Exchange-Rate Flexibility and the Efficiency of the Foreign-Exchange Markets

K,7 (#717 in RFD Series) |

INTER AWATIONAL FINANCE DISCUSSION PAPERS

EXCHANGE-RATE FLEXIBILITY AND THE EFFICIENCY OF THE FOREIGN-EXCHANGE MARKETS

by

Norman S, Fieleke

Discussion Paper No. 44, April 5, 1974

Division of International Finance

Board of Governors of the Federal Reserve System

The _analysis and conclusions of this paper represent the views of the author and should aot be interpreted as reflecting the views of the Board of Governors of the Federal Reserve Sys-em or its staff.

|

mmaenecemn '

Exchange-Rate Flexibility and the Efficiency of the Foroign-Exchange Markets

by Norman S. Fieleke*

Perhaps the major issue in the international monetary reform negotiations is how much market flexibility to allow in exchange rates between national currencies, Although there will surely be more flexibility in the future than under the Bretton Woods system, there has been much concern that a high degree of flexibility might somehow overburden the institutions of the foreign-exchange market, particularly the forward market, and thereby disrupt international commerce, While seldom clearly stated, the reasoning underlying this concern usually proceeds along the following lines." Substantial exchange-rate flexibility leads business management to expect greater exchange-rate variations, with the result that businesses seek to cover much more of their foreign-exchange exposure (i.e,, seek to "insure" against the greater exchangerate risk) by purchasing or selling foreign currency forward. However, foreign-exchange traders either cannot accommodate this greatly increased demand for their services, or can accommodate it only at substantial y higher cost. Consequently, business firms significantly reduce the volume of their international transactions.

This argument poses in an extreme form the interesting question of whether the foreign-exchange markets can function as efficiently

with a high degree of flexibility as with the relatively fixed rates

-2-

of the Bretton Woods system, There may not have been enough sustained experiences with highly flexible exchange rates to Provide a definitive answer, However, if we Can ascertain the immediate determinants of the cost of executing foreign-exchange transactions, Perhaps ye can then infer the influence of increased flexibility

on those determinants and thus on the transaction costs in which we

are interested. such inference, if Supported by the Scant direct

evidence on transaction costs under Sustained ‘flexibility, should

Provide us with at least a tentative conclusion, This is the approach taken by this Paper, which draws upon data generated by the 1971

\ experience with flexibility,

I, The Cost of Using the Foreign-Exchange Markets

In the absence of monopoly or market externalities, the social cost of the services of any class of middlemen, including foreignexchange traders, is represented by the difference between their receipts for the things they market and their payments for the same things, a difference commonly known as the markup," Unfortunately, there are no Published data on the markup which is actually Paid to foreign-exchange traders by those who use their services, However,

a fairly g00d &pproximation of at least the trend in the unit markup (or unit cost) can Probably be derived from data on the markup (or “asked minus ‘bid,’ or ' Spread’) which is quoted kon interbank foreign-

exchange transactions in New York.°

4 \ ‘

-3- 4

Data on this markup, expressed as a percentage, are plotted in Figures 1 and 2 for nine different currencies for the year 1971. For most of these currencies, the markup rose perceptibly in May, when a new, higher dollar value was fixed for the Swiss franc, and when the Dutch guilder and the German mark were allowed to rise in value in free market trading. (The Canadian dollar had been floating since June 1, 1970.) Except for the yen and the Belgian franc, these hightened markups then subsided by July to approximateiy the levels of January-April, Then came President Nixon's pronouncements of August 15, and by the end of August all the currencies represented in the charts, except the French franc, were being allowed to float, with varying degrees of freedom, to higher vaiues in terms of the dollar. In this month the markups on all the currencies except the Canadian dollar rose to levels that were very high by contrast with the more normal levels of January-April. Restrictions imposed by the Japanese Government at this time virtually terminated forward trading of the yen, but the abnormally high markups on other currencies fell considerably by November. Most markups, particularly the markup on the French franc, rose again in December when international meetings were held to establish new fixed rates of exchange,

These data suggest that the cost of using the foreign-exchange markets varied considerably during 1971. it is important to determine

the reasons for such variation.

es Vere ULI gsWOS 1 TvEweedatlo Maz pe AOJ

yt pysth UsUQ dum “uojsog JO yueg Teuor:

roe ®Tqerteae JOU 829M suoT ze ONb ‘EN 3SaTyq ay Aq paptaoad BIEP worz paqnduoy >09a4n0¢g

“SL IsnBny 493; o1qejjeae SUOI}B}ONb

uedrk P4emMsoy3 ON,

oa NVIGYNY)

ee, Ee es

Loo / A/ soo: 4Y3qQ1NO ‘Hoing ~~. / a ~ ! an | | ; * \ / £00° a \ i 3S3Nvave ae . \ ! zeke) ONVAS CvIo¥aWwWod) HONZYY \ / Py ° \ / , 7 Soo: \ i | v 900° Loo

800° LZ61 “y404 MAN ul SUdIIDIONYS yUogsayuy Sulso[ kopsouseg, Sa

P!Ig JUISO]> Appsaupany J9 seBo1aay Alyiuopw '1a—paxsy ‘3V MYOd SAV 06 SIIDNAYYTID g319313$ yoy —_Ala

_—_O

dld—daysv

“OTQeTTeAe FT pasn usy] 210M suotjejonb Aepsanyy to Aepsany pue ‘skepsaup >» swos r0J aTqeT Tear Jou e1eM suOT3eRONb ‘soToudzand Mey e& AOY “uo SOG JO Yue_ TeEUOTIEN Asxztg oy. Aq peptaoid ejep wory pagndwoy :adanos

