Models of Economic and Financial

Crises*

 

 

 

 

ROBERTO S. MARIANO

 

BULENT N. GULTEKIN

 

SULEYMAN OZMUCUR

 

TAYYEB SHABBIR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

R.S. Mariano is Professor of Economics and Statistics at the Department of Economics at the University of Pennsylvania; B.N. Gultekin is Professor of Finance at the Wharton School, University of Pennsylvania; S Ozmucur is Professor of Economics and Econometrics at the Department of Economics at Bogazici University, also  Department of Economics at the University of Pennsylvania; T. Shabbir is Chief of Research at the Pakistan Institute of Development Economics and Visiting Professor at the Department of Economics, University of Pennsylvania.

 

  * Paper to be presented at the Middle East Economic Association in the ASSA Meetings, January 7—9, 2000, Boston, MA.  The authors want to thank Sean Campbell for writing the FORTRAN algorithm, Professor Emre Alper of  Bogazici University and Boragan Aruoba of University of Pennsylvania for data collection. Financial support provided by the Economic Research Forum for the Arab Countries, Iran and Turkey (ERF Research Project No:  ERF99-US-4004) is also gratefully acknowledge.

 

 

 

1.0 Introduction

 

The decade of the 1990s was certainly marked by a rather unusual number of financial and economic crises. Amongst the prominent such crises were the near breakdown of the Exchange Rate Mechanism of the European Union in 1992-93, the Mexican (“Tequila”) Crisis in 1994-95 and, finally, the near catastrophic South East Asian Crisis of 1997 with its contagion influence spreading to Russia and South America in 1998. It was but natural that these crises should give rise to a number of important questions. A partial list of such queries might include questions regarding the causes of such financial and economic crises, an historical comparison of the causes as well as the impact of such crises, the contagion effects of 1the 1997 Asian Crisis, the ‘optimal’ level, speed and sequencing of financial liberalization especially that of global capital flows and, finally,  the appropriate role of macroeconomic management. Also, in the wake of these recent episodes, the very important question of the need and the feasibility of predicting of such crises has been brought to the forefront.

 

The objective of the present paper is to focus primarily on the last issue enumerated above, namely, the modeling of  an appropriate early warning system for the various financial and economic crises. Of course, the major impetus for this analysis has been provided by the near cataclysmic and rather unexpected financial and economic crisis that exploded in the summer of 1997 in the hitherto model economies of the South East Asia[1]. Following this crisis, several studies have explicitly dealt with the question of devising a viable “Early Warning System”. Obviously, such a system would have a great appeal for policymakers who may want to be ready with a hopefully effective set of preemptive measures for possible future use.

 

In terms of the structure of the remaining paper, following a brief discussion of the kinds of crises of interest, an overview of the relevant methodologies employed by the existing studies of the Early Warning Systems, this paper argues for Markov Switching modeling as a new methodological approach to the issue of predicting financial and economic crises. A prototype Markov Switching Model is then applied to the case of Turkey and its empirical results are then discussed. The paper concludes with some thoughts on the policy implications in the light of our model’s results.

 

2.0    Types of Crises

 

The different types of economic and financial crises that may be of interest range from the “garden variety” currency crises to banking crises, international debt crises and asset prices collapse such as in the stock or the real estate market.  In fact, IMF’s World Economic Outlook (1998) offers a succinct and useful characterization of the relatively more important types of crises.  It essentially identifies three kinds of crises: currency crises, banking crises (which may lead to systemic financial crisis with spillovers to real sector) and foreign or external debt crises.  A brief description of each follows:

 

According to the IMF’s definition, a currency crisis not only constitutes an actual, observed “severe” devaluation, but also includes cases where authorities are apparently able to fend off actual devaluation but only after substantial increase in domestic interest rate and/or expending international reserves.

 

The banking crises are perhaps the hardest to identify but are proxied by a large increase in the ratio of non-performing loans, illiquidity and impending or actual insolvency of banking and credit institutions.  These banking crises on occasion may degenerate into the “systemic financial crises” which are “potentially severe disruptions of financial markets that, by impairing markets’ ability to function effectively, can have large adverse effects on the real economy”.

