Capital Account Liberalization in Finland
Analysis of Structural Changes

The IMF Working Papers series is designed to make IMF staff research available to a wide audience. Almost 300 Working Papers are released each year, covering a wide range of theoretical and analytical topics, including balance of payments, monetary and fiscal issues, global liquidity, and national and international economic developments.

Abstract

The IMF Working Papers series is designed to make IMF staff research available to a wide audience. Almost 300 Working Papers are released each year, covering a wide range of theoretical and analytical topics, including balance of payments, monetary and fiscal issues, global liquidity, and national and international economic developments.

I. Introduction

This paper examines the structural determinants of Finland’s private sector net capital flows, and how the liberalization of these flows, which for the most part occurred during the second half of the 1980s and the early 1990s, affected the stability of the estimated structural parameters of the capital flow equation in Finland. Few studies to date have examined how the liberalization in Finland influenced the underlying determinants of capital flows. The study of Hãmãlãinen and Kovanen (1991) examines changes in the semi -elasticity of capital flows vis-à-vis changes in foreign interest rates due to elimination of capital controls, and notes that this elasticity had increased tenfold since 1984, following the introduction of certificates of deposits market in Finland. Kovanen (1992) examines how exchange rate expectations had become more important for the capital flows in the late 1980s, whereas the interest rate differential was important for capital flows during most of the 1980s. As the liberalization progressed, exchange rate expectations became more dominant in influencing capital flows.

Other studies that have analyzed general financial liberalization and the specific process of capital account liberalization have focused on money markets and exchange rate implications of the liberalization process. For instance, Starck (1988) studies the impact of capital account liberalization on the relationship between domestic and foreign interest rates, and shows that in Finland lending and deposit interest rates have become much more sensitive to international interest rates during the 1980s. The study of Kontulainen, Lehmussaari, and Suvanto (1990), as well as Lehmussaari, Suvanto, and Vajarme (1992), examine changes in exchange rate policy following liberalization, including the credibility of the pegged exchange rate regime. These studies note that both monetary and exchange rate policies became more linked since the mid-1980s as earlier the exchange rate policy was assigned an external role while monetary policy was geared toward internal stability. However, the credibility of the pegged regime was not substantially influenced by liberalization, as noted by Geadah, Saavalainen and Svensson (1992), who apply Svensson’s (1991) method to provide a test for the credibility of Finland’s exchange rate band. Their results suggest that the credibility of Finland’s exchange rate band could not be rejected within a 12-month horizon, except in the fall of 1991 following the peg to the European Currency Unit (ECU).

This study develops and estimates a model of international capital flows based on a portfolio equilibrium model of financial markets for a small open economy (Kouri and Porter (1974)). In this framework, capital flows are viewed as the mechanism by which domestic excess demand for or excess supply of money is removed. Consequently, the primary explanatory variables in the model are changes in the domestic income, current account balance of the balance of payments, domestic monetary variables, and foreign interest rates, all of which affect either the demand for or the supply of money in the economy. We derive a reduced-form structural equation for the capital flows in an attempt to avoid the simultaneity problem, and the resulting bias in the constant parameter estimates, that arises when domestic interest rates are treated as being exogenously determined. Further, since the model contains a monetary aggregate as an explanatory variable, we are able to estimate the effect of changes in the domestic money balances on the capital flows. In particular, an estimate of the extent to which monetary policy actions, as measured by changes in the Bank of Finland’s net domestic assets, are offset by capital flows is obtained as a result. The model also treats the exchange rate as an exogenous variable, hence it is applicable for a fixed exchange rate regime.

In order to capture changes in the structural estimates occurring during the period of capital account liberalization and thereafter, the estimation will be carried out recursively as well; the approach allows adjustments in the constant-parameter-estimates over time and also indicates the periods when structural breaks occurred. The information obtained from the recursive estimations is used to improve the fit of the standard ordinary-least-square estimates of the structural model.

Furthermore, we explore in some detail the role of exchange rate expectations and their determinants by estimating a structural equation for exchange rate expectations. The resulting equation is then inserted into the capital flow equation as an explanatory variable to test whether or not exchange rate expectations have a statistically significant, influence on the capital flows, particularly following liberalization. 2/ For the purpose of estimating a structural model for exchange rate expectations, we use the target zone literature, first developed by Krugman (1991). This approach is particularly appropriate for Finland where the official exchange rate regime was a basket peg combined with fluctuation limits. In practice, the studies have often used dummy variables to capture the changes in the capital flows due to speculative flows; this merely recognizes the difficulty of modeling empirically exchange rate expectations. 3/

To summarize, the estimation results support the theoretical framework adopted in the paper. The estimates for the exchange rate expectations are consistent with the target zone approach, and indicate significant mean-reversing property for the variable measuring lagged deviations from the central parity exchange rate. Other variables representing changes in the “fundamental” determinants of exchange rate changes also receive significant parameter estimates. Further, the fit of the estimated structural equation for private sector capital flows is quite good and estimated coefficients receive their expected signs and are statistically significant, except the interest rate variable and the proxy used for exchange rate expectations which receives an insignificant coefficient with an incorrect positive sign. Most of the structural breaks identified in the recursive estimation appear to relate to capital account liberalization, but also the liberalization that occurred in the domestic financial market in 1984 seem to have spilled over to the capital account. The estimation results indicate that domestic monetary policy autonomy was largely lost before full convertibility was achieved, suggesting that capital controls have not been fully effective.