"S e4nbiy LL6L ‘oeq “AON 390 ‘das ‘6ny Ainp aunp Aew ‘ad ‘sOW : YVIIOG NVIGUN a ~ TI —— —s ° ee ee TO ae -QNNOd wT wns & ~N y; — Oe 8 ees ess LOO 00: ONVYs (IVIDYANWOD) NVION3a fe ! it>) VYIT NVITVLI J i —— 800° 1Z6L “Y4OA MON Ul SUOHDIOND YUDQGsaju BHursoj>y ADpsaupam jo saBossay Apyuow —__P'S __ Pla—pex1sy >GaYvVMadOd SAVG 06 SSIDN3SYIND G3I1DI19S YO4S oid Qigd-d4dysv

II. Hypotheses Regarding Variation in Markups

The markup collected by foreign-exchange traders is a function of the demand for and supply of their services. Among the factors which determine the Supply price of these services, four would seem to be particularly important, First, the cost of exchanging cne currency for another probably decreases in the long run if the normal daily volume of trading increases, because in a large market with welldeveloped communications it is easier for foreign-exchange traders to match up specific bids with specific offers, or to buy and sell without influencing the price, In addition, the cost will be lower, the stronger the competition among the banks which serve the market, by contrast, foreign-exchange restrictions generally raise the cost of executing foreign-exchange transactions, as they tax or prohibit these transactions, snarl them in red tape, or arouse doubts about whether the parties to foreign-exchange contracts will be allowed to meet their obligations, Finally, because of the way business is done in the market, the cost of handling transactions is also enlarged by increases in uncertainty about the future level of exchange rates, or, more precisely, about_the future rate of change in exchange rates,

The manner in which exchange-rate uncertainty works its effects may not be immediately obvious, but can easily be illustrated, Assume that a foreign-exchange trader in one of the large banks is telephoned

by an important customer who wants to know the rates at which he could

-7- '

buy or sell pounds sterling in exchange for U.S, dollars; also assume that the trader neither Owns nor owes any sterling, that the customer does not reveal whether he wishes to buy or sell, and that the dollar Price of sterling has been falling rapidly. The trader knows the interbank bid and asked rates around which he might have sold or bought sterling for his customer just a minute ago, but he also knows that these rates are probably changing even as he talks with the customer, Because sterling has been falling rapidly in value, the trader will probably quote the customer a bid that is distinctly lower than the

bid he last observed in the interbank market, in order to reduce the risk of buying sterling from his customer at a price higher than that

at which he will be able to resell it. Yet the trader cannot be certain that sterling will continue to decline in vaiue, and he may well quote the customer an asking price that is little different from the last ‘asked’ in the interbank market. Thus the spread between the bid and the asked tends to widen when rates are changing rapidiy.*

Of course, there is no published measure of the extent to which foreign-exchange traders in fact realize such uncertainty-induced increases in quoted spreads, At least a portion of the quoted increase is probably realized, and this portion can be interpreted as a reward for the greater risk and effort associated with foreign-exchange trading at such times,>

The theory advanced here, then, is that the supply price (in the

form of a percentage markup) of foreign-exchange trading tends to be

awe”

-5-.,

lower,, the” greater is the.normal volume of trading and the greater is the competition among the banks, but that the markup tends to be higher, the greater is the intensity of foreign-exchange restricticns and the more uncertain is the rate of change in exchange rates, In the regression analysis which follows, we shall be interested in explaining variations in percentage markups over a period of only a few months, and we shall assume that there was no variation in the degree of bank competition during this short period, Consequently, this variable is dropped from further explicit consideration, Also, it happens that we cannot include the volume of foreign-exchange transactions as a distinct explanatory variable, because there are no data on this vartable and no reasonable proxies come to mind; this omission, while unfortunate, may not be serious, since percentage markups may be fairly impervious to fluctuations in volume per se in the short run,

Probably much more important than volume per se, at least in the period under examination, are the remaining two determinants of supply price: foreign-exchange restrictions and uncertainty over the rate of change in exchange rates.° It is reasonable to assume that both are uninfluenced by the level of the percentage markup, or, more generally, that both are determined exogenously, outside of a model which purports to explain the percentage markup earned by foreign-exchange traders. Therefore, the supply function is a reduced form which it is appro-

priate to estimate directly,’

-9-

Finally, it is assumed that realized markups show the same general variations, although probably not so extreme, as the quoted markups. Indeed, if this assumption did not hold true in 1971, it vould follow that the cost of using the foreign-exchange markets was not increased in that year either by the extensive exchange-rate revisions (and the accompanying uncertainty) or by the intersifications of exchangecontrols, or, more generally, that the cost of using the foreign-> exchange markets does not increase with even the less desirable forms

_~

of exchange-rate flexibility.

III. Test of the Hypotheses \

the testing of our hypotheses is difficult because there are no direct measures of the explanatory variables---a predicament, however, which is hardly novel, Fortunately, data are available for some reasonably good proxy variables, ‘For example, unusually large covered interest differentials are associated by some analysts with speculative runs and by others with government restraints over covered interest: arbitrage, and as a rule such large differentials probably are closely correlated both with stringent exchange restrictions and with the uncertainty that accompanies large-scale speculation,

While covered interest differentials thus warrant our attention, computation of the appropriate differentials sometimes demands data

that are not available; moreover, variations in ‘these differentials

probably seldom represent fully the variations in exchange-rate

-16-

uncertainty stemming from all major sources, Therefore, two other proxy variables for uncertainty will also be considered, The first such variable, suggested by the illustration in the preceding section, is change in the exchange rate, Second, uncertainty may also be enhanced by actions, meetings, or pronouncements of government 10

officials concerned with exchange rates,

With these considerations in mind, and assuming linearity, we

can specify the model as follows:

(1) we

Ht}

a + b)ED + by|Ar°/r°| + bac,

where M° = (4°~p°)/p°;

¢ it

i

A the asking price in dollars for a unit of foreign currency, B =the bid price,

and the superscript ° indicates spot; |

i

ED the “excess-over-normal': covered interest differential for a 90 day term, in terms of absolute vaiue and at an annual rate; tt

(A+B)/2; and

Le} ut

G = an official action, meeting, or pronouncement which appears likely to have a significant effect on market uncertainty,

