Currency and Banking Crises: The Origins and How to Identify Them

Abstract

Currency and banking crises have occurred periodically. Along with the growing integration and liberalisation of financial markets, there is an increasing number of currency and banking crises. This chapter shows that crises can occur in any good or bad economic conditions as they can be triggered by rational actions, panics or contagion effects. As the crises stem from various aspects, thus, they should be mitigated with different policies. Furthermore, this chapter discusses the exchange market pressure index to identify a currency crisis, the money market pressure index to identify a banking crisis and the financial market pressure index to identify the overall financial crises. Once the crises are identified, economists can start to investigate the determinants of the crises. While the probit/logit models are arguably the most popular approaches to investigating the determinants of crises, they fail to provide useful forecasts. On the other hand, while the signalling approach is considered the most successful method to forecast crises, the results are difficult to be interpreted. The empirical studies suggest that the currency and banking crises are typically preceded by a real appreciation and a lending boom, the signs of a boom period in the business cycle.

1. Introduction

Financial crises – in particular currency crises and banking crises – have occurred periodically. The IMF reported there were 158 currency crises and 54 banking crises during 1980–1995 [1]. The collapse of the Medici Bank in the 15th century is an example of a banking failure in the early days of the development of banking. The stories of bank failure continue for centuries. The latest series of financial crises started with the subprime mortgage crisis in the US in 2007–2008 which was then followed by the banking crisis in the UK in 2008–2009 and the Spanish banking crisis.

Along with the growing integration of financial markets, financial institutions such as banks have become increasingly vulnerable to financial turmoil elsewhere. Moreover, a financial turmoil that originates from a currency crisis often develops into a banking crisis, or vice versa, giving rise to what is known as a ‘twin crisis’.

This chapter explains the origins of currency and banking crises and explores different approaches to identifying the crises, as well as their determinants.

The structure of the chapter is as follows. The next section explains the origins of currency and banking crises and how to mitigate them. The third and fourth section explores different approaches to identifying the crises and their determinants, respectively. The final section concludes the chapter.

2. The origins of currency and banking crises

The models to explain the origin of currency and banking crises can be divided into three groups. The first group discusses the rational expectations of investors as the source of crises. According to this model, investors always observe the risk in their investments and act rationally. Thus, a bad macroeconomy fundamental or ineffective policy actions may trigger currency attacks or bank runs [2, 3].

The second group shows that a crisis can still occur at a good time. In this model, a crisis can be triggered by a random panic that influences investors’ behaviour to massively convert their assets, thus, the crisis is self-fulfilling [4, 5].

The last group explains that a crisis can also arise in the non-existence of panic at a good time. Following this model, a crisis in other places can spread through the financial market and creates a twin currency and banking crisis [6, 7].

Interestingly, the development of the currency crisis models has gone in the opposite direction to the banking crises development. The first generation model of the crises suggests that a currency crisis stem from rational actions of investor (the speculative attack model) and a banking crisis stem from a panic attack (the random withdrawal model). On the other hand, the second-generation models of the crises argue that a currency crisis is originated from a panic attack (the self-fulfilling prophecy model) and a banking crisis is originated from a ration action (the information-based bank runs).

2.1 The rational acts as the source of currency and banking crises

The rational acts as the source of currency and banking crises can be divided into two groups. In the currency crises model, the speculative attack model argues that investors always doubt the government’s ability to manage a fixed exchange rate when there is a current account deficit. When the foreign reserve is drying up to keep the exchange rate fixed, investors will attack the currency, leading to the breakdown of the fixed exchange rate regime. On the other hand, in the banking crises model, the information-based model shows that negative results in investors’ risk assessment of bank portfolios may influence a bank run and create a banking crisis.

2.1.1 Speculative attack on the currency

The speculative attack model shows that investors undertake an attack if they doubt the government’s capacity to keep the exchange rate fixed. Specifically, this condition occurs when the continuation of the current account deficit leads to a decline in foreign exchange reserves. As a result, the speculative attack causes the remaining reserves to move to investors, thus negatively affecting the currency [2, 8, 9].

The main contribution of the model is the idea that the speculative attack on the currency stems from a rational act rather than from investors’ panic. This model succeeded in explaining the currency crisis in Latin America just a few years after it was developed, prompting researchers to examine currency crises as rational events.

The model can be explained as follow. Let us recall the domestic money market equilibrium:

$$\mathit{m}–\mathit{p}=-\mathit{\alpha }\left(\mathit{i}\right),\mathit{\alpha }>\mathit{0}$$E1

Where m is the money supply, p is the price level, and i is the interest rate.

