Computational Economics最新文献

To comprehensively reflect the heteroscedasticity, nonlinear dependence and heavy-tailed distributions of stock returns while reducing the huge cost of parameter estimation, we use the Fama-French three-factor model to describe stock returns and then model the factor dynamics by using the ARMA-GARCH and Student-t copula models. A factor-based scenario tree generation algorithm is thus proposed, and the corresponding multi-stage international portfolio selection model is constructed and its reformulation is derived. Different from the current literature, our proposed models can capture the dynamic dependence among international markets and the dynamics of exchange rates, and what’s more important, make it possible for the practical solution of large-scale multi-stage international portfolio selection problems. Considering three different objective functions and international investments in the USA, Japanese and European markets, we carry out a series of empirical studies to demonstrate the practicality and efficiency of the proposed factor-based scenario tree generation algorithm and multi-stage international portfolio selection models.

Understanding financial contagion and instability, especially during financial crises, is an important issue in risk management. The emergence of alternative high-risk and speculative asset classes such as cryptocurrencies, make it imperative to effectively monitor the financial connectivity between heterogeneous asset classes across time, in conjunction with the associated risk, to avoid a substantial breakdown of financial systems during turmoil periods. To address this problem, this paper investigates the predictive capacity of time-varying graph connectivity measures on tail and systemic risk for heterogeneous asset classes. To this end, proper statistical and geometric rules are defined first, to infer the dynamic graph topology of asset returns. Then, a novel predictive signal is proposed to quantify and rank the predictive power of dynamic nodal and global graph measures. Finally, a minimum dominating set detection method is used to track the community structure of our asset classes over time and study its consistency with the time evolution of the top predictive measures. Our empirical findings show a remarkable variability of the predictive potential for the distinct connectivity measures, and reveal its importance in designing alerting mechanisms for risk management.

We develop a novel dynamic model for household debt and household income change studying the interaction between financial fragility and financial literacy. We compare the results to the U.S. data under several parameterizations. Households react pro-cyclically to income shocks and are better able to represent aggregate data when financial literacy is low.

Modern problems in Economics have tremendously benefited from the ever increasing amount of available information. Hence, most of the recent econometric approaches have focused on how to model and estimate relationships between covariates and dependent variables under this high-dimensional scenario. Particularly in the time series context, one usually aims to produce valuable forecasts of the dependent variables. In this paper our main goal is two-folded: i) employ several modern computationally highly intensive Machine Learning (ML) methods for achieving time series forecasting accuracy under a high-dimensional covariates setting; ii) propose a novel variation of the Elastic Net (ENet), the Weighted Lag Adaptive ENet (WLadaENet), which combines the popular Ridge Regression with a regularization method tailored for time series, the WLAdaLASSO (Konzen and Ziegelmann in J Forecast 35:592–612, 2016). To achieve our goal, we carry out Monte Carlo simulation studies as well as a real data analysis of USA inflation with a forecast range from January 2013 to December 2023. In our Monte Carlo implementations, the WLadaENet presents a solid performance both in terms of variable selection when the true model is sparse and in terms of forecasting accuracy even when the model is not sparse and nonlinearities are included. Our approach also performs reasonably well to forecast the USA inflation for different horizons ahead. Since the chosen period includes the Covid-19 crisis, a sub-period analysis is carried out, not leading to a uniformly best forecaster.

This article introduces the Risk Balancing Frontier (RBF), a new portfolio boundary in the absolute risk-total return space: the RBF arises when two risk indicators, the Tracking Error Volatility (TEV) and the Value-at-Risk (VaR), are both constrained not to exceed pre-set maximum values. By focusing on the trade-off between the joint restrictions on the two risk indicators, this frontier is the set of all portfolios characterized by the minimum VaR attainable for each TEV level. First, the RBF is defined analytically and its mathematical properties are discussed: we show its connection with the Constrained Tracking Error Volatility Frontier (Jorion in Financ Anal J, 59(5):70–82, 2003. https://doi.org/10.2469/faj.v59.n5.2565) and the Constrained Value-at-Risk Frontier (Alexander and Baptista in J Econ Dyn Control, 32(3):779–820, 2008. https://doi.org/10.1016/j.jedc.2007.03.005) frontiers. Next, we explore computational issues implied with its construction, and we develop a fast and accurate algorithm to this aim. Finally, we perform an empirical example and consider its relevance in the context of applied finance: we show that the RBF provides a useful tool to investigate and solve potential agency problems.

