|
on Econometric Time Series |
| By: | Ricardo P. Masini; Marcelo C. Medeiros; Eduardo F. Mendes |
| Abstract: | In this paper we survey the most recent advances in supervised machine learning and high-dimensional models for time series forecasting. We consider both linear and nonlinear alternatives. Among the linear methods we pay special attention to penalized regressions and ensemble of models. The nonlinear methods considered in the paper include shallow and deep neural networks, in their feed-forward and recurrent versions, and tree-based methods, such as random forests and boosted trees. We also consider ensemble and hybrid models by combining ingredients from different alternatives. Tests for superior predictive ability are briefly reviewed. Finally, we discuss application of machine learning in economics and finance and provide an illustration with high-frequency financial data. |
| Date: | 2020–12 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2012.12802&r=all |
| By: | Tae-Hwy Lee (Department of Economics, University of California Riverside); Shahnaz Parsaeian (University of Kansas); Aman Ullah (University of California, Riverside) |
| Abstract: | Hashem Pesaran has made many seminal contributions, among others, in the time series econometrics estimation and forecasting under structural break, see Pesaran and Timmermann (2005, 2007), Pesaran et al. (2006), and Pesaran et al. (2013). In our paper here we focus on the estimation of regression parameters under multiple structural breaks with heteroskedasticity across regimes. We propose a combined estimator of regression parameters based on combining restricted estimator under the situation that there is no break in the parameters, with unrestricted estimator under the break. The operational optimal combination weight is between zero and one. The analytical finite sample risk is derived, and it is shown that the risk of the proposed combined estimator is lower than that of the unrestricted estimator under any break size and break points. Further, we show that the combined estimator outperforms over the unrestricted estimator in terms of the mean squared forecast errors. Properties of the estimator are also demonstrated in simulations. Finally, empirical illustrations for parameter estimators and forecasts are presented through macroeconomic and financial data sets. |
| Keywords: | Structural breaks, Combined estimator |
| JEL: | C13 C32 C53 |
| Date: | 2021–01 |
| URL: | http://d.repec.org/n?u=RePEc:ucr:wpaper:202101&r=all |
| By: | Sayar Karmakar; Stefan Richter; Wei Biao Wu |
| Abstract: | A general class of time-varying regression models is considered in this paper. We estimate the regression coefficients by using local linear M-estimation. For these estimators, weak Bahadur representations are obtained and are used to construct simultaneous confidence bands. For practical implementation, we propose a bootstrap based method to circumvent the slow logarithmic convergence of the theoretical simultaneous bands. Our results substantially generalize and unify the treatments for several time-varying regression and auto-regression models. The performance for ARCH and GARCH models is studied in simulations and a few real-life applications of our study are presented through analysis of some popular financial datasets. |
| Date: | 2020–11 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2011.13157&r=all |
| By: | Hafner, Christian (Université catholique de Louvain, LIDAM/ISBA, Belgium); Herwartz, Helmut |
| Abstract: | A model for dynamic independent component analysis is introduced where the dynamics are driven by the score of the pseudo likelihood with respect to the rotation angle of model innovations. While conditional second moments are invariant with respect to rotations, higher conditional moments are not, which may have important implications for applications. The pseudo maximum likelihood estimator of the model is shown to be consistent and asymptotically normally distributed. A simulation study reports good finite sample properties of the estimator, including the case of a mis-specification of the innovation density. In an application to a bivariate exchange rate series of the Euro and the British Pound against the US Dollar, it is shown that the model-implied conditional portfolio kurtosis largely aligns with narratives on financial stress as a result of the global financial crisis in 2008, the European sovereign debt crisis (2010-2013) and early rumors signalling the UK to leave the European Union (2017). These insights are consistent with a recently proposed model that associates portfolio kurtosis with a geopolitical risk factor. |
| Keywords: | structural vector autoregressions, multivariate GARCH, portfolio selection, risk man- agement |
| Date: | 2020–01–01 |
| URL: | http://d.repec.org/n?u=RePEc:aiz:louvad:2020031&r=all |
| By: | D.S.G. Pollock; Emi Mise |
| Abstract: | This note describes the use of the Cholesky decomposition in solving the equation Ab = y when A = A0 is a symmetric matrix of full rank. A specialised version of the algorithm is provided for the case where A is a banded Toeplitz matrix, in which each band contains a unique repeated element and where the number of bands is considerably less than the order of the matrix, which is assumed to be large. This circumstance demands that steps should be taken to minimise the use of the computer's memory. An example is provided of the use of the algorithm in implementing a finite-sample Wiener-Kolmogorov filter aimed at removing the seasonal fluctuations from economic data. |
| URL: | http://d.repec.org/n?u=RePEc:lec:leecon:20/01&r=all |
