Twin Banking and Currency Crises and Monetary Policy

The final publication is available at Springer via http://dx.doi.org/10.1007/s11079-016-9391-2

Twin Banking and Currency Crises and Monetary Policy1 Ryota Nakatani, Ph.D.a a

Bank of Japan, 2-1-1 Nihonbashi-Hongokucho, Chuo-ku, Tokyo 103-8660, JAPAN E-mail: [email protected] Phone: +81-3-3279-1111 Fax: +81-3-5255-6758

Abstract Currency crises are found to be strongly associated with banking crises. This paper constructs a twin banking and currency crisis model by introducing the banking sector into the currency crisis model and examining the case in which the exchange rate risk is located in the banking system. The model shows that an unanticipated shock caused by the shift of investors’ expectations and/or a negative productivity shock can trigger a twin banking and currency crisis. To achieve both financial stability and economic stability, the central bank uses multiple monetary policy instruments. In contrast to the conventional policy recommendation in response to a currency crisis, i.e., interest rate hike, we find that when the exchange rate risk is located in the banking sector, the monetary policy option to prevent a twin crisis is to lower the policy interest rate and the reserve requirement ratio and raise the interest rate on reserves. Our results show that the location of the exchange rate risk matters for the choice of an appropriate monetary policy response during a crisis. Keywords: Currency Crisis; Banking Crisis; Twin Crisis; Monetary Policy; Reserve Requirement Ratio; Interest Rate on Reserves JEL Classification: E4; E5; F3; F41; G15; G21 1

This is the extended version of the author’s doctoral dissertation as submitted to and accepted by the University of California. I am grateful for helpful discussions and comments from the two anonymous referees, the editor George S. Tavlas, Michael M. Hutchison, Carl E. Walsh, Kenneth M. Kletzer, Michael P. Dooley, Johanna L. Francis and seminar participants at the Macro/International Finance workshop at the University of California, Santa Cruz. I especially appreciate greatly the support that I received from the following individuals: my advisor, Michael M. Hutchison, for guiding this research, offering sound advice and encouraging me; Carl E. Walsh for careful reading of this draft and providing detailed comments during development of this theoretical model to analyze the effects of monetary policy; and Kenneth M. Kletzer for commenting on and proofreading this research as well as advice for publication. The views expressed in this article are those of the author and do not reflect those of the Bank of Japan. 1


1 Introduction Currency crises are found to be strongly associated with banking crises, and a banking crisis tends to precede a currency crisis (Kaminsky and Reinhart 1999; Glick and Hutchison 2001). The most common way to model a twin banking and currency crisis in the literature is the use of the Diamond-Dybvig model.2 This type of model assumes two types of depositors (i.e., consumers): an impatient depositor (i.e., an early consumer) and a patient depositor (i.e., a late consumer), and it describes how bank runs occur (Goldfajn and Valdés 1997; Chang and Velasco 2001; Bleaney et al. 2008). Basically, the model relies on the shock caused by heterogeneity among depositors. However, in practice, problems in the banking sector may be triggered by other factors such as an expectational shift of international investors in financial markets rather than consumers’ withdrawal shock. Therefore, we develop a twin crisis model without relying on the heterogeneity of consumers. We extend the model developed by Aghion et al. (2000, 2001) (the ABB model) and show that an unanticipated risk premium shock caused by the shift of investors’ expectations in financial markets and/or a negative productivity shock in the real sector can trigger a twin banking and currency crisis. The ABB model analyzed the effects of credit constraints on currency crises by focusing on private foreign-currency debt. The authors explored how problems in the financial markets interact with currency crises, and how crises can have real effects on the economy. In their model, the presence of foreign currency debts reduces the profits and retained earnings of firms and results in lower investment and lower output in the real sector under a large currency depreciation. In their series of papers (Aghion et al. 2000, 2001, 2004), only firms are 2

See Glick and Hutchison (2013) for a comprehensive survey of the literature on currency crises and banking crises. 2


assumed to borrow from abroad and bear the risk of exchange rate. However, it is found that in some countries banks borrow from abroad and have a larger share in foreign currency debt than firms (Buch and Heinrich 1999). Radelet and Sachs (1998) pointed out that at the onset of the Asian crisis the effect of exchange rate movements on the banks’ foreign liability positions quickly eroded the capital base of banks and this in turn led to a severe credit crunch. Aghion et al. (2004) suggested that “An interesting extension of the analysis is to incorporate explicitly currency exposure at the banking level.” Therefore, we analyze the case in which commercial banks take the exchange rate risk. This is a very important extension of the model because both the mechanism of crisis and the policy implications differ from those of the literature. When banks bear the exchange rate risk, not only a currency crisis but also a banking crisis occurs when the country is hit by a negative productivity shock and/or risk premium shock. Therefore, to achieve two objectives, i.e., financial stability and economic stability, the monetary authority requires multiple policy instruments. Analyzing the effects of several monetary policy tools, we show that not only the interest rate policy but also the reserve requirement policy (i.e, the reserve requirement ratio and the interest rate on reserves) is important to prevent a twin banking and currency crisis. In addition, we show that the monetary policy response for the prevention of a crisis differs from the original ABB model when the exchange rate risk is located in the banking system. Namely, standard monetary tightening by raising the policy interest rate is no longer appropriate when commercial banks have foreign currency debts and take the exchange rate risk. Thus, our analysis demonstrates that the location of the exchange rate risk matters for choosing an appropriate

3


monetary policy response to prevent a crisis. This study contributes to the literature on twin crisis3 and monetary policy (especially the reserve requirement policy) discussed in the next section by enhancing our understanding of the role of location of the exchange rate risk on a crisis mechanism and providing insightful policy implications from both academic and macroprudential policy makers’ points of views.

