Background Indicators

Indicators of latent variables are usually assumed to be driven by the latent variable and some random noise. Background indicators are in contrast also systematically driven by variables outside the structural model of interest. This paper assesses instrumental variable estimates of effects of latent variables when a background indicator is substituted for the latent variable. It turns out that such estimates become inconsistent in empirically important cases. In certain cases the estimates capture causal effects of the indicator rather than effects of the latent variable. A simulation experiment that considers the effect of economic uncertainty on aggregate consumption illustrates some of the results.