Our concern is to explain the day-to-day variation observed in M° for eight different currencies while currencies were "floating"

during 1971. In equation (1) the constant term then is the minimum

-ll-

percentage markup consistent with the normal daily trading volume and the degree of competition among foreign-exchange traders in 1971, 17 The variable G actually consists of a set of dummy variab.es, During the prolonged ‘crisis: of 1971, meetings of high government officials often engendered press speculation as to whether the meeting would be followed by new government measures in the foreign-exchange market, Consequently, newspapers and other current records available to us were scanned to identify all meetings between finance ministers or heads of government of the countries in whose currencies we are interested. In the equation pertaining to each currency, all meetings involving the finance minister or head of government assoctated with that currency were represented by dummy variables, One variable was used to represent all meetings of the "Group of Ten’ countries; another was used to represent all meetings of the European Economic Community; finally, every other eligible meeting was represented by its own distinct variable. Each of these variables was assigned the value of one both for the last working day(s) before the meeting(s). it represented and for each day of the meeting(s) and was assigned the value of zero elsevhere. In addition, each pronouncement which was made by a finance minister or a head of government and which appeared likely to have a strong effect on market uncertainty was represented in the equation (or

equations) concerned by its own dummy variable having the value of one

Neaay |

“Measures selected were control measures, \

-12-

for the first working day that could have been affected by the pronouncement (and having the value of zero elsewhere). Finally, each policy measure which it seemed would stimulate apprehension (or reduce it) was also represented by a dummy variable, given the value of one only for the first working day that could have been influenced by the measure, Our selection procedure resulted in the incorporation of about seven dummy variables in the typical equation; all policy

It seems that the full uncertainty-effects of official pronounce ments and new control measures, and also of the variable JAr°/r° |, might be realized only with a lag. Foreign- xchange traders may frame their apprehensions about future rate changes on the basis not only of current changes but of changes in the recent past; and some time may be required for the dissipation of: concern generated by official statements and policy initiatives. Consequently, in the estimation of equation (1), one-period lags of these variables were introduced, and when the results suggested longer lags, the Almon polynomial distributed lag (PDL) technique was employed. />

With these dummy variables and one-period lags, then, equation (1) was originally estimated by the ordinary least squares technique!” for each of eight different currencies,” as the availability of data for Flotation periods would permit; and, unless the results warranted a

h PDL, the Cochrane-Orcutt approach was used in cases where the Durbin-

-13-

Watson statistic suggested autocorrelation of the residuals, +® An obvious variation of equation (1) is to substitute Ww? for M° and [Ar?°/99| for [Ar?/r°| (where the superscript 90 indicates 90 days forward). This variant was estimated by the same procedure as equation (1) for each of the same currencies except the yen, for which the necessary data on forward rates are not available. In the estimation of both M° and w?, the variable | (199-10) /r°| was used as a crude substitute for ED in the case of currencies for which data on ED are not available, on the grounds that itarge covered differentials are often accompanied by large forward premiums or discounts, !”

The statistical results are shown in Tables 1 and 2, which list the estimated equations in the order of their general acceptability to us in explaining spot percentage markups. Dummy variables for which the estimated coefficients were not significant at the 0.05 ievel have been dropped from the equations; a two-tailed test was used for the coefficients of dummy variables representing new controls (since it is’ often hard to form a reasonable hypothesis as to whether a particular control will initiaily enhance or reduce exchange-rate uncertainty), and a one-tailed test_was applied to the other dummy coefficients. Equations with dummy variables were also estimated without the dummies

(and are shown without them), since there is perhaps more doubt that

dummies capture the intended effects than that other variables do.

(ze*I-) (s£°9) — (70°@) oe'T= P ("B90 18) LT/Z1-9T/8 z000°0- 389810 6220°0 6£ "O=ze DUBIZ SBTMS (10°0-) (oy'y) (LE°%) €7L°7) = (OTT) 6L°I= P ("8GO 18) LT/Z1-9T/8 9000 °0- 2TZ00°O $00°O o82ET'O —LTIOO 1S °0=7¥ oueIZ SSTAS (Ze °€) (s9°L) (98°0) 91°%= P (7890 BET) LT/Z1-OT/S 1000°0 597910 SLI0'O = 7E‘0=,4 1apzyns yo3ng (00°24) ($9°9) (87°0-) 69°1= P ('S4° GET) LI/Z1-01/S ~000°0 9960°6 S100°0- 92° 0= 74 . yaew ues (€0°0) (68°1) (O€'7) so'z= PB (899 HZ) LT ZT-Et/s 6000°0 1°80'0 Z800°0 92 'O=7¥ oui; youerg (90°2) (67°S) (€8°T) Z6‘T= P (*S89° 22) LI/ZT-€2/8 €000°0 59STE “0 T1T80°0 v€°0= 78 DueAjZ ueT3Tag (s9°T) , (Love) (29"9) 90*z= P (890 £9) LT/ZI-EZ/8 TO00"O ZyTC*O 800T*O q BE °O=8 BATT ureteral (82°0) (€2°7-) (S6°7) (L£°4) yI'Z= P (7890 £9) LT/2I-€7/8 4 T000°0 8100°0- Ozzy'0 ©7S60°0 £7028 BATT LCT{eal ' (27°L) (88°S) €6°T= P .(*SGO 99) LT/ZT-L7/8 6z00°0 1979°0 q 9E°0=78 uak asaueder (Sz"0t) (€z°0T) (60°Z) (22°) pt0'Z= P (7890 99) LT/ZT-Lz/8 $z00°0 S0z0°0 9100°0 YELT“O q q “SLO, 8 uak asourder (46°S) (16%) = (€z"S) (45°S) 88°T= P (*S90 9L) LT/ZT-€2/8 . T000°0 ST¥0°O SEO" 10Z0°O 22°08, a punod "479 (90°9) — (66"L-) (EL 2-) — (OT"9) (58°S) A87°L) O€'c= P ("840 92) LT/cT-£7/8 T000°0 £000 O- z000~0-" 2L70°0 = .96£0°0 $Sz0°0 8L°0=74 prod -y°9 TL LL Le : i qdao1zsquy . : 3d 3 3 3 2 3 2 T-3, 3 c- T- 3 3 voOszTmM pojza3d outa 9 3° id €9 2° tT 1? a tat, 92 a Luyqing pur 9 [@-33_1-3;| JT-3 3-35 ,33-33| t) ry oo Oo” 06 pue Adu2zIing SOFRBA-Q pue x c—

