Date of Award

Summer 2026

Language

English

Embargo Period

7-6-2026

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Economics

Program

Economics

First Advisor

Ulrich Hounyo

Abstract

This dissertation studies the dynamic effects of economic shocks under uncertainty across three self-contained essays.

The first essay introduces a two-stage model averaging estimator for impulse response functions that flexibly combines local projection (LP) and vector autoregression (VAR) methods, as well as different statistical paradigms (Bayesian or frequentist). In the first stage, we compute weights within the LP and VAR groups. A second-stage weight then integrates the two groups by accounting for their structural differences and incorporating their respective strengths. Monte Carlo simulations illustrate a typical bias-variance trade-off: LP-based estimators tend to exhibit lower bias, whereas VAR-based estimators display lower variance. The proposed averaging estimators effectively balance these properties, leading to lower overall estimation error, particularly at intermediate horizons, and empirical applications further demonstrate that they yield smoother and more interpretable impulse response estimates than single estimators.

The second essay studies the finite-sample behavior of state-dependent impulse response estimators under uncertainty. Starting from a design where the true data generating process and its transition rule are known, misspecifications are added step by step, with performance evaluated by the continuous ranked probability score (CRPS) across single estimators and model-averaging schemes. Simulations show that CRPS is mainly determined by the transition process and the shock size: as shocks grow, LP estimators tend to outperform, though truncated-lag VARs can remain competitive, and Bayesian and frequentist estimators often move in opposite directions. The empirical analysis introduces three inflation-related regimes and compares the transmission of uncertainty shocks and monetary shocks across states. Uncertainty shocks are found to be state-dependent and build more slowly in favorable states, whereas the real effects of monetary policy shocks show little state dependence.

The third essay examines how US financial uncertainty shocks affect portfolio investment inflows in a group of 15 small open economies (SOEs) spanning advanced markets and regionally representative emerging markets. The analysis proceeds in two complementary studies, each conditioning on the state of global risk appetite to capture asymmetric transmission. The first study, using a threshold VAR, documents that uncertainty shocks transmit significantly to financial conditions, while capital inflow responses remain muted on average and adjust more rapidly than other variables. The second study, based on panel quantile regressions, reveals that shock effects on capital inflows are concentrated at the extreme left tail of the conditional distribution, providing direct evidence of tail risk in the SOE sample. Combining the evidence from both studies, two state-dependent mechanisms emerge. In the risk-off state, characterized by depressed global risk appetite, the negative shock effect on portfolio inflows is attenuated, consistent with an already-priced-in mechanism. Equity inflows further exhibit a bargain-hunting pattern where non-resident investors withdraw less from SOE equity markets, partially offsetting the negative shock effect. Both mechanisms are especially marked among SOEs with advanced and open capital markets.

License

This work is licensed under the University at Albany Standard Author Agreement.

Download

Share

COinS