THE WELFARE COST OF CAPITAL IMMOBILITY AND

Transcription

1 THE WELFRE COST OF CPITL IMMOBILITY ND CPITL CONTROLS nthony J. Makn & lexander Robson School of Economcs The Unversty of Queensland Brsbane 4072 ustrala ph: fax: emal: makn@economcs.uq.edu.au Department of Economcs The ustralan Natonal Unversty Canberra CT 0200 ustrala November 2002 Dscusson Paper No 318 ISSN Makn, 2002 Ths dscusson paper should not be quoted or reproduced n whole or n part wthout the wrtten consent of the author.

2 THE WELFRE COST OF CPITL IMMOBILITY ND CPITL CONTROLS nthony J. Makn bstract: Ths paper examnes the macroeconomc welfare effects of nterest rsk prema and controls that lmt nternatonal captal moblty. Usng extended loanable funds analyss, t frst demonstrates how perfect captal moblty maxmses natonal ncome, contrary to a prevalent vew that t s nmcal to economc welfare. s a corollary, the analyss then shows that captal controls, rrespectve of ther form, generally reduce natonal ncome and economc welfare by wdenng real cross-border nterest dfferentals. Captal controls n the form of quanttatve controls, such as the Chlean unremunerated reserve requrement system, and explct taxes on foregn nvestment flows mpose smlar welfare losses. However, quanttatve controls are relatvely more costly than optons to tax captal flows, due to revenue effects. JEL: F36, F40 The authors gratefully acknowledge the very useful research assstance provded by nthony Rosster.

3 1. INTRODUCTION Snce the demse of the Bretton Woods system of exchange rate management and consequent dsmantlng of a broad range of exchange controls, there has been an enormous growth n the volume of nternatonal captal flows. dvanced economes progressvely abolshed exchange controls from the early 1970 s onwards, whereas captal account lberalsaton n emergng economes accelerated most rapdly from the early 1990 s, accordng to an ndex of captal controls devsed by the Internatonal Monetary Fund (IMF, 1997, p.242). Meanwhle, lberalsed captal accounts have ncreased emergng economes vulnerablty to sudden nternatonal captal flow reversals of the magntude wtnessed n East sa and other emergng economes n the late 1990 s. In vew of the economc and fnancal dstress that short term captal flow reversals may cause, many economsts, such as Bhagwat (1998), Cooper (1998), Echengreen (1999), Krugman (1998), Rodrk (1998) and Wade (1998), favor the retenton of captal controls for emergng economes. Dscusson of captal moblty n the lterature 1 has focussed heavly on assessng condtons for t, and on ts mplcatons for the effectveness of short run macroeconomc polcy management. 2 Captal moblty has been measured aganst the standard nterest party condtons, as well as the extent of correlaton between domestc savng and nvestment (proposed by Feldsten and Horoka (1980)). Whle consderable attenton has been gven to examnng how captal moblty affects the potency of monetary and fscal polcy as stablsaton tools, relatvely lttle attenton has been pad to examnng how captal moblty 1 2 See surveys by Dumas (1995), Lews (1995), Marston (1995) and Isard (1995). See Mundell (1962) and Flemng (1962) and later surveys of ther model by Bruce and Purvs (1985) and Frenkel and Razn (1995), amongst others.

4 4 drectly affects natonal ncome as a measure of macroeconomc welfare. Ths paper frst demonstrates how nternatonal borrowng mproves natonal ncome and macroeconomc welfare. s a corollary, t then dentfes the welfare losses, or mplct costs, that arse when nternatonal captal moblty s less than perfect, due to offcal captal controls of varous knds. In prevew, these losses occur because captal mmoblty rases the external cost of captal, makng natonal ncome sub- optmal. However, the sze of the welfare loss depends on the knd of captal controls n place.

5 5 2. N EXTENDED LONBLE FUNDS FRMEWORK Internatonal captal flows are not purely fnancal phenomena snce nternatonal borrowng and lendng s ultmately ted to economc factors that determne savng and nvestment behavour. Intertemporal open economy models recognse ths, yet ther focus s external account determnaton wth reference to the behavour of savng, nvestment and ntertemporal consumpton under condtons of perfect captal moblty. 3 In what follows, captal moblty s related to savng, nvestment and the nternatonal flow of funds, consstent wth the ntertemporal approach. However, unlke ntertemporal models, the analyss s lmted to wthn-perod effects to dentfy the welfare costs of captal controls. 2.1 Captal utarky versus Perfect Captal Moblty Frst, we assume autarky and that domestc savng, S, the resdual from natonal ncome after prvate and publc consumpton plans have been satsfed, s fxed and nterest nelastc. Total nvestment spendng over a gven perod s funded out of avalable savng, wth the real nterest rate performng the balancng role. Demand for loanable funds s a functon of the real nterest rate: I = I () (1) where s the real nterest rate and I () < 0. In equlbrum under autarky, the domestc nterest rate s such that the market for loanable funds clears and I( ) = S (2) 3 See for nstance Frenkel and Razn (1987) and Obstfeld and Rogoff (1996).

