District manufacturing sentiment and state employment dynamics: local projections evidence from Federal Reserve survey data

Purpose This paper estimates the dynamic transmission of Federal Reserve district manufacturing sentiment indices to within-district state manufacturing employment, establishing whether district surveys contain geographically targeted labor market information beyond national indicators. Design/methodology/approach Local projections (Jordà, 2005) are applied to a panel of 16 U.S. states matched to five Federal Reserve district manufacturing surveys (Philadelphia, Richmond, Kansas City, New York and Dallas) over 1993m11–2026m6. Three survey subcomponents are examined – headline activity, new orders, and prices paid. District sentiment is further decomposed into a common component, shared across all five districts, and an idiosyncratic component specific to each state's own district, and both a placebo test and a horse race against rival districts' indices are used to validate the identification strategy. State fixed effects and a national manufacturing cycle control absorb permanent cross-state differences and aggregate shocks respectively. Findings District sentiment shocks generate positive, statistically significant, hump-shaped state manufacturing employment responses at all horizons h = 0 to h = 12, with 95% confidence bands strictly above zero throughout. The new orders subcomponent produces the most precisely estimated transmission channel (t > 4.0 at all horizons), followed by prices paid, identifying forward demand commitments as the primary mechanism. A placebo test in which states are randomly reassigned to a different district shows that the real own-district effect exceeds the 95th percentile of the placebo distribution at every horizon (empirical p-values 0.001–0.026), and a horse race against rival districts' indices confirms that the idiosyncratic, district-specific component of sentiment remains a significant predictor (t = 2.17–3.77) once the shared regional and national component is netted out. Research limitations/implications The crosswalk yields only 16 states across five districts, limiting the precision of clustered inference. The idiosyncratic sentiment component relies on a first-stage residualization against a common factor; although validated by placebo tests, imprecision at that stage could attenuate the reported coefficients. Time-fixed-effects estimates are noisier at short horizons. Coverage spans five of twelve Federal Reserve districts. Future research should extend the common-and-idiosyncratic decomposition to all twelve districts using Beige Book scores, disaggregate the analysis to metropolitan statistical areas, and apply the same identification strategy to other regional survey data where a shared cycle may similarly mask genuinely local signals. Practical implications Federal Reserve district surveys provide actionable subnational signals on employment dynamics that state-level policymakers cannot fully extract from national indicators such as the Institute for Supply Management (ISM) manufacturing index or industrial production data. In this context, demand-side interventions appear more effective in stabilizing regional manufacturing employment in the short to medium term than supply-side programs. These findings underscore the value of incorporating geographically disaggregated sentiment measures into policy design, particularly when addressing cyclical fluctuations in state-level labor markets. Social implications Findings imply that district manufacturing sentiment surveys carry geographically targeted employment signals that state labor market administrators currently underuse. New-orders improvements predict significant state manufacturing employment gains within one month, persisting for nearly a year – giving regional authorities a meaningful window to scale workforce development, unemployment insurance, and training capacity ahead of shifts. States more integrated into North American manufacturing supply chains respond faster, suggesting trade promotion and buyer-supplier facilitation programs generate labor market stabilization benefits beyond their direct export effects. Given ongoing United States-Mexico-Canada Agreement (USMCA) renegotiations, these results argue for embedding district-level sentiment monitoring into state and federal labor market forecasting infrastructure. Originality/value This is the first study to apply panel local projections to a state-district matched panel using Federal Reserve district manufacturing surveys as geographically assigned impulse variables. Results validate district surveys as subnational employment predictors beyond national indicators, and a novel decomposition of sentiment into common and idiosyncratic components, together with a placebo test and a horse race against rival districts, establishes that the district-specific component carries genuine informational content net of shared regional and national sentiment cycles.