Factor Exposure Analysis 118: Factor-based Return Attribution Analysis

Measuring asset allocation and selection decisions

January 2026. Reading Time: 10 Minutes. Author: Nicolas Rabener.

SUMMARY

  • Traditional attribution analysis is sector-based
  • However, this is not helpful for fixed income or quant strategies
  • A factor-based approach is universally applicable

INTRODUCTION

Most sophisticated investors seek to understand the drivers of both risk and return in their portfolios. Traditionally, they conduct performance attribution analysis to compare the portfolio’s returns against a benchmark index. The typical output decomposes performance into allocation, selection, and interaction effects, illustrating the impact of investment decisions.

However, few investors using tools such as the Bloomberg Terminal realize how dated the underlying methodologies are. The foundational work by Brinson and Fachler, who introduced sector-based performance attribution, was published back in 1985 – four decades ago. Since then, financial research has advanced considerably, and the analytical focus has shifted from sectors to factors.

In this research article, we explore factor-based performance attribution analysis, which offers a more modern and insightful view of portfolio performance.

CONTRIBUTION ANALYSIS

We construct five portfolios composed of U.S. equities and fixed income, with a fixed allocation of 60% to stocks and 40% to bonds. The fixed income allocation is consistently invested in U.S. investment-grade bonds, while the equity portion is allocated across several strategies: the S&P 500 (market capitalization-weighted and equal-weighted), small-cap stocks, momentum stocks, and energy stocks.

We then conduct a regression analysis using an elastic net model with a broad set of indices representing asset classes, sectors, and factors using daily returns of 2024. Importantly, we employ raw indices rather than residualized ones, as our previous research indicates that residualization often produces outputs that are more difficult to interpret (read Factor Exposure Analysis 116: Residualized Indices).

The regression results confirm that all five portfolios exhibit high betas to U.S. equities and fi