2 Motivation and Related Literature 2.1 Currency and Banking Crisis This paper is based on a specific currency crisis model (i.e., the ABB model). Currency crisis models can be classified into three generations (Glick and Hutchison 2013): first generation models that focus on inconsistencies between domestic macroeconomic policies (Krugman 1979; Flood and Garber 1984); second generation models that analyze an interaction between investors’ expectations and the central bank’s actual policy outcomes (Obstfeld 1996); and third generation models that emphasize the roles of various financial problems. Examples of such financial problems include: a balance sheet problem caused by foreign currency denominated debts under credit constraints (Aghion et al. 2001); the Diamond-Dybvig type bank run caused by a stochastic patience for consumption (Chang and Velasco 2001); a liquidity problem due to an interaction of domestic and international collateral constraints in the face of a production shock (Caballero and Krishnamurthy 2001); and moral hazard problems caused by the government guarantee (McKinnon and Pill 1996; Corsetti et al. 1999; Dooley 2000; Burnside et al. 2001; Dekle and 3

Note that there are different types of twin (or triple) crisis in the literature. For example, Miller and Vallée (2011) analyzed a crisis in foreign asset markets and a domestic banking crisis; Miller (2014) focused on a reserve currency debt crisis, a banking crisis and a currency crisis. 4


Kletzer 2002, etc.). In this paper, we use the third generation model developed by Aghion et al. (2000, 2001) for several reasons. First, as shown in their original paper, the ABB model can include the features of the first and the second generation models. Second, we can also include the possibility of multiple equilibria. Third, we can have short-run nominal rigidity, which is supported by empirical evidence, and see how financial frictions can cause currency crises. Fourth, the ABB model can graphically demonstrate the occurrence of a crisis with only two types of curves, which describe the financial markets and the real sector, and it can analyze the effects of both financial shocks and real shocks intuitively on the figure. Fifth, the ABB model is still the most useful model to analyze emerging economies in which crises often occur. During the global financial crisis in 2008-09, central bankers were concerned with the possibility of currency crises in some countries. Those concerns were large especially for emerging countries that had huge foreign debt in their economies. This ABB model illustrates the situation of those countries accurately. The following papers are related to our analysis. Aghion et al. (2004) incorporated the banking sector into the ABB model and derived implications for monetary policy during a currency crisis. Our research differs from their research in some respects. First, they only studied the case in which firms have foreign currency debts and not the case in which banks have foreign currency debts. However, as the economy develops and is globalized, not only firms but also banks can be exposed to international capital markets. Therefore, we analyze the model in which banks have foreign currency debt, and study how this differs from the model in which firms have foreign currency debt. Second, they assumed that there is no transaction cost for banks to receive deposits. By contrast, as we will explain below, we incorporate the cost structure of the banking

5


system into the model. Finally, they also assumed that reserves do not bear interest. But taking into account the increasing attention of the role of interest on reserves as a monetary policy instrument, we analyze the general case where central banks can use that interest rate as one of their monetary policy tools. Another type of model that investigates how shocks to the banking system affect the real economy uses a costly banking system. Edwards and Végh (1997) used economies of scope between loans and deposits and showed that high rates of devaluation led to lower credit and lower output. They also analyzed the role of reserve requirements as a countercyclical policy tool. Our analysis is consistent in the sense that the cost structure of the bank is a key factor in magnifying macroeconomic disturbances, but differs in some respects. In their model, households played an important role in the foreign exchange rate market because households hold internationally-traded bonds. By contrast, in our model, banks play an important role in taking exchange rate risks. In addition, they assumed zero interest rate on reserves, whereas we analyze a general case in which reserves earn a positive interest rate.

2.2 Reserve Requirement Policy In this paper, we analyze the monetary policy mix for the prevention of a twin banking and currency crisis when the monetary authority has several policy instruments. This is because when banks take the exchange rate risk, a banking crisis occurs concurrent with a currency crisis and hence a macroprudential policy, such as the reserve requirement policy, is also effective during the crisis. Specifically, we analyze three monetary policy instruments: the policy interest rate, the reserve requirement ratio and the interest rate on reserves. The reserve requirement policy is an

6


important policy that is quite often used by central banks especially in emerging economies. The literature on currency crises suggests that the monetary authority should increase the interest rate to prevent crises (Aghion et al. 2001; Flood and Jeanne 2005; Lahiri and Végh 2007).4 There are some publications that test the policy implications (Goldfajn and Gupta 2003; Goderis and Ioannidou 2008; Eijffinger and Goderis 2008; Eijffinger and Karataş 2012; Nakatani 2014). Most empirical literature has found evidence that higher interest rates led to stronger exchange rates. But the interest rate is not the only monetary policy tool for central banks. A lot of central banks are using more than one policy instrument to control their economies. Most central banks – over 90 % – oblige banks to hold required reserves against their liabilities (Gray 2011). The main policy instruments for central banks in emerging countries are the reserve requirement ratio and the policy interest rate. After the Lehman shock, the reserve requirement policy was actively used by emerging economies (Tovar et al. 2012). Recently, Cordella et al. (2014) found that more than half of developing countries have used the reserve requirement policy as a macroeconomic tool and most of them used it countercyclically.5 Nowadays, central banks in emerging countries change the reserve requirement ratio more often than the policy interest rate. For example, the People’s Bank of China (PBOC) altered the reserve requirement ratio 35 times between July 2006 and June 2011, whereas PBOC altered the one year lending interest rate 16 times in the same period. In addition, a central bank can use the interest rate and the reserve requirement ratio in 4

Braggion et al. (2009) analyzed an optimal monetary policy in a sudden stop, focusing on the evidence that in the wake of the 1997-98 financial crises, interest rates in Asia were raised immediately, and then reduced sharply. As they noted, a shortcoming of their model was the absence of investment. 5 Simulating Turkish data, Mimir et al. (2013) recently found that a countercyclical reserve requirement policy countervails the negative effects of the financial accelerator mechanism. 7


opposite ways. For example, the Central Bank of the Republic of Turkey raised the reserve requirement ratio aggressively and lowered the policy interest rate at the same time in the period from November 2010 to August 2011. Why is the reserve requirement so important for monetary policy these days? This is due to the fact that central banks usually have two policy objectives: price stability and financial stability. When there are multiple policy objectives, we cannot attain the goals with one monetary policy instrument, interest rate. In other words, multiple policy goals usually call for multiple instruments (Reinhart and Rogoff 2013). Thus, many central banks use the reserve requirement ratio for financial stability and the interest rate for price stability (or exchange rate stability especially in a currency crisis). Glocker and Towbin (2012) found that reserve requirements become more effective when there is foreign currency debt because interest rate policy is less effective in attaining both financial stability and price stability. Vargas and Cardozo (2012) showed that the use of reserve requirements is justified when monetary policy has several transmission channels (such as interest rate and exchange rate channels) and the central bank objective function includes financial stability. What is the theoretical background for the use of the reserve requirement policy?6 One conclusion we obtain from the literature is that raising reserve requirements can prevent financial imbalances by restraining credit growth in the upswing of the business cycle (Calvo et al. 1993), whereas lowering reserve requirements during a downturn can provide liquidity to the banking