(moTey SION UF pauTjyeqd) saTqeyaeaA Adoquep[dxg Jo squafoTszaog - SaaS ah: Sn i SOD SL SOUT = it Pa ED cnn re

ae TL6l UT Spoyieg pazoeTeg AOZ ‘Jajrey yUeqIaIUT YOR maN UT Ssautsng Jo BsSOTD Je SopoUsAAN) JUoeleszTq Ista UP SUOFIOBsueIy Jods uo sdnyiey esequesieg 10j suotjenby uoTssersey

*

T PTQeL

*Z e1qQB], JO pus Ae ajou des ‘BIeEp Jo sedINOs puv SaTqeyTABA UOTSsaIZaI ay Jo suoTIUTJep oJ :390N

-

aA (19°) (61°) (87°Z) (1S°€) (18°2) - 6£S0°0 S7£0°0 2920°0 1€Z0°0 1€70°O oUuGIy ssTMs (] . . _ (141) (yo'€) (2£9°€) (L6°2) (91°T) (19°€) (SZ °€) (50°) (S¢°T) ' €£000'0 $000°0 9000°0 9000°0 9810°0 0820°0 0z£0°0 90€0°0 —- 8200 duBay HBTMS (8S"1) (€8°7) (07'S) (00°€) : S0z0°0 0870'0 67S0°0 1170'O apy ERs ysqng (Sz°T) (€8°€) (€Z 1) ; 89S0°0 OSLT‘O g£80°0 ouvay uez3Teq 3, "3, 1-3, °5 Sax 9-43 €-33 2-32 I-32 , |S- 32-9732] | v-a2-€-33| | €-32-2-32| | @-32-T-32| | T-3a-32|

i TSMOT TOF SE BIB SOT}eI-Q pUB SQUSTOTJJIOO TeNpyTaATpuy “7 eerZep Jo Iqq Aq pazewy sa squatoTjJjea0o Bey Jo uns “OSLQETTPAG SPM ATQeTAeA stu JO sinseaw A10RIeFSTICS Ng

*S[ENPTset JO UOTJeTeIIOD04NB BoNper 0} pasn snbyuyoazZ 34NdIQ-sUeIYdOD,

(penutquo9) TU PtqeL

-l6-

(*s4o Tg) LT/ZT-9T/8

aueajy SSTMS

("89° 18) LT/ZT-9T/8 Duely SSTMS

("sqo SET) LT/21-G1/S zepTyns yo3ng

oa

(°'sqo 6€1) Li/ZT-0T/S yaeuw uewltoD

('sqo 49) Li/Z1-€¢/8 ouerzay YyOusrZ

(7890 6€) LT/ZI-€2/8 Duel UeTsTeg

(*sqo $$) LI/ZI-€2/8 CITT UeTIEIL

(*890 92) LT/ZT-€2/8 punod -yx°n

poyied owtz pue Aouarzang :

_ (LT'2%~) (s6°€) (£6°2) 01°0= P T100‘0- 3911? ‘0 6601 °0 q Lo°0=78 (7§'0-) (€€*2T) 6S") (06'€) (L1°2) EL'I= P 1000 °0- 20ZT0°0 6T00°0 s8Te "0 £770°0 q 18°0= 4 (€2°6) (LL) (18°) €9°I= P ¢000°0 56222 0 10S0'0 S€O=,¥ (11°9) (S0°€) (8€‘0) z6°1= P 9000°0 6120°0 8£00°0 S0°0=)8 (z8°0-) @L°T) (48'S) e9l ‘l= P 7000°0- 40610 6620°0 07°0=)4

(S¢°8) (€l°%) (S9°0) IL‘’I= P Z100°0 £02 “0 6E6T'O- = SE" 0= 8 (02° 4) (0%°€) (87° 2) 99°T= P oT00°O 78947°0 vyEe’O q €9°0=,4 (€g° OT) (847°) (66°S) p06°T= P 7600°0 6270°0 #S70°0 9€°0=,8 RR er qydao1aquy Ay 45 T7344 43 Jaq uos3ey 9 s oe 2 —uFqang T-3,_ 3 1,_3 "067-06 lo2-gtl aD ze

nr RN

SOFIeI- pue

(MoTaq BION UT peuTjeqd) seTqeTze, AroqeuETdxY J

a a ET EP

————— Oe EES

Joys2_ YUeqTsIUL HAOX MoN UT ssoutsng Jo 9xSoT)D Atted ed-06 uo sdnyiey s8ejusoieg Joy suoyfjzenby uoysse1s9y

TL6T UF Spotted Pez weT9S A0J

rrr ES

a O sjUaTOTZJI09

Je SapPoUSTANy JUoTessIq UsAesS UT SuUOTIOCSUeIL FIeNIOg

@ PTQeL

ee CE A A A

-l7-

to ‘syieg ‘weprajsuy ‘sTassnig ‘uopuoy uy a:e2 yUeEqieqUT a} aseuop YJuoM vey. ay ‘perzspysuoD Buteq

Aduaiind eyq uodn guypusdep ‘pure uopuoy uz vw3e1 Aysodep AeTToOpoing yuow aazuy ayy ere suot ze jnduos

dy? UF PeSN SojJe1 JYSsar9qUuT suL “TLET ‘Tpady—-Azenuer ‘sXkepsaupay toy [ep Us1szzTp eyQ Jo senyea

aINTOSqe ayI JO SZeIVAG ay ST [B}IUSLaJZTpP petasod |,Tewro0u,, ayy stays ‘ajze1 Tenuue ue Ye pue anTtea eanjTosqe jo suz93 UT ‘Shep YG JO WIT) =F AOJ [UTJUeIaJZTP JserequTE parsa0d | TewoU-1aA0c-sSsaoxs,, SUL =U

