Decoding economic determinants of presidential approval: insights from macro-level machine learning models

Key findings from this study

This research indicates that:

• Public and private sector activity measures show the strongest and most persistent associations with presidential approval across the study period.
• Presidential approval exhibits substantial lag structure, with economically relevant information concentrating at short horizons and around one-year delays.
• Housing and mortgage indicators produce delayed approval responses, while recession indicators generate short-run temporary dynamics rather than sustained effects.

Overview

This study characterizes the relationship between macroeconomic conditions and U.S. presidential approval from 1953 to 2023. The analysis employs machine learning variable selection methods applied to a large macro-financial dataset, then structures identified predictors into six economic categories for dynamic analysis within a Structural Vector Autoregression framework.

Methods and approach

The study combines penalized variable-selection methods—Elastic Net and Smoothly Clipped Absolute Deviation (SCAD)—to screen a comprehensive macro-financial dataset for economically relevant approval predictors. Selected predictors are organized into six economic categories and analyzed using Structural Vector Autoregression to examine dynamic co-movement patterns between macroeconomic conditions and approval ratings.

Results

Presidential approval exhibits strong persistence with retrospective dynamics, with economically relevant information concentrating at short horizons and approximately one-year lags. Public and private sector activity measures demonstrate the strongest and most persistent associations with approval, while monetary and labor market conditions contribute systematically but heterogeneously across specifications.

Housing and mortgage indicators produce delayed approval responses, whereas recession indicators display short-run dynamics consistent with temporary rally-around-the-flag patterns. The dynamic analysis reveals that approval co-moves with macroeconomic conditions in differentiated temporal patterns, with response timing varying substantially across economic categories.

Implications

High-dimensional variable screening combined with dynamic structural analysis provides a complementary approach to parsimonious models in presidential approval research. The findings establish that approval responds heterogeneously to different macroeconomic dimensions, suggesting that single-indicator models capture incomplete economic relationships. Future approval research may benefit from integrating variable selection methods with dynamic frameworks to avoid specification bias from arbitrary indicator selection.

The temporal heterogeneity in approval responses to economic conditions indicates that policy effects and economic shocks operate through distinct transmission channels. Understanding these differentiated dynamics may inform both election forecasting and the design of economic policy timing relative to electoral cycles.

Scope and limitations

This summary is based on the study abstract and available metadata. It does not include a full analysis of the complete paper, supplementary materials, or underlying datasets unless explicitly stated. Findings should be interpreted in the context of the original publication.