The study is structured as follows. Section II discusses developments in Finland’s exchange controls on capital account transactions, adjustments in the exchange rate policy, and reforms in the domestic financial sector. Section III reviews developments in the main economic variables. A simple framework is developed in Section IV, for the purpose of facilitating the empirical analysis of Section V. Section VI concludes the paper.

II. Financial Market Liberalization in Finland

1. Capital account of the balance of payments

Liberalization of international capital transactions in Finland did not start on a large scale until the second half of the 1980s—and was completed in the early 1990s—although steps toward more liberal treatment of certain capital transactions were already taken during the 1970s. However, these measures were limited in nature and were not followed consistently, as the earlier liberalization measures were occasionally reversed at a later date. Indeed, during the 1970s capital account transactions were usually tied to underlying real transactions, such as exports and imports. A particularly negative attitude was taken toward portfolio investments abroad due to the perceived speculative nature of these investments since they were primarily motivated by exchange and interest rate expectations. On the other hand, in the early 1970s, a favorable attitude was adopted by the authorities toward long-term borrowing from abroad, as the Bank of Fini and was mainly concerned about the terms of such financing. However, in order to offset the easing of domestic liquidity resulting from the more liberal treatment of the long-term foreign borrowing, a deposit requirement was introduced. The procedures applied to the long-term borrowing from abroad were further tightened during the mid-1970s, reflecting the worsening current account balance. Direct foreign investment abroad by Finnish enterprises, excluding portfolio investment, was treated liberally and applications for direct foreign investments were only rejected in special circumstances.

While the long-term capital account transactions were still subject to authorization, the efficiency of exchange controls had already been weakened by the early 1980s as the leads and lags associated with short-term capital movements, including foreign trade and forward exchange transactions which required no special authorization, tend to undermine the controls. 4/ A significant change as regards to liberalizing capital movements occurred in 1980 when the Bank of Finland withdrew from the forward exchange market and handed over the task of hedging importers and exporters foreign exchange risks to the banks. 5/ Further, the banks were now able to raise foreign funds freely to cover their forward exchange risks, while their overall risk position was managed through established limits for the net foreign exchange exposure. Overall, the authorities’ attitude toward long-term borrowing from abroad became progressively more liberal in the 1980s. An exception to this favorable trend was the decision to ban the sales of markka-denominated bonds to nonresidents in 1985 (the decision was reversed in 1990). Also, a negative attitude toward portfolio investments abroad continued until the reform of the Foreign Exchange Act in 1986, reflecting concerns over the volatility of the private portfolio investments and their adverse impact on the economy. 6/

In 1986, the Bank of Finland accelerated the process of dismantling the remaining capital account controls. Most important were the lifting of the controls on long-term foreign borrowing with a maturity of at least five years by domestic manufacturing and shipping companies for financing their own operations. The controls on trade-related financing credits were abolished as the foreign financing of long-term export receivables was allowed and the banks’ imports quotas were abolished. During 1987, direct investments abroad by Finnish companies, excluding financial and insurance companies, were freely permitted up to a certain limit, the upper limit on portfolio investments abroad was increased, and the right to raise foreign funds with maturities of at least five years was extended to all enterprises engaged in business activity. In 1988, companies were permitted to conduct direct investments abroad without an upper limit. However, the restrictions continued to apply to financial and insurance companies. The upper limit on portfolio investments in securities was abolished.

The usefulness of exchange controls as an instrument of exchange and monetary policy had decreased considerably by the end of the 1980s, due to the global integration of financial markets and the internationalization of Finnish enterprises. Because of the reduced role of exchange controls, all foreign exchange transactions were now permitted unless the Bank of Finland specifically determined that they were subject to authorization. In 1990, restrictions on foreign investments by the residents were removed, together with the liberalization of purchases and sales of stock-based derivatives. The remaining exchange controls, including short-term capital transactions and access to foreign loans by resident individuals and comparable entities, were liberalized in 1991. 7/ A more detailed summary of exchange control measures affecting the capital account of the balance of payments is given in Attachment II.

2. Exchange rate policy

Following the collapse of the Bretton Woods system, Finland was one of the first countries to adopt a system in which the domestic currency was pegged to a basket of currencies. 8/ The basket-pegging had been adopted as a guideline for exchange rate policy already during 1973, but was not formalized until November 1977 when the Currency Act was amended. A formal exchange rate band was then introduced by which the external value of the markka was expressed in terms of an exchange rate index reflecting the bilateral exchange rates of currencies most important for Finland’s foreign trade. Since its inception in 1977, the peg has only been altered twice. In 1984, the Soviet ruble was removed from the currency index when the index number became based on freely convertible currencies, and in 1991 the trade-weighted currency index was replaced by the peg to the ECU unit (information about the adjustment of the official exchange rate is provided in Attachment III). 9/ The peg to the ECU was terminated in September 1992 when the markka was allowed to float (Chart 1).

Chart 1.
Chart 1.

Finland: Official Exchange Rate

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).