 

Finally, the foreign debt crisis, somewhat of a relic of the 1980s, is characterized by “excessive” sovereign debt burden as measured in terms of ‘national ability to pay’ as indicated by foreign debt to GDP or foreign debt service to export earnings of a country.

 

The various types of crises described above not only differ in terms of their characterization but also other relevant aspects. For instance, the banking crises tend to precede currency crises, but they could occur simultaneously with systemic financial crises as in the case of the Asian Crisis of 1997-98.  The currency, banking and financial crises tend to be ranked in that order in terms of the severity as measured by lost output and time for recovery.[2]  Again, there may be important social costs besides the traditional financial costs of such crises (Shabbir (1999)).

 

While recognizing the important differences that characterize these various kinds of crises, it may be noted that studies of such crises exhibit methodological commonalties, which might as well be highlighted here.

 

3.0       Review of The Relevant Literature

 

Below we discuss in turn the relevant empirical and theoretical literature regarding the various kinds of financial crises.

 

 

 

 

 

 

3.1    Review of Empirical Literature

 

There are essentially two alternative methodologies that have been employed in the empirical studies of the early warning systems for different kinds of crises.

 

(a)      The relatively more popular approach is to use probit or logit models.  (As illustrated by Eichengreen and Rose (1998) for currency crisis, Demirguc-Kunt and Detragrache (1998) for prediction of banking crises and Edwards (1984) for an analysis of the determinants international debt reschedulings.)

 

(b)     Alternatively, the methodology adopted by Kaminsky and Reinhart (1996), and Kaminsky, Lizondo and Reinhart (1998) is known as the “signals” approach which essentially optimizes the signal to noise ratio for the various potential indicators of crisis.

 

 

Here we will briefly discuss only two representative studies – the first one dealing with currency crises which employs the ‘signals approach’ and the second one with banking crises that uses a logit/probit framework. [3] (Both these approaches are discussed in some detail later in this paper on the section on ‘Methodology’).

 

The representative study in the ‘signals’ approach genre is by Kaminsky et al. (1998) which presents a review of literature, discusses methodological issues and, in particular, reports on and extends Kaminsky and Reinhart (1996), who examine 76 currency crises from a sample of  fifteen developing and five industrial countries during 1970-1995. 

 

The so-called ‘signals’ model essentially involves monitoring the evolution of several key economic and financial variables or indicators that tend to exhibit an unusual behavior in the periods just preceding a crisis. More specifically, Kaminsky et al. (1998) examine over a dozen candidate indicators such as the real exchange rate, stock prices, natural output and bank deposits as the potential precursors of currency crises.  They analyze these indicators using the criteria to minimize the ratio of noise (“bad signals as a percentage of potential bad ones”) to signal (“good signals as a percentage of potential good ones”). They also judge these individual indicators in terms of their predicted lead time and persistence of their signals prior to the onset of the crisis.  More importantly, Kaminsky et al. (1998) chose the “best” two indicators in order to construct their “Leading Index of Currency Crises” by using a weighted average of monthly (year on year) percentage changes in the real exchange rate and the (negative of) monthly (year on year) percentage changes in gross international reserves.  The weights are chosen so as to equalize the conditional variance of these components.  This composite index is dubbed as the index of “exchange market pressure”.[4]  The periods in which this index exceeds its mean by more than three standard deviations are defined as crises periods.

 

3.1.1        Determinants of Currency Crises – Empirical regularities

 

The findings of the empirical regularities as culled from the twenty five studies that are reviewed by Kamisky et al. (1998) are reported in the following Table 1 where the various potential indicators or determinants of the currency crises have been grouped into broader classes such as those representing Capital Account or Financial Liberalization.

 

Table 1 shows the number of studies in which the particular indicator was found to be significant in at least one of the tests conducted. An assessment of the overall results as summarized in the above table does not provide “a clear-cut answer concerning the usefulness of each of the potential indicators of currency crisis”. This is in large part due to the disparate nature of the studies in terms of their relevant factors considered in the specification for the different studies, procedure to measure those variables and periodicity of the data. Also, at times, even when some variables were significant in the univariate tests they fail to be significant in the multivariate tests. However, on the positive side, Kamisky et. al (1998) favor drawing the following tentative conclusions from these group of studies:

 

1.      Given the fact, that the currency crises may be preceded by multiple economic and even political problems, the modeling of currency crisis prediction should involve a relatively broad range of indicators.

