Using quantile regression for duration analysis

Summary Quantile regression methods are emerging as a popular technique in econometrics and biometrics for exploring the distribution of duration data. This paper discusses quantile regression for duration analysis allowing for a flexible specification of the functional relationship and of the error distribution. Censored quantile regression addresses the issue of right censoring of the response variable which is common in duration analysis. We compare quantile regression to standard duration models. Quantile regression does not impose a proportional effect of the covariates on the hazard over the duration time. However, the method cannot take account of time-varying covariates and it has not been extended so far to allow for unobserved heterogeneity and competing risks. We also discuss how hazard rates can be estimated using quantile regression methods. This paper benefitted from the helpful comments by an anonymous referee. Due to space constraints, we had to omit the details of the empirical application. These can be found in the long version of this paper, Fitzenberger and Wilke (2005). We gratefully acknowledge financial support by the German Research Foundation (DFG) through the research project ‘Microeconometric modelling of unemployment durations under consideration of the macroeconomic situation’. Thanks are due to Xuan Zhang for excellent research assistance. All errors are our sole responsibility.     

A small but practically useful modification to the Hodrick–Prescott filtering: A note

The Hodrick–Prescott (HP) filtering is widely applied to decompose macroeconomic time series, such as real Gross Domestic Product, into cyclical and trend components. This paper presents a small but practically useful modification to this approach. The reason why this modified filtering is of practical use is that it provides not only identical trend estimates as the HP filtering but also extrapolations of the trend. We provide a proof based on a ridge regression representation of the modified HP filtering. This is mainly because it enhances our understanding of the approach.     

Quantile regression methods for censored gap time data

ABSTRACT

In longitudinal studies, subjects may potentially undergo a series of sequentially ordered events. The gap times, which are the times between two serial events, are often the outcome variables of interest. This study considers quantile regression models of gap times for censored serial-event data and adapts a weighted version of the estimating equation for regression coefficients. The resulting estimators are uniformly consistent and asymptotically normal. Extensive simulation studies are presented to evaluate the finite-sample performance of the proposed methods. An analysis of the tumor recurrence data for bladder cancer patients is also provided to illustrate our proposed methods.     

Inference on Filtered and Smoothed Probabilities in Markov-Switching Autoregressive Models

ABSTRACT

We derive a statistical theory that provides useful asymptotic approximations to the distributions of the single inferences of filtered and smoothed probabilities, derived from time series characterized by Markov-switching dynamics. We show that the uncertainty in these probabilities diminishes when the states are separated, the variance of the shocks is low, and the time series or the regimes are persistent. As empirical illustrations of our approach, we analyze the U.S. GDP growth rates and the U.S. real interest rates. For both models, we illustrate the usefulness of the confidence intervals when identifying the business cycle phases and the interest rate regimes.     

Panel Seasonal Unit Root Test: Further Simulation Results and An Application to Unemployment Data

Summary: In this paper the seasonal unit root test of Hylleberg et al. (1990) is generalized to cover a heterogenous panel. The procedure follows the work of Im, Pesaran and Shin (2002) and is independently proposed by Otero et al. (2004). Test statistics are given and critical values are obtained by simulation. Moreover, the properties of the tests are analyzed for different deterministic and dynamic specifications. Evidence is presented that for a small time series dimension the power is low even for increasing cross section dimension. Therefore, it seems necessary to have a higher time series dimension than cross section dimension. The test is applied to unemployment data in industrialized countries. In some cases seasonal unit roots are detected. However, the null hypotheses of panel seasonal unit roots are rejected. The null hypothesis of a unit root at the zero frequency is not rejected, thereby supporting the presence of hysteresis effects. * The research of this paper was supported by the Deutsche Forschungsgemeinschaft. The paper was presented at the workshop “Unit roots and cointegration in panel data” in Frankfurt, October 2004 and in the poster-session at the EC2 meeting in Marseille, December 2004. We are grateful to the participants of the workshops and an anonymous referee for their helpful comments.     

Time Series Properties of the German Production Index

Summary: The production index is an important indicator for assessing the cyclical state of the economy. Unfortunately, the monthly time series is contaminated by many noisy components like seasonal variations, calendar and vacation effects. Only part of those nuisance components are explicitly considered in the seasonal adjustment procedures used by statistical agencies. In this paper, we propose a more flexible specification for the seasonal and working day effects and introduce an indicator for the summer vacations effect. We allow for time-varying parameters and show that the resulting Unobserved Components Model delivers more reliable results for the trend and cycle components of the production index. * I am grateful to a referee and the participants of the ifo Lunchtime Seminar, the Pfingstkonferenz of the Deutsche Statistische Gesellschaft and the annual conference of the Verein für Socialpoltik for helpful comments.     