While the structure of the exchange rate arrangement remained broadly unchanged, the role of the exchange rate and exchange controls, however, has undergone considerable changes. As long as capital movements were tightly controlled and domestic financial markets were regulated, monetary policy had a substantial degree of autonomy and hence exchange rate could be used as an independent instrument of economic policy. 10/ Therefore, interest rates were geared toward demand management and structural objectives while decisions on exchange rate policy were primarily based on maintaining the competitiveness of the exportables sector. Although interest regulations and exchange controls were continued formally until the mid-1980s, their efficacy had weakened considerably much earlier. For instance, during the early 1980s, domestic enterprises and banks circumvented these controls following the development of the forward exchange market and the emergence of a short-term money market for grey money (see Section II.3).

Moreover, the Bank of Finland’s response to speculative attacks against the markka changed in the early 1980s. When the currency became subject to a strong speculative attack in 1982, the Bank of Finland devalued the markka exchange rate, first within the existing band and eventually by shifting the band position upward when the official exchange rate was devalued the second time. Speculative capital outflows reappeared in 1983; however, this time the Bank of Finland raised the call money interest rate which in turn pushed up the forward exchange rate and interest rates in the informal grey money market. This represented a shift from the previous competitiveness-oriented exchange rate policy, toward one by which the emphasis was on the stability of the exchange rate. It was made possible by the shift toward indirect instruments of monetary control. In fact, until 1992 when Finland moved to a floating exchange rate regime, the stability of the nominal exchange rate has been a proclaimed goal of the Finnish Government’s economic programs.

During the regime of the exchange rate peg to a basket, exchange rate policy was geared toward maintaining the exchange rate index below the central parity of the exchange rate band, except when the band position was shifted by the Bank of Finland. Often, the implicit band that guided the Bank of Finland’s interventions in the foreign exchange market was narrower than the announced exchange rate band. The rationale for maintaining the exchange rate closer to the stronger end of the exchange rate band, assuming that the band itself was credible, arose from the policy autonomy that it provided as the exchange rate within the band affected the expected future changes in the exchange rate. When the exchahge rate is close to the stronger end of the band, the market expects the exchange rate to rise in the future. This policy has allowed the Central Bank to maintain domestic interest rates above those in other countries, as a result of positive devaluation expectations.

3. Financial market reforms

Developments in the domestic financial markets, together with abolition of credit and interest rate controls, influenced the external capital account because with a more liberal exchange control system, it was easier for the commercial banks to take advantage of the yield differential between domestic and foreign interest-bearing assets. Furthermore, liberalization also broadened the scale and scope of financial market activities (including the emergence of new financial instruments) since it permitted the scope of financial market trading to expand, and forced competition in financial markets.

Until the mid-1980s, extensive exchange controls, particularly on capital transactions, allowed the Bank of Finland more independence in determining domestic interest rates. The elimination of capital controls started on a large scale only in the second half of the 1980s, although the significance of remaining exchange controls for domestic policy autonomy was already weakened before that. 11/ While a genuine market for short-term financial instruments had not emerged before 1987, this stage was preceded by a gradual movement toward interbank trading over the course of a ten-year period that started in the late 1970s. In particular, an important step toward the development of a market for short-term funds was the emergence of the “grey money market”. The initial spur for this market was given by the acceleration of inflation in the late 1970s and the negative real interest rates that resulted. Consequently, domestic enterprises were forced to seek outlets for higher returns on their net cash balances (this coincided with a large unmet demand for credit). Initially, the trading of cash funds was effected only between enterprises, but before long the banks had started to intermediate these funds through their trust departments and finance companies, which were beyond the scope of the Central Bank’s regulatory supervision. Consequently, the banks circumvented existing interest rate controls, while at the same time avoiding the stamp duties associated with these transactions. As lending by the banks’ became increasingly effected outside their balance sheets, it complicated the Bank of Finland’s credit control. After some initial hesitation, the Bank started to deregulate, step-by-step, the domestic financial markets.

Controls on lending interest rates were gradually removed during the period 1983-86. At the same time, the Central Bank moved from the direct controls of monetary aggregates toward market-oriented monetary management, including open market operations. For instance, during the 1970s monetary policy mainly operated directly through the borrowing quotas, and indirectly through controls on banks’ lending. By the early 1980s, a flourishing grey market had undermined the effectiveness of interest rate ceilings and credit restrictions, whereas increased integration of the Finnish economy into the international markets eroded the effectiveness of capital controls. Abrams (1988) notes that the reform of the regulations affecting banks’ funding began in 1980 when the Bank of Finland turned over the operation of forward exchange market in convertible currencies to the banks. While limitations were placed on the banks’ open foreign exchange positions, they were allowed to borrow and lend abroad to cover their exchange risks associated with the forward operations, thereby allowing the banks to exploit the differential between interest rates at home and abroad. Furthermore, the controls over banks’ lending decisions were minimal in general, but the interest rates that banks could charge their customers were tightly controlled, and were only liberalized in the late 1980s. The relaxation of conditions for the issuance of domestic certificates of deposits by banks in 1984 helped to develop the market for certificates of deposit in Finland, but the overall volume remained small until 1987. The market between the banks (Helibor market) started operating in 1986, when the Bank of Finland separated the overnight lending and deposit rates, but the final push for developing the interbank money market was given by the decision to eliminate the reserve requirement on certificates of deposits in 1987.

III. Data

During the period since the mid-1970s, except on a few occasions, the current account of the balance of payments has recorded a deficit, and the deficits grew quite large during the period following the capital account liberalization, particularly in the early 1990s. In particular, the current account deficit totaled about US$7 billion, or 1.4 percent of GDP, in 1990, but has since reversed to a small surplus (Chart 2). Despite the successive current account deficits, net capital inflows have been more than enough to finance these deficits, since the overall balance of the balance of payments has on average recorded a surplus (Table 1). The overall balance has been subject to large fluctuations during the second half of 1980s (Chart 3).

