Most recent issue published online in the International Journal of Computational Economics and Econometrics.
International Journal of Computational Economics and Econometrics
http://www.inderscience.com/browse/index.php?journalID=311&year=2024&vol=14&issue=1
Inderscience Publishers Ltd
en-uk
support@inderscience.com
International Journal of Computational Economics and Econometrics
1757-1170
1757-1189
© 2023 Inderscience Enterprises Ltd.
© 2024 Inderscience Publishers Ltd
editor@inderscience.com
International Journal of Computational Economics and Econometrics
https://www.inderscience.com/images/files/coverImgs/ijcee_scoverijcee.jpg
http://www.inderscience.com/browse/index.php?journalID=311&year=2024&vol=14&issue=1
-
Identifying trend nature in time series using autocorrelation functions and stationarity tests
http://www.inderscience.com/link.php?id=135644
Time series non-stationarity can be detected thanks to autocorrelation functions. But trend nature, either deterministic or either stochastic, is not identifiable. Strategies based on Dickey-Fuller unit root-test are appropriate to choose between a linear deterministic trend or a stochastic trend. But all the observed deterministic trends are not linear, and such strategies fail in detecting a quadratic deterministic trend. Being a confounding factor, a quadratic deterministic trend makes a unit root spuriously appear. We provide a new procedure, based on Ouliaris-Park-Phillips unit root test, convenient for time series containing polynomial trends with a degree higher than one. Our approach is assessed based on simulated data. The strategy is finally applied on two real datasets: money stock in USA and on CO<SUB align="right">2 atmospheric concentration. Compared with Dickey-Fuller diagnosis, our strategy provides the model with the best performances.
Identifying trend nature in time series using autocorrelation functions and stationarity tests
M. Boutahar; M. Royer-Carenzi
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 1 - 22
Time series non-stationarity can be detected thanks to autocorrelation functions. But trend nature, either deterministic or either stochastic, is not identifiable. Strategies based on Dickey-Fuller unit root-test are appropriate to choose between a linear deterministic trend or a stochastic trend. But all the observed deterministic trends are not linear, and such strategies fail in detecting a quadratic deterministic trend. Being a confounding factor, a quadratic deterministic trend makes a unit root spuriously appear. We provide a new procedure, based on Ouliaris-Park-Phillips unit root test, convenient for time series containing polynomial trends with a degree higher than one. Our approach is assessed based on simulated data. The strategy is finally applied on two real datasets: money stock in USA and on CO<SUB align="right">2 atmospheric concentration. Compared with Dickey-Fuller diagnosis, our strategy provides the model with the best performances.]]>
10.1504/IJCEE.2024.135644
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 1 - 22
M. Boutahar
M. Royer-Carenzi
CNRS, Centrale Marseille, Aix Marseille University, I2M, UMR 7373, Marseille, France ' CNRS, Centrale Marseille, Aix Marseille University, I2M, UMR 7373, Marseille, France
time series
stationarity
autocorrelation functions
unit root tests
Dickey-Fuller
KPSS
OPP test
trend detection
deterministic or stochastic trend
spurious unit root
2023-12-20T23:20:50-05:00
Copyright © 2023 Inderscience Enterprises Ltd.
14
1
1
22
2023-12-20T23:20:50-05:00
-
Machine learning-based business risk analysis for big data: a case study of Pakistan
http://www.inderscience.com/link.php?id=135648
In finance, machine learning helps the business by improving its abilities and flexibility to prevent risks, errors and to accept such challenges. This research analyses and forecasts the interest rate risk of Pakistan using machine learning models. It took a ten-year financial dataset of Pakistan investment bonds from the State Bank of Pakistan website. In this study, a framework was proposed and four different models were developed to forecast the interest rates: neural network, bootstrap aggregated regression trees, cascade-forward neural network, and radial basis neural network. Subsequently, these models were run under four different scenarios: forecasting with original, generated, LASSO extracted and weighted average features. In addition, the outcomes of these models were compared with four performance metrics: mean absolute percentage error, daily peak mean absolute percentage error, mean absolute error, and root mean square error. Overall, the results showed that radial basis neural network provided the best forecasting.
