Since the 2008-2009 global financial crisis, VaR (Value-at-Risk) techniques have become critical tools for monitoring and predicting the market risk and liquidity of financial assets. These financial risk modeling techniques, which have been recognized by the Bank for International Settlements (BIS) or the Basel Committee on capital adequacy and bank regulations, measure and prevent any potential losses that arise, not only from securities’ price changes and the interdependence between the different types of assets (stocks, currencies, interest rates or commodities), but also from their negative tail co-movements in bearish market conditions. . 

In the event of a financial crisis or market downturn, adequate liquidity risk modeling is advisable. In fact, the main advantage of VaR models is their focus on downside risk (i.e., the impact of the results of negative tails) and their direct interpretation in monetary terms. 

Nevertheless, particularly in times of financial turbulence, traditional VaR models do not properly consider nonlinear dependence between portfolio assets and become inefficient in illiquid market scenarios. Despite the advances in measurement models, obtaining precise market liquidity risk estimations and applying them to optimize portfolios continues to be a challenge for financial institutions.

Therefore, in the article “Multivariate dependence and portfolio optimization algorithms under illiquid market scenarios”, published in the European Journal of Operational Researchtogether with my colleagues José Arreola, from Rennes Business School (France), Theo Berger, from the University of Bremen (Germany) and Duc Khuong Nguyen, from IPAG Business School (France), we propose a model that considers multivariate dependence patterns in financial assets, and the assessment of their impact on performance and the optimal design of structured investment portfolios.

Liquidity risk assessment and prediction depends on many interconnected factors, such as the dependence between the asset prices and their temporary variations, the specific market frictions of the sector, the availability of financial and market data, stock market confidence, financial trading regulations in stressed markets, sudden market shocks that produce market downturns and contractions in capital inflows and outflows, and capital reserve levels of financial and trading institutions.

Our VaR model assesses risk in illiquid markets, considering the multivariate dependence of assets. We also examine the impact of changes in estimated liquidity risk on optimal portfolio assignment. To this end, our modeling approach combines LVaR (Liquidity Value-At-Risk) algorithms to measure liquidity risk, dynamic conditional correlation (DCC) t-copula models for dependence structure estimation and nonlinear optimization algorithms.

The objective of our research is to examine whether real-time optimization algorithms based on LVaR computation and dynamic conditional correlation (DCC) copulas for dependence estimation are capable of producing a better asset allocations of multiple assets in adverse market scenarios, considering operational and financial frontier limitations, evidenced largely by illiquidity shocks during the global financial crisis. 

This specific type of modeling is new in the literature and allows portfolio managers to anticipate the required liquidity horizons (closing periods) and determine robust allocation of multiple assets according to realistic market conditions. Furthermore, the obtained empirical results indicate that our modeling approach produces better efficient portfolios than competing optimization models (e.g., the traditional Markowitz´s mean-variance portfolio optimization approach). 

Since the 2008-2009 global financial crisis, VaR (Value-at-Risk) techniques have become critical tools for monitoring and predicting the market risk and liquidity of financial assets. These financial risk modeling techniques, which have been recognized by the Bank for International Settlements (BIS) or the Basel Committee on capital adequacy and bank regulations, measure and prevent any potential losses that arise, not only from securities’ price changes and the interdependence between the different types of assets (stocks, currencies, interest rates or commodities), but also from their negative tail co-movements in bearish market conditions. . 

In the event of a financial crisis or market downturn, adequate liquidity risk modeling is advisable. In fact, the main advantage of VaR models is their focus on downside risk (i.e., the impact of the results of negative tails) and their direct interpretation in monetary terms. 

Nevertheless, particularly in times of financial turbulence, traditional VaR models do not properly consider nonlinear dependence between portfolio assets and become inefficient in illiquid market scenarios. Despite the advances in measurement models, obtaining precise market liquidity risk estimations and applying them to optimize portfolios continues to be a challenge for financial institutions.

Therefore, in the article “Multivariate dependence and portfolio optimization algorithms under illiquid market scenarios”, published in the European Journal of Operational Researchtogether with my colleagues José Arreola, from Rennes Business School (France), Theo Berger, from the University of Bremen (Germany) and Duc Khuong Nguyen, from IPAG Business School (France), we propose a model that considers multivariate dependence patterns in financial assets, and the assessment of their impact on performance and the optimal design of structured investment portfolios.

Liquidity risk assessment and prediction depends on many interconnected factors, such as the dependence between the asset prices and their temporary variations, the specific market frictions of the sector, the availability of financial and market data, stock market confidence, financial trading regulations in stressed markets, sudden market shocks that produce market downturns and contractions in capital inflows and outflows, and capital reserve levels of financial and trading institutions.

Our VaR model assesses risk in illiquid markets, considering the multivariate dependence of assets. We also examine the impact of changes in estimated liquidity risk on optimal portfolio assignment. To this end, our modeling approach combines LVaR (Liquidity Value-At-Risk) algorithms to measure liquidity risk, dynamic conditional correlation (DCC) t-copula models for dependence structure estimation and nonlinear optimization algorithms.

The objective of our research is to examine whether real-time optimization algorithms based on LVaR computation and dynamic conditional correlation (DCC) copulas for dependence estimation are capable of producing a better asset allocations of multiple assets in adverse market scenarios, considering operational and financial frontier limitations, evidenced largely by illiquidity shocks during the global financial crisis. 

This specific type of modeling is new in the literature and allows portfolio managers to anticipate the required liquidity horizons (closing periods) and determine robust allocation of multiple assets according to realistic market conditions. Furthermore, the obtained empirical results indicate that our modeling approach produces better efficient portfolios than competing optimization models (e.g., the traditional Markowitz´s mean-variance portfolio optimization approach).