The experience of the 1970s and 1980s has demonstrated that the theory according to which free floating exchange rates would allow governments to orient their instruments of monetary policy toward domestic economic objectives, without paying attention to nominal exchange rates, has proved wrong. The main reason for this failure lies in the development of sizable and protracted deviations of real exchange rates from their equilibrium levels, which tended to worsen the configuration of current account imbalances. In such a context, the Spanish monetary authorities have defined their domestic objective in terms of nominal GDP growth and attempted to set accordingly their intermediate target—the growth of broad money—while at the same time striving to pursue an exchange rate objective. While those attempts may in the short run meet with success, they are potentially inconsistent. In particular, when a restrictive monetary policy combines with a more lenient fiscal stance to increase the interest rate and induce an incipient appreciation of the nominal and real exchange rate of the peseta, this tends to dampen inflation while widening current account and budget deficits. Unless lower inflation reverses in time the increases in interest and real exchange rates, the widening of those deficits may persist, exacting increasing costs. In addition, there is the question of to what extent the money supply and the exchange rate can be targeted independently. To insure compatibility between domestic and external objectives, the monetary authorities have resorted to a two-pronged approach: (1) interventions in the foreign exchange market have been directed to the fulfillment of the external objective while attempts have been made to sterilize their monetary impact; (2) capital controls—and at times credit rationing through moral suasion—have been set up to reduce the degree of capital mobility. An analysis of the experience with this approach might provide insights into the ability of the Spanish monetary authorities to maintain the peseta within the ERM broader band while retaining a measure of monetary independence.
The paper reviews the theoretical framework adopted in the literature to measure the effectiveness of sterilized intervention and applies this approach to Spain for the period 1980–89. Empirical evidence indicates that the central bank was successful in sterilizing foreign exchange interventions, thus preserving its monetary autonomy. To determine the role played by imperfect capital mobility in this regard, the main characteristics of Spanish capital controls maintained over the period 1986–91 are analyzed. An assessment of their effectiveness suggests that capital controls were instrumental in reconciling the domestic and external objectives of monetary policy in Spain over the 1980s. To conclude, the ability of the authorities to pursue such a policy in the context of the participation of the peseta in the ERM and the removal of all capital controls in February 1992 is questioned.
Intervention and Sterilization
Central bank interventions involve the sale or purchase of liquid assets in foreign currencies, mostly foreign bonds, on the spot or forward exchange markets against domestic currency.
These operations can be traced in the balance sheet of the monetary authorities, after accounting for forward operations1 included in the off-balance sheet items. The asset side is divided into two components, net domestic assets, composed of private and public bonds, and net foreign assets, which comprise the gross reserves held in various forms (gold, bonds, and deposits in foreign currencies), minus the liabilities owed to foreign monetary authorities. The monetary base, namely currency in circulation and bank reserves, and the net worth constitute the central bank’s liabilities.
| Central Bank Balance Sheet | |
| Assets | Liabilities |
| Net foreign assets | Currency in circulation |
| Net domestic assets | Banks’ reserves |
| Net worth | |
| Central Bank Balance Sheet | |
| Assets | Liabilities |
| Net foreign assets | Currency in circulation |
| Net domestic assets | Banks’ reserves |
| Net worth | |
In such a stylized framework, intervention on the foreign exchange market translates into symmetric changes in net foreign assets and in the monetary base and affects the distribution of the private sector portfolio between domestic and foreign bonds, the quantity of domestic money, and, therefore, interest and exchange rates. In this regard, an intervention has the same effects as an open market operation conducted through the sale or purchase of domestic assets. When intervention policy, pursued to achieve some exchange rate target (in terms of level or variability), appears incompatible with the attainment of domestic objectives (inflation, real growth), monetary authorities attempt to neutralize its monetary impact. This so-called sterilized intervention is carried out by selling or purchasing domestic bonds so as to restore the initial level of the monetary base and thus protect the money supply from external influences.