The modeling and forecasting of return volatility for the top three cryptocurrencies, which are identified by the highest trading volumes, is the main focus of the study. Eleven different GARCH-type models were analyzed using a comprehensive methodology in six different distributions, and deep learning algorithms were used to rigorously assess each model’s forecasting performance. Additionally, the study investigates the impact of selecting dynamic parameters for the forecasting performance of these models. This study investigates if there are any appreciable differences in forecast outcomes between the two different realized variance calculations and variations in training size. Further investigation focuses on how the use of expanding and rolling windows affects the optimal window type for forecasting. Finally, the importance of choosing different error measurements is emphasized in the framework of comparing forecasting performances. Our results indicate that in GARCH-type models, 5-minute realized variance shows the best forecasting performance, while in deep learning models, median realized variance (MedRV) has the best performance. Moreover, it has been determined that an increase in the training/test ratio and the selection of the rolling window approach both play important roles in achieving better forecast accuracy. Finally, our results show that deep learning models outperform GARCH-type models in volatility forecasts.

We develop a methodology that utilizes deep learning to simultaneously solve and estimate canonical continuous-time general equilibrium models in financial economics. We illustrate our method in two examples: (1) industrial dynamics of firms and (2) macroeconomic models with financial frictions. Through these applications, we illustrate the advantages of our method: generality, simultaneous solution and estimation, leveraging the state-of-art machine-learning techniques, and handling large state space. The method is versatile and can be applied to a vast variety of problems.

This paper knits the concepts of financial performance and financial distress in a unified framework. The machine learning algorithm of extreme gradient boosting (XGBoost) is employed to identify the set of factors predicting financial distress and performance and panel logistic regressions indicate the direction of influence and significance of these common factors. The XGBoost algorithm indicates the existence of some common factors, such as lagged net profit margin, growth of profit after tax, lagged assets turnover ratio, growth of sales and log of total asset. Additionally, past performance is found to impact current financial distress and vice-versa. The regression results shows that profit growth significantly improves financial performance while reducing corporate distress. This calls for a common framework to analyze these two phenomena for registered firms.

Algorithmic trading has become increasingly prevalent in financial markets, and traders and investors seeking to leverage computational techniques and data analysis to gain a competitive edge. This paper presents a comprehensive analysis of algorithmic trading strategies, focusing on the efficacy of technical indicators in predicting market trends and generating profitable trading signals. The research framework outlines a systematic process for investigating and evaluating stock market investment strategies, beginning with a clear research objective and a comprehensive review of the literature. Data collected from various stock exchanges, including the S&P 500, undergo rigorous preprocessing, cleaning, and transformation. The subsequent stages involve generating investment signals, calculating relevant indicators such as RSI, EMAs, and MACD, and conducting backtesting to compare the strategy's historical performance to benchmarks. The key findings reveal notable returns generated by the indicators analyzed, though falling short of benchmark performance, highlighting the need for further refinement. The study underscores the importance of a multi-indicator approach in enhancing the interpretability and predictive accuracy of algorithmic trading models. This research contributes to understanding of algorithmic trading strategies and provides valuable information for traders and investors looking to optimize their investment decisions in financial markets.

It is prone to overfitting and poor generalization ability for imbalanced small sample datasets in modeling. Auxiliary data is an effective solution. However, there may be data distribution differences between auxiliary data and small sample data, and the presence of noise samples affects the prediction performance. To address this issue, we propose an ensemble resampling based transfer AdaBoost (TrAdaBoost) algorithm for imbalanced small sample credit classification. The proposed algorithm framework has two stages: ensemble resampling dataset generation and weight adaptive transfer AdaBoost (WATrA) model prediction. In the first stage, neighborhood-based resampling technique is proposed to filter source data and reduce noise samples, followed by bagging resampling to balance the filtered source data. In the second stage, a weight adaptive TrAdaBoost model is utilized to address small sample with class imbalance issues and improve the effectiveness of the proposed method. We validate the proposed algorithm on two small sample credit datasets with class imbalance, and observe significant improvements in performance compared to traditional supervised machine learning methods and resampling methods based on the main evaluation criteria.