| By: | Matteo Iacopini; Carlo R. M. A. Santagiustina |
| Abstract: | Count time series obtained from online social media data, such as Twitter, have drawn increasing interest among academics and market analysts over the past decade. Transforming Web activity records into counts yields time series with peculiar features, including the coexistence of smooth paths and sudden jumps, as well as cross-sectional and temporal dependence. Using Twitter posts about country risks for the United Kingdom and the United States, this paper proposes an innovative state space model for multivariate count data with jumps. We use the proposed model to assess the impact of public concerns in these countries on market systems. To do so, public concerns inferred from Twitter data are unpacked into country-specific persistent terms, risk social amplification events, and co-movements of the country series. The identified components are then used to investigate the existence and magnitude of country-risk spillovers and social amplification effects on the volatility of financial markets. |
| Date: | 2020–12 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2012.13267&r=all |
| By: | Delouille, Véronique; Lefèvre, Laure; Mathieu, Sophie (Université catholique de Louvain, LIDAM/ISBA, Belgium); Ritter, Christian (Université catholique de Louvain, LIDAM/ISBA, Belgium); von Sachs, Rainer (Université catholique de Louvain, LIDAM/ISBA, Belgium) |
| Abstract: | In many applications, a control procedure is required to detect potential deviations in a panel of serially correlated processes. It is common that the processes are corrupted by noise and that no prior information about the in-control data are available for that purpose. This paper suggests a general nonparametric monitoring scheme for supervising such a panel with time-varying mean and variance. The method is based on a control chart designed by block bootstrap, which does not require parametric assumptions on the distribution of the data. The procedure is tailored to cope with strong noise, potentially missing values and absence of in-control series, which is tackled by an intelligent exploitation of the information in the panel. Our methodology is completed by support vector machine procedures to estimate magnitude and form of the encountered deviations (such as stepwise shifts or functional drifts). This scheme, though generic in nature, is able to treat an important applied data problem: the control of deviations in a subset of sunspot number observations which are part of the International Sunspot Number, a world reference for long-term solar activity. |
| Date: | 2020–01–01 |
| URL: | http://d.repec.org/n?u=RePEc:aiz:louvad:2020030&r=all |
| By: | Shaoxin Hong (Center for Economic Research, Shandong University, Jinan 250100, Shandong, Chin); Zhenyi Zhang (International School of Economics and Management, Capital University of Economics and Business, Beijing, Beijing 100070, China); Zongwu Cai (Department of Economics, The University of Kansas, Lawrence, KS 66045, USA) |
| Abstract: | n this paper, we propose the Cramer-von Mises type test statistic for testing heteroskedasticity in predictive regression when regressors are nonstationary. A Monte Carlo simulation study is conducted to illustrate the finite sample performance of the proposed test statistic and a real empirical example is examined. |
| Keywords: | Cramer-von Mises test statistic; Heteroskedasticity; Nonstationarity; Predictive regressions; Specification test. |
| JEL: | C12 C22 |
| Date: | 2021–01 |
| URL: | http://d.repec.org/n?u=RePEc:kan:wpaper:202101&r=all |
| By: | Jingwei Pan (Department of Empirical Economic Research, Technical University of Darmstadt) |
| Abstract: | Correlation indicates the strength of the linear relationship between two random variables and is therefore relevant for asset pricing, portfolio choice and risk management. In addition, forecasts of correlation dynamics allow for a better evaluation of the systemic risk and may give an initial signal about potential crises (Engle (2009)).This paper aims to evaluate daily correlation forecasts. For the calculation of the correlation forecasts, the BEKK model of Engle and Kroner (1995) and the DCC model of Engle (2002) are applied. Since there is no clear suggestion regarding the sampling scheme to estimate the realized correlations from intraday data (Andersen et al. (2006)), several experimental schemes with different sampling intervals are examined. Following Komunjer and Owyang (2012), a multivariate loss function which may be asymmetric is used to measure the distance between model correlation forecasts and realized correlations. The data sample contains intraday high-frequency and closing prices of the three major US indices: S&P 500, NASDAQ 100 and Russell 2000. Based on the results obtained so far, the following conclusions can be drawn: (i) Both models better predict correlations for the pair S&P 500 and NASDAQ 100 than for Russell 2000 and other two indices. (ii) The DCC model performs better than the BEKK model applying the symmetric loss function. (iii) On the basis of the correlation pairs between the Russell 2000 index and other two indices, the optimal degrees of asymmetry are negative for the BEKK forecast errors and positive for the DCC forecast errors in most cases. (iv) The degrees of asymmetry depend on the choice of sampling schemes for calculating the realized correlations. (v) Both models are unable to capture the sudden decrease of correlations during the crisis period. |
| Keywords: | Correlation forecasting, BEKK, DCC, asymmetric loss function |
| JEL: | C10 C52 G17 |
| URL: | http://d.repec.org/n?u=RePEc:sek:iefpro:11413234&r=all |