I am a Ph.D. Candidate in Economics at Princeton University.

I study how ** organizational and informational frictions ** at the microeconomic level shape ** macroeconomic fluctuations ** and how ** household heterogeneity ** influences aggregate outcomes.

I am currently on the 2024/25 academic job market.

My full CV can be found here and you can contact me at nikolakoudis@princeton.edu.

Incomplete Information in Production Networks

* (Ben Bernanke Prize in Financial and Monetary Economics)*

I develop a model of production networks in which firms choose inputs under incomplete information about sectoral productivity and aggregate demand disturbances. As in the rational expectations framework of Lucas (1972), firms use input prices as endogenous signals to infer broader economic conditions. Theoretically, I show that the effect of sectoral productivity shocks on aggregate output depends on the interaction between the economy's input-output structure and firm-level uncertainty. Specifically, shocks to upstream sectors have a larger impact on aggregate output during periods of high productivity uncertainty, while shocks to downstream sectors have a larger effect on output during periods of high demand uncertainty. When calibrated to historical U.S. data, the model generates a state-dependent measure of sectoral importance that diverges significantly from traditional metrics, particularly in economic downturns. On aggregate, the interaction between uncertainty and the U.S. economy's network structure dampened the impact of productivity shocks on aggregate output by 50% during Covid (a period of high measured demand uncertainty) relative to the Dot-Com bubble (a period of high measured productivity uncertainty) or non-crisis periods.

"A Theory of Supply Function Choice and Aggregate Supply," with Joel P. Flynn and Karthik Sastry. Paper, SSRN Link. April 2024.

*Revise and resubmit, American Economic Review*.

"Price Level and Inflation Dynamics in Heterogeneous Agent Economies," with Greg Kaplan and Giovanni L. Violante. Paper, Non-Technical Summary. August 2023.

*Revise and resubmit, Econometrica*.

"Prices vs. Quantities: A Macroeconomic Analysis," with Joel P. Flynn and Karthik Sastry. Paper. October 2024.

"Quick-Fixing: Near-Rationality in Consumption and Savings Behavior," with Peter Andre, Joel P. Flynn, and Karthik Sastry. Paper, SSRN Link. October 2024.

"The Economics of Segmented Housing Markets." Paper, SSRN Link. May 2024.

"An Assignment Model Approach to the Labor Share", with Narek Alexanian and Lukas Mann.

"Informational Gravity," with Elena Aguilar.

We develop a model of exporter dynamics in which aggregate uncertainty is endogenous to the exporting decisions of firms within a given market. Exporting firms provide an imperfect signal regarding the returns to exporting to rival firms within the same industry. This creates an informational externality arising through the extensive margin that can lead to informational "cascades" and multiple stochastic steady states in aggregate exports. The presence of this externality creates additional dynamic gains from the reduction in trade costs, which amplify the canonical static gains. The magnitude of these dynamic gains are decreasing in firm productivity dispersion and increasing in the (industry-level) elasticity of substitution. Our analysis suggests that temporary increases in trade costs can result in secular declines in aggregate exports through a permanent increase in firm-level uncertainty. We leverage text-based measures of uncertainty for OECD countries to test the theory.

"Rational Inattention: Signals and Precision."

This paper analyzes the continuous action rational inattention problem with binary states. I show that it is generally never optimal for an agent to acquire linearly dependent signals, which as a corollary implies that an agent will never choose more actions than there are states of the world. I demonstrate that the agent's optimal signal structure can be completely characterized through a single, univariate function. An intuitive condition on local convexity on this function determines the entire set of priors for which strictly positive information acquisition is optimal. Finally, I characterize state-dependent precision in information choice. In the widely used case of quadratic utility, I show that signal precision is invariant to the realization of the underlying state, thus providing a counterpart to the canonical Quadratic-Gaussian case when states are binary (Jung et al., 2019). I show that a necessary condition for state-dependent signal precision is variation in the convexity of marginal utility along the action space.