Capital Controls, Political Risk, and Interest Disparities

oy International Finance Discussion Papers Number 125

September 1978

CAPITAL CONTROLS, POLITICAL RISK AND INTEREST DISPARITIES

by

Michael P. Dooley and Peter Isard

NOTE: International Finance Discussion Papers are preliminary materials circulated to stimulate discussion and critical comment. References in i publications to International Finance Discussion Papers (other than an acknowledgment by a writer that he has had access to unpublished material) should be cleared with the author or authors.

a

‘}

—— oe

Capital Conrols, Political Risk and Interest Disparities by

Michael P. Dooley and Peter Isard*

Introduction and Overview

In his reinterpretation of the interest-rate parity theorem, Robert Aliber (1973) distinguishes between exchange risk and political risk as determinants of disparities between interest rates on different money-market assets. Interest disparities reflect exchange risk when assets are denominated in different currencies and/or political risk when assets are issued in different countries (i.e., under different legal jurisdictions).

It is now well established that assets differing essentially in only their currencies of denominatioz, such as Eurocurrency deposits in a particular financial center, exhibit interest differentials equal to the forward exchange premiums that must be paid to insure against exchange risk.+/ (See Aliber, 1973; Dooley, 1974; or Herring and Marston, 1976.)

In contrast, it is not well understood to what extent political

-

*/ The views expressed herein are solely those of the authors

and do not necessarily represent the views of the Federal Reserve System. We are indebted to Peter Clark and Frank McCormick for helpful suggestions.

1/ It is also well recognized that forward exchange "insurance" premiums are similtaneously determined with interest differentials as functions of policy instruments and/or other exogenous variables.

C

risk has contributed to disparities between interest rates on

assets denominated in the same currency but issued in different political

jurisdictions. Aliber defines the concept of political risk as

"the probability that the authority of the state will be inter-

posed between investors in one country and investment opportunities

in other countries" -- i.e., the probability that controls will .

be imposed on capital inflows or outflows. By the nature of risk,

this concept has nothing to do with existing capital controls

per se, but rather relates to the uncertainty of future capital

controls. Thus, interest differentials-due to the political

risk of future capital controls mst be distinguished from

disparities due to the effective tax that existing controls

place on interest earnings. - € The purpose of this note is to clarify that the

interest differential due to political risk depends not only

on the probability of capital controls but also on supplies

of outside (government) debt and the distribution of world

wealth. A simple model of portfolio behavior is used to

explain the interest differential between Euromark deposits

in Zurich (EDM) and interbank mark-denominated loans in Frankfurt

(GDM) for 3-month maturities between January 1970 and December 1974,

-3-

During this period Germany placed a series of controls on capital inflows (see the appendix for a chronology) and the interest differential (GDM - EDM) fluctuated from near zero at the start of 1970 to an annual rate of more than 10 per cent in April 1973, and then back to near zero after the controls were effectively removed. Our reading of the empirical evidence suggests that most of the swing was due to shifts in the tax that controls effectively imposed on nonresident earnings from assets held in Germany. At its peak we estimate that this

tax accounted for an interest differential of about 6 per cent per annum between February and October 1973. An additional differential of up to almost 2 per cent was apparently required, in the context of political risk, to induce nonresidents

to hold the excess supply of German outside debt.

Theoretical Underpinnings

Within the framework of the traditional theory of interest arbitrage, the interest differential between mark deposits in Germany and Euromark deposits outside Germany can be attributed to efforts by the German Lundesbank to hold the mark below the level to which private speculators expected it to appreciate. In this context speculative bids for forward marks, coupled with

central bank sales of spot marks, tended to open a covered

~-4- C differential in favor of mark-denominated deposits at German banks. Arbitragers purchased marks spot (from the Bundesbank), invested the mark balances in German bank deposits (or other claims on German residents), and sold the marks forward (to private speculators). As the stock of these arbitrage positions grew arbitragers became increasingly averse to increasing the share of their portfolios that was subject to political risk peculiar to nonresident claims on German residents. Individual arbitragers could have diversified and reduced their political risk by purchasing mark-denominated claims on non- German banks (i.e., Euromark deposits) instead of German bank deposits. But this would have forced Eurobanks either ( to hold uncovered mark liabilities subject to exchange risk, or to pay a premium to purchase marks forward, or themselves to purchase claims on German residents and accept the associated political risk. Consequently, Eurobanks would have discouraged mark depositers by offering lower yields. Thus the difference between Euromark rates and the interest rates available on claims against German residents can be attributed to the reluctance of nonresident arbitragers, including Eurobanks,

to acquire a larger stock of claims on German residents.