*uoqsog jo yueg TeuOTIEN

3Isifq 943 Aq pettddns o190m ejeq . -aotad pzq ayq sy g pue ‘qayreM yUeGIeQUT yIO_ MAN OUI UT SuUTSOTO ay. ye AdUaIAIND UszZaIOJ Jo AyuN e& AOZ szel[ TOP uT acqad Buy 7se aya S$ V er9ym *g/(q-y) :dnyrew a8equasieg

e 2SMOTTOJ SE 1B J puUe T SEeTqe]L UT SeTqeTAeA UOoTSseIZeI dy 103 BIep Jo sadAINOS pue sUOTIFUTJop euy :aqoN

(re °C), (79°€) (v2 °1) (oe "T) (2£°0) 61220 SSOT'0 £0900 SL€0°0 T9€0'0 DuBaz SSFMS (69° (79°y) «= (65°9) )=— (8° TT) (26°TT) (60°0) (90°47) (€9°4) (72°) (80°09) oT00°O §=z100°0 6100°O0 TE€00°0 8¥00°0 SZ700 '0 7790°0 6780°0 2790 °O ZZ00°0 Dueiz sstag (18°Z) (92°) (ov ° 4) (SO °€) 9670°0 ZL£S0°0 66S0°0 S9S0°0 AapTIns yoang (€6°t) (66°) (08°Z) (L7°Z) (87° 0-) 69¢9'0 Z020°0 €Z20'0 0€470°0 QLTO‘O- = SuBaz ueT Tey 7-3, €-3 e-3 T-3 3 G73, 4-3 23 @-3, [I-23 9” 9° 9° 9° 9° 06+ 067 & oe 06" 06" s-3, 75 2 47-3 -t5 522 -3 53 Po6*- 75671 | o63-fogt! § |fog-25g21 125g2-Toe2! | 'od2-082 |

iSMOTTOJ SB BIB SOFIBI-J pUe SJUPTOFJJeOO TeNSFATpUL *7 va1Zep yo qq Aq peiewy3sa squepfoTyyaoo Bet jo wns >

‘OTQETTBAR Sem aTqeyzea SFY Jo ainseam Ar02>0egSTIeS ON q

“STENPFTSIl JO UOTIETIII090RNe Vonper 0} pasn anbzuyoaz 34n010-aueAYI0D e

(penuyz3zuos) Zz PTqIeL

4 S8FIUaIING 19YIO aus JO suOFIBNTeASI pue ABT[OP ayy Jo vUoTIeENTeAap e YBNory see aBueyoxa Jo Quemusz{eaxz ydmoid e p1reMmoq y10M,, pInom Say. yey. Juemsounoucid juycf © penss} nopydmog pue uoXTN SJUepyseiIg yT Jaqueceq uO *SUO JO ONTeA ay2 SoeyeI IF YOTYM 10z ‘HT Aeqmedeq Inq sAEp [Te 10J O19zZ 03 Tenba Jes SEM eTQETAPA STYL i¢9 *LT-9T Asqueseq pue “Tt Aaquadaq-67Z Aequeaon ‘S9T-HT Jaquajdag uo 9UG aNTeA ay. Saxe Ay fe1eyMesTea o19z 02 Tenba Jas sea pue ‘sazz3uno. UL Jo. dnoig,, eyz Jo SaafIeqvesaide1 jo BuyJeew yea e10jaq Aep Ase, syR pur Jo Skep 1oz auO Jo aNTeA ay UaATS SPA aTqeFAeA STYL 279

. . - *sn 02 9TqeTFeAe Sspi0dsezx AayIO

pue sisdedsmou woriz 918 SaTqeyzeA nD, 29YIO TTS pue syYyI 1oZ eQeq “sJUapyse1 OJ BuypuaT Aoz ZufT1sis 03uy sjzysodap AdUaTIND Quapysai-—uou YOIFAS 02 SYUeqd Toy uMeAPYITA sem voTssyquied fosTyY “9/61 ST 32q903-'0 alojaq Buyanjew sayz yfanoss AZTAoy ne Teo, pue ‘pasqueren3—juamurzea03 ‘quawuzaa03 se [Tan se *sq9 BuyT1eys Teuofyyppe Zuyzseyoind woraz payyzqyyoid 02am sjuepyTsei-uoN *satIy~Aoyyne [eoo, pue suoyz4nzF Asuyt

-

yepoueuTZ YyIzm syyzsodap Juepysai—uvou [Teuofzyyppe peuueq pue ‘squepyTsei vele Buy{iazs—uou Jo szyzsodap yueq

-18-

Buy {193s sy. uF eseazouy Aue uo squewked yserezuy paryzqyyoad Quawmuzsaacy *y*p ey TE AsnBny ug *auo 03

Tenbsa Jas seh IF YOTYA TOF ‘TE YoenBny Ydeoxe skep TTB 10J O19z Jo enTeA ayi pousysse gem aTqUyIeA STYL +79

-

*uojsog jo yueg “ Teuoy3eN 3satg ey Aq pattTddns a1z9m ‘g +d uo pautzep AT [NJ vae YyOTYA ‘SseTqeyAeA asey. uo BIC : o¢1 pure ot

*HIOR MAN JO Yueg aAI*Soy TeIIpey

aya Aq pattddns azan STBFAUSAITIJTp pazsaod jo suo Tze qndwoo eyul SB T[eM Se eJep essay, ‘1eAeT omy

ao anoy ue itoj-aie ATUowMOD saqer aBueyoxs aTTya ‘paurzaouos Setz unos usyei1oz ayq uy Aep-pyw ynoqe jo

