Generalized reduced rank regression

I introduce a technique to estimate parameters in regressions with reduced rank parameters in a general setting. The framework can handle a general class of parameter restrictions and allows for specifications with heteroskedastic and autocorrelated regression errors. Applications of this technique include: estimation of structural equations, estimation of reduced rank matrices in cross-section, panel, and time-series analysis, including estimation of cointegration relations in time series and panels. – Estimation ; Reduced Rank Regression ; FIML, Panel-cointegration, Cointegration with Heteroskedasticity and Autocorrelation