From mid 1997 through mid 1999, the FOMC departed from its standard procedures by not raising the funds rate in response to increases in resource utilization rates, particularly as measured by increased labor market tightness. Greenspan's conception of policy explains this departure. Greenspan viewed policy as a forecasting exercise based on reduced-form relationships for predicting inflation, especially between unit labor costs and prices. However, these relationships changed unpredictably over time. As a result, policy had to be discretionary. This discretion took the form of “risk management.” Especially, Greenspan believed that monetary policy should counter the irrational expectations of investors, as long as actual and forecasted inflation were benign. Expectations were always central, but because they are not disciplined by a monetary policy rule, policy is necessarily discretionary.

In Measuring Business Cycles, Burns and Mitchell (1946) pioneered the atheoretical approach to forecasting used by business economists. They searched over the business cycle for empirical regularities, which they used to classify economic indicators as leading, contemporaneous, and lagging. Greenspan came out of this school. For him, monetary policy involved using empirically derived forecasting relationships in a way that allowed the FOMC's response to change when those forecasting relationships changed.

Greenspan (June 15, 2004, 11) wrote: “Policymakers have needed to reach beyond models to broader – though less mathematically precise – hypotheses about how the world works.” Greenspan (March 1997, 196) summarized his views: