Master / Diploma
Various models for time-series data are being presented on the basis of the concept of stochastic processes. The first part of the class discusses univariate linear time-series models (AR(p), MA(q), ARMA(p,q), ARIMA(p,d,q), ARFIMA (p,d,q)). The second part then presents GARCH-class models for time-dependent variability. It also examines tests and estimation methods in connection with return distributions of financial assets. Also multivariate time-series models with VAR components and error correction mechanisms are discussed. Finally, the course examines the concept of the cointegration, periodicity of time series and models in continuous time.
Recent empirical studies in economic research reveal an increasing interest into the investigation of individual data, which typically occur as qualitative variables. Conventional regression techniques provide inappropriate results in this case. The main component of the course is therefore the maximum likelihood method, with numerous applications. In particular, binomial models (logit / probit), multinomial models, regressions with censored and truncated variables (tobit models), models for count data, duration and hazard rate models as well as models for panel data are discussed.
As part of the module MA/D-WW-ERG-1901 or MA/D-WW-MG, our chair offers a seminar on Health Economics in English in the summer semester. Participants will be awarded 5 ECTS upon successful completion. Please find detailed information in OPAL.