 

2.      The variables that receive ‘ample’ support as useful predictors of currency crises include: international reserves, the real exchange rate, credit growth, credit to the public sector and domestic inflation. The results also lend support for including the trade balance, export performance, money growth rate, M2/International reserves, real GDP growth and the fiscal deficit as potential early warning indicators. On the other hand, the variables associated with the external debt profile or the current account balance did not fare well.

 

 

As mentioned earlier, besides presenting a relatively comprehensive  review of the some of the earlier work related to prediction of currency crisis, Kamisky et al (1998) also presents an extension of previous work which employs the ‘signals’ approach to identifying and predicting currency crises. Based on empirical results for a sample of fifteen developing countries and five industrial ones during 1970-95, the authors report that the variables with the best track record in anticipating crises include output, exports,

 

 

Table 1.  Performance of Indicators

                                                                                                Number of Studies              Statistically Significant

Sector                                    Variables                                     Considered                                    Results

Capital account                    international reserves                          11                                                 10

                                                short-term capital flows                      2                                                   1       

                                                foreign direct investment                    1                                                   1

                                                capital account balance                      1                                                   --      

                                                domestic-foreign interest

                                                   differential                                          2                                                   1       

 

Debt-profile                           foreign aid                                             1                                                   --

                                                external debt                                         1                                                   --

                                                public debt                                            1                                                   1       

                                    share of commercial bank loans         1                                                   --

                                                share of concessional loans               1                                                   1

                                                share of variable-rate debt                  1                                                   --

                                                share of short-term debt                     2                                                   --

                                                share of multilateral develop-

                                                   ment bank debt                                  1                                                   --

 

Current account                   real exchange rate                                12                                                 10

                                                current account balance                     6                                                   2

                                                trade balance                                        3                                                   2

                                                exports                                                   3                                                   2

                                                imports 1/                                              2                                                   1

                                                terms of trade                                        2                                                   1

                                                export prices                                         1                                                   --      

                                                savings                                                  1                                                   --

                                                investment                                            1                                                   --

 

International                         foreign real GDP growth                     1                                                   --

                                                foreign interest rates                           3                                                   1

                                                foreign price level                                2                                                   1

 

Financial liberal-

   ization                                 real interest rates                                  1                                                   1

                                                credit growth                                        7                                                   5

                                                lending-deficit interest                       

                                                   spread                                                 1                                                   --

                                                money multiplier                                   1                                                   1

 

Other financial                      parallel market premium                       1                                                   1

                                                central parity                                         1                                                   1

                                                position within the band                     1                                                   1

                                                money demand-supply gap                1                                                   1

                                                change in bank deposits                     1                                                   --

                                                central bank credit to banks               1                                                   1

                                                money                                                    3                                                   2

                                                M2/international reserves                  2                                                   2

 

 

 

 

Table 1.   Performance of Indicators (concluded)

                                                                                                Number of Studies              Statistically Significant

Sector                                    Variables                                     Considered                                    Results

Real sector                            inflation 2/                                             5                                                   5

                                                real GDP growth or level                     8                                                   5

                                                output gap                                            1                                                   1

                                                employment/unemploy-                     

                                                   ment 3/                                                3                                                   2

                                                change in stock prices                        1                                                   1

 

Fiscal                                      fiscal deficit                                           5                                                   3

                                                government consumption                  1                                                   1

                                                credit to public sector                         3                                                   3

 

Institutional/                         multiple exchange rates                       1                                                   --

Structural                               exchange/capital controls 4/              2                                                   1

                                                openness                                               1                                                   1

                                                trade concentration                             1                                                   --

                                                banking crisis                                       1                                                   1

                                                financial liberalization                          2                                                   1       

                                                months spent on peg                          1                                                   1

                                                past foreign exchange

                                                   market crisis 5/                                   1                                                   1

                                                past foreign exchange

                                                   market event 6/                                  1                                                   --

 

Political                                  government victory                             1                                                   --

                                                government loss                                  1                                                   1       

                                                legal executive transfer                       1                                                   1

                                                illegal executive transfer                     1                                                   1

1/  In the statistically significant results, the rate of growth of imports declines prior to a devaluation.

