I. Introduction

One of the major structural changes in the international economy during the past two decades has been the growing integration of capital markets in the industrial countries. This integration has reflected both the dismantling of capital controls and the removal of restrictions that have limited competition or asset price flexibility in domestic financial markets. The extensive liberalization of restrictions on cross-border financial transactions, in particular, stands in sharp contrast to the limited progress (or, in some cases, regression) that has been evident in reducing barriers to trade in goods and services.

As restrictions on domestic and external financial transactions have been lowered, financial markets have helped magnify the short-run spillover effects of domestic macroeconomic policies on neighboring countries. At the same time, the institutional changes brought about by financial liberalization may have weakened the authorities’ ability to respond to those spillover effects. In particular, questions have been raised about whether the effectiveness of traditional macroeconomic policy instruments has been reduced, whether exchange rate arrangements and intervention strategies need to be altered, and whether the coordination of macroeconomic policies among the major countries needs to be increased.

This paper examines one aspect of these policy issues--namely how should a country’s exchange rate policy be adjusted when the degree of integration between domestic and external financial markets increases as a result of both domestic financial liberalization and the relaxation of capital controls. ^2/ It is interesting to note that rather different approaches for solving this policy problem are being adopted in Europe and North America. The Delors Report (1989) has recommended that, in order for the members of the European Monetary System (EMS) to enjoy the full benefits of financial market integration, they should maintain a system of relatively fixed exchange rates between themselves. ^3/ In contrast, although the United States-Canada Free Trade Agreement contains provisions which significantly reduce or eliminate barriers to financial transactions between residents of the two countries, the Canadian authorities seem committed to a relatively flexible exchange rate policy vis-à-vis the United States and other countries. These different approaches leave unanswered the question of what exchange rate arrangements should be adopted by countries on the “periphery” of these large trading blocs, especially if these peripheral countries are also planning to reduce their capital controls and undertake domestic financial reforms.

Our analysis of the relationship between financial liberalization and exchange rate policy is divided into four sections. Section II briefly considers how the recent changes in capital controls and domestic financial regulations in the industrial countries have altered the terms and conditions under which borrowers can obtain funds from domestic and external financial markets. Section III develops a relatively simple model of real and financial sector linkages in an open economy that is used to provide a framework for analyzing how the changes in financial structure induced by the removal of capital controls and the liberalization of domestic financial market restrictions can influence optimal exchange rate arrangements. In this model, a financial constraint arises from the assumption that firms use credit from commercial banks to finance their working capital needs prior to the sale of their output. ^4/

In Section IV, this model is used to identify the exchange rate arrangements that best insulate the economy from domestic and foreign real and financial market shocks under three alternative assumptions about the economy’s financial structure: (1) when the economy has highly restrictive capital controls and places extensive restrictions on the domestic financial system; (2) when the authorities undertake a partial liberalization of domestic interest rates and the removal of some capital controls; and (3) when there is a complete removal of capital controls and the elimination of all restrictions on domestic interest rates. ^5/ To examine the role of creditworthiness considerations in influencing exchange rate arrangements, cases (2) and (3) are analyzed in the situations where international lenders either do or do not find it profitable to credit ration domestic borrowers. Two important results are obtained. First, as the economy’s financial structure is opened and liberalized the optimal scale of exchange rate intervention will change as the relative importance of different domestic and foreign shocks for output and price stability is altered. Moreover, although the optimal scale of exchange market intervention can vary with the financial structure, there is nonetheless a similarity in the response of the optimal degree of intervention to increases in the variances of domestic and foreign shocks across all financial structures. The final section of the paper summarizes our main conclusions and considers certain unresolved issues.

II. Liberalization and Structural Change in Major Financial Markets

Since the mid-1970s, the authorities in industrial countries have eliminated or weakened many of their restrictions on external and domestic financial transactions. ^6/ This liberalization of financial activities has reflected both the desire to produce greater competition and efficiency in domestic financial markets and a recognition that many large financial and nonfinancial institutions were increasingly turning toward the less regulated offshore markets to meet their financial needs. Although the specific financial liberalization measures have naturally differed across countries, there have been several common features: the elimination of barriers to market access; the removal of restrictions on the activities, products, and location of financial institutions; and greater reliance on market determined interest rates.

In the early 1970s, barriers to market access encompassed both capital controls limiting trade between domestic and external financial markets and limitations on domestic residents’ use of certain domestic markets. Capital controls and limitations on entry of foreign financial institutions into the domestic market ^7/ were part of the effort to isolate domestic financial systems from external developments. The limitations on access to certain domestic markets by domestic residents were typically designed to help protect specialized institutions or to prevent certain financial policies from being circumvented. ^8/ Although the lowering of restrictions on participation of foreign financial firms has often intensified competitive pressures in domestic markets, the removal of capital controls has been the most important change in the barriers to market access in many countries. ^9/

Restrictions on the activities, products, and location of financial institutions were often used to complement and reinforce the barriers to market access. Commercial and investment banking activities were separated in some countries partially due to concerns about minimizing the risks accepted by the commercial banking system. Limitations on the types of instruments that could be issued by certain institutions were also designed to enable other more specialized institutions to have adequate access to funds. The erosion of these restrictions has been most evident in those countries where a separation of commercial and investment banking has been maintained by law. ^10/

While ceilings on interest rates were employed in almost all industrial countries in the period up to the 1970s, there was a growing recognition that such ceilings could seriously interfere with the financial intermediation process, especially when inflation rates were rising and controlled interest rates were adjusted slowly. In particular, increases in market interest relative to prevailing ceiling interest rates on deposits could lead to a sudden withdrawal of deposits from certain institutions. Unless the borrowers from these institutions could find other sources of funds, their access to investment financing could be sharply and suddenly curtailed.

