SumOfCoefficientsPrior.tightness

SumOfCoefficientsPrior.tightness: float

Weight of each restriction row relative to a data row, \(\lambda_5\).

Larger binds the lag sums harder to one on the diagonal and zero off it; at the limit every equation is a unit root without error correction. Giannone, Lenza, and Primiceri parameterize the same rows by \(\mu = 1 / \lambda_5\), so their default of one coincides with this one and their “looser” is this field’s “smaller”. Must be positive.