Machine learning-based business risk analysis for big data: a case study of Pakistan
Mohsin Nazir; Zunaira Butt; Aneeqa Sabah; Azeema Yaseen; Anca Jurcut
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 23 - 41
In finance, machine learning helps the business by improving its abilities and flexibility to prevent risks, errors and to accept such challenges. This research analyses and forecasts the interest rate risk of Pakistan using machine learning models. It took a ten-year financial dataset of Pakistan investment bonds from the State Bank of Pakistan website. In this study, a framework was proposed and four different models were developed to forecast the interest rates: neural network, bootstrap aggregated regression trees, cascade-forward neural network, and radial basis neural network. Subsequently, these models were run under four different scenarios: forecasting with original, generated, LASSO extracted and weighted average features. In addition, the outcomes of these models were compared with four performance metrics: mean absolute percentage error, daily peak mean absolute percentage error, mean absolute error, and root mean square error. Overall, the results showed that radial basis neural network provided the best forecasting.]]>
10.1504/IJCEE.2024.135648
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 23 - 41
Mohsin Nazir
Zunaira Butt
Aneeqa Sabah
Azeema Yaseen
Anca Jurcut
Lahore College for Women University, Jail Road, Lahore, Pakistan ' Lahore College for Women University, Jail Road, Lahore, Pakistan ' Lahore College for Women University, Jail Road, Lahore, Pakistan ' Maynooth University Maynooth, Co. Kildare, Ireland ' University College Dublin, Belfield, Dublin 4, Ireland
machine learning
business risk analysis
interest rate risk
risk analysis
big data
forecasting models
Pakistan
2023-12-20T23:20:50-05:00
Copyright © 2023 Inderscience Enterprises Ltd.
14
1
23
41
2023-12-20T23:20:50-05:00
-
Improved stock price forecasting by streamlining indicators: an approach via feature selection and classification
http://www.inderscience.com/link.php?id=135655
Accurately predicting changes in stock prices is a complex and challenging task due to the multitude of factors influencing the stock market. Stock market analysts commonly rely on indicators for forecasting, but the interpretation of these indicators is often complicated and can result in inaccurate predictions. To enhance the precision of stock price forecasting, we propose a novel approach that incorporates feature selection algorithms and classification techniques. In fact, by identifying the most impactful indicators affecting each stock's price, the process of predicting stock prices will be significantly simplified. We conducted experimental tests on stock data from multiple companies listed in the Tehran Stock Exchange, spanning 2008 to 2021. Our findings demonstrate that reducing the number of features and indicators can significantly enhance the accuracy of stock price predictions in specific scenarios.
Improved stock price forecasting by streamlining indicators: an approach via feature selection and classification
Mohammad Javad Sheikhzadeh; Sajjad Rahmany
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 42 - 60
Accurately predicting changes in stock prices is a complex and challenging task due to the multitude of factors influencing the stock market. Stock market analysts commonly rely on indicators for forecasting, but the interpretation of these indicators is often complicated and can result in inaccurate predictions. To enhance the precision of stock price forecasting, we propose a novel approach that incorporates feature selection algorithms and classification techniques. In fact, by identifying the most impactful indicators affecting each stock's price, the process of predicting stock prices will be significantly simplified. We conducted experimental tests on stock data from multiple companies listed in the Tehran Stock Exchange, spanning 2008 to 2021. Our findings demonstrate that reducing the number of features and indicators can significantly enhance the accuracy of stock price predictions in specific scenarios.]]>
10.1504/IJCEE.2024.135655
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 42 - 60
Mohammad Javad Sheikhzadeh
Sajjad Rahmany
Department of Computer Science, School of Mathematics and Computer Science, Damghan University, Damghan, Iran ' Department of Computer Science, School of Mathematics and Computer Science, Damghan University, Damghan, Iran
stock forecasting
indicators
feature selection
classification
2023-12-20T23:20:50-05:00
Copyright © 2023 Inderscience Enterprises Ltd.
14
1
42
60
2023-12-20T23:20:50-05:00
-
Structural breaks detection using step-indicator saturation technique in state-space model
http://www.inderscience.com/link.php?id=135656
Recently, there has been a lot of interest in identifying structural breaks in economic time series. Failing to capture any structural breaks may have a pernicious effect on model estimation due to significant forecast errors after such breaks and inappropriate tests. Therefore, this study proposed a step-indicator saturation (SIS) technique as an extension of the general-to-specific (GETS) modelling framework for detecting any structural changes in time series. Monte Carlo simulations assessed the performance of the SIS in the local level model based on potency and gauge metrics using the 'gets' package in the R programming language. Sequential selection outperformed the non-sequential approach in the automatic GETS model selection procedure. Accordingly, this study applied the SIS technique to the Financial Times Stock Exchange (FTSE) Bursa Malaysia Hijrah Shariah and FTSE USA Shariah using a split-half approach and sequential selection. The retained indicators in the terminal model were selected based on the sequential and non-sequential algorithms. It was found that the retained indicators in both indices collided with the financial crises in 2008-2009. Overall, the proposed technique offers an effective approach to detect unknown locations, magnitudes, and structural break signs in a structural times series framework.
