Transformations  in
Business & Economics

LIQUIDITY RISK MANAGEMENT MODEL FOR LOCAL BANKS
Povilas Aniunas, Gailute Gipiene, Mantas Valukonis, Mindaugas Vijunas

ABSTRACT. The purpose of this research is to create a liquidity risk management model based on VaR methodology and other best practices that are suitable for local banks. Literature analysis led to the formation of liquidity risk management model assumptions. It was found that although liquidity risk management is frequently studied in the academic literature, most of researches are limited only to the proposal of one or several new liquidity indicators. There are only few attempts to manage liquidity risk by using new statistical methodologies (VaR). The balance among sufficient liquidity, safety and profitability should be the long-term goal of the bank. Nonetheless the authors often forget this and focus only on the accumulation of extremely high liquidity buffer. Therefore, there is a lack of models that would systematise the proposals of individual authors and provide complex liquidity risk measurement and management model based on the best practices, which are suitable for the local banks that are unable to carry out the asset securitisation and do not have access to funding from the parent bank. New liquidity risk management model was proposed based on the best practices described by using 10 indicators. It covers the following 4 aspects of sound liquidity management: compliance with the supervisory requirements, liquidity buffers, as well as the assessment of net funding gap and risk indicators. The model was employed to assess the liquidity risk of local Lithuanian bank. Although the bank complies with the prudential requirements set by the supervisory institution, liquidity risk management model showed that the bank has some serious problems. Backtesting confirmed that the confidence interval is sufficient and LaR conservatively predicts negative changes of bank's liabilities .

KEYWORDS:  banking, liquidity risk management, quantitative models, LaR, backtesting, model validation.

JEL classification:  C51, C52, G21, G32.