The money supply is calculated based on credit (d) and foreign reserves (r), therefore:

$$\mathit{m}=\mathit{d}+\mathit{r}$$E2

Assuming purchasing power parity holds, we can restate the price level (p) as a fraction of the foreign price level (p*) and the exchange rate (s), as follows:

$$\mathit{p}={\mathit{p}}^{\ast }+\mathit{s}$$E3

Imposing uncovered interest rate parity, we can substitute the interest rate (i) with the foreign currency interest rate (i*) and the change in the exchange rate (Δs), as follows:

$$\mathit{i}={\mathit{i}}^{\ast }+\mathsf{\text{∆}}\mathit{s}$$E4

In a fixed exchange rate regime, where s is equal to the future exchange rate (s^e), it implies that Δs = 0 and i = i*. By substituting Eqs. (2)–(4) into Eq. (1) with Δs = 0, it follows that:

$$\mathit{r}+\mathit{d}–{\mathit{p}}^{\ast }-{\mathit{s}}^{\mathit{e}}=-\mathit{\alpha }\left({\mathit{i}}^{\ast }\right)$$E5

Therefore, in a fixed exchange rate regime (assuming foreign currency interest rate and foreign price level are fixed), the credit grows at the same rate as the fall of the foreign reserve (∆d = −∆r). In the end, the foreign reserve will run out and force central banks to break the fixed exchange rate regime. Thus, the change in exchange rate policy will lead to speculative attacks which in turn lead to a crisis.

2.1.2 Information-based bank run

The information-based bank run model argues that the bank run is a logical consequence of a rational change of risk in bank portfolios [3].

In the model, there are three periods (T = 0, 1, 2) where agents have one short-term investment from T = 0 to T = 1 and one long-term investment from T = 0 to T = 2. All agents are identical at T = 0. The model imposes three assumptions: first, agents will adjust their preferences based on information on T = 1. Second, the returns on long-term investments are random. Third, long-term investment yields a zero payoff if liquidated at T = 1. Since there is no information about the returns of long-term investments, agents always observe their investments based on newly available information. If agents believe that the bank portfolio is at risk based on the latest available information, agents will withdraw their deposit. Consequently, the bank runs are information-based.

Information-based models view banks as providers of a valuable service (by creating non-marketable bank loans) rather than providers of liquidity insurance. However, non-marketable loans in the bank portfolio are difficult to monitor, thus creating asymmetric information between banks and agents.

Furthermore, agents with no interim information cannot observe the real value of a bank, thus they learn about a bank’s condition by observing other depositors. However, agents cannot distinguish whether the source of withdrawal is for consumption needs or a run by informed depositors. Therefore, risk-averse agents could assume the worst-case scenario which leads to panic [10]. In addition, a noisy signal and asymmetric information between agents could lead to bank run even when the fundamentals are good enough [11].

A bank run could be efficient since there is risk-sharing between agents. However, the liquidation cost would make a bank run becomes inefficient, so central banks should intervene to control the liquidation cost [12].

Furthermore, as banks are aware that some agents receive interim information and understand the implications of different types of contracts, thus a little change in the contract will discourage agents to conduct a bank run. However, different types of a contract will have different utilities, and banks, on purpose, sometimes would choose a contract that allows a bank run [13]. Furthermore, as the information-based bank runs stem from an asymmetric information, encouraging banks to regularly publish their financial report and a statement from a credible bank supervisor may reduce the risk of bank runs.

2.2 Panics as the source of currency and banking crises

Panics as the source of currency and banking crises can be divided into two groups. In the currency crises model, the self-fulfilling prophecy model argues that the herd behaviour of investors may cause panic and lead to the withdrawal of assets. As a result, the exchange rate tends to depreciate and translates into a crisis. On the other hand, in the banking crises model, the random withdrawal model shows that depositors can do a bank run due to random events because of the lack of information held by the depositors. When there is a massive bank run, most banks will suffer from liquidity problems since banks heavily invest in long-term illiquid assets which are costly to liquidate.

2.2.1 Self-fulfilling prophecy on the exchange market

The self-fulfilling prophecy model shows that herd behaviour of investors may cause panic and lead to the withdrawal of assets. As a result, the exchange rate tends to depreciate and translates into a crisis [4].

In this model, investors’ actions rely on the sequential observation of other investors’ actions. If an investor observes that many other investors sell the currency, then the investor will join the herd despite his own information. Thus, the equilibrium will move from no-attack to attack equilibrium [14].

Furthermore, a lack of information between investors can also lead to an attack and breakdown of the fixed exchange rate even though there is no coordinated action between investors [15]. In this example, investors always observe the state of the economy and consider other investors’ beliefs on the sustainability of a fixed exchange rate. Assuming other investors believe that the fixed exchange rate is unsustainable, investors will launch an attack if the cost of the attack is not too costly.