General linear estimators under the prediction error sum of squares criterion in a linear regression model

In this paper, the notion of the general linear estimator and its modified version are introduced using the singular value decomposition theorem in the linear regression model y=X β+e to improve some classical linear estimators. The optimal selections of the biasing parameters involved are theoretically given under the prediction error sum of squares criterion. A numerical example and a simulation study are finally conducted to illustrate the superiority of the proposed estimators.     

[HDDA] sparse subspace constrained partial least squares

ABSTRACT

In this paper, we investigate the objective function and deflation process for sparse Partial Least Squares (PLS) regression with multiple components. While many have considered variations on the objective for sparse PLS, the deflation process for sparse PLS has not received as much attention. Our work highlights a flaw in the Statistically Inspired Modification of Partial Least Squares (SIMPLS) deflation method when applied in sparse PLS regression. We also consider the Nonlinear Iterative Partial Least Squares (NIPALS) deflation in sparse PLS regression. To remedy the flaw in the SIMPLS method, we propose a new sparse PLS method wherein the direction vectors are constrained to be sparse and lie in a chosen subspace. We give insight into this new PLS procedure and show through examples and simulation studies that the proposed technique can outperform alternative sparse PLS techniques in coefficient estimation. Moreover, our analysis reveals a simple renormalization step that can be used to improve the estimation of sparse PLS direction vectors generated using any convex relaxation method.     

Testing for parameter constancy in the time series direction in panel data models

We propose tests for parameter constancy in the time series direction in panel data models. We construct a locally best invariant test based on Tanaka [Time series analysis: nonstationary and noninvertible distribution theory. New York: Wiley; 1996] and an asymptotically point optimal test based on Elliott and Müller [Efficient tests for general persistent time variation in regression coefficients. Rev Econ Stud. 2006;73:907–940]. We derive the limiting distributions of the test statistics as T→∞ while N is fixed, and calculate the critical values by applying numerical integration and response surface regression. Simulation results show that the proposed tests perform well if we apply them appropriately.     

A Bayesian wavelet approach to estimation of a change-point in a nonlinear multivariate time series

ABSTRACT

We propose a semiparametric approach to estimate the existence and location of a statistical change-point to a nonlinear multivariate time series contaminated with an additive noise component. In particular, we consider a p-dimensional stochastic process of independent multivariate normal observations where the mean function varies smoothly except at a single change-point. Our approach involves conducting a Bayesian analysis on the empirical detail coefficients of the original time series after a wavelet transform. If the mean function of our time series can be expressed as a multivariate step function, we find our Bayesian-wavelet method performs comparably with classical parametric methods such as maximum likelihood estimation. The advantage of our multivariate change-point method is seen in how it applies to a much larger class of mean functions that require only general smoothness conditions.     

Bayesian Forecasting of Many Count-Valued Time Series

Abstract

We develop and exemplify application of new classes of dynamic models for time series of nonnegative counts. Our novel univariate models combine dynamic generalized linear models for binary and conditionally Poisson time series, with dynamic random effects for over-dispersion. These models estimate dynamic regression coefficients in both binary and nonzero count components. Sequential Bayesian analysis allows fast, parallel analysis of sets of decoupled time series. New multivariate models then enable information sharing in contexts when data at a more highly aggregated level provide more incisive inferences on shared patterns such as trends and seasonality. A novel multiscale approach—one new example of the concept of decouple/recouple in time series—enables information sharing across series. This incorporates cross-series linkages while insulating parallel estimation of univariate models, and hence enables scalability in the number of series. The major motivating context is supermarket sales forecasting. Detailed examples drawn from a case study in multistep forecasting of sales of a number of related items showcase forecasting of multiple series, with discussion of forecast accuracy metrics, comparisons with existing methods, and broader questions of probabilistic forecast assessment.     

Robust difference-based outlier detection

Abstract

In this paper, we propose an outlier-detection approach that uses the properties of an intercept estimator in a difference-based regression model (DBRM) that we first introduce. This DBRM uses multiple linear regression, and invented it to detect outliers in a multiple linear regression. Our outlier-detection approach uses only the intercept; it does not require estimates for the other parameters in the DBRM. In this paper, we first employed a difference-based intercept estimator to study the outlier-detection problem in a multiple regression model. We compared our approach with several existing methods in a simulation study and the results suggest that our approach outperformed the others. We also demonstrated the advantage of our approach using a real data application. Our approach can extend to nonparametric regression models for outliers detection.     