2/  In one of the statistically significant results, an increase in inflation reduces the probability of an attack.

3/  In one of the statistically significant results, an increase in employment increases the probability of an unsuccessful attack.

4/  In the statistical significant result, the presence of capital controls increases the probability of an unsuccessful attack and reduces the probability of a successful attack.

5/  A past foreign exchange market crisis reduces significantly the possibility of an unsuccessful attack, and increases marginally the possibility of a successful one.

6/  Events include significant changes in exchange arrangements (such as devaluations, revaluations, decisions to float, and widening of exchange rate banks); crises overlap with events but include unsuccessful speculative attacks and excludes changes in exchange arrangements not associated with market pressure.                                                                     

 

 

 

 

 

 

real exchange rate deviations, equity prices and the ratio of broad money to gross international reserves.

 

Incidentally, Sami (1999) uses a variant of the Kaminsky et al signaling approach in order to analyze the Egyptian case. Based on monthly data from 1961:01 to 1999:03, Sami calculates the conditional probabilities for each value of a composite weighted index whose components are percentage change in real exchange rate and percentage charge in international reserves.[5] A crisis is characterized by a situation where the value of the above index exceeds its mean by two standard deviations.[6]

           

Further, in a similar vein as in Kaminsky et al, the IMF considers certain specific indicators of macroeconomic and financial vulnerabilities with attendant heightened probability of crisis occurrence.  In the context of the onset of the Asian Crisis of 1997-98, the IMF (ex-post) identifies a set of indicators (Table 2) which may have been present in the crisis countries in Southeast Asia (Indonesia, Korea, Malaysia, Philippines and Thailand).  It may be noted that while there is some overlap in terms of these ‘indicators of vulnerability’ and the ‘crises predictors’ identified by Kaminsky, et al. (1998), the IMF set is much broader in scope and more cautious regarding our ability to pinpoint a crisis as against merely being vulnerable to one.

 

 

 

3.1.2        Determinants of Banking Crises – Empirical Regularities

 

On the other hand, the representative study that uses the logit/probit framework is by Demirguc-Kunt and Detragiache (1996) and it deals with predicting banking crises.  Based on observations for 1980-94 for a large sample of developed and developing countries, it reported that banking crises tend to occur when the macroeconomic environment is weak especially when growth rate of GDP is low and inflation is high.  Also, high real interest rates, balance of payment deficits and presence of deposit insurance scheme were found to be significant precursors of banking crises.

 

It may be noted that both in the case of the studies of the currency as well as the banking crises, it may be noted that differences in methodologies, time periods covered and selection of countries, as well as disparate definitions of exchange market pressure, pose special challenges for arriving at generally applicable conclusions as to what set of leading indicators of currency and banking crises are likely to prove the most useful?  However, it is still possible to arrive at some tentative conclusions about the generally useful indicators of vulnerability.  Thus, currency crises tend to be preceded by an

 

 

Table 2.  Selected Indicators of Vulnerability

(end-1996)

 

 

________________________________________________________________________

 

Indicator*                                                         Indonesia      Korea      Malaysia      Philippines      Thailand

 

 

Macro Indicators                                

 

 

Inflation >5%                                                           Yes                 No                No                           Yes              Yes   

Fiscal deficit >2% of GDP                                      No                   No                No                           No                 No

Public debt >50% of GDP                                      No                   No                No                           Yes               No

Current account deficit >5% of GDP    No                   No                No                           No                 Yes

Short-term flows >50% current                             

   account deficit1                                                    Yes                Yes               Yes                         Yes               Yes

Capital inflows >5% of GDP                                  No                  Yes               Yes                         Yes               Yes

Ratio of short-term debt to international

   reserves >1 2                                                         Yes                Yes               No                           No                 Yes

 

 

Financial Sector Indicators

 

 

Recent financial sector liberalization                   Yes                Yes               No                           Yes               Yes