\ Se 91 Solel JSe19UF ‘atn1 e Sy *Sa eI JoxIeM yYURqISUT YIOZ maN vie pasn sajer eB8ueyoxg -“Jansyuerg

(penutquos) % PTqIeL

7 2

-19-

“SsyzUOW XFs uey ssaT Jo ATANQeM eB UITA spunz 03 ATuo pat{dde AyTezazuy pey ueq

ay. Ssqueweoetd ouezz [Te 03 TE Ane BOUTS PUPTIEZITMS OUT pamoTJZ pey eu spunj u3yfaioj uo ueq quowsed

JSSA9IUT BY puezXe 03 peerse UoTIePOOSSy ,sreyUeG ESTMS aya pug yueg TeuOTIeN SstMs ayy ZZ Asn3ny ug ‘auo

JO ONTCA BY UsATZ sem YF YyOTYM rox ‘77 JsnBny 3nq sktp [Te Aoz oraz JO ONTLA JYQ PeUsTsse sem aTqeyIAeA STUL

i

*sjuewjsnfpe ojei—aSueyoxe ofjpoeds £uyzjses3ns sea aWI ayy Jey pejiods1 sem ay €z asnBny ug

"aud 02 TenNbe Jes sem AF YyoTYA oz *€Z ysn8ny anq step [Te TOF OLGZ JO aNTeA JY UaATZ sem BTQqeTIeA sTUL

*sqysodap

BATT PTey-usfer0z TTe 03 yso1aqUT Jo JuaemAed-uou papuaqxe syueq TeFOAOMMOD UeTTeII ST Asequaceag up ‘auUO 03

Tenbs Jes sem 3F YOFYM oz ‘oT azaquasvaq ydaoxe sKke>? [TB OJ o1az Jo anTen ay pauSrsse sv

,penuyzuos) @ PTdPL

A aTQeTleA szuy

+]

S

“9

a)

-20-

The results shown in Tables 1 and 2 do not refute our hypotheses, except that official meetings, pronouncements, or new control measures appear to have directly affected the markups somewhat less often than we had supposed. It appears, as we had suspected, that the announcement of controls sometimes operates to reduce markups, no doubt by decreasing apprehension that current exchange rates might undergo abrupt change.” In other cases, the announcement apparently generates uncertainty about the impact of the control,

It is likely that the coefficients of determination would have been higher had more reliable data been available for the covered interest differentials, For example, the differentials were computed as of the forenoon in New York, while the only available percentage markup quotations which we could associate fairly accurately with any point in time were the quotations at the market closing. -° Moreover, the choice of interest rates to use in computing covered differentials, and the reliability of the quotations for those rates, also pose difficult problems; indeed, data were not available to permit a daily measure of the covered differential for the yen, the lira, or the Swiss

franc.

The unavailability of a good measure of the covered differential probably also accounts for the rather peculiar results (including the long lags) obtained in the equations for the markup on the Swiss franc.

Those equations suggest that exchange restrictions were very important,

-21-

but the dummy variables probably do not capture the continuing influence of the restrictions, while-a reLiable meastre of the covered differential

might well succeed in doing sc.

“IV. Conclusions

Instances of significant and sustained exchange-rate flexibility have been quite rare, so that there is very little direct evidence on whether foreign-exchange markets perform more efficiently or less efficiently with considerable exchange-rate flexibility, ‘This paper attempts to shed additional light on this important question by identifying the immediate determinants of the cost of executing foreign-exchange transactions. The regression resuits obtained are highly provisional, put they do suggest that the cost of using the foreign-exchange markets commonly rises with increases in two proxy variables: the covered interest differential and the rate of change in the exchange rate. Not only are these results interesting in théir own right, but they imply that a tentative conclusion about the relative costs of using the markets under different exchange-rate regimes could be based on inferences as to whether these two proxy variables would typically assume significantly higher values under one regime than under another. In this connection, a few general considerations are worth reciting.

Large-scale specuiation on a change in a governmentally managed

rate probably is a typical contributor to large covered interest

oe

MS

21 differentials, Such differentials are usually also associated with

government restrictions designed to influence an exchange rate; and,

as the events of 1971 illustrated, it is abrupt changes in these restrictions or in other government rate-fixing policies that often precede dramatic changes in rates. °° The common element in these cases---and, indeed, the essence of ‘fixed’ exchange rates---is strong government intervention, or changes in that intervention. In other words, strong government intervention probably often raises (sometimes with a long lag, as intervention may suppress a disequilibrium up to

a point) the values of the factors which determine the cost of foreignexchange trading. This inference is consistent with the direct evidence concerning the Canadian dollar, the one currency which at this writing has been allowed to float relatively freely for an extended period of time. As shown in Figures 1 and 2, the percentage markup on this currency remained remarkably stable throughout the unsettled year of 1971; the level of the markup was virtually the same as it had

been during the untroubled portion of Canada's most recent fixed-rate

period, 24

Of course, these considerations are hardly conclusive, and it is still conceivable that the long-run private cost of using the foreignexchange markets is lower under a system of adjustable pegs than under flexible rates. But since international commerce is not an end in itself, the question would then arise whether government fixing of exchange rates provided a subsidy to international commerce that was

9 not in the interest of society. ~>

«23- Footnotes

*Assistant Adviser, Board of Governors of the Federal Reserve System, This paper will be published in a forthcoming issue of the Journal of Financial and Quantitative Analysis. The author is indebted to the Federal Reserve Bank of Boston for financial support of this study, to Janet King for careful research assistance, and to Peter B. Clark for helpful comments. However, the author assumes full responsibility for the contents of the paper.

see H, Houthakker (1962), M, Friedman and R. Roosa (1967, pp. 39-41), P. Einzig (1965), G. Pelli (1970), and H. Fowler (1972).

“Various surveys have reported that tiie users of the ftoreignexchange markets are generally untroubled by the effects on the markets of a high degree of rate flexibility; but these surveys have provided no data regarding the influence of flexibility on the costs of the services provided by foreign-exchange traders. For such surveys, see N. Fieleke (1972 a,b), R. Fitzsimons (1971), and The Economist (1971).