Interest rate deregulation has proceeded at divergent rates in the major industrial countries. While interest controls were effectively removed in the mid-1960s in the Federal Republic of Germany, Regulation Q ceiling interest rates in the United States were phased out in the early 1980s. In Japan, the process of interest rate deregulation is still underway. The Japanese authorities have followed a program involving the introduction of new deposit instruments (e.g., money market certificates) carrying market-related interest rates with a continuing decline in the minimum issuance size for these instruments. Nonetheless, the interest rates on small bank deposits and postal savings deposits remain subject to controls.

This process of liberalization has fundamentally altered the financial structures of the major industrial countries and has forged new linkages between the financial markets in these countries. Increased competition has forced many financial entities to re-examine the range of financial services and products that they can profitably produce. The competition for new customers has often involved the creation of new financial products and services. While many of these innovations quickly pass from the scene, this competition has greatly facilitated the efforts of portfolio managers to achieve an internationally diversified portfolio and the search of corporate (and sovereign) treasurers for the lowest cost source of funds.

Despite the removal of capital controls and the liberalization of domestic financial markets, however, different categories of borrowers still often face divergent terms and conditions under which they can access foreign and domestic sources of funds. Access to borrowed funds is typically affected both by official restrictions on market access (including capital controls) and by the lenders’ evaluation of the borrowers’s creditworthiness. The liberalization measures undertaken in recent years imply that, to an important degree, creditworthiness considerations have now replaced capital controls and other official restrictions on market access as the principal limitation on an individual borrower’s access to funds. Creditworthiness considerations affect both the cost of borrowing (due to the presence of a premium reflecting default risk) and the availability of funds. In particular, access to some markets may be curtailed for certain classes of borrowers even in the absence of government restrictions. For example, only sovereign and large corporate borrowers which are regarded as good credit risks can issue on the Eurobond markets. Thus, while most borrowers can now seek funds from a broader range of domestic and international institutions than in the pre-reform period, the “small-country” assumption that they can borrow unlimited amounts at a fixed international interest rate would not be an appropriate characterization of the terms and conditions under which most borrowers can access international markets.

The presence of such market imposed limitations on the availability of funds is not inconsistent with the evidence that covered interest rate differentials have been arbitraged away on comparable short-term instruments and sharply narrowed on equivalent medium and long-term instruments. ^11/ The elimination of covered interest rate differentials indicates that a borrower with a given credit rating can obtain funds at a cost that is identical across markets in different countries and when using instruments denominated in different currencies. Nonetheless, this does not rule out the fact that borrowers with different credit ratings must pay different risk premiums; nor is it inconsistent with the fact that as the amount of debt issued by a single borrower increases, lenders would typically become more concerned about the borrower’s ability or willingness to pay. While these concerns would be reflected in higher borrowing costs, the lender may strictly limit the availability of funds to a borrower when the expected profit on an additional dollar of lending (even at a higher interest rate) is reduced to zero because the probability of default would be increased sufficiently to offset the higher interest income associated with the additional lending.

Whether such “equilibrium” credit rationing is a valid characterization of the situation confronting borrowers on international markets is open to debate. ^12/ Even if credit rationing was evident, it is unlikely that all borrowers would face the same degree of rationing. Nonetheless, the nature of the terms and conditions under which domestic borrowers can access international markets are factors that could have a strong bearing on the degree of exchange market intervention that would best stabilize domestic income or prices as a country undertakes a financial liberalization and opens to external financial transactions.

III. Basic Model

To identify the exchange rate policy that would impart the greatest stability to either domestic output or the price level in the face of various domestic and foreign shocks as domestic financial restrictions (e.g., interest rate ceilings) and controls on international capital flows are removed, this section develops a simple model which incorporates crucial linkages between real activity, domestic financial intermediation, and external financial and goods markets. What ideally is needed to analyze this policy issue is a multicountry model of large economies that specifies the linkages between the real and financial sectors both within and across countries. As the complexities involved in developing such models are well known, we have a much more modest first objective of focusing on the linkages between the real and financial sector in a single open country and considering how changes in those linkages affect exchange rate policy. Our model, therefore, necessarily abstracts from many of the complex real and financial sector linkages that exist in a modern industrial economy. Moreover, since our focus is on the design of exchange rate policy, we do not explicitly consider the case where the authorities have decided to commit to a fixed exchange rate arrangement (such as the EMS) and use other policy instruments (e.g., the required reserve ratio) to stabilize the economy. However, the model could be used to analyze this situation.

