But if the effect of WWI across different cities is random, and not correlated with the explanatory variables, then the study can still work despite WWI affecti

But if the effect of WWI across different cities is random, and not correlated with the explanatory variables, then the study can still work despite WWI affecting the target variable. It'll just be part of the error term.