|
on Econometrics |
| By: | Davidson, James; Sibbertsen, Philipp |
| Abstract: | This paper proposes simple Hausman-type tests to check for bias in the log-periodogram regression of a time series believed to be long memory. The statistics are asymptotically standard normal on the null hypothesis that no bias is present, and the tests are consistent. The use of the tests in conjunction with tests of significance of the long memory parameter is illustrated by Monte Carlo experiments. |
| Keywords: | long memory, log-periodogram estimation, Hausman test |
| JEL: | C12 C22 |
| Date: | 2005–06 |
| URL: | http://d.repec.org/n?u=RePEc:han:dpaper:dp-317&r=ecm |
| By: | Sibbertsen, Philipp; Krämer, Walter |
| Abstract: | We show that the power of the KPSS-test against integration, as measured by divergence rates of the test statistic under the alternative, remains the same when residuals from an OLS-regression rather than true observations are used. The divergence rate is independent of the order of integration of the cointegrating regressors which are allowed to be I(1 + dX) in our set up. |
| Keywords: | cointegration, power, long memory, KPSS-Test |
| JEL: | C12 C32 |
| Date: | 2005–06 |
| URL: | http://d.repec.org/n?u=RePEc:han:dpaper:dp-318&r=ecm |
| By: | Mehmet Dalkir (University of Kansas) |
| Abstract: | The method proposed in this chapter is making use of the bispectrum transformation to estimate the level of integration of a fractionally integrated time series. Bispectrum ransformation transforms the series into a two dimensional frequency space, and thus has higher information content compared to the Geweke-Porter-Hudak method. The bispectrum method is an alternative to the recently proposed wavelet method that transforms the original series into time-frequency (or time-scale) space. |
| Keywords: | Bispectrum, frequency domain, estimation, long memory |
| JEL: | C1 C2 C3 C4 C5 C8 |
| Date: | 2005–07–07 |
| URL: | http://d.repec.org/n?u=RePEc:wpa:wuwpem:0507001&r=ecm |
| By: | David Mayston |
| Abstract: | The desire to make public policy formation and decision-making more evidence-based increases the importance of assessing the estimation biases which may arise from the use of standard statistical techniques, such as ordinary least squares multivariate regression analysis, on which many existing studies rely. These biases may arise from the existence of multiple relationships between the variables of interest, in addition to those in the primary equation of interest. We investigate the nature of the dependency of the cumulative OLS endogeneity bias that results from these multiple additional interrelationships on their underlying structural parameters, and establish conditions under which the multiple additional relationships will cumulatively add to the extent of the cumulative bias in a predictable direction, rather than tending to offset each other. The analysis is extended to include partial regressions in which some variables are excluded from the OLS estimation. |
| Keywords: | Multivariate regression analysis; Cumulative endogeneity bias; Evidence-based policy |
| URL: | http://d.repec.org/n?u=RePEc:yor:yorken:05/11&r=ecm |
| By: | Joaquim J.S. Ramalho; Esmeralda Ramalho |
| Abstract: | The ability of six alternative bootstrap methods to reduce the bias of GMM parameter estimates is examined in an instrumental variable framework using Monte Carlo analysis. Promising results were found for the two bootstrap estimators suggested in the paper. |
| Keywords: | Endogenous Stratified Sampling, Bias correction, GMM, Parametric models |
| JEL: | C13 |
| Date: | 2005 |
| URL: | http://d.repec.org/n?u=RePEc:evo:wpecon:11_2005&r=ecm |