2. Structural relationships

The economic structure described above can be represented by the following set of relationships:

$\begin{array}{cc}\left(1\mathrm{a}\right)& {\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}=\overline{\mathrm{y}}+{\mu }_0+{\mu }_1\left({\mathrm{b}}_{\mathrm{t}}^{\mathrm{s}}-{\mathrm{p}}_{\mathrm{t}}\right)+{\mu }_2\left({\mathrm{r}}_{\mathrm{m},\mathrm{t}}-{\mathrm{Ep}}_{\mathrm{t}+1}+{\mathrm{p}}_{\mathrm{t}}\right)+{\mathrm{n}}_{\mathrm{t}}\\ & \hfill \begin{array}{cc}\left(\mathrm{under}\mathrm{credit}\mathrm{rationing}\right)& \end{array}\end{array}$

$\begin{array}{cc}\left(1\mathrm{b}\right)& {\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}=\overline{\mathrm{y}}+{\mu }_3-{\mu }_4{\mathrm{r}}_{\mathrm{bt}}+{\mu }_5\left({\mathrm{r}}_{\mathrm{m},\mathrm{t}}-{\mathrm{Ep}}_{\mathrm{t}+1}+{\mathrm{p}}_{\mathrm{t}}\right)+{\mathrm{n}}_{\mathrm{t}}\\ & \hfill \begin{array}{cc}\left(\mathrm{without}\mathrm{credit}\mathrm{rationing}\right)& \end{array}\end{array}$

$\begin{array}{cc}\left(2\right)& {\mathrm{p}}_{\mathrm{t}}={\mathrm{s}}_{\mathrm{t}}+{\mathrm{p}}_{\mathrm{t}}^{*}+{\epsilon }_{\mathrm{t}}\end{array}$

$\begin{array}{cc}\left(3\right)& {\mathrm{m}}_{\mathrm{t}}^{\mathrm{d}}={\mathrm{p}}_{\mathrm{t}}+{\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}+{\eta }_1\left({\mathrm{r}}_{\mathrm{m},\mathrm{t}}-{\mathrm{Ep}}_{\mathrm{t}+1}+{\mathrm{p}}_{\mathrm{t}}\right)+{\mathrm{v}}_{\mathrm{t}}\end{array}$

$\begin{array}{cc}\left(4\right)& \hfill {\mathrm{b}}_{\mathrm{t}}^{\mathrm{s}}=\psi \left(\mathrm{j}+{\mathrm{h}}_{\mathrm{t}}\right)+\left(1-\psi \right){\mathrm{f}}_{\mathrm{t}}\\ & \hfill \left[\mathrm{where}\mathrm{j}=ln\left[\frac{1-\mathrm{K}}{\mathrm{K}}\right]\right]\end{array}$

$\begin{array}{cc}\left(5\right)& {\mathrm{m}}_{\mathrm{t}}^{\mathrm{s}}={\mathrm{h}}_{\mathrm{t}}-\mathrm{k}\end{array}$

$\begin{array}{cc}\left(6\right)& {\mathrm{h}}_{\mathrm{t}}=\overline{\mathrm{h}}-\phi {\mathrm{s}}_{\mathrm{t}}\end{array}$

$\begin{array}{cc}\left(7\right)& {\mathrm{r}}_{\mathrm{b},\mathrm{t}}=\left(1+\theta \right)\left[{\mathrm{r}}_{\mathrm{f},\mathrm{t}}+{\mathrm{Es}}_{\mathrm{t}+1}-{\mathrm{s}}_{\mathrm{t}}\right]\end{array}$

$\begin{array}{cc}\left(8\right)& {\mathrm{r}}_{\mathrm{f},\mathrm{t}}={\mathrm{i}}_{\mathrm{t}}^{*}+{\rho }_{\mathrm{t}}+{\mathrm{g}}_{\mathrm{t}}\end{array}$

$\begin{array}{cc}\left(9\mathrm{a}\right)& {\rho }_{\mathrm{t}}={\gamma }_0+{\gamma }_1{\mathrm{r}}_{\mathrm{b},\mathrm{t}}-{\gamma }_2\left({\mathrm{r}}_{\mathrm{m},\mathrm{t}}-{\mathrm{Ep}}_{\mathrm{t}+1}+{\mathrm{p}}_{\mathrm{t}}\right)+{\gamma }_3\left({\mathrm{b}}_{\mathrm{t}}^{\mathrm{s}}-{\mathrm{p}}_{\mathrm{t}}\right)\\ & \hfill \begin{array}{cc}\left(\mathrm{under}\mathrm{credit}\mathrm{rationing}\right)& \end{array}\end{array}$