Once the monetary effects of the intervention have been so sterilized, the exchange rate can be affected only to the extent that the private sector does not consider domestic and foreign assets as perfect substitutes. In this case, investors would ask for changes in interest and exchange rates to accept the alteration in the composition of their portfolio thus allowing the central bank to achieve its exchange rate target. This same story can be looked at from the different perspective of the monetary approach of the balance of payments. Any attempt by the central bank to change the supply of credit to the economy while trying to stabilize the exchange rate would result in an offsetting capital movement to an extent depending on the degree of substitutability and mobility of domestic and foreign assets.2
The degree of effectiveness of sterilized intervention raises therefore two main issues: (1) Does it affect the exchange rate? (2) To what extent is sterilization a consequence of the central bank behavior or an offsetting reaction of the private sector to an ex ante excess or shortage of domestic money?
These issues have been dealt with in the literature on the portfolio balance model3 where demand for domestic bonds, foreign bonds, and base money by the private sector are determined by expected returns, expressed as a function of interest rate and exchange rate changes, income, and wealth. Interest and exchange rates adjust so as to insure equilibrium with the government supply of domestic bonds and base money. Within this framework, the behavior of the central bank is represented by two policy reaction functions that describe how the central bank supplies base money through the purchase of either domestic or foreign assets.
Kouri and Porter (1974), Artus (1976), and Obstfeld (1983) estimated this model, generally under a less elaborated form, for different countries and different periods over the 1960s and 1970s. Their results supported the thesis that sterilization was, to a large extent, the outcome of the central bank’s behavior; evidence on the ability of sterilized intervention to affect exchange rates was more mixed.
Other authors4 embarked on a different approach. Instead of building a complete portfolio model to measure the degree of sterilization, they attempted to estimate directly the degree of imperfect substitutability between domestic and foreign assets. They reasoned that finding evidence of imperfect substitutability would, by the same token, provide support for the thesis that sterilized intervention was effective. Their results provided little support for the hypothesis of imperfect asset substitutability in the recent period and were consistent with the conclusions obtained by most of the studies prepared in 1982 for the Versailles summit according to which sterilized intervention cannot affect exchange rates significantly and lastingly.
Intervention and Sterilization in Spain, 1980–89
To measure the degree of sterilization of foreign exchange intervention carried out by the Spanish central bank over the period 1980–89, two equations were estimated following the specification suggested in Kouri and Porter’s 1974 seminal paper and briefly outlined in the first section.
The first equation allows for the determination of the offsetting behavior of private capital flows in response to changes in domestic credit initiated by the central bank. In this equation, net private capital flows are determined by changes in domestic credit, in foreign interest rates, in domestic and foreign nominal incomes and wealth; the current account; and exchange rate uncertainty.
The leeway enjoyed by the central bank in the conduct of monetary policy is reflected in the offset coefficient linking private capital flows to changes in domestic credit. The offset coefficient is equal to - 1 if any change in domestic credit is offset by an opposite capital flow thus depriving monetary authorities of any degree of autonomy and to zero in the case of total independence.
The second equation allows for the determination of the fraction of foreign exchange intervention that is sterilized by the central bank. The equation is specified as an ad hoc reaction function of the central bank, relating changes in domestic credit5 to changes in the central bank’s foreign reserves; the nominal effective exchange rate; domestic inflation; and domestic demand pressures defined as deviations of actual output from its potential level.
Under the assumption that the central bank is willing to sterilize the monetary impact of all foreign exchange intervention, the sterilization coefficient linking changes in foreign reserves to changes in domestic credit is equal to - 1; its value is zero if no sterilization occurs.
These equations were estimated over the period Q1 1980 to Q3–1989.6 Estimates of the offset coefficient indicate that about 22 percent of changes in domestic credit were offset by private capital flows. This result suggests that private capital flows did not significantly undermine the ability of the central bank to achieve its domestic objectives. Estimates of the sterilization coefficient indicate that about 80 percent of the monetary impact of foreign exchange market interventions on the money supply were sterilized by the central bank during the 1980s.7 This result is consistent with those obtained in other studies on different countries.