-5-

A graphical representation of this story would show markets equilibrating on the finitely-elastic portion of the arbitrage schedule. There is an alternative story, however, which assumes that markets equilibrated on the infinitely-elastic (or riskneutral) portion of the arbitrage schedule in the context of capital controls already in place. This alternative story attributes interest disparities entirely (apart from random noise) to the effective tax that existing controls placed on nonresident interest earnings in Germany.

These conflicting stories suggest that estimation of the arbitrage schedule is the key to separating the interest disparity due to capital controls already in place from the interest disparity due to political risk associated with the prospect of additional (or tighter) controls. A problem with this approach, however, is that the arbitrage schedule has traditionally been viewed to describe the flow of arbitrage funds, which does not square well with the currently-accepted stock-equilibrium model of portfolio behavior. Our alternative approach is to consider the behavior of both the German private sector and nonresidents in

choosing their portfolio stocks, implicitly taking account of acbitrage possibilities. The Model

Consider a world divided into the German private sector,

which demands a net stock of Be mark-denominated claims against

German residents (including the German government); nonresidents,

who demand a net stock BC of mark-denominated claims against

German residents; and the German S0vernment, @8ainst which there exists a Stock of B mark-denominated Claims, Market-clearing in

this World requires

a (1) Bo +B = B Let Wo and "NR denote the "wealth" of the German Private sector and

of nonresidents; and let GDM, EDM and EDOL respectively denote

@ssets not denominated in marks, and assume that Capital Controls 2/ apply only to nonresidents! Claims against German residents ,£ Let

EZX denote the expected Tate of 4PPpreciation of the Mark 8gainst the dollar.

depend on relative expected yields and wealth ~. ignoring exchange ris} for the moment -- so that

. (2) Be = £(GDM-EDM CPMCEDOLAERK ,W_) Portfolio demands of nonresidents are viewed to depend on Similar variables, and in addition on the leve] of Sapitalcontrols CC, so that

R* 8 (GDIEZN-Epor EPYOEZX-EDOL,W CC)

When it is further assumed that the expected rate of appreciation of the mark equals the forward premium on the mark, which in turn is known to equal the excess of the Eurodollar rate over the

Euromark rate, we have

(4) E%X = EDOL-EDM and therefore (2a) Be = £(GDM-EDM,W,) (3a) Be = g(GDM-EDM,Wyp CC)

For purposes of avoiding a nonlinear specification

hypothesis, we assume that (2a) and (3a) have the linear forms

d (2b) Bo = 4) + a, (GDM-EDM) + awe d (3b) Bur = bs + b, (GDM- EDM) + boWyr + b,cCc

. 2 3/ with ay» a5 b, and by all positive.

Together, conditions (1), (2b) and (3b) imply

(5) GDM-EDM = cy +c ,B-c.W, - c,W + c,CC

1 2G 3 NR 4

with Cys Cy and cy all positive. Since we measure our wealth variables exclusive of claims on real assets, we impose the identity that the global net worth of paper assets is zero,

or that

; 4/ (6) BeW +We

Thus, condition (5) reduces to

(7). ‘GDM-EDM = Cc, + (c)-¢,)B - (cy-c,)W, + c,CC

3/ Implicitly we assume that ay and by are finite, reflecting aversion to political risk. 4/ We assume that B represents the total net liabilities of the German

government and that B, We and Wyre are measured in the same currency unit.

Note that nonresidents include official as well as private nonresidents, so the combined portfolio of nonresidents and tl.e German private sector includes no outside dollar-denominated assets. Hence, conditions (1) and (6) are consistent.

-8-

Condition (7) is our estimating equation. A more general development of the model would assume that exchange risk affects both German and non-resident portfolio demands, which would add another argument to equations (2) and (3), and hence another regressor in equation (7). Insofar as exchange risk increases private German demand for mark-denominated claims on German residents and reduces nonresident demand for mark-denominated claims on German residents, the sign of the net effect on global demand for claims on German residents (i.e., the global demand for outside German debt is ambiguous. Thus, exchange risk can either increase or reduce the interest

differential GDM-EDM.

C)

-9-

Data Sources and the Representation of Capital Controls

Our measure of GDM is tle 3-month Frankfurt interbank loan rate at or near the end of the month, as published by Morgan Guarantee Trust Company in World Financial Markets. EDM is the 3-month Euro- DM deposit rate in Zurich, at or near the end of the month, from the London Financial Times and internal records of the Swiss Bank Corporation. B is the cumlative sum, from an estimated end-of-1969 initial value, of the change in the German Federal Government's indebtedness plus the change in German official holdings of gold and external assets (from various issues of the Monthly Report of the Bundesbank, Tables VII.8 and IX.6a). Wo is the cumlative sum, from an estimated end-of-1969 initial value, of the change in the German Federal Government's indebtedness plus the German current-account surplus (same source, Tables VII.8 and 1X.1).