On the other hand, globalisation creates many investors who have identical decisions in selecting their portfolios. Driven by relative performance to other investors’ performances, investors select the same portfolio with other investors to match their performances and create herding behaviour which leads to attack equilibrium [16].

The model can be explained by imposing a conditional shift in domestic credit growth into the speculative attack model [9], where the growth is G₀ if there is no attack, while credit grows faster at G₁ if there is an attack.

Figure 1 simulates the attack on conditional policy shift. S₀* and S₁* represent “shadow exchange rate lines” correlating to the rate of credit growth at G₀ and G₁ respectively. S_f is a fixed exchange rate that intersects with shadow exchange rate line S₀* at point A and shadow exchange rate line S₁* at point C.

Figure 1.

Self-fulfilling prophecy with attack equilibrium.

Assuming “domestic credit” (G) is at the left side of point G^B (G ≤ G^B), “the shadow rate” is at point B if there is no attack and jumps to point C if there is an attack. In this simulation, the “shadow rates” (S*) are always below (or maximum at) the fixed exchange rate (S* ≤ S_f), thus giving no incentives to speculators to attack the fixed exchange rate.

Multiple equilibria can occur when “domestic credit” is in the range between G^A and G^B. The fixed exchange rate could be maintained if investors believe there is no chance to overcome the government (there is no immediate benefit). On the contrary, the “exchange rate” could shift to the upper “shadow rate line” (S₁*) if investors believe there will be an attack on the currency which leads to a breakdown of the fixed exchange rate. Consequently, all investors will sell domestic currency, leading to a collapse of the fixed exchange rate. However, there are multiple equilibria in this condition since the attack can only be succeeded if there is a large investor or coordinated action of small investors to launch an attack of sufficient size.

The drawback of the self-fulfilling model is the fact that the model implies the difficulty of predicting currency crises. It implies that policymakers have a limited role in managing the exchange rate since the triggers of currency crises are unclear, thus, they are difficult to forecast. However, as the attack can only succeed within a specific range of fundamentals, therefore, the policy maker can manage currency crises by managing the fundamentals [17].

2.2.2 Random withdrawal on the banking system

The random withdrawal model shows that depositors can do a bank run due to random events because of the lack of information held by the depositors. When there is a massive bank run, most banks will fail since banks heavily invest in loans that are costly to liquidate [5].

Influenced by the first-come-first-served rule, panic could occur when depositors think there will be a bank run, thus agents will try to withdraw their deposit as soon as possible before the bank collapses. Therefore, panic is self-fulfilling. Mervyn King, former Governor of the Bank of England, once said “it may not be rational to start a bank run, but it is rational to participate in one once it had started”. However, panic could be avoided if banks do not follow the first-come-first-served rule. Panic will lead to expensive liquidation costs and therefore can only occur when agents are risk-averse [18].

Since the first-come-first-served rule is an essential ingredient for a bank run, eliminating this rule will also eliminate the possibility of a bank run. As an alternative to this rule, suspension of deposit convertibility in the event of a bank run [5] and variation of contracts to accommodate the possibility of a bank run (an allow-bank run contract and a run-proof contract) [18] could be considered.

Furthermore, panic could occur because the institutional structure fails to provide liquidity [19]. To avoid panic, separated-local banks will prevent agents from conducting coordinated withdrawals. Problems in separated-local banks should be addressed by a local reserve bank. Therefore, panic is related to an institutional structure in the banking system when liquidity fails to be provided. However, panic could be avoided if banks can perform an interbank loan market. Furthermore, in order to prevent panic, policymakers should force separated-local banks to hold adequate reserves [20].

In the random withdrawal model, agents use the bank as insurance against risk to cover the uncertainty of consumption needs. To do this, banks provide liquidity and guarantee when agents liquidate their investments before maturity. In doing so, banks can increase welfare but are exposed to risk. Thus, they create the possibility of a self-fulfilling bank run.

The model has three periods (T = 0, 1, 2) where agents have one short-term investment from T = 0 to T = 1 and one long-term investment from T = 0 to T = 2. All agents are identical at T = 0 and learn their type at T = 1: being a type 1 agent or being a type 2 agent who cares only about consumption in T = 1 or T = 2, respectively. The salvage value of the long-term investment is equal to the initial investment if it is interrupted at T = 1.

There are two important assumptions in the model which can lead to bank panic: agents cannot claim physical assets in exchange for their deposit, and deposit withdrawals follow the first-come-first-served rule. Based on these assumptions, there will be two equilibriums: good equilibrium occurs when type 1 agents withdraw their deposit at T = 1 and type 2 agents withdraw at T = 2, and bad equilibrium occurs when there is panic. As the liquidation of a bank’s long-term assets is costly, thus, a bank will not survive if all deposits are withdrawn at once.