6

The empirical literature found mixed results about the effectiveness of the reserve requirement policy; some papers suggest that the reserve requirement policy has significant effects on economies (Loungani and Rush 1995; Gelos 2009; Ostry et al. 2012; Claessens et al. 2013; Mora 2014), while others show no or small effect (Valdés-Prieto and Soto 1998; De Gregorio et al. 2000; Concha et al. 2011). 8


system.7 An increase in reserve requirements raises lending interest rates, reduces deposit interest rates, and lowers bank stock prices because reserve requirements can be considered as taxes on bank profits (Reinhart and Reinhart 1999; Walsh 2012). In general, the reserve requirement functions as a tax on deposits (Black 1975; Fama 1980; Freeman 1987).8 Thus, the interest rate on deposits decreases when the reserve requirement ratio is raised. Reinhart and Reinhart (1999) developed the Dornbusch overshooting model to analyze the effects of the reserve requirement policy on capital flows. If a reserve tax is borne by depositors, an increase in reserve requirements lowers a deposit interest rate, which in turn lowers the real interest rate and results in depreciation of the real exchange rate.

3 Model This is a two period model. There are four agents in the economy: households, firms, commercial banks and the authority (the government and the central bank). Households, which own the firms and the banks, consume the goods, and deposit their money in the banks. Firms produce the only goods by borrowing under a credit constraint. Banks finance their lending operations with deposits and foreign currency debts. On the asset side, banks lend to both firms and the government, and hold cash as required reserves. The government collects tax from households and issues bonds to finance its spending. The central bank sets three policy variables:

7

Hoffmann and Löfﬂer (2014) find that central banks in emerging countries raise reserve requirements when interest rates in international funding markets decline or financial inflows accelerate to preserve financial stability, whereas they lower reserve requirements when funding from the advanced economies dries up to provide liquidity in the banking system. 8 Goodfriend (2002) and Walter and Courtois (2009) argue that the interest on reserves would eliminate distortions in financial markets due to the tax on reserves. 9


the policy interest rate, the reserve requirement ratio and the interest rate on reserves. The timing of events can be summarized as follows. The economy is initially in equilibrium before period 1 starts; all initial variables satisfy the equilibrium conditions and the Purchasing Power Parity (PPP) holds (see section 3.6). At the beginning of the first period, the price is preset before the actual exchange rate is known. Given this price, households, firms, commercial banks and the central bank choose their actions.9 Then, an unanticipated shock, which corresponds to a realization of the nominal exchange rate, occurs. This shock makes the price deviate from the PPP by the end of the first period. Subsequently, period 1’s output and profits are generated and the firms’ and banks’ debts are repaid. At this point, since the real exchange rate is different from the level that banks assumed at the beginning of the first period, banks may default depending on the sign of the unanticipated shock. Namely, if there is a positive unanticipated shock that leads to an appreciation of the currency, competitive banks can earn positive profits because the amount of repayments for foreign currency loans is lower than expected. Finally, a fraction of the net retained earnings of the firms after debt repayment is saved for investment in period 2 and this determines the level of output in period 2. By contrast, if there is a negative shock that induces a depreciation of the currency, banks would earn negative profits that make it difficult to repay the foreign currency debt and deposits. This negative unanticipated shock forces banks to default. This is what we call a banking crisis. If the default happens, firms cannot borrow money from banks and thus invest only with their retained earnings. In the absence of financial intermediation, the country’s output will shrink in the next period through the credit channel. This is what we call a twin crisis. 9

In the original ABB model, the central bank is assumed to act after the shock. We changed this assumption to understand how the monetary policy affects each economic agent’s behavior in general equilibrium. 10


3.1 Households The assumption about price setting is the same as in the ABB model. PPP is assumed to hold at the beginning of period 1. Following an unanticipated shock (that may be a shift of investors’ expectations), there are deviations from PPP that are corrected in period 2, i.e., P1  E1 and

P2  E2 where Pt is the price level in period t and Et is the nominal exchange rate (the price of foreign currency in terms of domestic currency) in period t . The price level of foreign countries is normalized to 1. At the beginning of the period 1, the representative household anticipates the firm’s and the bank’s real profits under the assumption that PPP holds. The expected real profit of the the bank, 1Be , is different from the actual real profit, 1B , which is determined at the end of period 1, because the shock makes the nominal exchange rate deviate from the level that PPP holds. By contrast, the expected profit of the firm equals the realized profit, 1F , since the unanticipated shock does not affect the firm’s profit in period 1. Thus, the household can anticipate the true price level but not the exchange rate. The household’s utility function is defined as uc1 , c2   ln c1   ln c2 where ct is consumption in period t and  is the discount factor. The household begins the first period with initial savings, s0 , the initial dividend from the firm,  0F , and the initial profit of the bank,  0B . The household’s real budget constraint in period 1 is





c1  s1  1  i0S s0 P0 P1  1    0F   0B

(1)

where s1 is savings in period 1, i0S is the interest rate on savings that is paid to the household at the end of the period 0 so that the household can use this interest income in period 1 and  is the tax rate on the dividend. The budget constraint thus says that, in the first period, the household’s

11


income consists of the real return on savings and profits from firms and banks. The household’s expenditures consist of consumption, savings, and the tax. The household’s optimization problem consists in choosing ct , st  for t  1,2 in order to maximize utility, subject to the budget constraint, given tax rate,  , initial variables, s0 , i0S , P0 ,  0F and  0B , price variables P1 and P2 , the saving interest rate i1S , and the expected

profits of the firm and the bank  1F and 1Be . Then we get the first order conditions of the household’s optimization problem that lead to the following Euler equation:

 1  i1S  1   2   c2 c1

(2)

where  2  P2  P1  P1 is the inflation rate in period 2. This equation suggests that consumption is a decreasing function of the interest rate on savings and an increasing function of the future inflation rate. The household also needs to satisfy the budget constraint. Using these conditions, we can derive the following saving function in period 1:



s1 s0 , i0S ,  0F ,  0B , 1 , i1S , 1F , 1Be , 2

1  1 







 1  i0S  1   2  1   1F  1Be  F B s0  1    0   0   . 1  i1S 1   1 

(3)

