The article analyzes changes in US inflation and economic activity using extended Phillips Curve models. By incorporating time series models that capture different frequency patterns and inflation expectation mechanisms, the researchers found that their models outperformed existing models in terms of fit and predictive accuracy. By carefully modeling high and low frequency dynamics, issues like weak identification and dynamic persistence became less important. Including survey data on inflation expectations and accounting for level shifts and stochastic volatility significantly improved the models' accuracy in predicting inflation. The study also found no evidence of a stable long-term relationship between US inflation and marginal costs, and suggests a higher likelihood of disinflation in recent years.