Umeå University's logo

Bias and MSE of the QME are studied by means of simulation. A bootstrap estimator of the QME covariance matrix is also included in the study. The simulation is based on travel distances reported in the Swedish Travel Habit Survey. The results are in accordance with the asymptotic properties of the QME. For example, the QME is better than other suggested estimators under asymmetric distributions of the error term. The results also suggest that the bootstrap technique is potentially useful for estimation of the QME covariance matrix.

Estimators of parameters in semi-parametric left truncated and right censored regression models are proposed. In contrast to the majority of existing estimators, the proposed estimators do not require the error term of the regression model to have a symmetric distribution. In addition the estimators use asymmetric “trimming” of observations. Consistency and asymptotic normality of the estimators are shown. Finite sample properties are considered in a small simulation study. For the left truncated case, an empirical application illustrates the usefulness of the estimator.

An estimator is proposed for semiparametric linear regression models with left truncated and right censored dependent variables. The estimator is derived from a moment condition following the principles of Newey [Newey, W.K., 2001. Conditional moment restrictions in censored and truncated regression models. Econometric Theory 17, 863–888] on conditional moment conditions. Consistency of the estimator is shown and simulation is used for illustration of the small sample properties.

Knowledge of consumer behaviour is important to companies and governmental authorities for, e.g, planning purposes. In this paper the relationship between interpurchase intervals of  cars and demographic and economic characteristics of the car owners are analysed. Data was collected within a panel study of households in Sweden. Due to the study design, data is left truncated and right censored. The degree of censoring was almost 80 percent which made estimation of the models challenging. Results show that income and age of the car owners are important for the length of the interpurchase intervals of cars. Moreover, the results imply that models recommended for interpurchase intervals of durable goods in general might be inappropriate for interpurchase intervals of cars. A generalized gamma distribution, which embeds several distributions, is used to model interpurchase intervals. Semiparametric models are also considered.