2.3 The contagion effects as the source of currency and banking crises

The contagion effects as the source of currency and banking crises can be divided into two groups. In the first model, the systemic risk model argues that a bank failure can create a systemic failure in the banking system through the money market. In the second model, the twin crisis model discusses how a currency crisis translates into a banking crisis or vice versa.

2.3.1 Systemic risk in banking system

The systemic risk model focuses on the propagation of a failure in one bank to other banks through financial transactions. Based on this model, interbank lending can overcome the moral hazard problem between the bank owner and depositors due to the supervision of peer banks. However, interbank lending also increases the risk of contagion for banks [6].

An interbank money market has a central role in developing systemic risk. If an interbank money market cannot support one illiquid bank, a systemic bank run may occur since agents may assume that there is not enough liquidity in the banking system. However, a problem in one bank is not sufficient to create panic. It can only be systemic when the problem occurs in a time of economic instability [21]. In addition, even though agents of one specific bank can have interim information; they do not have access to the interim information of other banks. Therefore, they will observe the number of bank failures as a proxy of interim information about macroeconomic conditions and other banks’ performance. In this sense, agents may conduct a bank run if they observe there are some bank failures [22].

One strand of study of systemic risk focuses on uncertainty over liquidity demand. Since agents are uncertain about where they want to consume, banks face the risk of withdrawal and the transference of agents’ deposits to other areas. To address this problem, banks create an interbank money market, thus there is no need to liquidate their long-term investments to meet agents’ cash demands. However, an interbank money market could make contagious bank failures when there is a gridlock in the payment system. Therefore, agents could panic when they fear there is no sufficient reserve among banks [23]. Furthermore, the interbank money market grows because of different liquidity shortages across regions. In this sense, the spread of contagion is influenced by types of claims in the interbank money market [24].

Another view of systemic risk studies the role of the unregulated banking system on systemic risk [25]. The study focuses on claims that bank failures are influenced by safety-net regulations, thus minimal regulatory intervention is required to regain financial stability. While financial market arrangement by a private institution (e.g. clearing house) is more efficient in preventing systemic shocks, a global liquidity shortage that triggers contagious runs may break down the arrangement. Therefore, an unregulated banking system is not immune to systemic risk.

2.3.2 Twin currency and banking crises

The linkages between the twin currency and banking crises are still ambiguous. It is hard to identify whether it is started by a currency or banking crisis for two reasons: First, banking and currency crises are sometimes driven by common factors [26]. Second, the currency attack and the bank run reinforced each other in a vicious circle [27].

The twin currency and banking crises model shows that the link between a currency and banking crisis lies in a problem for both foreign and domestic currency liquidity [7]. Investors start to attack the currency, either because of economic fundamentals or panic. Currency starts to depreciate and the pressure in the exchange market increases. To fund the attack, investors remove their money from banks and create pressure in the money market that can create a currency and banking crisis.

On the other hand, investors could also attack the bank, either because of random events or information-based. The cash is then used to attack the currency. To avoid sharp depreciation of the currency, the central bank starts intervening by selling foreign reserves and buying domestic currency. The money supply is contracted and pressure in the money market becomes higher. Banks start to have liquidity problems [28].

Investors will observe the central bank’s capability to intervene and decide whether to continue the attack. Investors will attack the currency if the central bank indicates its defence of the currency in limited foreign reserve. However, if the central bank decides to allow the currency to depreciate, negative news and fear of depreciation may create panic and a self-fulfilling prophecy.

The central bank’s intervention ceases when there is insufficient foreign reserve to sell or there is a lack of domestic currency to be bought, which then leads to a sharp depreciation of the currency. Indeed, the central bank could sterilise the intervention by buying domestic bonds. However, in many cases, the amount of available liquid and high-quality bonds is relatively limited compared to the value of an intervention (Figure 2).

Figure 2.

Relationship between currency and banking crises.

Due to the low value of the domestic currency (and the fall of financial asset prices), demand for the domestic currency to buy foreign currency is doubled. There is a liquidity shortage in the money market which leads to a high-interest rate. Some banks will have liquidity problems and become failed banks.

Furthermore, the second-round effect of currency depreciation starts to affect banks that are exposed to foreign liabilities [29]. Some banks have currency mismatches. The third round effect affects the bank’s debtors which leads to an increase in domestic and foreign currency non-performance loans (NPL). In addition, banks with foreign debt suffer from the increase in the cost of borrowing. Finally, both banks with and without foreign liabilities suffer from losses and have liquidity and insolvency problems.