3.2 Government and the Central Bank The government and the central bank are treated as one agent called “the authority.” The policy objective of the authority is to prevent a currency crisis and a banking crisis. To avoid a twin crisis, the central bank can use three monetary policy instruments: the policy interest rate ( i1 ), the reserve requirement ratio (1 ) and the interest rate on reserves ( i1R ) during the crisis period (i.e., period 1). The authority conducts open market operations using government bonds ( b1S ) to control 12


the policy interest rate in period 1. Commercial banks are the sole market participants in the bond market. The consolidated budget identity of the authority during the crisis period can be written as

i1R1d1  i1b1S  g1   0F

(4)

where g t denotes the government expenditure in period t . The authority is imposing an implicit tax on the interest rate on household savings by applying the positive reserve requirement ratio. This means that the household savings are taxed and distorted by the reserve tax. To avoid double taxation on savings, the government collects tax from the other source of household income, i.e., the dividend from the firm. Note that the amount of this tax base is determined in the previous period. The government collects this tax revenue from the household to finance interest payments on reserves and government bonds and the government expenditure. The government does not conduct any tax reform over the entire period; the tax rate is exogenously determined by the government before households choose their actions and the tax rate is constant over time. This is because we want to focus on and analyze the role of the monetary policy response for a banking and currency crisis rather than a tax policy response. In addition, it is not practical to assume that the government conducts tax reform when the country is combating a crisis. Thus, changes in monetary policy variables should be accompanied by an adjustment in government expenditure. The government expenditure is treated this way to satisfy the authority’s budget identity since the change of g1 (or the government budget deficit) is not a main focus in this third generation model.10 The central bank announces various interest rates when other agents choose their actions in period 1. Note that both reserves and government bonds are redeemed within the period. After

10

The government budget deficit plays a crucial role in the first generation models of currency crises. 13


the crisis period (i.e., in period 2), as we will see below, the ABB model describes the monetary policy using the money supply variable ( M 2S ) to apply the Interest-Parity-LM (IPLM) relationship. This is because the main focus of this paper is the monetary policy during the currency crisis, not the policy after the crisis.

3.3 Banks The role of banks is to take deposits from domestic households, borrow from foreign investors and lend to both firms and the government. The banking industry is assumed to be competitive. The representative bank raises its funds from the deposits of domestic households and external debt denominated in foreign currency. The inability of many countries to borrow abroad in domestic currency terms is referred to as the “Original Sin” (Eichengreen et al. 2007). Seeking the reason for Original Sin is beyond the scope of this paper. The essence of the ABB framework is that economic agents are not able to anticipate shocks and hence they do not hedge the risk. The bank gives loans to domestic firms and purchases government bonds each period. The bank must put a certain fraction of the deposits in a central bank’s reserve account to satisfy the reserve requirement obligation. The profit of the bank in period 1 is given by



1B  i1L l1S  i1 b1D  i1R1d1  i1D d1  i1F F1 E1 P1   d1 , l1S , F1



(5)

where itL is the interest rate on domestic lending, l tS is an amount of domestic lending, it is the policy interest rate that is applied to the interbank market,

b1D  1  1 d1  l1S  F1 E1 P1 .

(6)

is the net position of the bank on the interbank market where the bank trades the government

14


bonds with the central bank through open market operations,  t is the reserve requirement ratio,

d t is an amount of deposits, Ft is an amount of loan denominated in foreign currency, itR is the interest rate on reserves, itD is the interest rate on deposits, itF is the interest rate on foreign



currency loans, and  d t , ltS , Ft



is a cost function that represents banking technology. Loans

and bonds are within period loans and within period bonds. We assume that the bank incurs costs for maintaining and managing its deposits, loans and foreign debts11 whereas holding the reserve account balance at the central bank is charge-free. The bank’s cost function is assumed to be twice differentiable and satisfies the following standard assumption of convexity:

d  0 , dd  0 , l  0 , ll  0 , F  0 and FF  0 where d   d1 , dd   2  d12 , l   l1S , ll   2  l1S 2 , F   F1 and FF   2  F12 . We assume that the commercial bank holds reserves only to satisfy the reserve requirement obligation for the following reason. In practice, central banks set the interest rate on excess reserves at the level that is equal to or lower than the interest rate on required reserves such as zero. Thus, the interest rate on excess reserves is the lower bound on various interest rates and there is no incentive for commercial banks to hold excess reserves at an optimum.





The banks choose l1S , d1 , F1 to maximize their profits, given i1L , i1 ,1 , i1R , i1D , i1F , E1 P1 ,





and  d1 , l1S , F1 . Since the banks do not anticipate the shock, substituting the net position of the bank in the interbank market, they maximize the following expected profits under the assumption that PPP holds: 11

Such a formulation captures the idea that, in practice, banks must carry out a variety of costly activities such as evaluating creditors, managing deposits, maintaining ATM’s, and so on. 15




 

 





1Be  i1L  i1 l1S  i1 1  1   i1R1  i1D d1  i1  i1F F1   d1 , l1S , F1



(7)

The first order conditions of the banks’ profit maximization problem are thus: i1L  i1  l ,

(8)

i1  i1F  F ,

(9)

i1 1  1   i1R1  i1D  d .

(10)

The bank will adjust its volume of loans, foreign borrowing and deposits in such a way that the corresponding intermediation margins, i1L  i1 , i1  i1F and i1 1  1   i1R1  i1D , equal its marginal management costs. To analyze the effect of monetary policy further, we need to specify the cost function. Hereinafter, we analyze the case of constant marginal costs of intermediation ( l   L , F   F , d   D ) since a simple tractable characterization of equilibrium is obtained. The first order conditions of the bank’s profit maximization problem yield the following equations that characterize the relationship between interest rates.

i1L  i1   L ,

(11)

i1F  i1   F ,

(12)

i1S  i1D  i1 1  1   i1R1   D .

(13)

We need to define a banking crisis as follows. A banking crisis occurs when commercial banks earn negative profits and cannot meet debt obligations to international investors and/or domestic depositors. When this happens, commercial banks have no choice but to default.

