Econometric Time Series
http://lists.repec.org/mailman/listinfo/nep-ets
Econometric Time Series2015-08-30Yong YinEstimation of Fractionally Integrated Panels with Fixed Effects and Cross-Section Dependence
http://d.repec.org/n?u=RePEc:aah:create:2015-35&r=all
We consider large N, T panel data models with fixed effects, common factors allowing cross-section dependence, and persistent data and shocks, which are assumed fractionally integrated. In a basic setup, the main interest is on the fractional parameter of the idiosyncratic component, which is estimated in first differences after factor removal by projection on the cross-section average. The pooled conditional-sum-of-squares estimate is root-NT consistent but the normal asymptotic distribution might not be centered, requiring the time series dimension to grow faster than the cross-section size for correction. Generalizing the basic setup to include covariates and heterogeneous parameters, we propose individual and common-correlation estimates for the slope parameters, while error memory parameters are estimated from regression residuals. The two parameter estimates are root-T consistent and asymptotically normal and mutually uncorrelated, irrespective of possible cointegration among idiosyncratic components. A study of small-sample performance and an empirical application to realized volatility persistence are included.Yunus Emre Ergemen, Carlos Velasco2015-08-17Fractional cointegration, factor models, long memory, realized volatilityDetecting intraday financial market states using temporal clustering
http://d.repec.org/n?u=RePEc:arx:papers:1508.04900&r=all
We propose the application of a high-speed maximum likelihood clustering algorithm to detect temporal financial market states, using correlation matrices estimated from intraday market microstructure features. We first determine the ex-ante intraday temporal cluster configurations to identify market states, and then study the identified temporal state features to extract state signature vectors which enable online state detection. The state signature vectors serve as low-dimensional state descriptors which can be used in learning algorithms for optimal planning in the high-frequency trading domain. We present a feasible scheme for real-time intraday state detection from streaming market data feeds. This study identifies an interesting hierarchy of system behaviour which motivates the need for time-scale-specific state space reduction for participating agents.Dieter Hendricks, Tim Gebbie, Diane Wilcox2015-08Do We Need Ultra-High Frequency Data to Forecast Variances?
http://d.repec.org/n?u=RePEc:hal:wpaper:halshs-01078158&r=all
In this paper we study various MIDAS models in which the future daily variance is directly related to past observations of intraday predictors. Our goal is to determine if there exists an optimal sampling frequency in terms of volatility prediction. Via Monte Carlo simulations we show that in a world without microstructure noise, the best model is the one using the highest available frequency for the predictors. However, in the presence of microstructure noise, the use of ultra high-frequency predictors may be problematic, leading to poor volatility forecasts. In the application, we consider two highly liquid assets (i.e., Microsoft and S&P 500). We show that, when using raw intraday squared log-returns for the explanatory variable, there is a "high-frequency wall" or frequency limit above which MIDAS-RV forecasts deteriorate. We also show that an improvement can be obtained when using intraday squared log-returns sampled at a higher frequency, provided they are pre-filtered to account for the presence of jumps, intraday periodicity and/or microstructure noise. Finally, we compare the MIDAS model to other competing variance models including GARCH, GAS, HAR-RV and HAR-RV-J models. We find that the MIDAS model provides equivalent or even better variance forecasts than these models, when it is applied on filtered data.Georgiana-Denisa Banulescu, Bertrand Candelon, Christophe Hurlin, Sébastien Laurent2014-10-26Testing for Linearity in Regressions with I(1) Processes
http://d.repec.org/n?u=RePEc:ngi:dpaper:15-11&r=all
We propose a generalized version of the RESET test for linearity in regressions with I(1) processes against various nonlinear alternatives and no cointegration. The proposed test statistic for linearity is given by the Wald statistic and its limiting distribution under the null hypothesis is shown to be a X^2 distribution with a "leads and lags" estimation technique. We show that the test is consistent against a class of nonlinear alternatives and no cointegration. Finite-sample simulations show that the empirical size is close to the nominal one and the test succeeds in detecting both nonlinearity and no cointegration.Yoichi Arai, 2015-08The multivariate Beveridge--Nelson decomposition with I(1) and I(2) series
http://d.repec.org/n?u=RePEc:pra:mprapa:66319&r=all
The consumption Euler equation implies that the output growth rate and the real interest rate are of the same order of integration; i.e., if the real interest rate is I(1), then so is the output growth rate and hence log output is I(2). To estimate the natural rates and gaps of macroeconomic variables jointly, this paper develops the multivariate Beveridge--Nelson decomposition with I(1) and I(2) series. The paper applies the method to Japanese data during 1980Q1--2013Q3 to estimate the natural rates and gaps of output, inflation, interest, and unemployment jointly.Murasawa, Yasutomo2015-08-28gap; natural rate; trend--cycle decomposition; unit rootCovariate-augmented unit root tests with mixed-frequency data
http://d.repec.org/n?u=RePEc:ptu:wpaper:w201507&r=all
Unit root tests typically suer from low power in small samples, which results in not rejecting the null hypothesis as often as they should. This paper tries to tackle this issue by assessing whether it is possible to improve the power performance of covariate-augmented unit root tests, namely the ADF family of tests, by exploiting mixed-frequency data. We use the mixed data sampling (MIDAS) approach to deal with mixed-frequency data. The results from a Monte Carlo exercise indicate that mixed-frequency tests have better power performance than low-frequency tests. The gains from exploiting mixed-frequency data are greater for near-integrated variables. An empirical illustration using the US unemployment rate is presented.Cláudia Duarte2015Multivariate Dynamic Copula Models: Parameter Estimation and Forecast Evaluation
http://d.repec.org/n?u=RePEc:usg:sfwpfi:2015:13&r=all
This paper introduces multivariate dynamic copula models to account for the time-varying dependence structure in asset portfolios. We firstly enhance the flexibility of this structure by modeling regimes with multivariate mixture copulas. In our second approach, we derive dynamic elliptical copulas by applying the dynamic conditional correlation model (DCC) to multivariate elliptical copulas. The best-ranked copulas according to both in-sample fit and out-of-sample forecast performance indicate the importance of accounting for time-variation. The superiority of multivariate dynamic Clayton and Student-t models further highlight that positive tail dependence as well as the capability of capturing asymmetries in the dependence structure are crucial features of a well-fitting model for an equity portfolio.Aepli, Matthias D., Frauendorfer, Karl, Fuess, Roland, Paraschiv, Florentina2015-07Multivariate dynamic copulas, regime-switching copulas, dynamic conditional correlation (DCC) model, forecast performance, tail dependence.Autocorrelation robust inference using the Daniell kernel with fixed bandwidth
http://d.repec.org/n?u=RePEc:yor:yorken:15/14&r=all
We consider alternative asymptotics for frequency domain estimates of the long run variance, in which the bandwidth is kept fixed. For a weakly dependent process, this does not yield a consistent estimateof the long run variance, but the standardized mean has t limit distribution, which, for any given bandwidth, appears to be more precise than the traditional Gaussian limit. In presence of fractionally integrated data, the limit distribution of the estimate is not standard, and we derive critical values for the standardized mean for various bandwidths. Again, we find that this asymptotic result provides a better approximation than other proposals like the Memory Autocorrelation Consistent (MAC) estimate. In multivariate set up, fixed bandwidth asymptotics may be also used to provide a characterization to the limit distribution of estimates of cointegrating parameter which differs substantially from the conventional Narrow Band asymptotics.Javier Hualde, Fabrizio Iacone2015-08long run variance estimation, long memory, large-m and fixed-masymptotic theory