Chart 2.
Chart 2.

Current Account Balance

(As percent of GDP)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).
Table 1.

Finland: Balance of Payments

(millions of U.S. dollars)

article image

Source: International Financial Statistics. Figures are annual averages for the periods.

Private capital flows include the capital flow of banks and other private sector, as well as the errors and omissions.

Chart 3.
Chart 3.

Overall Balance of Payments

(In millions of U.S. dollar)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).

In particular, private sector capital flows have become highly volatile since the mid-1980s, following liberalization, as the yield differentials at home and abroad have been able to influence the flows more freely (Chart 4). The increased volatility of the private flows reflect partly the increasing significance of portfolio inflows, particularly after the mid-1980s (Chart 5), while direct foreign investments (FDIs) abroad have shown more stability, but have nonetheless grown. However, the net direct investments flows have since the late 1970s been dominated by net outflows, suggesting that while the yields on domestic and foreign assets may be important for the portfolio investment flows, direct investment flows are affected by other factors as well.

Chart 4.
Chart 4.

Private Sector Capital Flows

(As percent of GDP)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).
Chart 5.
Chart 5.

Finland: Portfolio and Direct Investment Flows

(In millions of U.S. dollars)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).___ = Portfolio—- = Direct Investment

The impact of liberalization of the capital account on the balance of payments is apparent in Table 1 which shows that annual net inflows of capital grew from about US$1 billion during 1975-84 to almost US$4 billion during 1985-94. Net direct investment outflows increased from a low annual level of US$71 million during 1975-84 to an average of US$1.3 billion during 1985-94. At the same time, net portfolio inflows of nonresidents to Finland have more than compensated for these direct investment outflows, increasing from an annual average of US$400 million during 1975-84 to almost US$4.7 billion per year during 1985-94. Net portfolio inflows started to expand rapidly after 1984. Other private inflows, including net borrowing by the banks and the nonbank sector have been relatively stable (about US$0.6 billion per year) both before and after the liberalization; however, this reflects the fact that the net inflows of the late 1980s were reversed during the 1990s, following the recession, while yearly variations in these flows have been relatively large.

Monetary and credit aggregates were also strongly affected by capital account liberalization (Charts 6 and 7), particularly since the mid-1980s. In the second half of the 1980s broad money grew almost 15 percent annually, and credit extended to the private sector expanded by about 20 percent per year during the same period. The establishment of an interbank money market in 1986, together with a more liberal capital account, made the domestic money market interest rates highly sensitive to changes in international rates.

Chart 6.
Chart 6.

Credit to the Private Sector

(As percent of GDP)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).
Chart 7.
Chart 7.

Broad Money Stock

(As percent of GDP)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).

IV. Theoretical Framework

A simple theoretical framework is used here to illustrate the linkages between interest rates at home and abroad, and the important role played by capital movements in this process. The model follows the portfolio balance approach of Kouri and Porter (1974), by which domestic and foreign holdings of interest-bearing assets are mainly influenced by the respective yields on these assets while capital flows work to clear excess demand for or supply of money balances in the model. It is assumed that there are three types of financial assets with stable demand functions in the economy: base money, domestic interest-bearing assets and foreign interest-bearing assets. Allocation of wealth between these assets depends upon domestic and foreign interest rates, income, and exchange rates. 12/

1. Private capital flows

More formally, the demand for domestic base money is mainly influenced by the level of interest rate at home (a function of the foreign interest rate). Equation (1) below essentially says that the demand for domestic money balances as percent of total wealth is influenced by the interest rate; 13/

MdY1const.-μi(1)

This equation can be transformed into Equation (1’) by differencing Equation (1) and reorganizing it, so that the left-hand side denotes the change in the demand for base money as percent of the beginning-of-period wealth.

ΔMdY1=μΔi+[(MY)(ΔYY1)]1(1)

The supply of base money is given by Bank of Finland’s balance sheet identity; that is, the change in the money supply equals the change in the Bank of Finland’s domestic and foreign net assets;

ΔMs=ΔNDA+ΔNFA(2)

Equilibrium gives the domestic interest rate;

Δi=(1/μ)[(MY)(ΔYY1)]1-(1/μ)(ΔNDA+ΔNFAY1)(3)

In this equation, changes in net foreign assets are defined as the sum of current account balance and the capital account balance, which includes the private sector capital flows, foreign direct investment flows, and official capital flows (ΔNFA = ΔCA + ΔKAP + ΔFDI + ΔKAO), thereby linking domestic and foreign interest rates.

We assume that private capital flows will be sensitive to changes in the relative yields to domestic and foreign assets, adjusted for expected changes in the exchange rate;

ΔKAPY1const.+gΔ[i-i*-(ΔeEe)](4)

where ΔeE denotes the change in the expected future exchange rate (this is unobservable), while g measures the elasticity of capital flows with respect to changes in the relative yields. If domestic and foreign interest-bearing assets are perfect substitutes and capital flows are unrestricted, then g would be very high (theoretically, infinite), since any differences in the relative returns would be arbitraged out immediately by the market. This parameter is of interest, since it is likely to increase in response to capital account liberalization. Moreover, since the domestic interest rate (i) is an endogenous variable, the formulae for the domestic interest rate is inserted into the capital flow equation, thus avoiding the simultaneity problem in the estimation. Using equation (4), we obtain a reduced form equation for the private sector net capital flows;

ΔKAPY1=const.δ1[Δi*+Δ(ΔeEe)]-δ2[ΔNDAY1+ΔCA+ΔFDI+ΔKAOY1-((MY)(ΔYY1))1](4)

where δ1 = g (1-θ), δ2 = θ, and θ=gg+μ between one and zero. In this equation, exchange rate expectations are endogenous. They will be discussed in more detail in the following section.