The initial values of B and We are not important to the regression results, since errors in these initial values affect only the estimated intercept parameter.

We experimented with two different representations of the capital controls variable CC. As described in the appendix, controls on capital inflows were imposed or tightened in 5 major doses and a number of subsequent modifications .— Accordingly, one of our representations of CC is a step function constructed with five zero-one dummy variables. .

Our other representation of CC assumes that the

effective tax imposed by controls increased continuously

5/ Not all of the doses vere directed at bank deposits, but interest rates on bank deposits clearly responded to controls that directly affected interest rates payable on alternative assets.

- 10 -

between the first dose of controls in April 1970 and the last dose in February 1973. Here we follow Wilton's (1975) technique for representing structural shifts, as modified by Reid (1977), which allows us to specify CC as a polynomial of any degree m. We have chosen m=3 as the minimum value that allows for an inflection point. Given that the capital controls applied only to increases in nonresident claims on Germans, we assumed

in both representations that the effective tax imposed by controls dropped to zero when capital began to flow strongly

out of Germany in the fourth quarter of 1973.

Empirical Results

Table 1 presents our regression results. Both specifications of CC yield good fits. The estimated coefficients of B and We have correct signs (as will be discussed below); and the coefficients on the capital control variables are appropriate, with the exception of the insignificant negative coefficient on the first step (April 1970 through April 1971) in equation 2. Critical single-tail t-values are 1.3 for 90 per cent confidence tests and 2.4 for 99 per cent confidence tests.

The two equations respectively attribute interest differentials of 5.7 (=10.1-33.0+28.6) and 6.16 per cent per annum to capital controls in place during their tightest interval between February and October 1973. On average during this period

the two equations respectively estimate that capital controls

in place explained 74 and 81 per cent of the interest differential.

C

“#161 aequeseq y8nory? 0461

Kaenuer woajy suopzearasqo A[YyQuow 09g Jo S3stTsuos potjaod atdues sy_L “San,TeA-}2 91e sfTsayqueted ut SlsqUNN ‘UOTJETIIIOO [eTAVS JapAO-3SATZF AOF YI9II0D 0 posn Sem dAnposord 3yNdAV-auRAYDOD 24]

"Wa JO SUOTI[Tq UT 210 am pue g ‘wnuue aad ued aod ut painseow ST oTquTIeA quapuadep oy, = /®B (79°8) EL6T °390 YBN0zYA E/61 “49d WOAF OT°O9 + q ° a °M (82°2) (08° €) f6'l ° = " = YY E161 “wer YSnozya zZ61 euNnp woz zt'z + | 777 = OFF tte = (867°) g(19°I-) (2EL°0) (ZL 1)

ZL61 APH YSN0AYA ZL6T “APW WOAZ CZH" + MZ70° - Al€1O’ + 87°24 = Wda-NdD =(°72

(766°) ZL6T °99d YBNOAYI TL6T ACW WOAZ TTY” +

(267° -) IL61 T¥ady ysnorzy3 OL6T TFAdy worz 69° -,

"€Z61 “AON UT BuTuUzsaq

pue 0/61 YOAeW OF AOTAd OC ST pue fE/G6I °390 pue ‘qag useMjeq [ 3B SuTeUIeT *€/61 “qed UT [ 02 OL6T YOABW UT O worz (yquow ted ¢¢/{ Aq) A[avauT[ sesvatouy 7 e1a4A

(0° ) 884° = OHY g8°l = “Md (€6°%) (€T'2%-) C9S*T) Q(4L*I-) (881) (10°) G68" = ¢29°82 + ,20°€€ ~ ZI°OL + “MLvSO" - GESEO’ + 99° = NdA-WAD = ("T Z S}[Nsoy uUoTsSssis9y sf e[quy

-1-

- ]2 -

The remainder of the interest differential can be attributed to political risk, under the assumption that the probability of additional capital controls was uncorrelated with the level of capital controls in place during our sample period.& The equation specifications emphasize that the interest differential attrubuted to political risk depends not only on the probability of additional controls on capital inflows, but also on the stock of German government liabilities (B) and on the distribution of the counterpart net assets (here labeled "wealth") between the German private sector and non-residents (i.e., that size of W. relative

G

to B). As B increases holding W. constant, additional German

G debt is pushed into non-resident portfolios and the interest differential due to non-resident exposure to political risk eC increases. And as We increases holding B constant, thereby

reflecting a shift in wealth from non-residents to the German

private sector, the interest differential due to non-resident

“exposure to political risk is reduced. Thus, the signs of the

estimated coefficients on B and Wo are appropriate when the

6/ This latter assumption allows us to avoid attributing any of the

interest differential to the covariance between existing controls and . the probability of additional controls. This is consistent with the

fact that we do not treat the probability of additional controls as

a variable in our model, and thus we implicitly assume it to be constant ~ throughout the sample period, independent of the level of existing controls. Even if we could measure variations in the probability of additional controls, however, it would be difficult to separate the anticipatory encouragement to capital inflows from the ultimate discouragement that would follow an increase in the probability of additional (nonretroactive) controls.