Using difference-based methods for inference in nonparametric regression with time series errors

Summary. We show that difference-based methods can be used to construct simple and explicit estimators of error covariance and autoregressive parameters in nonparametric regression with time series errors. When the error process is Gaussian our estimators are efficient, but they are available well beyond the Gaussian case. As an illustration of their usefulness we show that difference-based estimators can be used to produce a simplified version of time series cross-validation. This new approach produces a bandwidth selector that is equivalent, to both first and second orders, to that given by the full time series cross-validation algorithm. Other applications of difference-based methods are to variance estimation and construction of confidence bands in nonparametric regression.     

MODELING AND ESTIMATING VARIANCES IN REGRESSION

ABSTRACT

It is well known that ignoring heteroscedasticity in regression analysis adversely affects the efficiency of estimation and renders the usual procedure for constructing prediction intervals inappropriate. In some applications, such as off-line quality control, knowledge of the variance function is also of considerable interest in its own right. Thus the modeling of variance constitutes an important part of regression analysis. A common practice in modeling variance is to assume that a certain function of the variance can be closely approximated by a function of a known parametric form. The logarithm link function is often used even if it does not fit the observed variation satisfactorily, as other alternatives may yield negative estimated variances. In this paper we propose a rich class of link functions for more flexible variance modeling which alleviates the major difficulty of negative variances. We suggest also an alternative analysis for heteroscedastic regression models that exploits the principle of “separation” discussed in Box (Signal-to-Noise Ratios, Performance Criteria and Transformation. Technometrics 1988, 30, 1–31). The proposed method does not require any distributional assumptions once an appropriate link function for modeling variance has been chosen. Unlike the analysis in Box (Signal-to-Noise Ratios, Performance Criteria and Transformation. Technometrics 1988, 30, 1–31), the estimated variances and their associated asymptotic variances are found in the original metric (although a transformation has been applied to achieve separation in a different scale), making interpretation of results considerably easier.     

The Effects of Ignoring Level Shifts on Systems Cointegration Tests

Summary: In this paper I analyse the effects of ignoring level shifts in the data generating process on systems cointegration tests that do not accommodate level shifts. I consider two groups of Likelihood Ratio tests based on procedures suggested by Johansen (1988) and Saikkonen and Lütkepohl (2000b). The Monte Carlo analysis reveals that ignoring level shifts reduces the tests’ sizes to zero and causes an important drop in the small sample power for increasing shift magnitudes. This suggests that one should apply test procedures, which take account of level shifts. * This paper is a revised and summarized version of Chapter 3 of my PhD thesis (Trenkler, 2002). I would like to thank two anonymous referees for helpful comments on the submitted paper. Furthermore, I am grateful to Christian Müller, Ralf Brüggemann, and Helmut Lütkepohl for many useful suggestions and comments on an earlier version of the paper and the corresponding chapter of my thesis. The research was supported by the Deutsche Forschungsgemeinschaft (DFG) through the SFB 373 “Quantification and Simulation of Economic Processes” and the SFB 649 “Economic Risk”.     

Item count technique with no floor and ceiling effects

Abstract

Conclusions of sample surveys dealing with matters of privacy are often disputed. Many people either refuse to participate or provide false answers. Indirect questioning techniques are designed so that reliable estimates can be produced while the privacy of the participants is protected. One such method, the Item Count Technique is gaining increasing popularity. However, the original version of the method does not fully protect the privacy of the participants. In this paper we propose a new version of the technique which better protects the privacy of the respondents and does not have the so-called floor and ceiling effects.     

Ergodicity and truncation bounds for inhomogeneous birth and death processes with additional transitions from and to origin

ABSTRACT

In this paper, we present the extension of the analysis of time-dependent limiting characteristics the class of continuous-time birth and death processes defined on non-negative integers with special transitions from and to the origin. From the origin transitions can occur to any state. But being in any other state, besides ordinary transitions to neighboring states, a transition to the origin can occur. All possible transition intensities are assumed to be non-random functions of time and may depend on the state of the process. We improve previously known ergodicity and truncation bounds for this class of processes that were known only for the case when transitions from the origin decay exponentially (other intensities must have unique uniform upper bound). We show how the bounds can be obtained if the decay rate is slower than exponential. Numerical results are given in the queueing theory context.     

Truncated Poisson Regression for Time Series of Counts

We consider partial likelihood analysis of a truncated Poisson regression model for time series of counts. We focus our attention on the study of asymptotic theory for the maximum partial likelihood estimator of a vector of regression parameters. Simulations and data analysis integrate the presentation.