is he year in which he curren MAS policy regime was pu in place, according o MAS (undaed, 2001). The saisics indicae ha, over 1 One should no infer, however, ha adjusmens in he exchange rae are necessarily implemened by open marke purchases in he foreign exchange marke. Excep when ineres raes are a (or near) zero, hey could alernaively be implemened by purchases in he domesic money marke. 3

5 he span from 1981 o he middle of 1997, he S$ appreciaed in value by abou 80 percen relaive o he policy baske, despie a large drop in This appreciaion was needed o preven inflaion since foreign inflaion was proceeding a a rae higher han he Singapore arge and also because rapid produciviy growh in Singapore was bringing abou an ongoing appreciaion in real erms due o he Balassa-Samuelson effec. Afer a fall during he Asian financial crisis of , he value of he S$ levelled off and has no changed much since. Even so, he value of he S$ remains abou 70 percen higher in 2005 han in 1981 in erms of he (rade-weighed) baske. The mos common formulaic descripion of moneary policy procedures for more ypical economies is provided by some varian of he Taylor rule, inroduced by John Taylor (1993), which relaes periodic adjusmens in a money-marke ineres rae in response o exising (or prediced) inflaion and oupu-gap measures. A common formulaion is (1) R = r + p + µ 1 ( p π*) + µ 2 (y y) + η µ 1, µ 2 0 where R is he ineres rae, p is he curren inflaion rae, π* is he arge inflaion rae (a which he cenral bank wishes o keep inflaion on average), and y y is he oupu gap, i.e., he percen (or fracion) by which real oupu exceeds he naural rae of oupu ha represens an efficien, marke-clearing level. The erm η represens random policy influences by he cenral bank, which in principle should be very small. In comparison o (1), he Singapore policy rule migh be represened as follows: (1 ) e = e p + µ 1 ( p π*) + µ 2 (y y) + η, µ 1, µ 2 0. Here e is he nominal exchange rae, expressed as foreign currency unis per uni of homecounry money (e.g., yen/dollar if he Unied Saes is aken as he home counry). Accordingly, e is he average rae of appreciaion of he currency (perhaps negaive) ha 4

6 reflecs he sum of he long-run rae of appreciaion of he real exchange rae plus he average inflaion rae abroad. Clearly, moneary policy designed o reduce inflaion when i is above is arge value would call for an increase in e under his rule, raher han an increase in R. This desired increase could in principle be brough abou by he cenral bank by conducing an open-marke sale of foreign exchange, alhough in normal circumsances i could be effeced by a sale of shor-erm securiies, as would usually be he case wih he Taylor rule (1). 2 I should be emphasized ha he policy behavior described by (1 ) is no inended o keep he exchange a any paricular value oher han whaever would be consisen wih he inflaion and oupu gap arges specified on he righ-hand side of he relaionship. Is here any reason o believe ha in realiy MAS behaves in a manner similar o he rule (1 )? In ha regard, MAS Saff Paper No. 31, 2004, wrien by Eric Parrado of he IMF, uses monhly daa for o esimae a rule of he form (1 ) bu wih inclusion of an addiional e -1 erm o reflec smoohing of he exchange rae. (Also, his preferred equaion uses he expeced inflaion rae nine monhs ino he fuure, raher han he curren rae.) Parrado s insrumenal-variable esimaes are as follows: 3 e = E -1 p (y y ) e -1 (0.009) (0.55) (0.14) (0.022) R 2 = 0.86 J-sa p-value = 0.85 Clearly hese provide srong suppor for he suggesion made above. I mus be said ha he MAS normally does no describe is policy in his manner, insead emphasizing he BBC aspecs (baske, band, crawl) regarding he exchange-rae 2 If foreign exchange and domesic shor-erm securiies were perfec susiues, hen a purchase (of a given size) of eiher would have he same effec. This paper s analysis presumes ha hese wo asses are close bu no perfec subsiues. 3 Here he figures in parenheses are sandard errors, he R 2 saisic is unadjused, and he repored p-value is Hansen s J saisic for esing he hypohesis ha he assumed orhogonaliy condiions are valid. 5