3.4 Firms and the Wealth curve The representative firm uses capital input k1 to produce one type of goods y1 that can be

16


sold to domestic households in period 1. We assume the production function yt  f  At , k t   At k t where At is the total factor productivity in period t . Assuming that the

working capital k1 fully depreciates within one period, the firm maximizes its real profit net of loan repayments





1  y1  1 i1L l1D

(14)

where l1D is an amount of domestic currency loans in period 1. Apart from the original ABB model, all loans are within period loans; the firm borrows at the beginning of each period and repays loans at the end of the period. As in the ABB model, firms are facing the credit constraint and an amount of capital is determined as follows. We assume that whenever profit is positive, the firm retains a proportion

1   

of profit and uses it to finance its future investment.12 Thus, the retained earnings in

period 1 is defined as 1   1 . The remaining proportion  of profit is distributed to domestic households; the households receive a fraction of the firm’s profit that equals

1F  1

(15)

where 1F is a realized profit. In the presence of the credit constraint, the firm can at most borrow an amount proportional to the retained earnings from the commercial bank:

l1D   1   1 . We assume that this constraint is binding. We focus on the case in which the credit multiplier is constant over time and firms do not have a choice to default.13 As in the case of households, the firm assumes that PPP holds. Since capital fully depreciates within one period, investment in the 12

The firm will always save a constant fraction of the profits under the assumption of the logarithmic preference (Aghion et al. 2004). 13 See Aghion et al. (2004) for the case in which firms have a choice to default. 17


current period equals the capital in the next period. Therefore, under the credit constraint, the equation of motion for capital, which equals investments, can be written as follows:

k 2  1   1   1 for E1  E1*

(16a)

where E1* is the level of the nominal exchange rate that satisfies PPP in period 1 ( E1*  P1 ). As long as an unanticipated shock leads to an appreciation of the currency, competitive banks can earn positive profits and engage in financial intermediation because the amount of repayments for foreign currency loans is lower than expected. Note that an amount of the working capital in period 1 is determined in the same way:

k1  1   1    0

(17)

where 1    0 is the initial retained earnings that the firm holds at the beginning of period 1. In the credit-constrained economy, the output of the supply-side in period 2 is characterized by the following production function:

y 2S  f A2 , 1   1   1  for E1  E1*

(18a)

Thus, y2S is predetermined at time 2. This equation characterizes the relationship between output in period 2 and the nominal exchange rate in period 1; this is called the Wealth curve. For the same reason, the output in the first period can be written as y1S  f A1 , 1   1    0  .

(19)

By contrast, when an unanticipated shock that induces a real currency depreciation occurs, commercial banks should default due to the negative profit and firms cannot borrow from banks. This means that there is a jump in the Wealth curve because the credit multiplier suddenly becomes zero.

18


k2  1   1 for E1  E1*

(16b)

y 2S  f A2 , 1   1  for E1  E1*

(18b)

Thus, firms can invest only with their retained earnings, and output in period 2 should always shrink. These two situations can be drawn as in Figure 1. Figure 1: Wealth Curve of Banking and Currency Crisis Model

E1

Wealth Curve

E1*

A2 1   1

y2

A2 1   1   1

Using the equation for the profit of the firm, the credit constraint, and the output function, we can write the loan demand function of the firm in period 1 as follows:











l1D i1L , A1 ,  0   1    f A1 , 1   1    0  1   1    1  i1L .

(20)

In contrast, the output of the demand-side in period t  1,2 is determined as

    c  I  c  1   1   y  1  i l  for E  E y  c  I  c  1   y  1  i l  for E  E . y1D  c1  I1  c1  1   1    y1D  1  i1L l1 ,

y 2D

2

D 2

2

2

D 2

2

2

D 2

2

L 2

L 2

2

2

* 1

1

1

* 1

(21) ,

(22a) (22b)

where I t denotes an investment in period t . Solving the equations above, we finally obtain the output of the demand-side:

19




   c  1   1   1  i l  1  1   1    for E  E y  c  1   1  i l  1  1    for E  E . y1D  c1  1   1    1  i1L l1 1  1   1   ,

y 2D

L 2

2

D 2

L 2

2

2

1

2

1

* 1

* 1

(23) ,

(24a) (24b)

3.5 IPLM Curve The interest rate on foreign currency debt and the deposit interest rate satisfy the following uncovered interest parity condition since banks are indifferent between borrowing from abroad and taking deposits from domestic households.





1  i1D  1  i1F E2e E1

(25)

where E2e is the expected nominal exchange rate at the beginning of period 2. Evidently, if i1D is increased, but E2e and i1F are unchanged, then E1 must fall (appreciate): the orthodox relationship. A money market equilibrium can be expressed by an LM equation:



M 2S  P2 m D y2 , i2 , 2 , i2R



(26)

where a real money demand ( m D ) is a standard function of output and the monetary policy instruments. Using the PPP assumption P2  E2e  E2 , the money market equilibrium in period 2 becomes





M 2S  E2e m D y2 , i2 ,  2 , i2R .

(27)

Combining the IP condition (25) with the LM equation (27), the IPLM curve in this twin crisis model can be written as E1 

1  i1F M 2S . 1  i1D m D y2 , i2 , 2 , i2R



20



(28)


We can see the negative relationship between E1 and y 2 , suggesting a negative slope of the IPLM curve. Concordant with Aghion et al. (2001, 2004), we define a currency crisis as follows. We refer to the equilibrium with low output and a depreciated domestic currency (i.e., a high nominal exchange rate) as the ‘currency crisis’ equilibrium.

3.6 Defining the Equilibrium From here we define the equilibrium when i F is exogenous, which is a standard assumption in a small open economy model, because we will derive the policy implication in this setting later. Recalling that the economy is initially in equilibrium where PPP holds; the initial variables

 ,  0

F 0

,  0B , y0 , i0L , i0S , i0 , 0 , i0R , s0 , l0 , b0 , F0 , g0 , A0



satisfy

the

equilibrium conditions:





 0  y0  1  i0L l0 ,  0F   0 ,





 0B  i0Ll0  i0 b0  i0R 0 d 0  i0S s0  i F F0   s0 , l0 , F0  0 , y0  f A0 , 1   1    0 

b0  1  0 s0  l0  F0 , i0L  i0   L ,

i0  i F   F , i0 1  0   i0R 0  i0S   D ,





     1    f A , 1   1     1   1   1  i , s0  1    0F 1  i0S   1    1  i0S   i0S ,

l0

0

0

21

L 0

following


i0R0 d0  i0 b0  g0   0F , where the subscript 0 denotes that the variable is in the initial equilibrium. Note that four initial variables i0 ,  0 , i0R , A0  are exogenous variables. Departing from the initial equilibrium, a new equilibrium at the end of period 1 is a set of



allocations c1 , c2 , k1 , k 2 , y1 , y2 , ~ s1 , l1 , b1 , F1 , g1 , 1 , 1F , 1Be , 1B exchange

 ,  0

F 0

rates

i , i L 1

S 1

, E1



and a set of interest and



,

given



,  0B , i0S , s0 , P0 , P1 , P2 , A1 , A2 , i1 , 1 , i1R , i F , M 2S , E2e , ,  ,  ,  ,  such that:

1. c1 , c2 , s1 solve the household’s problem and satisfy equations (1), (2) and (3).