Recent capital account liberalization                   No                  Yes               No                           No                  No

Credit to the private sector >100% of GDP         No                  Yes               Yes                         No                 Yes

Credit to the private sector,

   real growth  >20%                                                No                   No                Yes                         Yes                No

Emphasis on collateral when making

   loans                                                      Yes                Yes               Yes                         Yes                Yes

Estimated share of bank lending to the

   real estate sector >20% 3/                                   Yes                Yes               Yes                         Yes                Yes

Stock of nonperforming loans >10% of

   total loans                                                              No                  No                No                           No                   No

Stock market capitalization (as percent

   of GDP)                                                                   40                   30 310                          98                    56

 

 

Source:  IMF, International Financial Statistics; World Economic Outlook database; World Bank, IFC Emerging Market database.

 

*The cut-off points are based on the relevant literature that attempts to predict currency and banking crises (Kaminsky, Lizondo, and Reinhart (1997) for currency crisis, and Hardy and Pazarbasioglu (1998) for banking crisis).

 

1 Defined as the sum of net portfolio and other investments in the financial accounts.

 

2 As of June 1997.

 

3 At end-1997, includes indirect exposure through collateral.

 

overvaluation of the real exchange rate, rapid domestic credit growth, expansion of credit to the public sector, a rise in the ratio of broad money to foreign exchange reserves, an increase in the domestic inflation rate, a decline of FDI flows, and an increase in industrial country interest rates.  Less important factors in this regard are a widening of the trade deficit, and increase in the fiscal deficit, deterioration in export performance, as well as a slowdown in real GDP growth.  It may be noted that current account and fiscal deficits do not seem to emerge as important indicators.[7]  On the other hand, regarding the banking crises, these are often preceded by large inflows of short-term capital (‘hot money’), rapid expansion of domestic credit (which may result from inadequately sequenced and/or supervised) financial liberalization, recessions, and declines in asset prices such as stocks and real estate.  The various case studies suggest that often financial sector liberalization without adequate prior strengthening of the regulatory structure not only sets the stage for a banking crisis but also makes it more difficult to cope with it if one erupts.

 

3.2      Review of Theoretical Literature

 

Using the case of the currency crises as an important illustration of the financial crises in general, this section presents a brief overview of the theoretical literature on the causes of currency crises with a special reference to identifying the potential early warning indicators.

            The historical development of the theoretical literature can be grouped in three “generations” of models --- each reflecting the distinct mechanism that is espoused as the major cause of such crises. We will discuss these models in turn.

 

3.2.1 First Generation or ‘Fundamentals’ Models

 

Epitomized by Krugman(1979), the first generation models tend to focus on the role of economic and financial ‘fundamentals’ such as the unsustainable fiscal policies in the face of the fixed exchange rate as the major cause of an eventual currency crisis. Given a fixed exchange rate regime, the persistent need to finance government budget deficits through monetization would surely lead to a reduction in the international reserves held by the Central Bank. Since such reserves are finite the speculative attack on the currency is the eventual outcome of this scenario.[8]

 

This rather simple model suggests certain ‘fundamental’ imbalances such as the gradual decline in international reserves, growing budget deficits and domestic credit growth as the potential early warning indicators of speculative attacks. As noted in Abiad(1999), in addition to these indicators, other models in the spirit of the first generation models suggest current account deficit and real exchange rate overvaluations as additional early warning indicators. These reflect the alternative mechanisms that will force the monetary authority to eventually abandon the peg in the face of an expansionary fiscal policy either by leading directly to a worsening of the current account through a rise in the import demand or indirectly through a rise in the relative price of the nontradables (and the subsequent overvaluations of the real exchange rate).8

 

3.2.2        Second Generation or ‘Self-fulfilling Prophecy’ Models

 

The development of the so-called Second Generation models of the currency crises were motivated by the EMS currency crisis in 1992-93 where some countries such as the UK and Spain suffered crises despite having adequate international reserves, manageable domestic credit growth and non-monetized fiscal deficits --- characteristics that ran counter to the necessary conditions asserted by the first generation models. Obstfeld (1994) and Krugman (1998) addressed the concerns raised by these counter-examples.