7 ha have been emphasized in he work of Williamson (1999) see Khor, Robinson, and Lee (2004). Bu if he band and is crawl are designed primarily so as o achieve arges for p and y y, hen his amouns o he same hing (as is argued above). 3. Open-Economy Model Le us now indicae how eiher of he foregoing policy rules could be combined wih a formal model, of an economy open o foreign rade, for he purpose of moneary policy analysis. One paricular example of such a model is he one uilized by McCallum and Nelson (1999) or McCallum (2005), which differs from many oher opimizing models by reaing impors as raw maerials for he producion process raher han finished consumer goods. I is a small-open-economy model ha can be summarized wih he following equaions: (2) c = E c +1 + b 0 b 1 r + v (3) y = ω 1 c + ω 2 g + ω 3 x (4) im = y σq + cons (5) q = s p + p* (6) x = y* + σ*q + cons (7) (8) y = (1 α ) [ α a σα q ] + cons p = (1 +β) [ βe p + p ] +κ(y y ) + u (9) R R* = E s + 1+ξ (10) r = R E p + 1 A very brief descripion of each will be provided. Equaion (2) is a consumpion (c ) Euler equaion, reflecing ineremporal opimizaion, while (3) is a log-linearized approximaion o 6

8 an ideniy ha splis oupu y no value added! ino hree componens: consumpion, governmen consumpion g, and expors x. 4 Nex, in (4) impor demand im is given by cos minimizaion for a producion funcion of he CES ype wih σ as he elasiciy of subsiuion beween impors and labor. An analogous relaion (6) governs demand from abroad for home-counry expors. Equaion (5) defines he log of he real exchange rae q in relaion o he log of he nominal exchange rae s and he logs of home and foreign price levels, p and p*. Equaion (7) specifies he naural rae (i.e., flexible-price) value of he log of real oupu, y, wih his value depending upon a sochasic erm a ha reflecs he resuls of echnology shocks (assumed o follow an exogenous AR(1) process wih auocorrelaion parameer 0.95) and he real price of impored inpus o producion. A varian of he Calvo model of nominal price sickiness appears as (8) while (9) represens uncovered ineres rae pariy, wih a sochasic disurbance. 5 Finally, (10) is he Fisher ideniy ha defines he oneperiod real rae of ineres r in relaion o he nominal rae R and expeced inflaion. Togeher wih he Taylor syle policy rule (1), his model provides 10 srucural equaions o generae values of he sysem s 10 endogenous variables, namely, c, y, g, x, im, p, s, q, R, and r. Thus we can very simply esablish he main poin of his secion, which is ha adopion of he s policy rule (1 ) would no aler he liss of endogenous and exogenous variables. Consequenly, i follows ha use of s as he policy-rule insrumen, raher han he more sandard R, is perfecly sensible and coheren. Which of he wo insrumen/indicaor variables would be more desirable will be deermined by quaniaive aspecs of he economy under consideraion. 4 Domesic invesmen would also be included in a model ha disinguishes beween consumpion and invesmen spending. 5 This disurbance incorporaes our assumpion ha foreign and domesic securiies are no perfec subsiues. 7

9 To make such a deerminaion for he model given above, quaniaive values have o be assigned o each of he model s parameers, including hose ha describe he sochasic behavior of he exogenous variables and shocks ha impinge upon he sysem. In McCallum (2005) I have calibraed he model (2)-(10) o represen a ypical indusrial economy, seing he average raio of impors (and expors) o producion (no value added) a For Singapore he comparable figure is approximaely I will be of ineres o compare he performance of policy rules (1) and (1 ), wih smoohing added in each case, under hese wo specificaions of he economy s degree of openness, wih he oher aspecs of he calibraion kep he same. The relevan comparison is provided in Table 1. There X/Y denoes he raio of he economy s expors (and impors) o producion, which is varied over a wide range in Table 1: Effecs of Openness on Policy Rule Performance Cell enries are sandard deviaions of p, y%, R, s X/Y = 0.01 X/Y = 0.15 X/Y = 0.30 X/Y = 0.60 R rule (1) s rule (1 ) he differen columns. For a given calibraion of he model, described in McCallum (2005), he wo rows of cells repor he variabiliy of inflaion, he oupu gap, he ineres rae, and he exchange rae s rae of appreciaion. Wih all variables measured as percenage 6 The model also includes a feaure represening habi formaion in consumpion behavior. 7 Singapore expors (X) and impors (M) are each abou 1.5 imes as large as GDP, implying a value of 0.6 for M/Y. To see his, noe ha GDP = Y M, so Y/M = 1 + GDP/M. 8