6 6 Consequently, the domestc savng schedule s drawn vertcally, whereas the net domestc demand for nvestment purposes s a derved demand, depcted as a downward slopng schedule n nterest rate-loanable funds space n Fgure 1. Wth perfect captal moblty, a small economy s domestc borrowng requrement over and above avalable domestc savng s fully met by foregn lenders (nvestors) at the exogenous real world nterest rate,. Therefore, let S F be the foregn lendng schedule, where S F = S F (). Moreover, S F () =, as the foregn lendng schedule s nfntely elastc. Snce domestc frms wll only be wllng to borrow from abroad f >, assume ths condton also holds. The market for loanable funds must clear, so wth perfect captal moblty I ( ) S + = S (3) F where from the home economy perspectve, S F s foregn captal nflow n the form of borrowng. 4 S Domestc, World Interest Rates d g f b e c ρ SF S F I a d Loanable funds Fgure 1 Internatonal Captal Moblty and Macroeconomc Welfare 4 Ths paper restrcts attenton to borrowng, although foregn captal nflow can of course also nclude foregn ownershp of equtes ssued by resdent enterprses.

7 7 Domestc nvestment therefore exceeds domestc savng at to the extent of foregn borrowng. Ths ex ante foregn borrowng requrement s shown by dstance fc n the fgure. Hence, f external debt s ntally nl, t reaches level fc by perod end. s the real world nterest rate s lower than the real autarky nterest rate, and snce I () < 0, we must have I() > I( ), so that nvestment under autarky s always lower than when nternatonal borrowng s permtted. Here, and n what follows, we abstract from the effect that changng exchange rate expectatons have on nterest dfferentals by assumng that foregn lendng s denomnated n the currency of the lenders, thereby nullfyng exchange rate rsk from foregners perspectve. Ths s consstent wth the practce of advanced economy lendng to emergng economes, the bulk of whose loans are denomnated n foregn currency terms. lternatvely, t s possble to assume that exchange rate expectatons are statc throughout. Internatonal captal moblty s therefore perfect n ths context f foregn lenders satsfy the excess domestc demand for funds and real nterest party prevals. 2.2 Income Gans from Foregn Borrowng Fgure 1 also reveals how foregn borrowng rases natonal ncome, consstent wth McDougall s (1960) neoclasscal foregn nvestment model. The margnal product of captal determnes the slope of the nvestment demand schedule, so that gven, extra unts of foregn fnanced captal, tmes ther margnal product, add to GDP to the extent of the area abcd. However, of that the rectangular area, afcd s pad to foregn lenders, leavng a net natonal ncome gan equvalent to the trangular area fbc. Internatonal captal moblty

8 8 therefore enables lower domestc nterest rates and hgher natonal ncome, provded the productvty of the extra foregn-fnanced captal exceeds ts cost. More formally, the welfare gan under perfect captal moblty s: [ I( ) S ] d = I( ) d S.[ W = ] P (4) Note that, snce I() > S for all nterest rates between and, we must have W P > 0, so that the welfare gan from nternatonal borrowng s always postve. Interest pad to foregn nvestors s equal to [I - S ]. 2.3 The Costs of Captal Immoblty If foregn lenders perceve hgh foregn debt as a sgn of heghtened country rsk and dmnshed credtworthness, they demand an nterest premum, ρ, to compensate. Ths explans the convex foregn lendng schedule rsng from the world nterest rate, n Fgure 1. The more averse foregn nvestors are to rsng foregn debt, the steeper the slope of the S schedule and the hgher the rsk premum and nterest dfferental wll be. t some F pont, foregners could judge the level of lendng rsk prohbtve, such that the foregn lendng schedule becomes vertcal. Hence, the foregn lendng schedule s no longer perfectly elastc. Reflectng the assumpton that the rsk premum s an ncreasng functon of the stock of borrowng outstandng, t obeys, S F = S F () wth S F () > 0,. 5 The rsk premum, always postve, s the dfference 5 Of course, f the ntal level of debt exceeded zero, the foregn lendng schedule would rse from a pont above.

