Determinants of the timing of bank failure in North Cyprus.

Purpose--The purpose of this paper is to investigate the determinants of the timing of bank failure in North Cyprus over the period of 1984-2002 using a discrete-time logistic survival analysis.

Design/methodology/approach--The empirical methodology employed in the paper allows for the determination of the factors that influence the time to bank failure. The model links the time of bank failure to a set of bank-specific factors and macro-environment that may have exacerbated the internal troubles of the financial institutions.

Findings--An empirical examination of the results on survival analysis reveal that the three variables, namely: low asset quality (total loan as a percentage of total assets), low liquidity (total liquid asset as a percentage of total assets), and high credit extended to the private sector (ratio of the private credit to gross domestic product) are the main factors that explain the survival time of banks in North Cyprus.

Research limitations/implications--For further research this paper may better distinguish time to bank failure if it extends the time period and if it uses exchange pressure from Turkey that may have a direct effect on bank failure in North Cyprus.

Practical implications--Nowadays bank failure is an important problem in the world. Using time technique to investigate bank failure will help to learn the factors that determine time to bank failure, which will further help to take precautions and prevent the cost of bank failure.

Originality/value--The analysis would appear to be the first to provide evidence and investigate the time to bank failure in the North Cyprus banking sector.

Keywords Microeconomics, Macroeconomics, Banks, Business failure, Cyprus

Paper type Research paper

1. Introduction

The North Cyprus economy has experienced two banking sector distress periods. The first took place in 1994 and the second took place between the years 2000 and 2002. In 1994, the economic fundamentals in Turkey were deteriorating. Particularly, there was a continuing devaluation of the Turkish Lira (TL), which resulted in a serious currency crisis [1]. As there is a close monetary and economic link between Turkey and North Cyprus, as a consequence of the financial distress experienced in Turkey in 1994, banks in North Cyprus were also affected. In 1994 two banks (namely: Everest Bank Ltd and Mediterranean Guarantee Bank Ltd) were placed under the control of the Turkish Republic of Northern Cyprus (TRNC) Ministry of Finance. Later, these banks had to be bailed out by the government. Mediterranean Guarantee Bank Ltd became a public bank and the Everest Bank Ltd was taken over by a private owner.

In Turkey the International Monetary Fund supported the pegged exchange rate base anti-inflation programme implemented in December 1999. However, after 14 months, the programme had to be abandoned, with the collapse of the TL. In 2000, five banks, namely: the Cyprus Credit Bank Ltd, Cyprus Liberal Bank Ltd, Everest Bank Ltd, Kibris Yurtbank Ltd, and Cyprus Finance Bank Ltd, were put under the Saving Deposit Insurance Fund (SDIF), and then these banks were closed in the year 2001. The bankruptcy of these five banks started a serious banking crisis in North Cyprus. Criminal investigations have been conducted to investigate the management, and the total loss of these five troubled banks was reported to be around 112 trillion TL. Another four banks, namely: Cyprus Commercial Bank Ltd, Yasa Bank Ltd, Tilmo Bank Ltd, and Asia Bank Ltd, were put under the SDIF in 2001, and Cyprus Industrial Bank Ltd was put under SDIF in 2002. Furthermore, Finba Ltd was taken over by Artam Bank Ltd in 2000 and Med Bank Ltd and Hamza Bank Ltd were taken over by Seker Bank Ltd in the years 2001 and 2002, respectively. During the period of 2000-2002 ten financial banks were forced by the Government of North Cyprus to suspend their operation. During 1999 there were 37 surviving banks in North Cyprus. However, towards the end of 2002 ten of these banks were revoked from operation, two banks were taken over by other bank, and only 25 banks remained. The increase in the failure of commercial banks in North Cyprus increased attention on efforts to investigate the determinants of bank failure.

Survival models attempt to explain the duration of a particular event. Graphical methods are useful for displaying data on duration and for preliminary analysis of survival that may suggest the survival patterns of the banks. For this reason, as a first step of survival analysis the study presents the Kaplan and Meier (1958) Product-Limit Estimator, an estimate of the distribution of bank failure duration. This method is especially useful for estimation and graphical survival curves. The next step of survival analysis concerns method that deals with events measured or occurring in discrete-time survival analysis for analysing the length of time until the occurrence of event [2]. In discrete-time methods for modelling the time to an event the functional form can either be a logistic model or complementary log-log model. The methodology employed in this paper is a logistic survival model that uses a latent variable modelling framework. The results obtained in this methodology permit us to verify the determinants of time to bank failure. The results of both the Kaplan-Meier (KM) method and discrete-time logistic survival analysis are obtained by processing a computer software package (STATA 8).

2. Empirical literature

Although extensive literature exists on explaining the probability of bank failure, much less attention has been given to predicting the timing of bank failure. To the best of the author's knowledge Lane et al. (1986), Whalen (1991), Cole and Gunther (1995), Henebry (1996), Laviola et at (1999), Wheelock and Wilson (1995, 2000), and Molina (2002) are the only published empirical studies that offer model to predict the time to failure. In their models they apply Cox proportional hazard (CPH) model (1972) to predictions of time to bank failure.

Lane et al. (1986) apply the CPH model to the prediction of bank failure for 464 banks (130 banks failed) in the USA during the period 1979-1983. Lane et al. classified 21 financial ratios under the five categories of the capital, asset quality, management, earning, and liquidity (CAMEL) rating and applied a stepwise procedure for combining the backward and forward elimination technique both one year and two years prior to bank failure. Results reveal that a proportional hazard model is an effective early warning tool that identifies financial distress prior to the actual failure date. In particular, the ratio of commercial loans to total loans, the ratio of loans to deposits, the ratio of total equity capital to total assets and of operating expenses to operating income significantly determines the one-year ahead of time to failure.

Consistent with Lane el at (1986) and Whalen (1991) use the CPH model to estimate the time to failure in the USA between 1987 and 1990. The main difference between Lane et at and Whalen is their sample selection and time period. Whalen utilized a larger sample of 1,500 randomly selected banks. Moreover, Whalen employed the local economic condition variable. Similar to Lane at at (1986), Whalen suggests that the CPH model provides an effective early warning for banks time to failure. In particular, variables such as the ratio of total loans to total assets, the ratio of net income to average total assets and the ratio of capital to average total assets are significant.

Another study, by Cole and Gunther (1995), investigated the time to failure utilizing a different statistical technique. Using quarterly US data during the period 1985-1992, Cole and Gunther suggest that the factors influencing the probability of bank failure may be different from those explaining the time to failure. Cole and Gunther separate the determinants of bank failure from the survival time of failing banks in the USA using a split-population survival model (log-logistic distribution), logit survival model and CPH model. Findings suggest that when the logit survival model is utilized, capital (ratio of total equity capital to total assets), leverage (ratio of loans to total assets), troubled assets (ratio of non-performing loans to total loans) and net income (ratio of net income to total assets), liquidity (ratio of certificate of deposits to total assets), and asset size are important in explaining the time to bank failure.

Henebry (1997) utilized a CPH model for bank failure prediction in the USA during the period of 1985-1988. The results show that from the selected accounting ratios; the ratio of primary capital to total assets, non-performing loans to total loans and total loans to total assets were significant in predicting time to bank failure.

Wheelock and Wilson (2000) also employed a proportional hazard model in the USA during the period 1984-1993, using variables that are in the context of …