10 deviaions from seady-sae values, quarerly bu in annualized unis, he figures for inflaion and he oupu gap represen roo-mean-square deviaions from arge. Thus small values are more desirable han large values. Going across he op row, we see ha wih an ineres rae insrumen rule poorer performance is realized wih highly open economies. Moving from X/Y of 0.15 o 0.60, o be specific, resuls in approximaely no change in inflaion variabiliy bu a major increase in oupu gap variabiliy. Effecs on he variabiliy of ineres and appreciaion raes are minor. In he second cell row, by comparison, he exchange rae rule is increasingly effecive in sabilizing inflaion and oupu as he degree of openness is increased. Thus for he model a hand, i is clearly he case ha an increased degree of openness makes use of he exchange rae rule relaively more aracive. Does he very high level of openness refleced by X/Y = 0.6 also make rule (1 ) more aracive in absolue erms? From he las column of Table 1 we see ha in ha case variabiliy of inflaion is (slighly) higher bu variabiliy of he oupu gap is (grealy) reduced by use of he exchange rae rule. The answer will hen depend upon he weigh assigned by he relevan objecive funcion o oupu-gap variabiliy. 8 If he value were 0.1 for he laer relaive o inflaion variabiliy (in erms of variances), hen he exchange rae rule (1 ) would be preferable. Weighs somewha lower han 0.1 are no uncommon in he lieraure, however, so he absolue superioriy of (1 ) is no a foregone conclusion. Also, i is possible ha he variabiliy of R and s or s would be aken ino accoun by he relevan cenral bank. Accordingly, no conclusion of he absolue ype can be made on he basis of our simple sudy. For his ype of comparison, a more precise numerical calibraion of he model 8 If X/Y were assumed o be 0.75, however, he exchange rae rule would would resul in inflaion and oupu sandard deviaions of 2.08 and 2.15, boh smaller han he values 2.56 and 7.11 provided by he ineres rae rule. 9

11 and a more careful consideraion of he appropriae objecive funcion would have o be developed. These are asks ha are beyond he scope of his paper. 4. Relevance for Chinese Policy The pas wo secions have developed a characerizaion of Singapore moneary policy and illusraed is relaive effeciveness for economies wih very high raios of rade o domesic producion. In ligh of hese findings, we now reurn o he suggesion ha China s new exchange rae policy is or will be similar o ha of Singapore. 9 In ha regard, he firs poin o noe is ha China does no have a very high raio of rade o domesic producion so a major reason why Singapore s sysem is aracive does no apply o China. Nex i is imporan o recognize ha he MAS policy regime is enirely differen from a fixed exchange rae. In fac, he MAS does no even rely upon auonomous arge values for he exchange rae, bu merely values ha are believed o be helpful (emporarily) in achieving MAS objecives for inflaion (and possibly oupu). Some wriings, such as Williamson (1998, 1999), have argued ha such an arrangemen is no a pure floa of he exchange rae, bu i is no clear wha is mean by ha objecion. As Rober Mundell frequenly emphasizes, he erm floaing exchange rae, is highly incomplee as a descripion of a policy regime. 10 The erm fixed exchange rae consiues a descripion of moneary policy, since moneary policy and exchange rae policy are no wo independen eniies, bu floaing exchange rae is a highly incomplee policy specificaion. 11 I is my argumen ha, for he MAS, he exchange rae is imporan primarily in is role as an inermediae informaion or insrumen variable ha is involved in he procedures used o achieve he objecive of low inflaion, augmened perhaps by oupu gap consideraions. The Singapore 9 Thus we are using our normaive analysis o provide suggesions abou wha will be done in pracice. 10 See, for example, Mundell (2001, p. 13). 11 Acually, here are significan varians of fixed-rae regimes: currency unions, currency boards, pegs, ec. 10