Structural breaks detection using step-indicator saturation technique in state-space model
Farid Zamani Che Rose; Mohd Tahir Ismail; Nur Aqilah Khadijah Rosili
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 61 - 80
Recently, there has been a lot of interest in identifying structural breaks in economic time series. Failing to capture any structural breaks may have a pernicious effect on model estimation due to significant forecast errors after such breaks and inappropriate tests. Therefore, this study proposed a step-indicator saturation (SIS) technique as an extension of the general-to-specific (GETS) modelling framework for detecting any structural changes in time series. Monte Carlo simulations assessed the performance of the SIS in the local level model based on potency and gauge metrics using the 'gets' package in the R programming language. Sequential selection outperformed the non-sequential approach in the automatic GETS model selection procedure. Accordingly, this study applied the SIS technique to the Financial Times Stock Exchange (FTSE) Bursa Malaysia Hijrah Shariah and FTSE USA Shariah using a split-half approach and sequential selection. The retained indicators in the terminal model were selected based on the sequential and non-sequential algorithms. It was found that the retained indicators in both indices collided with the financial crises in 2008-2009. Overall, the proposed technique offers an effective approach to detect unknown locations, magnitudes, and structural break signs in a structural times series framework.]]>
10.1504/IJCEE.2024.135656
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 61 - 80
Farid Zamani Che Rose
Mohd Tahir Ismail
Nur Aqilah Khadijah Rosili
Department of Mathematics and Statistics, Faculty of Science, Universiti Putra Malaysia, 43400 UPM Serdang, Selangor Darul Ehsan, Malaysia ' School of Mathematical Sciences, Universiti Sains Malaysia, 11800 USM Penang, Malaysia ' Faculty of Computing and Engineering, Quest International University, 30250 Ipoh, Perak, Malaysia
structural breaks
step-indicator saturation
SIS
Monte Carlo
model selection
state-space model
general-to-specific
GETS
2023-12-20T23:20:50-05:00
Copyright © 2023 Inderscience Enterprises Ltd.
14
1
61
80
2023-12-20T23:20:50-05:00
-
American financial markets dependencies: a vine copula approach
http://www.inderscience.com/link.php?id=135659
Regular vine copulas are used to evaluate the dependence between American financial markets (Argentina, Brazil, Canada, Chile, Colombia, Mexico, Peru, and the USA) from August 16, 2011, to April 21, 2022. The behaviour of marginal distributions is described by AR(1)-TGARCH models with errors distributed as an asymmetric skewed Student's t, which are adequate to model returns and their volatility. The conditional dependency between pairwise countries is estimated for the covid period, and three subperiods are analysed, pre-covid, covid, and post-covid. It is found that the contagion routes between the different American countries have the USA as the root node.
American financial markets dependencies: a vine copula approach
Arturo Lorenzo-Valdes
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 81 - 97
Regular vine copulas are used to evaluate the dependence between American financial markets (Argentina, Brazil, Canada, Chile, Colombia, Mexico, Peru, and the USA) from August 16, 2011, to April 21, 2022. The behaviour of marginal distributions is described by AR(1)-TGARCH models with errors distributed as an asymmetric skewed Student's t, which are adequate to model returns and their volatility. The conditional dependency between pairwise countries is estimated for the covid period, and three subperiods are analysed, pre-covid, covid, and post-covid. It is found that the contagion routes between the different American countries have the USA as the root node.]]>
10.1504/IJCEE.2024.135659
International Journal of Computational Economics and Econometrics, Vol. 14, No. 1 (2024) pp. 81 - 97
Farid Zamani Che Rose
Mohd Tahir Ismail
Nur Aqilah Khadijah Rosili
Department of Mathematics, UPAEP University, 17 Sur # 901, Barrio de Santiago, 72410, Puebla, Mexico
vine copulas
TGARCH
dependence
2023-12-20T23:20:50-05:00
Copyright © 2023 Inderscience Enterprises Ltd.
14
1
81
97
2023-12-20T23:20:50-05:00