2. Exchange rate expectations

As noted previously, the Kouri-Porter type portfolio-balance model treats the exchange rate as an exogenous variable, determined by a fixed exchange rate regime. Furthermore, due to the inherent difficulty with modeling exchange rate expectations, Kouri and Porter (1974) and others have used dummy variables to capture changes in exchange rate expectations associated with the realignments of the official parity. In this study, on the other hand, we utilize the recent literature of target zones to model exchange rate expectations. Since Finland maintained a pegged exchange rate regime during 1977-92, by which the official exchange rate was maintained within a target zone, theoretical work on target zones provides a natural basis for modelling of expectations concerning future exchange rate changes. Recent work on target zones is basically founded on the work of Krugman (1991), and assumes that exchange rate variable, as any other asset price, depends on both a set of current fundamentals and expectations about future values of the spot exchange rate. An important assumption of the Krugman framework is that the target zone is fully credible and consequently the central bank will only defend the target zone, through marginal interventions, when the exchange rate locates either at the lower end or the upper edge of the band. In addition, a feature of the Krugman model is that the exchange rate within the band displays mean reversion, i.e., the expected future exchange rate within the band is closer to the long run mean of the exchange rate within the band the further away in time that it is. Also, the uncovered interest rate parity is assumed to hold, which implies that the risk premium is zero.

However, the predictions of Krugman’s model have been overwhelmingly rejected by empirical studies. Consequently, the model set out above has been extended in a number of ways. 14/ Host importantly, the central parity of an exchange rate band will occasionally realign in practice, through discrete jumps, while remaining constant between the realignments. Hence, expected changes in the exchange rate must reflect these discrete adjustments in the band position. The expected change in the exchange rate can be decomposed into a change in the central parity, called the expected rate of realignment, and a change in the exchange rate relative to the central parity, called the expected rate of currency depreciation. These modifications change fundamentally the relation between the interest rate differential and exchange rate change, since the inclusion of the time-varying realignment risk offers a reason why there need not be a fixed relation between the interest rate differential and exchange rates. The exchange rate within the band would still display strong mean-reversion.

Svensson (1990) examines the exchange rate risk premium within a target zone, with a devaluation/realignment risk. He demonstrates that the risk premium is the sum of two separate components, one arising from stochastic exchange rate movements within the band, and the other arising from the stochastic devaluation/realignment risk due to realignments. The change in the spot exchange rate can therefore be decomposed into the depreciation of the currency within the target band and the expected rate of change of the central parity. Since the first component is not directly observable, it may be obtained as the predicted value of the regression where the exchange rate change is regressed against its lagged level. The expected rate of change of the central parity, on the other hand, can be calculated on the basis of the difference between the nominal interest rate differential and the expected depreciation within the target band. Svensson (1991) found that the extent of depreciation within the target band is often sizeable, about ±5 percent per year. Bertola and Cabballero (1992) demonstrate that the relationship between the exchange rate and the fundamentals is steeper in a band which is not fully credible than is the corresponding relationship in a freely floating regime.

Thomas (1993) estimates the expected rate of depreciation for a set of countries by regressing the change in the exchange rate on current exchange rate, measured in relation to the central parity, and dummies that measure the effects of realignments. He notes that this allows the data to decide to what extent the exchange rate variable exhibits mean reversion or mean diversion in the sample. His results show mean-reversion in the sample countries (France, Italy and the United Kingdom). Furthermore, coefficients for the dummies are positive, reflecting the tendency for the exchange rate to be located at the lower end of the band after a realignment, while over time the currency depreciates to achieve mean reversion. Thomas also estimates the expected change in the central parity, using the differential between domestic and foreign inflation rates, relative unit labor costs, deviations of unemployment rate from NAIRU, government debt to GDP ratio (domestic and foreign), exchange rate deviations from the lower limit of the band, and the change is official reserves, as proxies for the fundamental. However, these variables show relatively weak significance levels in the estimation.

Helpman, et al. (1994) analyze performances of the exchange rate bands of Chile, Israel, and Mexico, using Krugman’s target zone approach. They estimate the realignment expectations using a simple model, consisting of dummies for realignments, a variable that measures the deviations of the exchange rate from the central parity and its second and third moments (these capture the nonlinear movements in the exchange rate deviations), a lagged deviation expectation, and several macroeconomic variables (such as income growth, change in reserves, and change in the real effective exchange rate). While the variable measuring the deviations from the central parity, as well as its second and third moments, receive statistically significant parameter estimates in these countries, the fundamental variables perform less well.

Bartolini (1993) studies the market expectations of a devaluation of the Irish pound during 1987-93. Using the approach of Rose and Svensson (1991) and Svensson (1992), expected deviation of the spot rate vis-à-vis the central parity are estimated using a linear regression model in which a set of fundamentals was used as explanatory variables.

