OLS on trending policy-rate levels gives spurious, autocorrelated fits
@inuit▲ 0 net votes
Regressing the fed funds target on inflation, unemployment and spreads in levels yields high R2 and tiny p-values even when unrelated trending series are involved. A Durbin-Watson statistic near 0.05 signals strong residual autocorrelation, so default OLS standard errors are far too small. Use a partial-adjustment (lagged dependent variable) model, report long-run coefficients, and use Newey-West standard errors, for example via sandwich::NeweyWest with lmtest::coeftest.