9 9 between the nterest rate foregn lenders demand under mperfect captal moblty and the nterest rate under perfect captal moblty. Hence, = d + ρ (5) where s the equlbrum domestc nterest rate. d Now the market for loanable funds must stll clear, so n equlbrum, I ( ) = S( ) (6) d d or, usng the above expresson for d, I( + ρ) = S ( + ρ) (7) Snce ρ > 0, and snce the demand for loanable funds s decreasng n the nterest rate, I( +ρ) < I() (8) Hence, under mperfect captal moblty, nvestment s lower than wth perfect captal moblty. Foregn debt related rsk therefore causes macroeconomc welfare losses snce potental natonal ncome gans from foregn borrowng are not realsed. Wth reference to Fgure 1, the welfare loss s area fgec. Note however that foregn borrowng stll confers a net welfare gan of gbe, provded the equlbrum nterest rate allowng for rsk s less than the autarky rate. lthough nternatonal captal mmoblty lmts an economy's growth, t also follows that the hgher the nterest rsk premum, the slower foregn debt accumulates, suggestng that rsng nterest rsk prema stablse foregn debt levels. More formally, the welfare gan from foregn borrowng under mperfect captal moblty, W I, s

10 10 W I = + [I() - S ρ ] d (9) The costs, or welfare losses, of mperfect captal moblty can now be quantfed smply as, L = + ρ [I() - SF ()] d (10) Ths loss s always postve, snce I () > S F () for all nterest rates n between and + ρ.

11 11 3. CPITL CONTROLS In the above benchmark cases, foregn nvestors lent funds through ther purchases of debt nstruments, wthout offcal restrctons of any knd mposed by the borrower economes. We now examne the macroeconomc welfare costs of mposng such restrctons. In practce, such controls range from those amed at lmtng the quantum of captal nflows to those n the form of taxes on captal nflows. What becomes evdent s that rrespectve of the type of captal control, the mnmum lendng rate demanded by foregn lenders, or alternatvely the mnmum yeld expected on bonds ssued by the borrowng economy, wll always be hgher than the prevalng world nterest rate, wth adverse mplcatons for natonal ncome. 3.1 Quanttatve Restrctons Frst we consder the welfare costs of va measures that restrct the quantum of captal nflows. The most common means by whch the domestc monetary authortes may lmt captal nflows s through mandatory unremunerated reserve requrements (URR). In the past, URR s have been most notably mplemented by Chle, but also by monetary authortes n rgentna, Brazl, Columba, Costa Rca, Czech Republc and Mexco. URR requres that a set percentage of funds borrowed from abroad be deposted wth the central bank for a mnmum perod. s no nterest s pad on the depost, ths effectvely makes the reserve requrement an mplct tax on captal nflows. 6 6 Under the Chlean system, foregn nvestors also had the opton of payng the central bank an amount equal to the forgone nterest wthout actually depostng funds, makng the tax on captal flows explct. See Neely (1999), De Gregoro et al (1999) and Ulan (2000)for related dscusson.

12 12 Reserve requrements therefore act to rase the mnmum nterest rate at whch foregn nvestors would lend to fnance addtonal domestc nvestment. Internatonal lenders are faced wth the choce of () lendng F unts for n perods to the economy wth the URR at a rate of q 0 or () buyng bonds offerng a yeld of. Wth opton (), f V s the future value of F at perod n, then V and F must be related accordng to the relatonshp, F V = (11) n 1+ ) q ( 0 Total captal nflow to the economy mposng the reserve requrement wll be F (1+s), where s represents the fracton of the nflow requred to be deposted wth the central bank for m perods, assumng m n. 7 t the end of perod m, the unremunerated reserve depost sf s refunded by the central bank and s lent at the rate. t perod n, both opton () and () nvestments mature. To determne the relatonshp between present values of the two nvestments are equated, yeldng: q 0 and the exogenous world nterest rate, the sf V (1 + s) F = + (12) m n (1 + ) (1 + ) q 0 By substtutng n the relatonshp between V and F, then rearrangng, the followng relatonshp between and s obtaned. q 0 m {( 1+ ) 1} m ( 1+ ) 1 s n = (1 + ) (13) q > 0 The above expresson suggests that the ntal mnmum lendng rate q 0 must rse as s and m rse or as n falls. However, for the purposes of estmatng welfare effects, we assume that 7 s a specal case, under the Colomban regme, the reserve requrement depended nversely on the maturty of the foregn loan.