Following the above discussion, an expected change in the exchange rate may be decomposed into two components, one measuring the realized deviations in the exchange rate from the central parity, whereas the other measures the anticipated deviation from the central parity. More formally, the expected rate of change in the exchange rate can be modelled as follows:

Δetet1=Σjαjdj-βXDEVt1+λZt1k+et1(5)

where d denotes a dummy variable (to capture the breaks in the structural equation associated with discrete adjustments in the central parity and otherwise), XDEV= (e - c)/e denotes deviations of the spot exchange rate from the central parity, and Z refers to the “fundamentals” representing expected deviations from the central parity, (eE - c)/e. 15/ The fitted values of equation (5), forwarded by one quarter and differences, are used in the capital flow equation as a proxy for the change in the exchange rate expectations. 16/

V. Estimation Results

The theoretical framework developed in Section IV is used to estimate equations for the private sector capital flows and the expected exchange rate change, using quarterly data of the Finnish economy for the period 1978Q2-92Q2. For the purpose of estimation, net private sector capital flows include net flows of deposit money banks and the other private sector, portfolio flows, and net errors and omissions of the balance of payments, reported in the International Financial Statistics. 17/ Direct investment flows are not included in the private sector net capital flows because, as noted above, these flows are likely to be influenced by other factors than the yield differentials suggested in the theoretical framework. Net capital flows of the official sector are treated as exogenous. The official exchange rate basket, based on trade weights, is used for the exchange rate variable in the period prior to mid-1991 when the official exchange rate became based on the ECU basket. The ECU basket is used thereafter and it is also used for the floating period (the interest rate variable mirrors the construction of the exchange rate variable). 18/

1. Exchange rate expectations

A structural equation for exchange rate expectations is estimated using equation (5), which is based on the target zone approach discussed in Section IV. Hence, changes in Finland’s spot exchange rate are influenced by changes in economic fundamentals, which reflect expected deviations from the central parity, and deviations from the central parity within the band (which are observable). Two estimation results are reported, one including no dummy variables, while the other including all statistically significant dummies.

When no dummy variables are included, the structural equation is able to explain about 30 percent of variation in the exchange rate variable, and all explanatory variables are highly significant (Table 2). As expected, the estimation indicates that the exchange rate displays the mean-reverting property, captured by XDEVt-1. The fundamentals also perform well, but the residuals fail the X2 test of normality.

Table 2.

Finland: Exchange Bate Expectations Sample period 1978Q2-92Q2

article image
H.C.S.E. = Heteroscedastic consistent standard errorHETER = test of unconditional heteroscedasticityARCH = test of autoregressive conditional heteroscedasticityGROWTH = (MY)(ΔYY1)

= significant at 5 percent level

= significant at 1 percent level

For the purpose of identifying structural breaks in the exchange rate equation, recursive estimation technique included in the PC-GIVE program is used. 19/ This method allows estimation of the structural parameters in a step-by-step fashion, recursively, thereby identifying any structural breaks in the structural equation. 20/ 21/ Using a recursive Chow-test (Chart 8), the break points can be identified (the significance of each point is tested with the likelihood ratio-test which is an F-test of one-step up Chow test). Accordingly, the statistically significant breaks occur in 1986Q2, 1989Q2, 1991Q4, and the break points are related to the corresponding realignments of the central parity. The estimated residual sums of squares (RSS) is shown in Chart 9, suggests that the model fits data well until mid-1991. Indeed, the estimation indicates that the decision to peg the markka to the ECU basket in 1991Q3 was not fully credible, as evidenced by increasing residual sums of squares. 22/

Chart 8.
Chart 8.

Recursive One-Step Up Chow Test

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).—- = 5 percent significant level.
Chart 9.
Chart 9.

Recursive Residual Sums of Squares

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).

Using information from the recursive estimation to identify the break points in the time series, equation (5) was re-estimated and dummy variables included for significant break points (Table 3). 23/ First, it is apparent that all the dummy variables are highly significant in the estimation and hence improve the fit considerably. Most of the other parameter estimates remain statistically significant, except the parameter estimate for changes in the Bank of Finland’s net domestic assets which is not significant and the official foreign currency reserves which is only significant at a 5 percent level. The residuals are not normally distributed, and there is heteroscedasticity in the residuals. This is not unexpected in a sample relating to the target zone (Svensson (1992)). These estimates for exchange rate expectations are used in the capital flow equation. Chart 10 shows how they differ in terms of their ability to track actual exchange rate movements.

Table 3.

Finland: Exchange Rate Expectations Sample Period 1978Q2-92Q2

article image
H.C.S.E. = Heteroscedastic consistent standard errorHETER = test of unconditional heteroscedasticityARCH = test of autoregressive conditional heteroscedasticityGROWTH = (MY)(ΔYY1)The dummy variables are constructed as follows:e.g.,DM824=1whent=1982Q4=0elsewhere

= significant at 5 percent level

= significant at 1 percent level

Chart 10.
Chart 10.

Estimates for Exchange Rate Expectations

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).____ = Based on results reported in Table 2.—- = Based on results reported in Table 3.

