14 Simultaneous Equations, SUR and Vector Autoregressions
Economic variables are usually determined jointly, and many questions concern a system of equations rather than one: supply and demand, consumption and income, wages and prices, the several components of a demand system. This section treats systems in three ways. Structural simultaneous equations models, the founding problem of econometrics at the Cowles Commission, raise the question of identification. Seemingly unrelated regressions exploit correlation between the disturbances of equations that are otherwise separate. And vector autoregressions, introduced by Sims (1980) as an alternative to structural models, treat every variable as endogenous and let the data speak; with cointegrated variables they become vector error correction models, and Johansen’s method tests for the number of cointegrating relations.
14.2 Identification
14.3 Estimating a system
14.4 Seemingly unrelated regressions
14.5 Vector autoregressions
14.6 Cointegrated systems: the vector error correction model
14.7 Systems in software
14.8 Practice Problems
Questions on this section
Stuck on something here? Ask below and it stays attached to this topic.