The model shows that a successful intervention by the central bank may still lead to a banking crisis through the liquidity shortage channel. Furthermore, a successful attack on currency could lead to a banking crisis in two channels: on the one hand, sharp currency depreciation directly creates a currency mismatch for banks with foreign liquidity exposure. On the other hand, sharp currency depreciation affects the economy and decreases debtors’ financial performance which leads to increasing NPL both for domestic and foreign currency loans. In addition, banks with foreign debt also suffer from the increasing cost of borrowing. Therefore, twin currency and banking crises should appear simultaneously.

3. The identification of currency and banking crises

The identification of currency and banking crises can be divided into three groups. In currency crises, the Exchange Market Pressure Index is arguably the most popular approach to identify the crises. In the banking crises, there is a growing interest to employ the Money Market Pressure Index to identify the crises. While the above indexes gain popularity in the currency and banking crises literature, there may be a benefit to investigating the crises as twin currency and banking crises due to interconnection in the financial market. Thus, the Financial Market Pressure model arguably provides better insight into the overall pressure on the financial market.

However, economists should employ the market pressure-based approach with caution due to the absence of consensus on how to weigh the variables and how to define the thresholds. As different weights and thresholds may lead to different crisis databases, economists may be tempted to adjust the models to fit their preferences.

3.1 Identification of currency crises

In general, there are two methods to identify currency crises. Early studies of currency crises generally use the depreciation of the exchange rate at a certain level as a basis to identify the crisis. There is no consensus about the threshold of currency depreciation to identify currency crises. For example, some economists define a currency crisis as when a currency depreciates more than 25% and there is an increase of 10% in the rate of depreciation [30].

However, using exchange rate depreciation to identify the currency crises may be biased when the central bank intervenes in order that the exchange rate does not depreciate despite considerable pressure on the currency. Even though central banks announce that they are employing an information targeting framework or a free float exchange rate regime, it is commonly acknowledged that central banks do intervene in the foreign exchange market to smooth exchange rate volatility or to maintain the exchange rate in a certain band due to the fear of floating [31].

For that reason, most recent studies use Exchange Market Pressure Index (EMPI) as the basis to identify currency crises. This model is originally developed as a monetary model to calculate the amount of foreign exchange intervention to meet a desired exchange rate target [32].

EMPI illustrates that the pressure on the exchange rate is not only reflected in the depreciation ($∆{\mathrm{e}}_t$) but also on the amount of central bank intervention through the spot market ($∆R_t$) and sometimes through the interest rate. In the event of an intervention by central banks to slow the depreciation rate, EMPI shows higher pressure in the exchange market despite there being only limited depreciation in the exchange rate.

As the variables included in the EMPI have different volatility, careful consideration of the weight associated with each variable is therefore required, so that no particular variable can distort the EMPI. Currently, there are three different popular weighting methods in the market pressure-based approaches [33, 34, 35], all of which use the standard deviation to weight the variables.

$${\mathit{\text{EMPI}}}_t=∆{\mathrm{e}}_t-\frac{{\sigma }_e}{{\sigma }_R}∆R_t$$E6

$${\mathit{\text{EMPI}}}_t=\frac{1}{{\sigma }_e}∆{\mathrm{e}}_t-\frac{1}{{\sigma }_R}∆R_t$$E7

$${\mathit{\text{EMPI}}}_t=\frac{\frac{1}{{\sigma }_e}}{\left(\frac{1}{{\sigma }_e}+\frac{1}{{\sigma }_R}\right)}∆{\mathrm{e}}_t-\frac{\frac{1}{{\sigma }_R}}{\left(\frac{1}{{\sigma }_e}+\frac{1}{{\sigma }_R}\right)}∆R_t$$E8

where ${\mathrm{\sigma }}_{\mathrm{e}}$ is the standard deviation of the exchange rate and ${\mathrm{\sigma }}_{\mathrm{R}}$ is the standard deviation of foreign reserve.

There is no consensus on the threshold, for example, some economists define currency crises as EMPI exceeding three standard deviations or more above the mean.

3.2 Identification of banking crises

There are two popular methods to identify banking crisis episodes. The first method is based on events, such as bank performance, government bailout, widespread bank failures, the extent of bank runs, and professional analysis to specify bank crises [36, 37].

Despite its popularity, it is difficult to identify the start date of banking crises using this event method. For example, the government’s bailout normally occurs at the peak of a crisis, sometimes it involves a political process that delays the bailout.

To address the drawback of the event methods, some economists develop a Money Market Pressure Index (MMPI) as a quantitative approach to identify banking crises. The new method argues that pressure in the banking system should be reflected by the change in the interest rate. However, the central bank may conduct a monetary operation to manage the interest rate. Thus, the pressure in the banking system cannot be shown by the interest rate alone. To address this issue, the interest rate should be offset by the monetary operation [38].