3a fuller justification for this measure will be developed in following passages of this paper. At this point we simply note that in the U.S. economy transactions prices commonly change in the same general way as quoted prices, and that an increase in the markups which banks charge each other on ‘wholesale’ transactions will surely be parai:eled by an increase in the markups which they charge their

nonbank, “retail customers. For an endorsement of this markup as an

= Dhe

indicator of social cost, and for a criticism of another indicator commonly used by exchange-market participants, see fF. Machlup (1970), “Por a discussion of price formulation in the securities markets, see G.P.E, Clarkson (1965), >Perhaps the best published evidence as to whether risk-bearing

is rewarded can be found in the study by L. Fisher and J, Lorie (1964).

Most theorizing on the subject seems to agree with Irving Fisher's

~assertion that Usually . . . an uncertain income will be vaiued at

less than its mathematical expectation (quoted by K, Arrow, 1971, p. 24).

Ore is consistent with this view that foreign-exchange traders, in discussing these matters, express much more concern over exchange restrictions and exchange-rate uncertainty than over fluctuations in volume as such.

7With respect to the demand price offered (in the form of a percentage markup) for the services of foreign-exchange traders, it seems unlikely that foreign-exchange restrictions or uncertainty about exchange-rate changes wouid be considered endogenous in any demand function that might reasonably be constructed. Because the supply function is a reduced form, there is no need, for our purposes, to consider the demand function any further, Cf. J. Johnston (1963, p. 234).

Statements by ‘insiders’ support the assumption made in the text.

For example, the International Fronomic letter of First National City

Fy V \

-25-

Bank (1971, p. 13) asserts, ‘When uncertainty about exchange rates grows and large fluctuations are anticipated, exchange dealers protect themselves against loss by increasing the spread between buying and selling rates. The result is to raise materially the cost to businesses engaged in international trade,’ Also see W. Page (1971).

For support for the former view, see H. Grubel (1966, pp. 16-17, 20, and 35:3; for the latter view, see E. Sohmen (1966, pp. 3-9, 30, and 1970, pp. 312-13). There is a growing body of literature which seeks to explain the existence of covered interest differentials; for example, see H, Stoll (1972) and L. Officer and T. Willett (1970) and the works cited therein.

10, good example is the statement on June 19, 1972, by Denis Healey, the British Labor party's spokesman on finance. His prediction that the pound sterling would be devalued by July or August was uidely credited with hastening that event (New York Times, 1972). It should be noted that official actions, meetings, or pronouncements might cause foreign-exchange traders to suspect a rate change but provide little basis for predicting the magnitude or the timing of the change, with the result that the other explanatory variables considered in the text would not adequately represent such officially generated uncertainty; an illustration would be the announcement of a control the detai-s of which were not immediate-y specified.

11

The “normal covered differential for a particular foreign

currency (vis-a-vis the U.S. dollar, of course) is assumed to be the

Nae?

-26-

average differential (without regard to sign) for January-April, 1971. That is, this period was taken as a representative “crisis-free," or "normal," period during which covered differentials were attributable to influences other than large-scale speculation or stringent exchange restrictions. For additional information on the construction of ED, see the note accompanying Table 2.

12that is to say, the constant is the average percentage markup which would have been quoted at times (in 1971) when there was an absence of change in the exchange rate, of stringent exchange restrictions, of large-scale speculation, and of unsettling noises by government officials, given the long run normal daily trading Volume and the degree of bank competition,

13Refore the "final" regression equations were selected, PDLs of | Ar°/r°| were inserted into each equation not already incorporating such a PDL, as a last check on the lag structure of this variable.

14although our basic supply price function was in reduced form, the proxy variables used in the estimating equation may invite some bias, since rising transaction costs (including M) contribute to larger covered differentials (represented by ED). To construct a model that could actually be esttmated without the possibility of such bias would probably be impossible, given the present state of the art and the limitations of the available data. In our view, the questions investigated in this paper are important enough that the possibility

of biased results should not forestall the investigation, provided the

-27-

bias may not be severe and we are alerted to interpret the results with caution. That any bias will not be severe is suggested by the fact that there were frequent and widespread reports of bursts of speculation and of new exchange controls (the factors to be represented by ED) during the period under analysis, so that causation probably ran predominantly from ED'to M, as posited by the model. It is also

worthy of note, at this point, that the simple correlation between

-ED and | Ar°/r°| is low or nil for each currericy for which data on

ED are available, ranging in value from -0.02 for the mark to 0.34 for the Belgian franc, for the period being examined.

15a ithough the markup on the Canadian dellar is plotted in Figures 1 and 2, no regressions were run to explain this markup since it scarcely ever rose above the "normal" level.

l6tt was not possible to utilize the Cochrane-Orcutt and PDL techniques simultaneously.

1, an earlier version of this paper, the variable | (190-19) /r9| was included in the equation for each currency (except the yen, for which the necessary data are not available), on the reasoning that foreign-exchange traders might expect more rapid change in the spot rate (and thus widen their bid-asked spreads), the larger the near-term forward premium or discount. However, the several traders who read

that version were unanimous and unequivocal in rejecting the suggestion

\ that they accepted the forward rate as a reliable forecast of the future

~28-

spot rate. (By contrast, H. Working (1961, pp. 161-62) has argued that "Futures prices tend to be highly reliable estimates of what should be expected on the basis of contemporarily available information , ,°, JR)

18y¢ may be, of course, that the influence of such offictal activities is captured fairly well by the other explanatory variables, Similarly, government intervention in the form of purchases or sales of foreign exchange would presumably influence the percentage markup indirectly by affecting the rate of change in the exchange rate or perhaps by affecting the forward premium or discount; but the mere presence of the government in the market might have a direct and distinct influence. To test for this latter influence, we included in several regression equations a measure of central bank market transactions (without regard to sign) in the currency concerned, based on internal Federal Reserve data. The coefficients of all such measures were not significant at even the 0.10 level, although the data available to us were not comprehensive,

19this is not to say, of course, that the long-run effect of the control itself is to reduce the markup; indeed, we would expect the opposite. It should be noted that, with one exception, no control measure appears to have been announced on a working day earlier than the day the control took effect, so that, with this one exception, the announcement and the introduction of the control affected market

uncertainty at the same time,

20Te should also be noted that the exchange-rate quotations used in computing any particular covered differential sometimes pertain to a time of day several hours different from that for the interest rate quotations which were used. (See the note accompanying Table 2.)