$\begin{array}{cc}\left(9\mathrm{b}\right)& \begin{array}{cc}{\rho }_{\mathrm{t}}={\gamma }_4+{\gamma }_5{\mathrm{r}}_{\mathrm{b},\mathrm{t}}-{\gamma }_6\left({\mathrm{r}}_{\mathrm{m},\mathrm{t}}-{\mathrm{Ep}}_{\mathrm{t}+1}+{\mathrm{p}}_{\mathrm{t}}\right)& \end{array}\\ & \hfill \left(\mathrm{under}\mathrm{no}-\mathrm{credit}\mathrm{rationing}\right)\end{array}$

$\begin{array}{ccc}\left(10\right)& {\mathrm{r}}_{\mathrm{m},\mathrm{t}}={\mathrm{r}}_{\mathrm{b},\mathrm{t}}-\omega +{\chi }_{\mathrm{t}}& \\ & & \left(\mathrm{under}\mathrm{full}\mathrm{liberalizationg}\right)\end{array}$

where: μ₁, μ₂, μ₃, μ₄, η₁, r₁, r₂, r₃, r₄, r₅, r₆, > 0, 0 ≤ ϕ ≤ 1, and 0 ≤ | φ | ≤ ∞.

and: ^18/

Yt	=	level of output in period t
Pt	=	price level in period t
EPt+1 - Pt	=	expected inflation rate
${\mathrm{p}}_{\mathrm{t}}^{*}$	=	foreign price level in period t
ht	=	monetary base in period t
mt	=	stock of money in period t
${\mathrm{b}}_{\mathrm{t}}^s$	=	supply of domestic credit in period t
ft	=	foreign credit to domestic banks measured in domestic currency
st	=	exchange rate measured as the domestic price of a unit of foreign currency
Est+1 - st	=	expected future change in the exchange rate
rb,t	=	domestic loan rate in period t
rm,t	=	domestic deposit rate in period t
rf,t	=	foreign interest rate charged to domestic banks in period t
${\mathrm{r}}_{\mathrm{t}}^{*}$	=	foreign risk-free rate in period t
ρt	=	risk premium charged by foreign lenders to domestic banks in period t
K	=	required reserve ratio

View Table

Yt	=	level of output in period t
Pt	=	price level in period t
EPt+1 - Pt	=	expected inflation rate
${\mathrm{p}}_{\mathrm{t}}^{*}$	=	foreign price level in period t
ht	=	monetary base in period t
mt	=	stock of money in period t
${\mathrm{b}}_{\mathrm{t}}^s$	=	supply of domestic credit in period t
ft	=	foreign credit to domestic banks measured in domestic currency
st	=	exchange rate measured as the domestic price of a unit of foreign currency
Est+1 - st	=	expected future change in the exchange rate
rb,t	=	domestic loan rate in period t
rm,t	=	domestic deposit rate in period t
rf,t	=	foreign interest rate charged to domestic banks in period t
${\mathrm{r}}_{\mathrm{t}}^{*}$	=	foreign risk-free rate in period t
ρt	=	risk premium charged by foreign lenders to domestic banks in period t
K	=	required reserve ratio

n_t, ε_t, v_t, g_t and χ_t = random shocks.

E is the expectations operator. When placed in front of a variable it represents the expected value of such variable conditioned on information available up to period t. Expectations are fully rational in the sense that the expected exchange rate and price level are consistent with the model’s structural relationships.

Equations (1a) and (1b) represent semi-reduced form supply function for domestic output when firms either are or are not credit rationed. The derivation of these relationships, which reflect both production functions and the conditions for labor market equilibrium is presented in Appendix I. ^19/ When firms are credit rationed, the supply of domestic output (${\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}$) would be a positive function of the stock of real credit (${\mathrm{b}}_{\mathrm{t}}^{\mathrm{s}}-{\mathrm{p}}_{\mathrm{t}}$). This reflects the fact that since firms must make wage payments in advance of their sales of output, credit rationing limits the amount of labor they can employ. In this situation, the firms’ demand for labor would be determined by the amount of real credit available and the real wage rate they must pay. In addition, output would depend positively on real production shocks (n_t) and the expected real return on deposits (r_{m, t} - Ep_t+1 + P_t). Since the households’ supply of labor is assumed to be a positive function of the real wage and the expected real return on deposits, ^20/ a higher expected real return on deposits would increase the supply of labor and thereby the supply of output.

Equation (1b) represents the supply of output when domestic firms are not credit rationed. In this case, firms can achieve their desired levels of borrowing and output at the prevailing set of prices, the wage rate, and interest rates (see Appendix I). Although the real stock of credit would no longer impose an exogenous constraint on their activities, the level of the nominal loan rate would affect the amount of borrowing the firms would undertake to satisfy their need for working capital. ^21/ As a result, the firms’ demand for labor and for working capital loans would be a negative function of the real wage rate and the loan rate. In this situation, the conditions for labor market equilibrium imply that the supply of output would depend negatively on the loan rate. Moreover, just as in the credit rationing case, the supply of output would also depend positively on real production shocks and the expected real return on deposits.

