As climate warms, atmospheric wind patterns are predicted to shift poleward. However, different climate models disagree on the magnitude of the shift, and we lack a clear understanding because too many factors are at play. To separate the different factors, we extend a technique that "locks" a given effect to quantify the effects of (1) increased surface evaporation as the sea surface warms up, and (2) changes in cloud patterns contributing to anomalous warming in the atmosphere. We apply the new method in a simplified Earth-like model with global ocean to quantify the impacts on the three effects on the circulation shift in response to quadrupled CO2 concentration. In addition we show that across the future warming projections of 33 state-of-the-art climate models, about 60% of the model spread in the shift of southern hemispheric summertime surface westerlies can also be attributed to the same two factors (or more specifically, their contrasting changes between the subtropical and subpolar regions). Therefore, we should focus on improving the understanding and the model representation of these two factors for better predicting the wind pattern shift in the future.