2. Sterilization of capital flows

A particular issue arises from changes in the net domestic assets of the Bank of Finland, an explanatory variable in the capital flow equation. It is possible that the monetary authority tries to adjust its net domestic asset position in order to offset the liquidity effects arising from capital flows, that is, either to fully or partially sterilize changes in the net foreign asset position. Furthermore, monetary policy may be used to achieve a balance of payments target; then the Central Bank would increase rather than decrease the net domestic assets of the Central Bank when responding to an inflow of capital. 24/ These considerations would give rise to a bias in the parameter estimates when a single-equation estimation technique is used. In the case of sterilization the offset coefficient is biased toward minus one, and in the case of targeting the balance of payments the bias is toward zero (Kouri and Porter (1974)).

The presence of sterilization can be tested by regressing the changes in the net domestic assets against the capital flows (possible including other variables to improve the overall fit). It should be noted, however, that a simultaneity problem does not arise in the case of current account variable, since the current account balance represent an autonomous source of foreign exchange and therefore the change in the monetary base is brought about through its impact on the balance of payments.

A negative, statistically significant, parameter estimate for the net capital flows suggests that the Central Bank has indeed sterilized private capital flows during the estimation period (Table 4). However, the estimate of the elasticity of the net domestic assets with respect to capital flows is not very large, -0.462, implying partial sterilization. When the estimation is carried out recursively (Charts 11 and 12), the estimate of the capital flow elasticity declines slightly over time, but its t-value rises significantly since 1986. This could indicate that the Central Bank’s interventions for the purpose of sterilizing capital flows have become more systematic in the second half of 1980s. In the estimation of the capital flow equation, the net domestic asset variable will be treated as an exogenous variable, since the possible bias is relatively small. 25/

Table 4:

Finland: Estimation of Sterilization Coefficient Sample Period 1978Q2-92Q2

article image
H.C.S.E. = Heteroscedastic consistent standard errorHETER = test of unconditional heteroscedasticityARCH = test of autoregressive conditional heteroscedasticity

= significant at 5 percent level

= significant at 1 percent level

Chart 11.
Chart 11.

Recursive Estimate of the Capital Flow Elasticity

(including ±2 standard error bars)

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).
Chart 12.
Chart 12.

Recursive Estimate of the T-Value of the Capital Flow Elasticity

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).

3. Private sector capital flows

In this section, estimation results for private sector capital flows are reported, with the exchange rate expectations being those obtained in Section IV.1. 26/

An empirical counterpart of equation (4’) is given by 27/:

ΔKAPtYt1=γ0-γ1Δit*-γ2Δ(Δet+1Eet)-γ3ΔNDAtYt-1γ4[ΔCAt+ΔFDIt+ΔKAOtYt1]+γ5(MtYt)(ΔYtYt1))t1+et2(6)

The purpose is to analyze structural changes in the estimated coefficients of the capital flow equation due to the liberalization of capital account controls. As before, the estimation proceeds in two stages. First, the recursive estimation is used to reveal the structural break points in the capital flow equation. 28/ Estimation result without any dummies is given in Table 5. While all parameter estimates except the proxy for exchange rate expectations receive correct signs, the estimated parameters of foreign interest rate and exchange rate expectations are not statistically significant. Furthermore, the parameter estimates of 73 and 74 are statistically not different from unity.

Table 5:

Finland: Estimation of Private Capital Flows Sample Period 1978Q2-92Q2

article image
H.C.S.E. = Heteroscedastic consistent standard errorHETER = test of unconditional heteroscedasticityARCH = test of autoregressive conditional heteroscedasticity

= significant at 5 percent level

= significant at 1 percent level

Using the one-step up Chow test (Chart 13), it can be seen that a number of breaks occurred during the 1980s and early 1990s. While the structural break of 1984Q2 relates possibly to the development of the domestic market for certificates of deposits (Section II.3.), it nonetheless has significant spill-over effects for private sector capital flows. On the other hand, the break points that occurred during the second half of the 1980s are more closely related to the liberalization of the capital account (including quarters 1987Q1, 1988Q2, and 1990Q2). However, the break point in 1991Q4 is not apparent since it may be related to the general instability in the foreign exchange market. Chart 14 plots the estimated residual sums of squares which indicate that the structural adjustment in the private capital flows has been relatively smooth, unlike the sharply rising RSS apparent in the estimated equation for the exchange rate in the early 1990s. While abstracting from most of the recursively estimated parameter estimates and their respective t-values, the recursive parameter estimate of NDA variable is shown in Chart 153 in equation (6)). It is apparent that the estimated constant-parameter shifts and its t-value rises after 1984 (Chart 16), but returns to its original level in 1990 when there is another shift in the constant-parameter estimate. 29/ This parallels the pattern observed for the sterilization coefficient in the previous section, possibly reflecting the increased role of the Bank of Finland’s sterilizations during a period of strong capital inflows. With respect to the other parameter estimates, the estimated coefficient of 74 has remained broadly constant, subject to small variation. However, the parameter estimate of 75 has risen steadily during the estimation period, reflecting the closer link between domestic demand for money and capital flows following the liberalization of capital controls.

Chart 13.
Chart 13.

Recursive One-Step Up Chow Test

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).—- = 5 percent significance level.
Chart 14.
Chart 14.

Recursive Residual Sums of Squares

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).
Chart 15.
Chart 15.

Recursive Estimate of the Elasticity With Respect to the Domestic Variable

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).
Chart 16.
Chart 16.

Recursive Estimate of the T-Value of the Elasticity With Respect to the Domestic Credit Variable

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).