The central conclusion emerging from these estimates is that during the 1980s the Bank of Spain was able to maintain a rather independent monetary policy, while pursuing an exchange rate target. One can surmise that the achievement of domestic and external objectives, otherwise incompatible, was made possible by imperfect capital substitutability or mobility, or both. The next sections review the main characteristics of the Spanish capital controls responsible and their apparent impact on capital mobility.
Foreign Exchange Regulations
Concerned that free capital flows might disrupt the rather unsophisticated Spanish financial markets and reduce their monetary autonomy, the authorities maintained a very tight control on both capital inflows and outflows, under the provision of the Act on Foreign Investment introduced in 1974, up to their entry in the European Community in 1986. Over the period 1986–92, Spain progressively phased out all capital controls generally well ahead of the schedule retained in the EC directives.8 The most important impediments to capital flows maintained over this period can be classified into four sets.
The first set limited the ability of nonresidents to speculate against the peseta by restraining their access to Spanish money and capital markets. Financial loans in pesetas granted by residents to nonresidents and purchases by residents of securities denominated in pesetas and issued by nonresidents were tightly controlled or forbidden. These restrictions were alleviated, however, in two respects: (1) international bodies were allowed to issue bonds denominated in pesetas on the Spanish financial market as early as 1987, but the bonds needed to be approved by the Ministry of Finance; (2) Spanish banks could lend up to 50 percent of their liabilities in convertible pesetas to nonresidents after March 1990. It is worth noting that these remaining constraints could easily be circumvented by forward operations,9 totally liberalized after July 1989. All restrictions on financial loans in pesetas to nonresidents were phased out in April 1991.
The second set of measures implemented to dampen short-term capital outflows involved the interdiction for residents to hold deposits in foreign currencies abroad or in the Spanish banking system, except for those balances related to international trade or denominated in ECU. These restrictions were eliminated in two steps. As of April 1991, residents were allowed to open accounts in foreign currencies in Spanish authorized banks. As of February 1992, the interdiction to hold monetary assets abroad was phased out.
The third set of regulations monitored the issuance of foreign assets on Spanish stock exchanges. These regulations reflected the authorities’ concern that the domestic capital market might be destabilized by the free access of foreign borrowers due to its segmentation, lack of depth, and rather unsophisticated level of development. In July 1989, following the reform of Spanish stock exchanges, the admission of foreign securities on the Spanish capital market was freed with respect to capital controls. Their admission is, however, still submitted to the agreement of the Comision Nacional del Mercado de Valores (CNMV), the body regulating stock exchanges.
The last group of restrictions concerned transitory measures designed to prevent undue appreciation of the peseta associated with resident foreign borrowing. Thus, while the Bank of Spain liberalized one-year loans in foreign currencies up to the limit of Ptas 750 million in March 1986 and doubled this ceiling in March 1987, upward pressure on the peseta in 1988 and 1989 prompted the authorities to restore more restrictive measures. As of June 1988, all new foreign borrowing by residents with an average maturity of less than three years and an amount greater than Ptas 1,500 million needed to be authorized, and a 30 percent non-remunerated deposit requirement was imposed after February 1989 on all foreign borrowing. For loans intermediated by domestic banks, the deposit requirement was reduced to 20 percent of the loan. The deposit requirements were phased out at the end of 1989 for the domestic banks and in March 1991 for other Spanish residents.
From this brief review of the evolution of Spanish capital controls since 1986, it appears that the most important restrictions were oriented toward the monitoring of short-term capital inflows, in response to the tendency of the peseta to appreciate.
Effectiveness of Capital Controls
The main objective of capital controls is to regulate inward or outward capital flows so as to preserve the autonomy of monetary policy in a context of exchange rate stabilization. The effectiveness of capital controls in this regard can be measured directly by their apparent impact on the capital account of the balance of payments and indirectly by the degree of financial market segmentation.