MMPI can be formulated as:

$${\mathit{\text{MMPI}}}_t={\omega }_1∆{\gamma }_t+{\omega }_2∆i_t$$E9

Where $∆{\gamma }_t$ is changes in reserves to bank deposits ratio, $∆i_t$ is changes in short-term real interest rate, and $\omega$ is the weight between variables.

Although the MMPI was introduced before the Global Financial Crisis of 2007–2008, it gained popularity as an approach to identifying banking crises in the aftermath of the Global Financial Crisis due to its reliability and simplicity. The index shows that while it is based on only two variables, the MMPI can identify banking crises in both developed and emerging markets [39]. In addition, the MMPI provides a clear indication of the start and end dates of banking crises - a feature that is not available in the ‘so-called’ event approach.

3.3 Identification of twin currency and banking crises

While there are many techniques for identifying currency and banking crises, those techniques mainly examine the crises as isolated crises. However, as the financial system is interconnected, identifying a crisis as an isolated crisis may lead to incomplete information about overall pressure in the financial system. For example, high pressures in the exchange market can be distributed to the money market, or vice versa, and create a mild pressure in both markets. Thus, the exchange market and the money market may seem stable amid high pressures on the overall financial market, which creates hidden crises [40]. Understanding these hidden crises will be the key to future mitigation policy. Therefore, there are significant benefits in considering currency and banking crises as a twin crises, rather than isolated crises.

To understand the model, let us start with the monetary equilibrium as follows:

$${\mathit{M}}^{\mathit{s}}={\mathit{M}}^{\mathit{d}}$$E10

where:

M^s = total money supply issued by the Central Bank.

M^d = total demand for money.

On the one hand, money is supplied by the central bank, which creates money by buying foreign reserves (F_t) and domestic assets (D_t). On the other hand, the demand for money can be represented as a function of price (P_t), income (Y_t) and interest rate (R_t). Thus, eq. (10) can be rewritten as:

$$F_t+D_t=P_t{Y}_t^{{\beta }_t}exp\left(-{\alpha }_t{R}_t\right)$$E11

where:

${\beta }_t$= income elasticity >0 at time t.

${\alpha }_t$= interest rate coefficient > 0 at time t.

As the money created by buying foreign reserves can be measured by multiplying foreign reserves by the exchange rate (${\mathit{ER}}_t$), eq. (11) can be represented as follows:

$$\left(F_t.{\mathit{ER}}_t\right)+D_t=P_t{Y}_t^{{\beta }_t}exp\left(-{\alpha }_t{R}_t\right)$$E12

The real measure of monetary equilibrium can be obtained by deflating the changes in the money supply by the total money supply created by the Central Bank.

$$f_t+d_t={\pi }_t+{\beta }_t{y}_t-{\alpha }_t{r}_t$$E13

where:

$f_t$ = ${\mathit{FR}}_t$ .${\mathit{ER}}_t$ /M_t.

$d_t$= ∆$D_t$ /M_t.

π_t = ∆P_t/P_t.

y_t = ∆Y_t/Y_t.

r_i (t) = ∆R_t/d_t.

Eq. (13) can be re-arranged to separate financial system and macroeconomic indicators as follow:

$$f_t+d_t+{\alpha }_t{r}_t={\pi }_t+{\beta }_t{y}_t$$E14

In an open economy, the monetary equilibrium is affected by other countries. Thus, the interaction between country a and country b can be determined by employing the International Fisher Effect to the monetary equilibrium:

$$\left(f_a-f_b\right)+(d_a-d_{b)}+\alpha \left(r_a-r_b\right)=\left({\beta }_a{y}_a-{\beta }_b{y}_b\right)+\left({\pi }_a-{\pi }_b\right)$$E15

By adjusting the change of foreign reserves and the change of price by the rate of appreciation of currency a in terms of currency b ($e_{\mathit{ab}}$), eq. (15) can be rewritten as:

$$f_a+e_{\mathit{ab}}+d_a+\alpha r_a=\left({\beta }_a{y}_a+{\pi }_a\right)+\left(f_b+d_b+\alpha r_b-{\beta }_b{y}_b-{\pi }_b+e_{\mathit{ab}}\right)$$E16

The ($f_a+e_{\mathit{ab}}+d_a+\alpha r_a$) is referred to as the Financial Market Pressure, $\left({\beta }_a{y}_a+{\pi }_a\right)$ as the Macroeconomic Pressure, and $\left(f_b+d_b-{\beta }_b{y}_b-{\pi }_b+e_{\mathit{ab}}+\alpha r_b\right)$ as the External Pressure. Thus, in the final form, eq. (16) can be represented as:

$$\mathit{\text{Financial Market Pressure}}=\mathit{\text{Macroeconomic Pressure}}+\mathit{\text{External Pressure}}$$

The above model shows that, in a close economy, the relationship between the financial market pressure and the macroeconomic pressure should be one-on-one. However, external pressure may force the relationship to break apart.

