Search-and-matching models give the match-specific productivity shock a central role in job creation, job destruction, and worker turnover, but because output is rarely observed at the match level, match quality has been difficult to measure directly. I infer it from worker flows in a matched panel of workers and employers built from LinkedIn profiles (Revelio Labs), 2007 to 2025. A Hidden Markov Model classifies employment spells into seven latent match states based on spell duration, exit outcome, and the worker's employment history. For each spell, I construct a match innovation, the part of the realized outcome that could not have been predicted when the match formed, as an empirical analogue of unexpected match quality, and aggregate these innovations to the firm level. Among Compustat firms, conditional on firm controls, a 10-percentage point shift in match composition from "churning" matches prone to separation toward internal growth tracks predicts 3.9 percentage points higher employment growth, a 1.5 percentage point higher probability of survival, and 4.8 percent higher sales. These effects are comparable to prior estimates for a one standard deviation increase in TFP across U.S. manufacturing establishments, suggesting that match quality is an important source of the firm level shocks that drive growth and survival.

This paper examines how systematic biases and heterogeneity in beliefs about the state of the economy shape labor market dynamics. I present evidence that households form dispersed, sluggish expectations, with more optimistic workers demanding higher wages. I develop a search-and-matching model with sticky wages and endogenous separations where workers and firms face information frictions about aggregate productivity. The gap between firm and worker beliefs drives unemployment volatility, while delayed firm learning raises persistence. Heterogeneity in workers' learning rates explains observed differences in employment transitions: workers with more sluggish beliefs are hired at higher wages and face greater separation risk in downturns.

Using household expenditure data and a new occupation-based measure of skill intensity, I document two facts. First, higher-income households allocate a larger share of their spending to skill-intensive goods and services, suggesting that income growth can raise the skill premium through demand. Second, holding real income fixed, consumption has shifted toward these goods and services over time, pointing to additional roles for technological change and relative prices. To quantify the sources of the rise in the U.S. skill premium, I develop a model that combines capital–skill complementarity, final-good-specific production technologies, and nonhomothetic preferences. I provide new estimates of capital–skill complementarity at the final-good level, showing that equipment complements high-skill labor and substitutes for low-skill labor. Equipment accumulation accounts for 83.5% of the predicted rise in the skill premium, and productivity-driven consumption upgrading for 8.4%.

This paper examines how financial constraints shape firms’ hiring and job-posting behavior. Leveraging micro firm-level data on vacancy postings and financial ratios, we document that firms with higher debt-to-asset ratios exhibit shorter vacancy durations. We argue that financially constrained firms are less selective in recruitment, resulting in lower match quality. To formalize this mechanism, we develop a search-and-matching model with heterogeneous firms facing financial frictions. In the model, constrained firms adopt lower reservation thresholds for match productivity, leading to higher worker turnover. Using this framework, we study the long-run productivity effects of policies that relax firms’ credit constraints.