13 13 m=n=1. gan, f nvestors contnue to be averse to rsng external ndebtedness, the equlbrum nterest rate wll be q1, nclusve of a rsk premum, and the macroeconomc welfare effects wll be as shown n Fgure 2 below. The welfare loss from captal mmoblty s area fhjc, whereas the net gan compared wth the autarky state s area hbj. Domestc, World Interest Rates q1 d q0 b h g f S j e q+ρ S F c ρ S F S F I a d Loanable funds Fgure 2 Macroeconomc Welfare Effects of Unremunerated Reserve Requrements Mathematcally, domestc resdents receve a welfare gan over the autarky state of, W q = = ] (14) [ I( ) S ] d I( ) d S [ q 1 q1 q1 Snce 1 >, ths welfare gan s lower than that under perfect captal moblty and lower q than under mperfect captal moblty n the absence of a reserve requrement. The overall net welfare loss s gven by: q 1 WP Wq = [ I( ) S ] d > 0 (15) and the loss specfcally due to the URR regme s:

14 14 W q 1 W = [ I( ) S d (16) I q ] 3.2 Taxes on Foregn Lendng lternatvely, captal controls may be n the form of explct proportonal taxes on prncpal loaned or nterest earned by foregn lenders (sometmes called wthholdng taxes). ssume foregners have F to nvest and can choose between () holdng foregn currency denomnated debt nstruments ssued by the borrower economy that mature at the end of perod n, payng a pre-tax rate of return of t0 or () holdng bonds at the alternatve world nterest rate of. If the proportonal rate of tax on foregn lendng s t, then arbtrage should ensure that, F n n (1+ ) = F (1+ ) (1 t ) (17) t 0 whch yelds the followng expresson 8 for t0. 1 n 1 (1 t ) t = (18) n n > (1 t ) (1 t ) Ths expresson shows that wth a dscrmnatory tax on foregn lendng, the mnmum return t0 demanded by foregn nvestors to fund domestc nvestment has to be hgher than, the hgher s t, or the lower s n. gan, for the purpose of gaugng the wthn perod macroeconomc welfare effects, however, we smply assume n = 1, n whch case the above expresson reduces to 8 More complcated formulatons are possble, such as ncludng regular repayments on the loan, but these do not provde a closed form soluton to the nterest rate relatonshp.

15 15 t0 + t = (19) 1 t or, t (20) t + 0 provded the dscrmnatory tax rate s not excessvely hgh. Moreover, f foregn nvestors reman averse to the economy s rsng external ndebtedness as dscussed earler, the equlbrum domestc nterest rate wll be t1 as shown n Fgure 3. The loss specfcally due to the tax on captal outflows s the foregone natonal ncome hjcf less the taxaton revenue gan for the economy, approxmated by hjmk. S t +ρ S F S ρ F Domestc, World Interest Rates t1 d to h k f m j c I S F Loanable funds Fgure 3 Welfare Effects of Taxes on Captal Inflows More formally, the economy experences a net economc welfare gan compared to autarky equal to, t ] W = [ I( ) S d (21) t1

16 16 Because t 1 > d >, ths welfare gan s lower than that accrung wth mperfect captal moblty and no dscrmnatory taxes, and less agan than perfect captal moblty would bestow. The macroeconomc welfare loss specfcally due to the tax on captal nflow s, W I t t [ I( ) S ] d + ( t1 d [ I( S W = 1 ) ) ] (22) d Other thngs equal, ths loss s less than would arse under a URR captal control regme.