The financial market pressure model suggests that the twin currency and banking crisis is characterised by a sharp depreciation of the exchange rate and the rise of the interest rate. To address the issues, central banks may sell their foreign reserve to stabilise the exchange rate and expand the base money. Thus, the exchange rate and the interest rate may seem stable amid high pressure in the financial system. In order to measure the total pressure in the financial market, the change in the exchange rate and the interest rate should be offset by the central bank’s intervention in the exchange market and domestic money market.

4. Determinants of currency and banking crises

Following the models to identify currency and banking crises that have been described in the previous section, many empirical studies have been made to find the determinants of the currency and banking crises.

There are two popular methodologies to investigate crises. First, the multivariate probit/logit model is arguably the most popular methodology to analyse currency and banking crises [41]. This model uses the event of crisis as a dummy dependent variable with a value of one if there is a crisis and a value of zero if there is no crisis. As independent variables, a set of macroeconomic indicators is used. These binary models are occasionally also combined with the panel method when investigating a large sample of countries. One advantage of the method is that asymmetries and other non-linearities can be straightforwardly tested [42]. However, despite their popularity, binary models fail to provide useful forecasts [43].

The second strand of literature uses a non-parametric methodology to predict the currency and banking crises. One commonly used non-parametric methodology to examine currency crises is the signalling method [33]. After the crises are identified, the threshold of each variable is determined. A signal is flared when a variable exceeds a given threshold level. The variables are then investigated to calculate the correct signal, missing signal, wrong signal, or correctly do not produce a signal. Noise-to-Signal Ratio is then used to understand the ability of variables to predict systemic banking crises. As a lower Noise-to-Signal Ratio represents a low frequency of false signals, thus, the threshold level can be adjusted to find the lowest Noise-to-Signal Ratio (Table 1).

	Crisis	No Crisis
Signal was issued	Correct Signal (A)	False Signal (B)
No Signal was issued	Missing Signal (C)	Correctly no signal (D)

Table 1.

The classification of signals.

Noise-to-Signal Ratio can be obtained by the following formula:

$$\mathit{\text{Noise}}-\mathit{\text{to}}-\mathit{\text{SignalRatio}}=\frac{\frac{B}{\left(B+D\right)}}{\frac{A}{\left(A+C\right)}}$$E17

The signalling method is considered the most successful method to forecast financial crises [43]. However, the signalling method has one main drawback. It evaluates the variables individually. Thus, we need to create a composite index to measure the result. However, it is difficult to interpret the index as it is highly variable [44].

Furthermore, the most recent study employs innovative techniques such as Markov switching models [45], artificial neural networks and genetic algorithms [46], and binary recursive trees [47].

In general, the above methodologies find that the currency and banking crises are typically preceded by a real appreciation and a lending boom [35, 48]. Those two variables are signs of a boom period in the business cycle.

In a boom period, the economy typically enjoys high growth, high export and large capital inflow. High capital flow is usually dominated by hot money which is invested in portfolio instruments such as stocks and bonds. Thus, stock and bond prices start to increase [49]. On the other hand, these also lead to a real appreciation of currency [50, 51].

If real appreciation continues, exporters start to lose competitiveness which leads to decreasing export, increasing imports and a current account deficit. On the other hand, overvalued currency also provides an incentive for investors to attack the currency. Thus, the economy fundamentally becomes fragile.

Funded by capital flows, banks start pushing their lending, leading to a significant increase in speculative financing. On the other hand, to avoid the adverse effect of real appreciation, the central bank starts to intervene by buying foreign currency and selling domestic currency. Both foreign reserves [52] and domestic money supply increase. Abundant liquidity encourages banks to push their lending and creates a lending boom. The bank’s liquidity ratio starts to decrease, and the banking system becomes weaker.

Current account deficit pressures currency to depreciate. If foreign investors start pulling out their money, the currency depreciates faster, along with the fall of asset prices [53]. Furthermore, liquidity becomes tight, and interest rates are increasing.

Soon, firms and households will have difficulty paying the loan. Current account deficit and high non-performing loans will lead to banking crises and massive capital outflow. As a result, the currency will crash.

As fast currency depreciation is devastating, thus, the central bank tries to intervene to smooth the volatility (in the free float rate regime) or to defend the currency (in the fixed-rate regime). The success of the central bank’s intervention depends on two things: the amount of foreign reserve and the amount of domestic liquidity. Even though the central bank collects sufficient foreign reserves during a boom period, the intervention may fail if there is not enough domestic currency in the market to be bought (Figure 3).