Next information from the recursive estimation is used to improve the overall performance of the constant-parameter estimates. First, all dummy variables are included for the significant break points identified in the recursive estimation. Second, each explanatory variable is broken down into two separate series, reflecting the possible changes in the coefficients due to the liberalization of the capital account. 30/

The final estimation results are reported in Table 6 which suggest that the liberalization has indeed increased the sensitivity of capital flows to changes in the underlying determinants of the flows. The parameter estimates of the interest-rate elasticity are not statistically significant. 31/ Also, the proxy for the exchange rate expectations failed to produce a parameter estimate that would be statistically significant and negatively signed, as predicted by the theoretical model. This reflects, in part, the inherent difficulty in supplementing for unobservable behavior. 32/ The demand for base money (as reflected in the income variable) becomes more important for capital flows, consistent with the portfolio-balance approach, and the income elasticity has almost doubled, from 0.050 to 0.091. Further, the estimated elasticities of γ3 and γ4 are -0.833 and -1.021, respectively, and are not statistically different from one as suggested by the model. 33/ Overall, the structural model fits data well and the residuals behave as expected. 34/

Table 6:

Finland: Estimation of Private Capital Flows Sample Period 1978Q2-92Q2

article image
H.C.S.E. = Heteroscedastic consistent standard errorHETER = test of unconditional heteroscedasticityARCH = test of autoregressive conditional heteroscedasticityDMM=1fortheperiod1986Q2andonward=0elsewhere

= significant at 5 percent level

= significant at 1 percent level

4. Floating exchange rate period

The exchange rate of the Finnish markka was floated in September 1992 (Attachment III). Since the Kouri-Porter type portfolio-balance approach assumes that the exchange rate is fixed, the previous model structure is not applicable for the floating exchange rate period. In fact, under a managed floating both the exchange rate and the net foreign assets of the Bank of Finland become endogenous policy variables that have to be modeled explicitly. We abstract from this exercise in this study since this would require a much more complex model structure. In addition, limited data available for the floating period and the instability in the foreign exchange market during 1992-93 suggest that it would be difficult to obtain stable structural relations empirically.

However, since it is of interest to analyze the behavior of structural models using out-of-sample data, we generate forecasts of the exchange rate and capital flow models for the floating exchange rate period. 35/ The data seem to confirm the earlier theoretical argument that the target zone model is not appropriate for the floating exchange rate period since the model fails to capture the dynamic pattern of actual exchange rate change (Chart 17). On the other hand, the forecasting ability of the capital flow model is much better and the model captures the dynamic pattern of the capital flow variable well. However, some systematic underestimation of the capital inflows is apparent in the end of the forecasting period (Chart 18). 36/

Chart 17.
Chart 17.

Exchange Rate Expectations Actual and Forecasted

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).___ = Actual—- = Forecasted
Chart 18.
Chart 18.

Private Sector Capital Flows Actual and Forecasted

Citation: IMF Working Papers 1995, 131; 10.5089/9781451933291.001.A001

Sources: IMF, International Financial Statistics (various issues); and the Bank of Finland Monthly Bulletin (various issues).____ = Actual—- = Forecasted

VI. Conclusions

This study analyzes structural changes that occurred in Finland during the 1980s and early 1990s, following the liberalization of capital controls and reforms elsewhere in the domestic financial system. In addition to the constant-parameter estimates that are provided, the study utilizes recursive estimation techniques to identify breaks in the estimated parameters during the course of financial liberalization. The recursive estimation technique turns out to be quite useful in identifying structural breaks and revealing variations in the constant-parameter-estimates over time. This information is then used to improve the fit of ordinary least-square estimates.

The study employs the target zone approach to model the exchange rate expectations. The parameter estimates obtained are consistent with the chosen approach, indicating mean-reversing property of the variable measuring lagged deviations from the central parity rate. Other variables representing changes in the fundamental determinants of exchange rate changes receive significant parameter estimates in the majority of cases, except official reserve holdings, which is only significant at 5 percent level, and the Bank of Finland’s NDA which is not significant. The dummy variables relate to the structural breaks identified in the recursive estimation and are mostly linked to realignments in the central parity.

The fit of the estimated structural equation for private sector capital flows is good and all estimated coefficients receive expected signs and are statistically significant, except the interest rate variable and the proxy for the expected exchange rate change which is insignificant and receives an incorrect positive sign. Most of the structural breaks identified in the recursive estimation relate to capital account liberalization, but the liberalization that occurred in the domestic financial market in 1984 appears to spill over to the capital account in the form of a shift in the private capital flows. Overall, the adjustment of the private capital flows to the liberalization is smoother than is the case with the exchange rate variable.

The study examines the changes in the structural parameters in response to the capital account liberalization. While some of the parameter estimates respond to the liberalization in a predictable way, the scale of the increase in these estimated parameters is smaller than expected on the basis of the theoretical model. The elasticity of the private capital flows with respect to changes in the current account variable and the net domestic assets of the Bank of Finland remain broadly stable, despite progressive liberalization. These parameter estimates are not significantly different from one which suggest capital controls have not been fully effective, thereby undermining the Bank of Finland’s monetary autonomy even before the elimination of capital controls. On the other hand, the parameter estimate for the income variable indicates that the link between the demand for domestic base money and the private capital flows has become more important.

Capital Account Liberalization in Finland: Analysis of Structural Changes
Author: Mr. Arto Kovanen