max

Example 1, revisited (Instrumental Variable Regression, [38, 26, 28]). The method of instrumental variables (IV) provides a general solution to the problem of an endogenousexplanatoryvariable. Withoutlossofgenerality,considerthemodeloftheform Y "g0pXq`ε, Erε|Zs"0, (2revisited) where g0 is the unknown function of interest,Y is an observable scalar random variable,X is a vector of explanatory variables,Z is avector of instrument variables, andεis the noise term. Three observable variablesX,Y,Z (denoted by filled circles) andoneunobservableconfounding variableε. Note in this case the operatorsA1 and A2 are not compact [11] due to common elements.