This paper develops a novel two-step method to establish Generation Shift Keys (GSKs) on any desired aggregation level. Estimates are derived from econometric models, utilizing regression coefficients to measure how net position deviations affect power outputs, isolated from the influence of renewable energy generation and load. This enables GSK estimations based on actual common dispatch decisions, decoupled from fixed generating unit parameters and merit-order assumptions. The method's first step, the Generation Share Key (GShK), is applied to the four German TSOs. The obtained results are robust, suggesting the implementation for other countries with sufficiently stable markets. The second step is conducted for the generating units of 50Hertz's control region. For most analyzed units, size and variability of the estimated GSK values are appropriate regarding capacity and primary energy source. Also, the analysis discusses possibilities to account for the susceptibility of GShKs and GSKs to prevailing market situations.