Working Papers:
Technology News and Business Cycles (January 2026)
Abstract: I study the macroeconomic effects of technology news shocks using a novel frequency-based index of technology coverage from a major U.S. news outlet for the period 1981Q1-2019Q4. The index is used to construct an external instrument that identifies technology news shocks in a structural VAR. The identified shock does not affect productivity on impact but accounts for a significant share of its long-run growth. A positive technology news shock induces an expansion in real activity, with output, consumption, hours, and investment rising before any productivity gains occur. The shock is also associated with a strong positive response of stock prices and a mild disinflationary effect. Technology news shocks emerge as an important, though secondary, source of output fluctuations at business-cycle frequencies.
The Dynamics of Unemployment and Hours with Rigid Wage Contracts (with Karl Harmenberg and Erik Öberg)
Abstract: Do wage-contract rigidity and variable pay matter for the Diamond-Mortensen-Pissarides theory of unemployment dynamics? Hagedorn and Manovskii (2008) suggested using the volatility of real wages and steady-state level of tightness to infer the worker bargaining weight and outside option, and found that with this strategy, the DMP model does explain the observed volatility of unemployment. We assess whether this finding is robust to amending the model with rigid variable-pay contracts, following Broer, Harmenberg, Krusell, and Öberg (2023). For given parameters, neither contract rigidity nor variable pay affect unemployment dynamics, but they do affect wage volatility in response to a productivity shock. With a realistic degree of contract rigidity, and variable-pay contracts calibrated to explain intensivemargin fluctuations in hours worked, the calibrated model cannot explain the observed volatility of unemployment.
Abstract: This paper examines whether traditional fiscal multipliers are mismeasured because of the way government output is valued in the U.S. National Income and Product Accounts (NIPA). Conventional methods value public services at their input costs, thereby neglecting the broader economic benefits households derive from these services. To address this issue, I compare the government spending multipliers estimated under two alternative methods for valuing government output. The baseline method follows standard national accounting practices, while the alternative method revalues government output by applying market-based markups to input costs, thereby constructing a proxy for the shadow price of public services. Theoretically, I show that if a market analog is both a close substitute in consumption and comparable in cost structure to the government-provided service, its market markup can serve as an informative adjustment to the input-cost measure, better approximating the unobserved willingness to pay for that service.
Building on these insights, I focus the empirical analysis on public education, a sector with a market counterpart. I construct an adjusted government education output series using market markups from the educational services sector and estimate cumulative spending multipliers under both valuation approaches. Under the baseline method, the multiplier peaks at 5.66, while the markup-based approach yields multipliers between 8% and 160% higher, depending on the horizon.
Work in progress:
The Propagation of Fiscal Policy Shocks through Production Networks