17 17 4. CONCLUSION Hgh captal moblty mproves macroeconomc welfare n advanced and emergng economes because t narrows real nterest dfferentals and rases domestc nvestment. In ths way, nternatonal fnancal flows play an mportant role n the process of economc growth by enablng domestc captal accumulaton to be hgher than otherwse. n mportant, though htherto neglected corollary s that captal mmoblty stemmng from captal controls drectly causes macroeconomc welfare losses. Usng an extended loanable funds framework, ths paper has shown that exchange controls of dfferent knds reduce macroeconomc welfare by rasng the external cost of captal. Captal controls n the form of taxes on nflows are preferable to quanttatve controls known as unremunerated reserve requrements, snce taxes mpose smaller welfare losses due to revenue effects. Ths result mmcs the well-known result from nternatonal trade theory that t s better to mpose tarffs, rather than quotas, on mported goods and servces. t the same tme, ths paper has abstracted from problems that may arse n practce wth the ntermedaton of funds through fnancal nsttutons, as well as nformaton asymmetres between domestc borrowers and nternatonal lenders and moral hazard problems arsng from offcal guarantees to lenders, explct and mplct. It has also mplctly been assumed that captal controls are not evaded, though emprcal evdence provded by Dooley (1996) and Edwards (1998) suggests that n practce evason has been wdespread. Captal controls are advocated as a means of mnmsng nternatonal captal flow reversals that occur due to nformaton and moral hazard problems. However, the above analyss

18 18 suggests these problems are best addressed at ther prmary source, the domestc economy s fnancal system, not by means of captal controls as a second best polcy opton.

19 19 REFERENCES Bhagwat, J. (1998) The Captal Myth: The Dfference Between Trade n Wdgets and Dollars, Foregn ffars 77(3), May/June, Bruce, N. and Purvs, D. ((1985) The Specfcaton of Goods and Factor Markets n Open Economy Macroeconomc Models n Handbook of Internatonal Economcs, v.2, msterdam: Elsever. Cooper, R. (1998) Should Captal ccount Convertblty Be a World Objectve? n P. Kenen (ed) Should the IMF Pursue Captal ccount Convertblty? Prnceton Essays n Internatonal Fnance, No.207. De Gregoro, J., Edwards, S. and Valdes, R. O. (1999) "Captal Controls n Chle: n ssessment." Paper presented at the Interamercan Semnar on Economcs, Ro de Janero, Brazl. < Dooley, M. (1996) Survey of the Lterature on Controls Over Internatonal Captal Transactons IMF Staff Papers, 43(4), Dumas, B. (1995) "Partal Equlbrum vs General-Equlbrum Models of Internatonal Captal Market Equlbrum" n R. van der Ploeg (ed.) Handbook of Internatonal Macroeconomcs, London: Blackwell. Edwards, S. (1999) "How Effectve are Captal Controls?", Journal of Economc Perspectves, 13(4), Echengreen, B. (1999) Toward a New Internatonal Fnancal rchtecture, Insttute for Internatonal Economcs, Washngton DC. Feldsten, M. and Horoka, C. (1980) Domestc Savng and Internatonal Captal Flows Economc Journal, 90 (2), Flemng, J. (1962) Domestc Fnancal Polces under Fxed and under Floatng Exchange Rates IMF Staff Papers, 9(4), Frenkel, J. and Razn,. (1992) Fscal Polces and the World Economy, 2 nd edton, MIT Press, Masachusetts. Internatonal Monetary Fund (1997) Internatonal Captal Markets: Developments, Prospects and Key Polcy Issues, Internatonal Monetary Fund, Washngton, D.C. Internatonal Monetary Fund (1999) World Economc Outlook, Washngton, D.C. Isard, P. (1995) Exchange Rate Economcs, Cambrdge Unversty press, New York. Krugman, P(1998) Savng sa: It s Tme to Get Radcal Fortune, 138, Lews, K.K. (1995) "Internatonal Fnancal Markets" n G. Grossman and K. Rogoff (eds) Handbook of Internatonal Economcs, msterdam: North-Holland.

20 20 Marston, R.C. (1995) Internatonal Fnancal Integraton, New York: Cambrdge Unversty Press. MacDougall, G. (1960) The Benefts and Costs of Prvate Foregn Investment from broad: Theoretcal pproach The Economc Record, 36(73), Neely, C. J. (1999) "n Introducton to Captal Controls", The Federal Reserve Bank of St Lous Revew, 81(6), Obstfeld, M. and Rogoff, K. (1996) Foundatons of Internatonal Macroeconomcs, MIT Press, Boston. Rodrk, D. (1998) Who Needs Captal ccount Lberalzaton? n Should the IMF Pursue Captal ccount Convertblty? n P. Kenen (ed) Prnceton Essays n Internatonal Fnance No Ulan, M. K. (2000) "Revew Essay: Is a Chlean-Style Tax on Short-Term Captal Inflows Stablzng?", Open Economes Revew, 11(2), Wade, R. (1998) The san Debt-and-development Crss of 1997-?: Causes and Consequences, World Development, 26(8):