2. l1D solves the firm’s problem of maximizing its profit 1 , which is determined by equation (14) and a proportion of which is distributed to the household as a dividend  1F by equation (15), subject to the credit constraint and its production technology (19), which determines y1 with k1 defined by (17), and satisfies the equation (20). Since the firm’s problem determines the amount of capital in the next period k 2 by condition (16ab), it also determines the level of the output in the next period y2 by condition (18ab).





3. d1 , l1S , F1 solve the commercial bank’s problem of maximizing the expected profit 1Be , which is determined by equation (7) (and results in 1B ex post from equation [5]), and satisfy the balance sheet condition (6), which determines the commercial bank’s demand of bonds b1D , and the first order conditions (8), (9) and (10).





4. b1S , g1 solve the authority’s problem and satisfy equation (4). 5. E1 satisfies the IPLM condition (28). D S D S * 6. Markets clear: s1  d1  ~ s1 , b1  b1  b1 , and l1  l1  l1  0 E1  E1 and l1  0

22


E1  E1* . We can solve this model because we have 21 endogenous variables (if we count s1 and d1 as different variables and count l1 and b1 in the same manner) and 21 equilibrium conditions (if we count [16ab] as one condition and count [18ab] in the same manner).

3.7 Twin Crisis and Monetary Policy The initial equilibrium, which is defined by the intersection of the IPLM and Wealth curves, is graphically shown in Figure 2. Figure 2: Equilibrium before the Shock

E1

Wealth Curve

IPLM Curve

y2 Based on the ABB model, Nakatani (2014) empirically found that both productivity shocks in the real sector and shocks in the financial markets occur during currency crises. Thus, we consider the case in which both a negative productivity shock and a financial shock occur simultaneously in period 1 (the results do not change when we only consider one shock). A financial shock can take the form of three types of shocks to risk premium: (1) an increase in the perceived exchange rate risk premium (same as the original ABB model), (2) an increase in the risk premium on domestic assets that changes the foreign currency cost of the debt of banks to foreign creditors due to the 23


interest parity condition (Bergman and Jellingsø 2010), or (3) an increase in the international interest rate that may be caused by the foreign monetary policy. Taking into account the fact that the deposit interest rate is a function of three monetary policy variables, the new IPLM curve in each case can be written as

E1 

M 2S 1 iF  E D R D R 1  i1 i1 , 1 , i1 m y 2 , i2 ,  2 , i2



   M 1  i 1     1  i i ,  , i  m  y , i ,  , i  F

E1

D 1

S 2

D

1

1

R 1

D

2

2

2

R 2

M 2S 1 i F  F E1  1  i1D i1 , 1 , i1R m D y 2 , i2 ,  2 , i2R



 



(29a)

(29b)

(29c)

where  E is the foreign exchange risk premium after the shock,  D is the risk premium on domestic assets after the shock, and  F is the risk premium on international interest rate after the shock. In all cases, the increase in the risk premium shifts the IPLM curve upwards as depicted in Figure 3. In addition, the negative productivity shock moves the Wealth curve to the left. Now we have a multiple equilibria situation that contains a twin (i.e., banking and currency) crisis equilibrium with low y 2 and high E1 . Note that the output level of a twin crisis is small but still positive because firms can invest with their retained earnings when banks bear the exchange rate risk.

24


Figure 3: Equilibria after the Shock without Policy Response

E1

Wealth Curve Negative Productivity Shock

IPLM Curve Risk Premium Shock

Negative Productivity Shock

y2 What kind of monetary policy should the central bank choose in a twin crisis? The Wealth curve implies that the monetary policy to prevent the crisis is to lower the lending interest rate so that the curve moves to the right to preserve economic stability (Figure 4). This means that the central bank needs to lower the policy interest rate from equation (11), otherwise the countervailing effects from the interest rate hike (from equations [14] and [18ab]) lead to an increase in the interest burden on domestic loans and hence result in a contraction in output. This result is related to arguments by Furman and Stiglitz (1998), who asserted that higher interest rates seriously erode the net worth of debtors, leading them to contract investment and production. On the other hand, the IPLM curve implies that the monetary policy to prevent a crisis is to raise the deposit interest rate so that the curve moves downward to preserve financial stability (Figure 4). Since the policy interest rate should be lowered to shift the Wealth curve, the central bank resorts to the reserve requirement policy to move the IPLM curve. Equation (13) shows that the central bank should raise the interest rate on reserves and lower the reserve requirement ratio. Thus, in this twin banking and currency crisis model, the interest rate defense should be conducted through the deposit interest rate with the reserve requirement ratio and not the policy interest rate. The 25


economic intuition is as follows. A decrease in the reserve requirement ratio and an increase in the interest rate on reserves lead to a decrease in the reserve tax imposed on deposits. This reduction in the distortion caused by the required reserves makes deposits more attractive to the commercial bank and hence it increases the deposit interest rate. This increase in the deposit interest rate induces a currency appreciation from the IPLM condition (29abc). In conclusion, the monetary policy that prevents a twin banking and currency crisis to preserve financial and economic stabilities is to (1) lower the policy interest rate, (2) lower the reserve requirement ratio, and (3) raise the interest rate on reserves. Figure 4: Equilibrium after the Shock with Monetary Policy Response

E1

Wealth Curve

Interest Rate Policy Reserve Requirement Policy

Interest Rate Policy

IPLM Curve

y2 Why are our policy implications different from those of the original ABB model? Aghion et al. (2004) introduced the banking sector into their model and analyzed the effects of the reserve market. In their model, the central bank needs to raise the policy interest rate to induce currency appreciation and lower the discount window rate to offset the countervailing effects of the interest rate hike in lowering output. This is due to the fact that the IPLM curve was characterized by the lending interest rate since the firm bears the exchange rate risk in their model. Thus, there is no 26


choice but to raise the policy rate in spite of the countervailing effects. In other words, in the original ABB model, an increase in the policy interest rate will induce currency appreciation because the firms want to borrow more in foreign currency than in domestic currency. By contrast, since the commercial banks bear the exchange rate risk in our model, the IPLM condition is determined by the deposit interest rate and hence the lending interest rate is irrelevant to the shift of the IPLM curve. In other words, we only need to care about the shift of the Wealth curve when we consider the effects from the change in the lending interest rate, which only the policy interest rate can affect. The economic intuition is that when commercial banks bear the exchange rate risk, an increase in the policy interest rate has adverse macroeconomic effects on the real economy because it will only increase the interest burden on firms. Next, we examine the case in which the credit constraint is derived from the moral hazard problem and the credit multiplier depends negatively on lending interest rates. As in the ABB paper, we assume that there is a cost to hide the production value so that the firm defaults, and this cost is proportional to the amount of funds invested k t . We also assume that the lender can still collect the due repayment with the monitoring probability  when the firm chooses to default. The firm decides not to default as long as the following incentive constraint is satisfied:









yt  1  itL ltD  yt kt   1  itL ltD .

