Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Tensor-grid discretization of VARs is inefficient. In particular, when there are just a few variables or the VAR components are correlated, this approach creates large inefficiencies because some discretized states will be visited with only vanishingly small probability. I show how to construct an efficient grid by either pruning these low-probability states or working directly with sparse grids. Efficient grids vastly improve accuracy for a given grid size, or, conversely, vastly reduce the number of states required to attain a given level of accuracy.