In this paper we conduct an empirical analysis on the performance gains of using high frequency data in portfolio selection. Within a CRRA-utility maximization framework, we suggest the construction of two different portfolios: a low and a high frequency portfolios. For ten different risk aversion levels, we compare the performance of both portfolios in terms of several out-of-sample measures. Using data on fourteen stocks of the CAC 40 stock market index, from January 1999 to December 2003, we conclude that the “fight” is always “won” by the high frequency portfolio for all the considered performance evaluation measures.

In this paper we perform a quantitative analysis, under the IFRS 9 framework, on the tradeoff of classifying a financial asset at amortized cost versus at fair value. We define and implement a banking impairment model in order to quantify the forward-looking expected credit loss. Based on the suggested impairment model we conduct a backtest on the 10-year Portuguese Government bonds, for the time period from January 2003 to December 2019. The Portuguese bonds’ history constitutes a very rich data set for our experiment, as these bonds have experienced significant downgrades during the 2011-2014 financial crisis. We suggest a quantitative and systematic approach in order to find efficient allocations, in an income/downside comprehensive income bi-dimensional space. Resorting to stochastic simulation, we show a possible approach to mitigate the estimation error ingrained in the proposed bi-objective stochastic model. Finally, we assess the out-of-sample performance of some of the suggested efficient allocations.

In this paper, we devise a forward-looking methodology to determine efficient credit portfolios under the IFRS 9 framework. We define and implement a credit loss model based on prospective point-in-time probabilities of default. We determine these probabilities of default and the credits’ stage allocation through a credit stochastic simulation. This simulation is based on the estimation of transition matrices. Using data from 1981 to 2019, in a non-homogeneous Markov chain setting, we estimate transition matrices conditional on the global real gross domestic product growth. This allows considering the effects of the economic cycle, which are of great importance in bank management. Finally, we develop a robust optimization model that allows the bank manager to analyze the tradeoff between the annual average portfolio income and the corresponding portfolio volatility. According to the proposed bi-objective model, we compute the efficient credit portfolios constructed based on 10-year maturity credits. We compare their structure to those generated by the IAS 39 and CECL accounting frameworks. The results indicate that the IFRS 9 and CECL frameworks generate efficient credit portfolios whose structure penalizes riskier-rated credits. In turn, the riskier efficient credit portfolios under the IAS 39 framework concentrate entirely on speculative-grade credits. This pattern is also encountered in efficient credit portfolios constructed based on credits with different maturities, namely 5 and 15 years. Moreover, the longer the maturity of the credits that enter into the composition of the efficient portfolios, the more the speculative-grade credits tend to be penalized.