Equation (2) assumes that purchasing power parity holds apart from a white noise shock. Thus, changes in the domestic price level (p_t) are determined by movements in the exchange rate (s_t), the international price of goods (${\mathrm{p}}_{\mathrm{t}}^{*}$) and a random shock (ε_t). ^22/ The demand for money, represented by equation (3) is taken to be a positive function of the level of nominal income (${\mathrm{p}}_{\mathrm{t}}+{\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}$) and the expected real return on money (r_{m, t} - Ep_t+1 + p_t) and is affected by a random shock (v_t). ^23/

To simplify our analysis, we will assume that there is a direct relationship between the stock of base money (h_t), the money supply (m_t), and the stock of domestic credit (${\mathrm{b}}_{\mathrm{t}}^{\mathrm{s}}$). In particular, banks are to be subject to a required reserve ratio of K percent on domestic deposits. ^24/ Since the authorities do not pay interest on required reserves and banks are taken as maximizing expected profits, they do not hold any excess reserves. ^25/ Moreover, since households are assumed to hold all of their money in the form of interest earning deposits rather than currency, the entire monetary base is held as reserves by commercial banks. In this situation, the supply of credit is given by equation (4), ^26/ and the stock of domestic money is given by equation (5) (in log form).

Exchange rate policy is represented by the reaction function in equation (6) which relates the stock of base money to an exogenous component and to movements in the exchange rate. The parameter φ would take on the value zero with a flexible exchange rate and (in an absolute value sense) would be infinite with a fixed exchange rate.

Equations (7)-(10) describe the economy’s interest rate structure. Three alternative structures are examined. For the economy that is closed to external financial transactions and whose financial system is heavily regulated, the domestic loan (r_b,t) and deposit (r_m,t) interest rates would be constrained by ceilings established by the authorities (denoted by a bar over the interest rate). For the partially liberalized and opened economy, it is still assumed that the domestic deposit interest rate is constrained by an interest rate ceiling. However, the domestic loan rate would be linked to the international loan rate that international lenders charge domestic banks. Equation (7) states that deviations of the domestic loan interest rate from interest rate parity are solely accounted for by the remaining capital control (θ) which is assumed to take the form of a proportional tax on borrowing from abroad. As represented by equation (8), the interest rate charged domestic banks by international borrowers (r_f,t) would reflect the risk-free international interest rate (${\mathrm{i}}_{\mathrm{t}}^{*}$), any random component of the risk free rate (g_t), ^27/ and the risk premium (ρ_t) that foreign lenders would attach to reflect the probability that the domestic banks would default on their obligations.

As was argued earlier, the probability that the domestic banks would default on their obligations reflects the probability that the domestic firms would be unable to service their debts. In this situation, it is shown in Appendix II that the determinants of the risk premium that would reflect the probability of default would differ depending on whether or not international lenders credit ration domestic banks. When there is credit rationing, this risk premium would be positively related to both the level of the domestic loan rate and the real stock of credit and negatively related to the expected real return on deposits (equation (9a)). The positive effects of the loan rate and the real stock of credit arise because higher values of these variables increase the scale of the domestic firms’ debt-service obligations and thereby increase the probability that an adverse shock to production would leave firms unable to service those debt obligations. In contrast, a higher expected real deposit rate would, in a general equilibrium context, increase the supply of labor ^28/ and reduce the real wage confronting firms, thereby allowing them better to service their debts. When there is no credit rationing, firms can achieve their desired level of borrowing, which will be a function of the level of the loan rate and the real wage (Appendix I). In this situation, equation (9b) indicates that the risk premium would be a positive function of the loan rate (since the firm’s debt-service obligations would be higher) and a negative function of the expected real return on deposit (since this would increase the supply of labor and reduce the real wage facing the firm).

Finally, in the case of a completely open and liberalized economy, all capital controls are removed, and depositors would be free to substitute between holding domestic deposits and foreign assets. If households are risk neutral, they would adjust their holdings of these assets until the expected risk-adjusted yield on each instrument was identical. Just as foreign lenders, domestic depositors would want to earn a return on domestic deposits that would compensate for the risk that domestic banks would default on their obligations to foreign creditors and domestic depositors. ^29/ To simplify, we will assume that depositors’ evaluation of the banks’ default risk is the same as that of the foreign creditors. As indicated in equation (10), this implies that the domestic deposit rate (r_{m, t}) would equal the nominal loan rate (r_{b, t}) less real intermediation costs (ω). In addition, the deposit rate is assumed to be affected by a random shock (χ_t). ^30/

To close the model, it is assumed that the output and money markets clear and that expectations are formed rationally in the sense that economic agents use all the relevant and available information in forming their forecasts of the future price and exchange rate levels. ^31/

IV. Exchange Rate Policy under Alternative Financial Structures

Our model can be used to consider the type of exchange rate policy (as characterized by the value of φ) that will best stabilize output or the domestic price level under alternative financial structures when the economy is confronted with shocks affecting foreign prices (ε_t), the supply of domestic output (n_t), the domestic demand for money (v_t), the international loan rate (g_t), and the deposit rate (χ_t). ^32/ Output instability is measured in terms of the variance of output around the expected steady state level of output ($\mathrm{E}\left({\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}-\mathrm{E}{\left({\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}\right)}^2\right)$). In this case, the authorities’ policy problem can be represented as:

Since the expected values of all shocks in the model are zero, $\mathrm{E}\left({\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}\right)$ would be the solution for ${\mathrm{y}}_{\mathrm{t}}^{\mathrm{s}}$ when all of the error terms are dropped from the equation system.