Changes in the capital account after 1986 suggest, at first sight, the effectiveness of previous controls. Specifically, comparison of the main items of the capital account during the periods 1983–85 and 1986–88 (Table 10) points to an important development in the relative size of direct and portfolio investments by nonresidents after a significant easing of capital controls. These investments, scaled by reference to the average of the sum of merchandise exports and imports, increased by about 200 percent between the two periods under review. Another illustration of the apparent effectiveness of capital controls is provided by the impact of the nonremunerated deposit requirement imposed on foreign borrowing in 1989 to dampen capital inflows. Net foreign borrowing by residents—provided directly from abroad or indirectly by the Spanish banks—declined from Ptas 511.9 billion in 1988 to Ptas 73.1 billion in 1989. These examples, however, can be viewed from a different perspective in which capital controls operate less by rationing than by inducing substitution with more costly sources of financing.10 In the first example, one can argue that even without the removal of capital controls the huge increase observed in long-term capital inflows after 1985 would have occurred in any event—through the screening device of a discretionary policy leading to higher costs of financing and capital misallocation—as it was needed to offset the mounting imbalance between domestic saving and investment. In the second example, the already largely liberalized capital regulation offered many avenues for substitution as suggested by the increase in long-term capital inflows from Ptas 10,200 billion in 1988 to Ptas 17,200 billion in 1989.11
Long-Term Capital Flows
(In percent of changes of the average of the sum of exports and imports)
Long-Term Capital Flows
(In percent of changes of the average of the sum of exports and imports)
| 1983–85 | 1986–88 | Changes in Percent | |||
|---|---|---|---|---|---|
| Nonresident | |||||
| Direct investment | 3.8 | 7.5 | 97.4 | ||
| Portfolio | 1.1 | 6.1 | 454.5 | ||
| Resident | |||||
| Direct investment | 1.1 | 1.9 | 72.7 | ||
| Portfolio | 0.7 | 0.5 | -28.6 | ||
| Net position (nonresident minus resident) | |||||
| Direct investment | 2.7 | 5.6 | 107.4 | ||
| Portfolio | 0.4 | 5.6 | 1,300.0 | ||
| Total | 3.1 | 11.2 | 261.3 | ||
Long-Term Capital Flows
(In percent of changes of the average of the sum of exports and imports)
| 1983–85 | 1986–88 | Changes in Percent | |||
|---|---|---|---|---|---|
| Nonresident | |||||
| Direct investment | 3.8 | 7.5 | 97.4 | ||
| Portfolio | 1.1 | 6.1 | 454.5 | ||
| Resident | |||||
| Direct investment | 1.1 | 1.9 | 72.7 | ||
| Portfolio | 0.7 | 0.5 | -28.6 | ||
| Net position (nonresident minus resident) | |||||
| Direct investment | 2.7 | 5.6 | 107.4 | ||
| Portfolio | 0.4 | 5.6 | 1,300.0 | ||
| Total | 3.1 | 11.2 | 261.3 | ||
An indirect but more relevant approach to estimate the effectiveness of capital controls is to determine to what extent they have resulted in segmentation between domestic and foreign financial markets thus allowing for the conduct of an independent monetary policy. Estimates of the model discussed earlier suggest that offsetting capital flows did not undermine significantly the achievement of the domestic target of monetary policy in Spain over the 1980s. However, it was not possible to tell if this outcome was a consequence of imperfect assets substitutability or impaired capital mobility. An analysis of the differences between the Euro and domestic interest rates of the peseta (Chart 6) provides evidence that impaired capital mobility was the main culprit as it severed the links between the domestic and foreign money markets of the Spanish currency.12 The spread between the two interest rates, first positive in 1986, turned negative at the beginning of 1987 and stayed that way during most of the years 1987–89, reaching at times a level close to 6 percentage points. Under the assumption of perfect capital mobility, this spread should have been eliminated.13 Its persistence was a reflection of two kinds of measures. First were those implemented to dampen speculation against the peseta by hampering the ability of nonresidents to borrow or sell pesetas forward. The market segmentation engineered by these restrictions allowed for changes in the Euro interest rate of the peseta consistent with the expected depreciation of the currency14 while the domestic interest rate was kept down to fulfill internal macroeconomic objectives.15 Second were those implemented to contain the appreciation of the peseta by reducing the profitability of short-term investments in pesetas either directly by limiting the remuneration of nonresident deposits16 or indirectly by reducing the forward discount of the peseta. The effectiveness of this last measure was enhanced by discretionary limits imposed on the short open foreign exchange position of Spanish banks.17 During the period 1978–89, characterized by high pressures in favor of the peseta, these restrictions accounted for the negative spread between the Euro and domestic interest rate of the peseta.
