This paper reexamines Bomberger's (1996) empirical support for the use
of measured disagreement across survey respondents for the Livingston
CPI series as a proxy for inflation uncertainty. We draw attention to
violations of modeling assumptions for maximum likelihood estimation
that result from the overlapping nature of the forecasts. Building upo
n the work of Rich, Raymond and Butler (1992), we employ a Generalized
Method of Moments (GMM) estimation procedure and document that disagr
eement does not provide significant predictive content for inflation u
ncertainty after accounting for Autoregressive Conditional Heteroskeda
sticity (ARCH) effects in the data.