 





 

This constraint yields the credit multiplier  itL   1    1  itL 

which depends

negatively on the lending interest rate. Since we assume perfect competition among banks, the firm has full bargaining power upon loan contracts. Does this relaxation of the assumption of credit multiplier change the policy implications? The answer is no, since an increase in the policy

27


interest rate will dampen the economy by shifting the Wealth curve to the left through both a decreased credit multiplier and reduced net worth of the firm due to the increased interest rate repayments. In this setting, an appropriate monetary policy response is to lower the policy interest rate. Note that the reserve requirement policy is effective only for the deposit interest rate and cannot affect the Wealth curve. Thus, the appropriate monetary policy mix is the combination of lowering the policy interest rate and the reserve requirement ratio and raising the interest rate on reserves. Finally, we consider the case in which not only commercial banks but also firms borrow from abroad in foreign currency and thus the exchange rate risks are located in both firms and banks. In this case, firms are indifferent between borrowing from domestic banks and abroad. This means that the lending interest rate enters in the IPLM condition derived from the arbitrage by firms. For this reason, the policy interest rate can have a direct effect on exchange rates through interest parity conditions of firms and banks. Namely, an increase in the policy interest rate induces an appreciation of domestic currency. However, the increase in the policy rate has also adverse macroeconomic effects on the real economy by reducing the net worth of the firms (as mentioned earlier) and leading to lower investment and output. Therefore, in this case the appropriate monetary policy response should be determined by the authority by comparing the merit of an interest rate hike on currency appreciation and the consequences of countervailing effects induced by the interest rate hike. When the monetary authority can use the reserve requirement policy, the policy mix suggested by the case of twin crisis can be one choice.

28


4 Conclusion Prevention of currency and banking crises is an important goal for policy makers, especially for the monetary authority. In this paper, we developed a twin banking and currency crisis model by extending the third generation model of currency crises developed by Aghion et al. (2000, 2001, 2004). We extended the ABB model so that the exchange rate risk is located in the banking sector, and we provided the mechanism through which the crisis affects the real economy. This is the first theoretical research in which commercial banks take foreign currency loans and bear the exchange rate risk in the framework of the ABB model. This framework provides a new model in which not only a currency crisis, but also a banking crisis, can be concurrently triggered by an unanticipated shock such as an expectational shift of international investors in financial markets and/or a negative productivity shock in the real sector rather than consumers’ withdrawal shock that most of banking crisis models have relied on. In this setting, we showed that the monetary authority needs to use multiple monetary policy instruments to preserve financial and economic stabilities. In practice most central banks are using not only the policy interest rate, but also other policy instruments, to stabilize financial markets. As argued in the literature review (section 2.2), one of the most widely and often used policy tools is the reserve requirement policy, which can be an effective tool to stabilize the banking system. Therefore, by introducing the banking sector and the reserve system and analyzing the behavior of commercial banks, we examined the combination of appropriate interest rate policy and reserve requirement policy that can prevent a twin crisis. Thus, the key innovation of our twin crisis model is the types of shocks, the location of the risk and multiple monetary policy instruments for the prevention of the crisis. In this crisis model, we

29


showed that the monetary authority should use the reserve requirement policy for financial stability and the interest rate policy for economic stability. Namely, the monetary policy to prevent a twin crisis is to lower the policy interest rate and the reserve requirement ratio and increase the interest rate on reserves. This is in stark contrast to the literature on currency crises that insists that the policy interest rate should be hiked during the crisis. Note that one of the concerns of policy implications derived from the original ABB model was the countervailing effect that the interest rate hike lowers output. However, in our twin crisis model, the central bank no longer faces the dilemma of countervailing effects because the lending interest rate does not enter the IPLM condition. Instead, the deposit interest rate determines the IPLM curve. Therefore, the monetary authority can use the reserve requirement policy to prevent a currency and banking crisis by changing the deposit interest rate. One of the policy implications suggested by this model is as follows. If the international interest rate increases as a result of monetary tightening by the Federal Reserve Board, central banks in emerging countries in which economies have foreign currency debts in the banking sector should lower their policy interest rates and reserve requirement ratios and raise the interest rates on reserves. In this way, our analysis suggests that it is necessary for central bankers to assess the location of the exchange rate risk when conducting monetary policy during crises.

30


References Aghion P, Bacchetta P, Banerjee A (2000) A simple model of monetary policy and currency crises. Eur Econ Rev 44:728-738 Aghion P, Bacchetta P, Banerjee A (2001) Currency crises and monetary policy in an economy with credit constraints. Eur Econ Rev 45:1121-1150 Aghion P, Bacchetta P, Banerjee A (2004) A corporate balance-sheet approach to currency crises. J Econ Theory 119:6-30 Bergman UM, Jellingsø M (2010) Monetary policy during speculative attacks: are there adverse medium term effects? N Am J Econ Financ 21:5-18 Black F (1975) Bank funds management in an efficient market. J Financ Econ 2:323-339 Bleaney M, Bougheas S, Skamnelos I (2008) A model of the interactions between banking crises and currency crises. J Int Money Financ 27:695-706 Braggion F, Christiano LJ, Roldos J (2009) Optimal monetary policy in a ‘sudden stop.’ J Monet Econ 56:582-595 Buch CM, Heinrich RP (1999) Twin crises and the intermediary role of banks. Int J Financ Econ 4:313-323 Burnside C, Eichenbaum M, Rebelo S (2001) Hedging and financial fragility in fixed exchange rate regimes. Eur Econ Rev 45:1151-1193 Caballero RJ, Krishnamurthy A (2001) International and domestic collateral constraints in a model of emerging market crises. J Monet Econ 48:513-548 31