 

            The main innovation of these second generation models lies in identifying the role that the ‘expectations’ of the market agents may play in precipitating currency crises. More specifically, Krugman (1998) notes three elements that characterize the second generation models of such crises. First, there must be reasons such as a preference for price stabilization for which the government would want to defend the peg. Second, at the same time there must be reasons why the government would like to end the peg, say, for the sake of stabilization policy to combat high unemployment. Finally, these models maintained that the relative cost of defending a peg is directly proportional to the people’s expectations that such a peg may be abandoned.[9]

 

                In these second generation models, several aspects of the key role played by the expectations may be noted. First, in these models the government policy both affects expectations and is affected by it ---- this simultaneity gives rise to the possibility of ‘self fulfilling prophecy’ aspects of these models and may generate multiple equilibria. The second interesting aspect of this debate, which has not yet been fully explored in the literature, is concerned with the exact manner in which these expectations are formed or determined. If these expectations are purely psychological based or are formed in the ideal environment of perfect information (regarding the government objective function) and perfect coordination (amongst market participants), one may have difficulty in finding a “tight relationship  between fundamentals and crises as sometimes crises may take place without a previous significant change in fundamentals”.[10] However, if, as is likely, market participants have incomplete information about the government’s objective function or its ability to defend the fixed exchange rate and suffer from imperfect coordination (due to important non-linearities in market coordination), fundamental factors such as the current account deficit or credit growth can be the important determinants of the expectations and thus (albeit indirectly) of the likelihood of the occurrence of crises. Thus the second generation models, couched in terms of a government’s objective function and peoples expectations, suggest such early warning indicators of currency crises as the high unemployment, inflation, current account deficit, financial sector vulnerability and domestic debt.[11]

 

3.2.3 Third Generation or ‘Contagion’ Models

 

The third generation models are based on the notion of ‘contagion’ where the mere occurrence of a crisis in one country increases the likelihood of a similar crisis elsewhere. As described in Masson (1998), three related scenarios can be identified to represent the paradigm of contagion:  ‘monsoonal effects’, ‘spillover effects’ and ‘pure contagion effects’.

 

The monsoonal effects refer to the case when a common external shock affects all the countries in a region or a group. For instance, the oil price shock (in the 1970s), unexpected increase in the world interest rate (as in the sovereign debt crisis of the 1980s) and more recently, appreciation of the reference currency (U.S Dollar) relative to the other key currencies (Yen, say) which played a big part in the Asian Crisis of 1997. On the other hand, in the case of the ‘spillover effects’ , an unexpected devaluation in a given country may adversely affect the relative international competitiveness of a group of competitors thus initiating a chain reaction in terms of a series of currency devaluations in the affected countries. (For instance, Thai Baht’s devaluation made the exports of the competitors in the region relatively more expensive and raised the speculation of ‘competitive devaluations’ in Malaysia, Indonesia and South Korea). Finally, in the case of the ‘Pure Contagion Effects’, the foreign investors may decide to color all the countries which are perceived to be in a similar situation as the initial victim of a currency crisis with the same (negative) brush—this is tantamount to a ‘herd behavior’. In terms of the potential indicators of early warning suggested by the third, generation models, some of the viable indicators will be variables such as the ‘world interest rate, growth rate of trading partners and measures of relative international competitiveness in the export sector.

 

 

4.0 Methodology

 

As mentioned earlier, the recent efforts at devising an early warning system for an impending financial crisis have taken the form of two related approaches.  The first approach estimates a probit or logit model of the occurrence of a crisis with lagged values of early warning indicators as explanatory variables.  This approach requires the construction of a crisis dummy variable that serves as the endogenous variable in the probit or logit regression.  Classification of each sample time point as being in crisis or not depends on whether or not a specific index of vulnerability exceeds an arbitrarily chosen threshold.  For example, for currency crises, the index of vulnerability is sometimes based on a weighted average of percentage changes in nominal exchange rates, gross international reserves and short-term interest rate differentials (e.g. local versus US rates when dealing with crises in the Philippines).  Explanatory variables typically would be variables in the real sector of the economy, financial variables, external sector and fiscal variables.  This approach has the advantage of providing a framework for statistically measuring the magnitude and significance of the effects of various potential explanatory variables on the onset of a crisis.  The estimated model also allows the estimation of the probability of occurrence of a crisis in the future given projected or anticipated values of the explanatory variables.  Negative aspects of the approach partly derive from the following:

 

1.      The model does not address the independence of crisis occurrence from period to period – except indirectly through serial correlations that exist in the explanatory variables.

