Is the risk aversion parameter in the simple intertemporal consumption CAPM "small" as in Hansen and Singleton (1982,1983), or is it that its reciprocal, the intertemporal elasticity of substitution, is small, as in Hall (1988)? This paper attributes the disparate estimates of this fundamental parameter not only to failures of instrument admissibility as do Hall (1988) and Hansen-Singleton (1996), but rather to failures of instrument relevance. That is, the disparate estimates reflect near nonidentification due to the unpredictability of asset returns and consumption growth. One natural identifying restriction from the risk aversion perspective leads to estimates that are low and stable over both time and model specifications. An equally natural identifying restriction from the intertemporal substitution perspective leads to estimates of the reciprocal that are also low and stable.