Three Month Euro Rate and Domestic Intrest Rate
(In percent)
Sources: Bank of Spain and Chase Manhattan Bank—Spain.1 In pesetas.2 In Euro-pesetas.Three Month Euro Rate and Domestic Intrest Rate
(In percent)
Sources: Bank of Spain and Chase Manhattan Bank—Spain.1 In pesetas.2 In Euro-pesetas.Three Month Euro Rate and Domestic Intrest Rate
(In percent)
Sources: Bank of Spain and Chase Manhattan Bank—Spain.1 In pesetas.2 In Euro-pesetas.Conclusion
In 1989, the conflict between the domestic and external objectives of monetary policy was brought to the fore. Large capital inflows induced by the more restrictive stance of monetary policy adopted to cool down the activity put the authorities in a dilemma either to accept an appreciation of the peseta leading to a further deterioration in the current account or to intervene on the foreign exchange market and jeopardize monetary policy. The dilemma was heightened by the participation of the peseta in the wide band of the ERM in June 1989. In this context, the authorities reintroduced some capital controls—while pursuing the dismantling of most of them according to EC directives—and implemented credit ceilings. The strategy of reinforcing a measure of imperfect capital mobility, in a way consistent with the pattern observed in the 1980s, was effective. However, such a policy cannot be pursued after 1992 when the integration of EC financial markets is expected to be achieved.
Appendix A Measure of Foreign Exchange Market Efficiency: The Case of the Spanish Peseta
This annex provides estimates of the degree of substitutability between short-term deposits on the Euromarket of the main currencies on the Euro-market by testing the Uncovered Interest Parity (UIP) conditions.18 UIP conditions state that interest rate differentials between two currencies or equivalently their forward exchange margins cannot differ on average from the rationally anticipated changes of the future spot rate if foreign exchange markets are efficient and economic agents are risk neutral. If it were not the case, economic agents could make riskless profits, for instance, by borrowing the currency with low interest rate to invest in the currency with high interest rate.
UIP was tested on monthly and quarterly data under the following form:
where e is the observed first difference of the logarithm of the exchange rate of the peseta in t + 1—measured at end of quarters or over three-month intervals; i and ie are the contemporaneous three-month Euro interest rates of the peseta and of the relevant foreign currency; u is an error term with the usual statistical properties. For UIP to hold, a and b should be significantly different from zero and 1, respectively.
Euro interest rates have been retained to remove the influence of imperfect capital mobility affecting some domestic interest rates. Except for the monthly Euro rate of the peseta provided by Chase Manhattan Bank’s Spanish branch for the period 1986–90, all other series are from the IMF’s International Financial Statistics (IFS) data base.19
The UIP test was applied on monthly data for the peseta exchange rates against German mark (DEM) and French franc (FRF) and on quarterly data vis-à-vis the same currencies plus the Belgian franc (BEC) and the U.S. dollar (USD).