 

2.      Additional serial correlations may even be introduced inadvertently through the explicit manner in which the crisis dummy variable is constructed.  For example, the use of exclusion windows (where the crisis variable automatically is set to zero for k periods immediately following a time point rated to be in crisis) establishes perfect correlation between a crisis time point, and the next k periods following it.  In general, any serial correlation in the crisis dummy variable which is not taken into account in the probit or logit regression would cause the estimates of the model to be inconsistent.

 

3.      Another source of inconsistency:  errors in the construction of the crisis dummy variable leading to misclassification of time points – either a false signal of a crisis or a missed reading of a crisis.

 

4.      The method does not provide a direct measure of the weakness or intensity of the signal of each explanatory variable regarding the onset of a crisis.

 

 

The second method uses a signaling approach to get a more direct measure of the importance of each candidate explanatory variable.  The approach constructs a similar binary variable from each explanatory variable – thus imputing a one (for crisis) or a zero (no crisis) signal from each explanatory variable at each point in time in the sample. A signal-to-noise is then computed for each explanatory variable over the whole sample period – as a quantitative assessment of the value of the variable as a crisis indicator.  This signal-to-noise ratio is defined as the ratio of the success rate of crisis predictions relative to the false alarm rate.  More specifically, let nij be the sample frequencies (for each explanatory variable) defined as follows:

 

 

 

                        Actual

          Prediction

No Crisis

Crisis

No Crisis

n11

n12

Crisis

n21

n22

 

Then, the signal-to-noise ratio for the indicator variable is

 

            [n22/(n21 + n22)]/[n21/(n11 + n21)].

 

This approach allows a direct ranking of variables as crisis indicators and provides a quick focus on the source of the crisis (assuming an encompassing set of indicators).  But the approach does not take into account strong correlations among indicators, provides no framework for statistical testing or calculation of crisis probabilities in the future, and is still open to misclassification errors that can bias the conclusions of the analysis.

 

In this paper we propose the Markov Switching Model as the new approach to predicting financial crisis and apply it to the case of Turkey. This methodology avoids the potential misclassification errors in the probit data, addresses the serial correlations inherent in crisis occurrence, allows for measuring and testing significance of indicator variables, delivers forecast probabilities of future crises conditional on projected future values of indicator variables, and short-run forecasts of key macroeconomic variables.

 

Our proposed approach constructs a quantitative prediction model that consists of two parts:

 

1.      A Markov chain model of the unobservable financial health of a country, say, St.  We argue that what we observe are indicators of this latent attribute of the country.  Initially, we assume two states:  normal (St=1), and critical (St=0).  We further assume that this Markov chain is of order 1, with transition probabilities that are time varying through dependence on observable indicator variables.  Part of our empirical analysis will deal with identifying the appropriate set of indicator variables, thus identifying early warning indicators of a crisis.  Experimentation starts with the indicator variables suggested by earlier studies.

 

2.      A vector autoregressive (VAR) model of key macroeconomic variables—such as GDP or industrial production, inflation, interest rate and exchange rate.  This VAR model differs from the usual one in the sense that it includes the unobservable state variable, St, as an additional endogenous variable.  With the inclusion of St, we introduce the notion that the VAR system behaves in a different fashion depending on whether financial conditions are normal (St=1) or critical (St=0).  We reflect this in our model by allowing VAR parameters to change in value over time as financial conditions become normal or critical.

 

 

In summary, we are proposing a Markov Switching VAR Model that allows intercepts, lag coefficients and error variances in the VAR model to stochastically switch over time according to the value taken by the Markov chain.

 

The model is described in further detail below.

 

Let St - Markov chain of order 1 with transition probabilities pt and qt.