Alternatively, the authorities might desire to stabilize the price level. Price instability is calculated similarly as the variance of the price level around its expected steady state level (E(p_t-E(p_t))²). In this case, the authorities policy problem becomes:

While the authorities would presumably be interested in both output and price stability (with various degrees of importance attached to each objective), we will separately examine--as two polar cases--the ϕs that will yield either the greatest output stability or the greatest price stability. This will allow us to identify how exchange rate policies heavily weighted in favor of one or the other of the objectives would differ and possibly conflict.

As indicated in Table 1 (and in Appendices III to VI), the optimal degree of exchange market intervention (as given by the scales and signs of the φs) for either of the two polar objective functions depends on both the nature of the economy’s financial structure and the terms and conditions under which domestic borrowers can access international financial markets. In addition, these factors determine the types of shocks to which the φs are sensitive. However, given the nature of the authorities’ price or output stability objective, the response of the φs to increase in the variances of the various shocks (i.e., the signs of the partial derivatives of the φ with respect to the shock variance) tends to be similar across financial structures.

1. A fully regulated financial system

An economy with a fully regulated financial system is assumed to be closed to external financial transactions and to have domestic interest rates subject to binding ceilings (i.e., ${\mathrm{r}}_{\mathrm{m},\mathrm{t}}={\overline{\mathrm{r}}}_{\mathrm{m}}$ and ${\mathrm{r}}_{\mathrm{b},\mathrm{t}}={\overline{\mathrm{r}}}_{\mathrm{b}}$). In this case, equations (1a) and (2)-(6), as well as the assumption of output and money market equilibrium, characterize the economy’s structure. ^33/ As a result, the optimal value of the exchange rate intervention parameters that would best stabilize either output or the price level, are affected by shocks to foreign prices, real output, and domestic money demand but are unaffected by the behavior of international interest rates. The irrelevance of foreign interest rate shocks in this case reflects the assumption that domestic residents are totally cut off from international financial markets by capital controls.

a. Output stability

As shown in Appendix III, the value of the exchange market intervention parameter (ϕ) that minimizes output instability may be either positive or negative depending on whether the supply of output is more responsive to changes in either the real supply of credit or the real return on time deposits (see equation (1a)). ^34/ The reasons for this ambiguity can be illustrated by considering the economy’s response to an unexpected shock that increases the demand for money. This excess demand for money would lead the exchange rate to appreciate and the price level to fall. As the domestic price level declines, the real supply of credit would rise (thereby stimulating output). However, since the domestic price level would have fallen below its long-term expected level, this would create the expectation of a future rise in prices, and this would lower the expected real return on deposits ^35/ (thereby reducing output). If the supply of real credit effect dominates (μ₁ > μ₂), for example, then a positive money demand shock, which would lower the domestic price level, would also raise output. In contrast, if μ₁ < μ₂, a positive money demand shock would lower both the price level and output. When the authorities want to stabilize domestic output, it is therefore not surprising that their exchange rate policy would have to differ depending on whether a money demand shock or a foreign price shock raised or lowered income. To avoid a long taxonomy of cases in what follows, we will focus on the formulation of exchange rate policy when μ₂ > μ₁.

Table 1.

Optimal Exchange Market Intervention

^1/See Appendix III.

^2/See Appendix IV.

^3/See Appendix V.

^4/See Appendix VI.

Table 1.

Optimal Exchange Market Intervention

			φy	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\epsilon }^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\mathrm{v}}^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\mathrm{n}}^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\mathrm{g}}^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\chi }^2}$
I.	Fully controlled economy 1/		>0 if μ2 > μ1	<0	>0	<0	--	--
II.	Partially liberalized economy that is
	1.	Credit rationed 1/	>0 if μ2 > μ1	<0	>0	<0	--	--
	2.	Not credit rationed 2/	≷0	≷0	>0	<0	<0	--
III.	Fully liberalized economy that is
	1.	Credit rationed 3/	≷0	≷0	≷0	≷0	≷0	≷0
	2.	Not credit rationed 4/	≷0	≶0	≷0	≷0	≷0	≷0
			φp	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\epsilon }^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\mathrm{v}}^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\mathrm{n}}^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\mathrm{g}}^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\chi }^2}$
I.	Fully controlled economy 1/		>0	<0	>0	>0	--	--
II.	Partially liberalized economy that is
	1.	Credit rationed 1/	>0	<0	>0	>0	--	--
	2.	Not credit rationed 2/	≷0	<0	>0	>0	>0	--
III.	Fully liberalized economy that is
	1.	Credit rationed 3/	≷0	<0	>0	>0	>0	>0
	2.	Not credit rationed 4/	≷0	<0	>0	>0	>0	>0

^1/See Appendix III.

^2/See Appendix IV.