-13-

political risk concerns future interest payments to non-residents (i.e., capital inflows).

The two equations provide slightly different estimates of the time path of the interest differential attributable to political risk. Evaluated at the initial values of B and Wo» the first three terms of equations 1 and 2 respectively attribute interest differentials of -0.1 and 1.1 per cent per annum to political risk at the beginning of the sample period. Between January 1970 and the end of April 1973, when the interest differential reached its peak of slightly more than 10 per cent per annum, Germany's stock of international reserves increased - rapidly and B increased four times as much as Woe This pushed claims on Germany into non-resident portfolios and increased the interest differential due to political risk. Equation 1 estimates that the interest differential due to political risk peaked at 1.8 per cent per annum at the end of July 1973; equation 2 putsthe peak at 1.6 per cent at the end of September 1972. In both equations the direction of political risk changed during the last quarter of 1974, and equations 1 and 2 respectively attribute interest differentials of -0.6 and -0.4 per cent

per annum to. political risk at the end of December 1974.

- 14 -

Appendix: A Chronology of German Capital Controls 1/

I. On April 1, 1970 the Bundesbank reintroduced a special reserve ratio on the growth of banks' liabilities to non-residents. With the exception of a four month period, September through December 1971, when liabilities of both residents and non-residents carried equal special reserve ratios, bank liabilities to non-residents were subject to higher reserve requirements than bank liabilities to residents. This program served two purposes. First, it

induced German banks to pay lower deposit rates to non-residents than to residents. (This effect of the program probably was less important after May 1971 when controls were tightened to make “payment ‘of interest on deposits held by non-residents subject

to prior approval by the Bundesbank.) Second, it absorbed reserves and thereby "sterilized" the increase in the monetary base resulting from bank-reported capital inflows.

II. On May 10, 1971 interest payments on non-resident bank deposits exceeding DM 50,000 were made subject to prior approval by the Bundesbank, which was not normally granted.

III. On March 1, 1972 the Federal Government introduced a cash deposit requirement (Bardepot) of 40% on most types of new credits

of non-residents to German non-banks in excess of DM 2 million per

7/ Based on various issues of the Monthly Report of the German Bundesbank,

- 15 -

individual. The cash deposit, held by the Bundesbank,did not pay interest. The deposit was increased to 50% effective on July 1, 1972, and the exempt amount was simultaneously reduced to DM 0.5 million. The exemption was further reduced to DM .05 million

on January 1, 1973. On January 30, 1974 the cash deposit requirement was reduced to 20% and the exemption raised to

DM .1 million. In mid-September 1974 the cash requirement

was eliminated retroactively from August 1, 1974,

IV. On June 29, 1972 the Federal Government decreed that the purchase of fixed-interest securities by non-residents was subject to prior authorization. Fixed-interest securities included all maturities of bonds: for example, all bank bonds, mortgage bonds, communal bonds, industrial bonds, and public authority bonds. The authorization requirement was, in practice, equivalent to prohibition of such purchases. The authorization requirement for all but short term securities (less than four years to maturity) was terminated on January

30, 1974.

V. On February 5, 1973 the Federal Government extended its prior authorization requirement to the acquisition of domestic shares and mutual funds by non-residents, and to the raising of loans abroad by residents, including trade credits. Controls now applied to almost all capital transactions with non-residents, and no

longer just to transactions in fixed-interest securities. These

additional measures were terminated on January 30, 1974.

- 16 ~-

References

Aliber, Robert Z., "The Interest Rate Parity Theorem: A Reinterpretation," Journal of Political Economy, 81 (No. 6), 1973.

Dooley, Michael P., "A Model of Arbitrage and Short-term Capital Flows,'' International Finance Discussion Paper No. 40, Federal Reserve Board, 1974.

, "Note on Interest Parity, Eurocurrencies and Capital Controls," International Finance Discussion Paper No. 80, Federal Reserve Board, 1976.

Herring, Richard J. and Richard C. Marston, "The Forward Market and Interest Rates in the Eurocurrency and National Money Markets," in Carl H. Stem, John H. Makin, and Dennis E. Logue, eds., Eurocurrencies and the International Monetary System, Washington, American Enterprise Institute for Public Policy Research, 1976.

Reid, Frank, "Dummy Variables with a Transitional Phase," Canadian Journal of Economics, 10 (No. 2), 1977.

Wilton, David A., "Structural Shift With an Interstructural Transition Function," Canadian Journal of Economics, 8 (No. 3), 1975.