A perusal of the results reported in Table 11 suggests that UIP cannot be rejected on monthly data since the a and b coefficients do not differ significantly from their theoretical values. However, this result should be viewed with caution since the use of monthly interest rates to forecast three-month changes in the exchange rate results in a moving average process on the residuals.20
Estimates for the Period May 1986-January 1990
Estimates for the Period May 1986-January 1990
| F(2.86)=2.87 |
| R2=0.06 |
Estimates for the Period May 1986-January 1990
| F(2.86)=2.87 |
| R2=0.06 |
On quarterly data, estimates displayed in Table 12 indicate that UIP could not be rejected in the case of the EMS currencies—with a qualified positive response for the French franc. It is worth stressing also the weakness of the standard economic indicator, R2. It confirms a well-known phenomenon in the empirical literature, that is, the forward rate is not a good predictor of future exchange rate changes, even if it is an unbiased one. These results are consistent with those obtained by Galy (1988) on market efficiency in the case of EMS currencies and by Fama (1984) concerning the identification of contemporaneous disturbances in UIP tests.
Estimates for the Period Q3–1980 to Q3–1989
Estimates for the Period Q3–1980 to Q3–1989
| F(4.140)=9.29(*) |
| R2=0.21 |
Estimates for the Period Q3–1980 to Q3–1989
| F(4.140)=9.29(*) |
| R2=0.21 |
The central bank can intervene directly on the forward exchange market by outright sales or purchases of foreign currencies or indirectly by combining spot operations with currency swaps. In either case, the balance sheet of the central bank would not be affected. These operations would be recorded in off-balance sheet items.
A sale of domestic assets implemented by the central bank to bring down domestic credit induces an increase in the interest rate, an offsetting capital inflow, and an incipient appreciation of the exchange rate. To thwart this appreciation, the central bank would have to buy foreign assets, thus restoring the initial level of the monetary base.
Kouri and Porter (1974) extended the portfolio balance model developed by Tobin (1989) for a closed economy to an open economy.
Changes in domestic credit are measured at the level of the consolidated banking system.
Detailed results were published in IMF internal papers and are available from the author.
Stability tests fail to detect a possible change in regime following the entry of Spain in the EMS in the case of the first equation. In the case of the reaction function of the central bank, however, it seems that the sterilization behavior has been stronger after 1985.
Note that after 1992, Spain would still have the possibility to restore capital controls under the “safeguard clause” but with the consent of the EC Commission.
Borrowing pesetas for three months and selling them spot against dollars is equivalent to a forward sale of pesetas against dollars for the same maturity.
For instance, an importer who is not allowed to cover his position in foreign currency on the forward market can switch the denomination of his contract to domestic currency and let the foreign exporter cover his position on the Euro-market where the discount on the domestic currency is likely to be larger than on the domestic market.
Note that foreign borrowing by the public sector and loans granted by foreign firms to their Spanish subsidiaries and classified as direct investments were not affected by the nonremunerated deposit requirement.
A spread between the Euro and domestic rates of a given currency can persist, even with perfect capital mobility, reflecting differences in reserve requirements applied on residents’ and nonresidents’ deposits.
Uncovered Interest Parity (UIP) tests suggested that the interest rate differentials between the peseta and main ERM currencies on the Euro-market were consistent with rationally expected exchange rate changes—see appendix.
Similar restrictions were imposed in the case of the lira and the French franc (see Giavazzi and Giovannini (1989) and Bruneel and others (1985>)).
Spanish banks were not allowed, up to March 1990, to pay interest on nonresidents’ deposits in convertible pesetas for balances higher than Ptas 10 million.
When nonresidents are purchasing pesetas on the forward market against a given currency, Spanish banks cover their position by borrowing the currency and selling it on the spot market against pesetas, thus taking a short open position in the spot market. The 20 percent nonremunerated deposit imposed between February and November 1989 on increases in this short position raised by 25 percent the cost of foreign borrowing for the banks, which reduced accordingly the swap rate offered to nonresidents and therefore the incentives to buy pesetas. Note that banks are not incurring a net foreign exchange position since the on-balance-sheet short position is offset by an opposite forward position.
UIP was tested using the seemingly unrelated regressions procedures proposed by Zellner (1962) following the algorithms presented by Judge and others (1988).
The three-month Euro-peseta interest rate on quarterly data for the period 1980–85 was not directly available. It was calculated on the basis of the three-month forward margin of the peseta against the U.S. dollar and the three-month Euro-dollar rate, assuming that the covered interest parity paradigms always held.