12 sysem is, herefore, one varian of a floaing exchange-rae regime. For a sysem of he ype jus described, i is crucial ha adjusmens in he cenral bank s seing of he exchange rae mus be made promply and vigorously. This does no mean han no high-frequency smoohing is admissible, bu ha (say) quarerly averages need o conform reasonably well o he dicaes of he policy rule. The laer, moreover, has in he case of Singapore resuled in sizeable adjusmens, as discussed above. In he case of China, by conras, we know ha he regime of a fixed exchange rae was given up in July 2005 wih grea relucance. 12 Furhermore, i seems clear han he Chinese auhoriies no necessarily he People s Bank of China, bu governmen auhoriies more generally remain srongly concerned wih he demand from abroad for Chinese expors. In par, his is because of a perceived need o keep he demand for labor high and rising so as o preven poliical unres. Accordingly, i would appear ha he likelihood, ha China is now or will be managing is exchange rae (and moneary policy) in a manner similar o he pracice of Singapore, is very small. Some ype of floaing exchange-rae regime may well be adoped evenually for he Chinese economy, so as o permi moneary policy acions ha are appropriae domesically. Bu in ha case he sysem would probably be unlike Singapore s, since China s degree of openness is much smaller. In sum, here seems o be lile or no reason o believe ha a sysem like Singapore s has been adoped by China or will be adoped by China in he near fuure. 12 Perhaps his fac accouns for he prevalence of press commenaries suggesing ha he RMB vs. baske exchange rae will be held fixed, despie he announcemen ha i will floa. (See, for example, Chinese Puzzles, The Economis 376, Augus 13, 2005, p. 61, which saes: In heory, if he dollar falls agains he oher currencies, he People s Bank of China should le he yuan [sic] rise agains he dollar in order o hold he overall value of he baske seady. ) 11

13 References Areddy, James T., Mary Kissel, Andrew Browne, and Michael M. Phillips, 2% Soluion: China Les Yuan Rise vs. Dollar, Wall Sree Journal, July 22, Buron, John, Singapore Ses Model for China s Renminbi Reform, Financial Times, July 22, Khor, Hoe Ee, Edward Robinson, and Jason Lee, Managed Floaing and Inermediae Exchange Rae Sysems: The Singapore Experience, Moneary Auhoriy of Singapore, MAS Saff Paper No. 37, December McCallum, Benne T., Theoreical Analysis Regarding a Zero Lower Bound on Nominal Ineres Raes, Journal of Money, Credi, and Banking 32, 2000, , A Moneary Policy Rule for Auomaic Prevenion of a Liquidiy Trap, NBER Working Paper 11056, January Also in Takaoshi Io and Andrew Rose, eds., Moneary Policy wih Very Low Inflaion in he Pacific Rim. Universiy of Chicago Press for NBER, forhcoming. McCallum, Benne T., and Edward Nelson, Nominal Income Targeing in an Open- Economy Opimizing Model, Journal of Moneary Economics 43, June 1999, Moneary Auhoriy of Singapore, Singapore s Moneary Policy Framework, undaed. Singapore s Exchange Rae Policy, February 2001., Moneary Policy Operaions in Singapore, January

14 Mundell, Rober, One World, One Money? (Debae wih Milon Friedman) Opions Poliiques, May 2001, Parrado, Eric, Singapore s Unique Moneary Policy: How Does i Work? Moneary Auhoriy of Singapore, MAS Saff Paper No. 31, June Taylor, John B., Discreion Versus Policy Rules in Pracice, Carnegie-Rocheser Conference Series on Public Policy 39, December 1993, Williamson, John, Crawling Bands or Monioring Bands: How o Manage Exchange Raes in a World of Capial Mobiliy, Inernaional Finance 1, Ocober 1998, , Fuure Exchange Rae Regimes for Developing Eas Asia: Exploring The Policy Opions, presened a conference organized by he Insiue of Policy Sudies, Singapore, June 21-22,