Dynamic Projection of Federal Fiscal Revenue Under Reciprocal Tariffs and Macroeconomic Assessment of Manufacturing Reshoring Potential

Abstract

In response to the U.S. proposed "reciprocal tariffs" policy scheduled for implementation in 2025, this study constructs a quantitative analytical framework integrating machine learning forecasting with macroeconomic comprehensive evaluation. First, utilizing approximately 70,000 trade and tariff data points from 2020 to 2024, the study employs a log-linear model to estimate the tariff-trade elasticity coefficient at approximately -0.75, revealing the significant suppressing effect of tariff changes on import volumes. Building upon this, an ensemble learning model combining gradient-boosted decision trees and random forests is introduced to conduct dynamic simulations of fiscal revenue across different time horizons. Findings indicate that raising the average effective tariff to 50% would cause tariff revenue to surge by 181% in the short term. However, due to trade diversion and contraction, long-term growth would narrow to 32%. Subsequently, the study establishes a macroeconomic impact assessment system encompassing dimensions such as trade balance, price levels, GDP, and employment, and derives a "manufacturing reshoring potential index." Empirical analysis indicates that while high tariffs could temporarily reduce the trade deficit by approximately $12.8 billion and yield a marginal positive GDP contribution, the resulting surge in household consumption expenditures, steep rise in strategic resource costs, and reciprocal countermeasures from partner nations limit the Manufacturing Reshoring Potential Index to 33.3/100—a modest level. This study confirms that tariff measures alone are insufficient to achieve large-scale, resilient manufacturing reshoring.