Figure 3.

The cycle of currency and banking crises.

The above relationship also shows that there is a physical link between the currency crisis and the banking crisis, as there is a vicious cycle in economic activities. The growth may invite capital flows. However, the capital flow also stimulates growth. On the other hand, the currency attack encourages investors to withdraw their money to fund the attack. Thus, the bank run is inevitable. However, in the event of a bank run, many investors reinvest their fresh cash speculatively in foreign currency. Thus, the banking and currency crises reinforce each other in a vicious cycle. The vicious cycle suggests that the initial crisis in the twin currency and banking crisis is hard to determine.

In an empirical study, the lending boom is often represented as financial sector indicators (e.g. M2 multiplier, domestic credit/GDP, real interest rate), and real appreciation is often represented as external sector indicators (e.g. export, term of trade, real exchange rate, import, international reserve).

5. Conclusions

Currency and banking crises have centuries of history, yet it is still difficult to mitigate future crises. The origins of currency and banking crises can be divided into three groups, which are rational actions, panic and contagion effect.

The first-generation model of currency crises started in the early 1980s with the speculative attack model which suggests the rational expectations of investors as the source of currency crises. In this model, investors are assumed to doubt the government’s ability to manage a fixed exchange when there is a current account deficit. When the foreign reserve is drying in order to keep the exchange fixed, investors will attack the currency, leading to the breakdown of the fixed exchange rate regime.

When the speculative attack model could not explain the de facto breakdown of the European exchange rate mechanism which led to currency crises in Europe in 1992–1993, the second-generation model of currency crises which focuses on the self-fulfilling model emerged. In this model, the herd behaviour of investors may cause panic and lead to the withdrawal of assets. As a result, the exchange rate tends to depreciate and translates into a crisis. Many aspects of herd behaviour such as coordinated action of investors, sequential observation of other investors’ actions and information cascade are investigated. However, herd behaviour is not the whole story since investors are unlikely to ignore their new information where potential capital gain does not depend on other investors’ actions.

On the other hand, early generations of banking crises focused on random withdrawal. This model shows that depositors can do a bank run due to random events such as sunspots or economic projections due to the lack of information held by the depositors.

Unclear triggers of the bank run in the random withdrawal model encourage the emergence of the information-based model. This model shows that the bank run is a logical consequence of a rational change of risk in bank portfolios.

However, in the aftermath of the Asian financial crises of 1997–1998, new models emerge that claim currency and banking crises can still occur in the absence of panic and a well-perform economy due to contagion effects.

According to the model, failure in one bank can spread to the whole banking system through money market currency and banking crises still can occur in the absence of panic and a low-risk environment due to a contagion effect.

The systemic risk claims that interbank lending increases the systemic risk for banks. Therefore, failure in one bank can lead to the failure of many other banks. Furthermore, due to the interconnectedness of financial markets, the twin crisis model shows that a currency crisis can easily translate into a banking crisis or vice versa.

In terms of the identification of crises, early studies of currency crises generally use the depreciation of the exchange rate at a certain level as a basis for determining the crisis. However, this may be biased when the central bank intervenes so that the exchange rate does not depreciate despite considerable pressure on the currency. For that reason, most recent studies use Exchange Market Pressure Index as the basis for the determination of currency crises. This model illustrates that the pressure on the exchange rate is not only reflected in the depreciation but also in the amount of central bank intervention through the spot market (and sometimes through interest rate).

On the other hand, the definition of a banking crisis is more complicated than a currency crisis. There are various methodologies to define a banking crisis. These methodologies consider various factors such as bank performance, government bailout, widespread bank failures, extensive bank runs and professional analysis to specify bank crises. To address the complexity of the event approaches, the money market pressure index was developed to help determine the banking crisis.

In terms of methodology, while the multivariate logit model is arguably the most popular methodology for analysing currency and banking crises, it fails to provide a good forecast. On the other hand, while the signalling method is considered the most successful method to forecast financial crises, it is difficult to interpret the result as it is highly variable. The most recent study employs innovative techniques such as Markov switching models, artificial neural networks and genetic algorithms, and binary recursive trees. However, while they are much more complicated, the projection powers are still not significantly improve.

Furthermore, the empirical studies suggest that the currency and banking crises are typically preceded by a real appreciation, which is often represented as financial sector indicators (e.g. M2 multiplier, domestic credit/GDP, real interest rate) and a lending boom, which is often represented as external sector indicators (e.g. export, term of trade, real exchange rate, import, international reserve).

Acknowledgments

Thank you to Professor Catarina Figueira for her useful comments and guidance on this work.