^3/See Appendix V.

^4/See Appendix VI.

View Table

Table 1.

Optimal Exchange Market Intervention

			φy	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\epsilon }^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\mathrm{v}}^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\mathrm{n}}^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\mathrm{g}}^2}$	$\frac{\partial {\phi }_{\mathrm{y}}}{\partial {\sigma }_{\chi }^2}$
I.	Fully controlled economy 1/		>0 if μ2 > μ1	<0	>0	<0	--	--
II.	Partially liberalized economy that is
	1.	Credit rationed 1/	>0 if μ2 > μ1	<0	>0	<0	--	--
	2.	Not credit rationed 2/	≷0	≷0	>0	<0	<0	--
III.	Fully liberalized economy that is
	1.	Credit rationed 3/	≷0	≷0	≷0	≷0	≷0	≷0
	2.	Not credit rationed 4/	≷0	≶0	≷0	≷0	≷0	≷0
			φp	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\epsilon }^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\mathrm{v}}^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\mathrm{n}}^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\mathrm{g}}^2}$	$\frac{\partial {\phi }_{\mathrm{p}}}{\partial {\sigma }_{\chi }^2}$
I.	Fully controlled economy 1/		>0	<0	>0	>0	--	--
II.	Partially liberalized economy that is
	1.	Credit rationed 1/	>0	<0	>0	>0	--	--
	2.	Not credit rationed 2/	≷0	<0	>0	>0	>0	--
III.	Fully liberalized economy that is
	1.	Credit rationed 3/	≷0	<0	>0	>0	>0	>0
	2.	Not credit rationed 4/	≷0	<0	>0	>0	>0	>0

^1/See Appendix III.

^2/See Appendix IV.

^3/See Appendix V.

^4/See Appendix VI.

If the authorities are primarily interested in stabilizing output, then φ_y would generally be positive when μ₂ > μ₁ (Table 1). ^36/ In this case, any shock that produces an exchange rate appreciation would then lead the authorities to increase the stock of base money (see equation (6)). It has already been noted that, when μ₂ > μ₁, any decline in the price level (which would be associated with an exchange rate appreciation) would lead to a fall in output. If the authorities therefore increase the stock of base money as the exchange rate appreciates, this would increase the supply of real money and credit thereby stimulating output, which would help offset the initial decline in output.

As indicated in Table 1, any increases in the variances of the underlying shocks would naturally influence the scale of optimal exchange market intervention. For example, an increase in the variance of foreign price shocks (a higher ${\sigma }_{\epsilon }^2$) would lower the optimal φ_y (i.e., a move toward greater exchange rate flexibility). This result reflects the fact that a positive foreign price shock would initially raise the price level and increase both the expected real return on deposits and output. ^37/ Since the higher price level would reduce the real money supply, the resulting excess demand for money would lead to an appreciation of the exchange rate. As the exchange rate appreciates, the initial rise in the price level (due to the positive foreign price shock) would be partially offset. As a result, the expected return on deposits and output would fall. The magnitude of these offsetting effects rises as the degree of exchange rate flexibility increases. Therefore, as the variance of foreign price shocks increases, the authorities would find it optimal to rely on greater exchange rate flexibility to offset these shocks.

In contrast, an increase in the variance of money demand shocks (a higher ${\sigma }_{\mathrm{v}}^2$) would lead to the adoption of less exchange rate flexibility (see Appendix III). This result differs from the traditional conclusion that a high variance of “monetary” shocks raises the desirability of greater flexibility of exchange rates. ^38/ Our result differs from the traditional conclusion because many of the earlier analyses gave the authorities the object of stabilizing consumption and have taken output as exogenous and fixed (apart from a random component). ^39/ In our analysis, however, monetary shocks directly affect output through the assumption that real credit is needed to finance the production process. In particular, output is affected by changes in the real stock of credit and the expected real deposit rate as well as by random output shocks (equations (1a) and (1b)). As a result, the exchange rate (and price level) movements that are needed to restore money market equilibrium in the face of a money demand shock could affect the real return on deposits and thereby output. In fact, the closer that φ_y is to zero (a flexible exchange) the greater would be the effects of a money demand shock on the expected real return on money. In this situation, the authorities could help stabilize output by undertaking greater exchange market intervention (increasing φ_y) as ${\sigma }_{\mathrm{v}}^2$ rose. As a result, more of any positive money demand shock would be satisfied through increases in the money supply (and therefore increases in foreign exchange reserves) rather than declines in output or a fall in the price level.