21in this connection, to the extent that ED represents large

scale speculation, the low correlation between | dr/r| and ED (cf. fn. 11) suggests that speculation was not massive when the authorities allowed exchange rates to vary.

22The rapid rate changes of 1971 and the associated increases in transaction costs were clearly the heritage of earlier governmental fixings at disequilibrium levels, and it is far from obvious that such high transaction costs would continue with sustained (even impure) floats. On the contrary, Figures 1 and 2 show that transaction costs for "floating" currencies trended sharply downward throughout the period of the general "float," that is, from August through November, as exchange rates were allowed to move nearer their market-equilibrium levels.

23It is not claimed that the Canadian dollar has been allowed to float perfectly freely. But if increases in reserves or exchange restrictions are used as the criteria, there is no doubt that after June 1, 1970, the Canadian dollar has generally been allowed to respond much more freely to market influences than most other currencies have

been.

-30-

24cf, Fieleke (1972b, pp. 180-86). 25ce, A. Lanyi (1969). Herbert Grubel has suggested to me that exporters and importers would pay more in total markups under highly flexible than under fixed rates, assuming that the quoted unit markups were to remain about the same, because under flexible rates more trade would be transacted through the forward market, where unit markups are | higher than in the spot market. This interesting point is, of course, ‘distinct from the issue, addressed in this paper, of whether exchange-

market intermediaries perform as efficiently under flexible as under

fixed rates.

-A]- \ References

Almon, S., "The Distributed Lag Between Capital Appropriations and Expenditures,'' Econometrica, January 1965, 33, 178-96.

Arrow, K. J., Essays in the Theory of Risk-Bearing, Chicago:

Markham Publishing Co., 1971.

Clarkson, G.P.E., "A Theory of Stock Price Behavior," Industrial Management Review, Spring 1965, 6, 93-103.

Einzig, P., "The Potential Volume of Forward Exchange Facilities," Banca Nazionale del Lavoro Quarterly Review, December 1968, 21, 397-414,

Fleleke, N., "Exchange-Rate Fleyihtlitw and the Forward-Exchane Markets: Some Evidence from the Recent Experience with the German Mark," New England Economic Review, May/June 1972, 2-10. (a).

» "The Hedging of Commercial Transactions Between U.S. and Canadian Residents: A View from the United States," in Canadian - United States Financial Relationships, Boston: Federal Reserve Bank of Boston, 1972, pp. 171-91. (b).

Fisher, L. and Lorie, J.H., "Rates of Return on Investments in Common Stocks,'' The Journal of Business, January 1964,

37, 1-21. Fitzsimons, R.B., ‘Who Are the ‘Currency Speculators'?" The Banker,

November 1971, 1277-83.

Fowler, H.H., Partner, Goldman, Sachs & Co., "Some Scenarios for

International Monetary Negotiations," remarks in San Francisco,

April 14, 1972.

Friedman, M. and Roosa, R., The Balance of Payments: Free Versus Fixed Exchange Rates, Washington, D.C.: American Enterprise Institute for Public Policy Research, 1967.

Grubel, H.G., Forward Exchange, Speculation, and The International Flow of Capital, Stanford: Stanford University Press, 1966.

Houthakker, H.S., "Exchange Rate Adjustment," in U.S., Congress, Subcommittee of the Joint Economic Committee, Factors Affecting Ise United States Balance of Payments, 87th Cong., 2d. sess., Washington, 1962, pp. 292-93.

Johnston, J., Econometric Methods, New York: McGraw-Hill Book Company, Inc., 1963.

Lanyi, A., The Case for Floating Exchange Rates Reconsidered,

Essays in International Finance No. 72, Princeton: Princeton University, 1969.

Nachlup, F., "The Forward-Exchange Market: Misunderstandings between Practitioners and Economists,"' in Approaches to Greater Flexibility of Exchange Rates, ed. by George N. Haln,

Princeton: Princeton University Press, 1970, pp. 297-306.

<336

Officer, L. H. and Willett, T.D., "The Covered-Arbitrage Schedule: A Critical Survey of Recent Developments," Journal of Money, Credit, and Banking, May 1970, 2, 247-57.

Page, W., President, Morgan Guaranty Trust Co. of New York, "Address at a luncheon meeting of Italian business leaders in Milan, Italy," October 27, 1971.

Pelli, G., "Why I Am Not in Favor of Greater Flexibility of Exchange Rates,'' in Approaches to Greater Flexibility of Exchange Rates, ed. by George N. Halm, Princeton: Princeton University Press, 1970, pp. 203-93.

Sohmen, E., "Exchange Risks and Forward Coverage in Different Monetary Systems,'' in Approaches to Greater Flexibility of Exchange Rates, ed. by George N. Halm, Princeton: Princeton University Press, 1970, pp. 311-15.

» The Theory of Forward Exchange, Princeton Studies in

International Finance No. 17, Princeton: Princeton University,

1966.

Stoll, H.R., "Causes of Deviation from Interest-Rate Parity,"

Journal of Money. Credit, and Banking, February 1972, 4, 113-17. Working, H., "New Concepts Concerning Futures Markets and Prices,"

The American Economic Review, May 1961, 51, 160-63.

"An International Banking Survey,'' The Economist, November 27, 1971. First National City Bank, "Monetary Uncertainty ‘ind World Trade," International Economic Letter, November 1971.

New York Times, June 21, 1972.