2015 Assessment of the Walleye Pollock Stock in the Eastern Bering Sea

The general modeling approach remained the same. A more fully developed treatment of uncertainty in current-year fishery mean weights-at-age and those used for near term projections was included. This approach estimated the variability of cohort-effects and year-effects using a simple model developed separately from the assessment (where the year and cohort effects are treated as random variables). This model was tuned to the observed fishery mean weights-at-age with observation variances estimated from
a bootstrap resampling of fishery observer data. These same data were then included within the assessment model with the year and cohort effects estimated (but here as “fixed effects”) setting the variance terms to the values estimated externally. This allows for a more appropriate accounting for and propagation of uncertainty in values for mean weights used for estimating Fmsy uncertainty. This resulted in the risk-averse buffer between ABC and OFL (computed as 1-ABC/OFL) changing from 13% to about 22%, depending on selectivity configuration.