Alternatively, if the variance of output shocks (${\sigma }_{\mathrm{n}}^2$) increases, then output instability would be minimized by reducing φ_y. A positive output shock would not only raise output but would also lead to an appreciation of the exchange rate (since it would create an excess demand for money). With μ₂ > μ₁, any exchange rate appreciation would lower output and thereby help offset the effects of the positive shock to the supply of output. A decline in φ_y would amplify the offsetting appreciation of the exchange rate and would therefore amplify the offsetting effect on output.

b. Price stability

When the authorities are principally concerned with maintaining price stability, exchange rate policy also involves interventions that would reduce (increase) base money as the exchange rate depreciates (appreciates). φ_p is therefore positive (see Appendix III). Such a strategy would help stabilize the price level by creating money market conditions that would lead to exchange movements that offset any initial appreciation or depreciation. For example, an initial appreciation of the exchange rate would lead the authorities to increase the monetary base. This increase in the monetary base would create an excess money supply that would produce an offsetting depreciation of the exchange rate. In the sense that both φ_y and φ_p are positive, then there is no fundamental conflict between the exchange market policies that will promote output and price level stability in the case of a fully regulated financial system.

As in the case when output stability was the objective, the authorities would find it optimal to reduce the scale of their intervention as the variance of foreign price shocks (${\sigma }_{\epsilon }^2$) increases. This allows exchange rate movements to have greater scope to offset the effect of the foreign price shock on the domestic price level (Table 1). However, as the variances of demand for money shocks (${\sigma }_{\mathrm{v}}^2$) or real output shocks (${\sigma }_{\mathrm{n}}^2$) increase, the authorities’ intervention in the exchange market would become more active (φ_y would rise). Since a positive money demand shock (or a positive supply of output shock) would create an excess demand for money, this would lead the exchange rate to appreciate which would drive down the domestic price level. To stabilize the domestic price level, the authorities would increase base money so as offset the money demand shock (or the effect of a positive shock on that demand for money).

One conflict between an exchange rate policy designed to stabilize output and that designed to stabilize prices thus occurs in the optimal response to an increase in the variance of output shocks. When output stability is the objective, the authorities would reduce φ_y in order to allow exchange rate movements greater scope for offsetting the initial effects of the shock on output. However, larger exchange rate adjustments would conflict with the price stability objective.

V. Conclusions and Unresolved Issues

Our analysis indicates that the removal of capital controls and liberalization of the economy affects exchange rate policy both by broadening the set of shocks affecting the economy and by altering the nature of the impact of a given shock on domestic output and prices. For example, shocks to international interest rates would have an increasingly important effect on the economy (and therefore the formulation of exchange rate policy) as domestic borrowers and depositors can more freely engage in external financial transactions. While the optimal scale of exchange market intervention can vary with the nature of the economy’s financial structure, there is nonetheless a similarity in the response of the optimal degree of intervention to increases in the variance of the various domestic and foreign shocks across all financial structures, as long as the authorities’ basic price or output stability objective remains unchanged. At the most fundamental level, these similarities reflect the influence of the financial constraint facing firms that they must finance their working capital needs prior to the sale of their output. This creates real and financial sector linkages that determine the economy’s response to the various domestic and real shocks. While changes in the economy’s capital controls and domestic financial regulations can amplify (or reduce) the effects of a given shock on prices or output, the presence of the financial constraint ensures that the nature of that impact (i.e., whether it is positive or negative) is stable across financial structures.

In addition, our analysis indicates that the combination of domestic financial liberalization and the removal of capital controls would not in general create a preference for one particular type of exchange rate arrangement. The optimal change in exchange rate arrangements would reflect both the changes in the financial structure of the economy as liberalization takes place and the relative sizes of the various shocks that impinge on the economy.

Finally, our analysis indicates that the extent of the changes in optimal exchange rate arrangements that would occur as financial liberalization takes place would be strongly influenced by the nature of the terms and conditions under which domestic borrowers can access international financial markets. These terms and conditions would reflect both the nature of the country’s capital controls and the decisions of international lenders regarding the creditworthiness of domestic borrowers. As noted earlier, creditworthiness considerations influence both the interest cost of obtaining external credits and the availability of those credits if lenders become sufficiently concerned about the likelihood of default. As a liberalization proceeds, creditworthiness considerations therefore replace official restrictions as the key limitation on market access. This also implies that shifts in market perceptions of the creditworthiness of a country’s borrowers could imply a sharp change in the nature of a country’s optimal exchange rate arrangements, especially if this change in perceptions leads to the imposition of credit rationing.

There are a number of important issues that we have not been able to address in our analysis because of certain simplifying assumptions. Some of these issues have considerable policy relevance. For example, although we have assumed that the relationship between the stock of base money and the money supply would not be affected by the removal of capital controls or the domestic financial liberalization, this has clearly not been the case in most industrial countries. The money multiplier and the velocity of money have often been influenced by the removal of interest rate ceilings and the availability of new financial instruments. In addition, we have not allowed for the feedback effects between countries at different stages of domestic financial liberalization and employing different degrees of capital controls. Such feedback effects could be particularly important for countries on the periphery of the European Community, especially as 1992 approaches. Our analysis has also abstracted from the role of securities and equity markets. While incorporating the influence of these markets is clearly important, the requirement that firms fund their working capital needs (from whatever financial source) prior to production would ensure the existence of strong real and financial sector linkages no matter how complex the financial structure. Finally, we have ignored the effect of different types of fiscal systems on capital flows and the location of financial and real activity. Examining the effects of these factors on the relationship between financial structure and exchange rate policy leaves considerable scope for future work.