Calvo GA, Leiderman L, Reinhart CM (1993) Capital inflows and real exchange rate appreciation in Latin America: the role of external factors. IMF Staff Pap 40:108-151 Chang R, Velasco A (2001) A model of financial crises in emerging markets. Q J Econ 116:489-517 Claessens S, Ghosh SR, Mihet R (2013) Macro-prudential policies to mitigate financial system vulnerabilities. J Int Money Financ 39:153-185 Concha A, Galindo AJ, Vasquez D (2011) An assessment of another decade of capital controls in Colombia: 1998-2008. Q Rev Econ Financ 51:319-338 Cordella T, Federico P, Végh CA, Vuletin G (2014) Reserve requirements in the brave new macroprudential world. World Bank Policy Research Paper No. 6793 Corsetti G, Pesenti P, Roubini N (1999) Paper tigers? A model of the Asian crisis. Eur Econ Rev 43:1211-1236 De Gregorio J, Edwards S, Valdés RO (2000) Controls on capital inflows: do they work? J Dev Econ 63:59-83 Dekle R, Kletzer KM (2002) Domestic bank regulation and financial crises: theory and empirical evidence from East Asia. In: Edwards S, Frankel JA (eds) Preventing currency crises in emerging markets. University of Chicago Press, Chicago Dooley MP (2000) A model of crises in emerging markets. Econ J 110:256-272 Edwards S, Végh CA (1997) Banks and macroeconomic disturbances under predetermined exchange rates. J Monet Econ 40:239-278 32


Eichengreen B, Hausmann R, Panizza U (2007) Currency mismatches, debt intolerance, and the original sin: why they are not the same and why it matters. In: Edwards S (ed) Capital controls and capital flows in emerging economies: policies, practices and consequences. University of Chicago Press, Chicago Eijffinger SCW, Goderis B (2008) The effect of monetary policy on exchange rates during currency crises: the role of debt, institutions, and financial openness. Rev Int Econ 16:559-575 Eijffinger SCW, Karataş B (2012) Currency crises and monetary policy: a study on advanced and emerging economies. J Int Money Financ 31:948-974 Fama EF (1980) Banking in the theory of finance. J Monet Econ 6:39-57 Flood RP, Garber PM (1984) Collapsing exchange-rate regimes: some linear examples. J Int Econ 17:1-13 Flood RP, Jeanne O (2005) An interest rate defense of a fixed exchange rate? J Int Econ 66:471-484 Freeman S (1987) Reserve requirements and optimal seigniorage. J Monet Econ 19:307-314 Furman J, Stiglitz JE (1998) Economic crises: evidence and insights from East Asia. Brook Pap Econ Act 2:1-135 Gelos RG (2009) Banking spreads in Latin America. Econ Inq 47:796-814 Glick R, Hutchison MM (2001) Banking and currency crises: how common are twins? In: Glick R, Moreno R, Spiegel M (eds) Financial crises in emerging markets. Cambridge University Press, Cambridge 33


Glick R, Hutchison MM (2013) Models of currency crises. In: Gerard C (ed) The evidence and impact of financial globalization. Elsevier Inc., Oxford Glocker C, Towbin P (2012) Reserve requirements for price and financial stability: when are they effective? Int J Cent Bank 8(1):65-113 Goderis B, Ioannidou VP (2008) Do high interest rates defend currencies during speculative attacks? New evidence. J Int Econ 74:158-169 Goldfajn I, Gupta P (2003) Does monetary policy stabilize the exchange rate following a currency crisis? IMF Staff Pap 50(1):90-114 Goldfajn I, Valdés RO (1997) Capital flows and the twin crises: the role of liquidity. IMF Working Paper No. 97/87 Goodfriend M (2002) Interest on reserves and monetary policy. Fed Reserv New York Econ Policy Rev, May 2002:77-84 Gray S (2011) Central bank balances and reserve requirements. IMF Working Paper No. 11/36 Hoffmann A, Löfﬂer A (2014) Low interest rate policy and the use of reserve requirements in emerging markets. Q Rev Econ Financ 54:307-314 Kaminsky GL, Reinhart CM (1999) The twin crises: the causes of banking and balance-of-payments problems. Am Econ Rev 89(3):473-500 Krugman P (1979) A model of balance-of-payments crises. J Money Credit Bank 11:311-325 Lahiri A, Végh CA (2007) Output costs, currency crises and interest rate defence of a peg. Econ J 117:216-239 34


Loungani P, Rush M (1995) The effects of changes in reserve requirements on investment and GNP. J Money Credit Bank 27:511-526 McKinnon RI, Pill H (1996) Credible liberalizations and international capital flows: the ‘overborrowing syndrome.’ In: Ito T, Krueger AO (eds) Financial deregulation and integration in East Asia. University of Chicago Press, Chicago Miller V (2014) A crisis transmission channel for reserve currency countries: a cautionary tale. Open Econ Rev 25: 809-818 Miller V, Vallée L (2011) Central bank balance sheets and the transmission of financial crises. Open Econ Rev 22: 355-363 Mimir Y, Sunel E, Taşkɪn T (2013) Required reserves as a credit policy tool. BE J Macroecon 13:823-880 Mora N (2014) Reason for reserve? Reserve requirements and credit. J Money Credit Bank 46:469-501 Nakatani R (2014) The effects of financial and real shocks, structural vulnerability and monetary policy on exchange rates from the perspective of currency crises models. UTokyo Price Project Working Paper No.061, Graduate School of Economics, University of Tokyo (Available online at http://www.price.e.u-tokyo.ac.jp/img/researchdata/pdf/p_wp061.pdf) Obstfeld M (1996) Models of currency crises with self-fulfilling features. Eur Econ Rev 40:1037-1047 Ostry JD, Ghosh AR, Chamon M, Qureshi MS (2012) Tools for managing financial-stability risks

35


from capital inflows. J Int Econ 88:407-421 Radelet S, Sachs JD (2000) The onset of the East Asian financial crisis. In: Krugman P (ed) Currency crises. University of Chicago Press, Chicago Reinhart CM, Reinhart VR (1999) On the use of reserve requirements in dealing with capital flow problems. Int J Financ Econ 4:27-54 Reinhart CM, Rogoff KS (2013) Shifting mandates: the federal reserve’s first centennial. Am Econ Rev 103(3):48-54 Tovar CE, Garcia-Escribano M, Martin MV (2012) Credit growth and the effectiveness of reserve requirements and other macroprudential instruments in Latin America. IMF Working Paper No. 12/142 Valdés-Prieto S, Soto M (1998) The effectiveness of capital controls: theory and evidence from Chile. Empirica 25:133-164 Vargas H, Cardozo P (2012) The use of reserve requirements in an optimal monetary policy framework. Borradores de Economía 716i. Walsh CE (2012) Discussion of ‘reserve requirements for price and financial stability: when are they effective?’ Int J Cent Bank 8(1):115-126 Walter JR, Courtois R (2009) The effect of interest on reserves on monetary policy. Fed Reserv Richmond Econ Brief 09-12

36