Featured Image

How a Multi-Factor Attribution Framework Can Provide a Deeper Insight Into the Sources of Relative Performance

Risk, Performance, and Reporting

By Ravinder Dosanjh  |  April 20, 2020

Traditional performance analysis techniques such as benchmarking, peer grouping, and Brinson attribution have generally been used as intuitive indicators of a portfolio’s performance. However, they risk leaving managers blind to the areas of unintentional exposure and the true underlying drivers of return.

Using a multi-factor attribution framework or risk-based performance attribution allows managers to see and quantify the contribution from various factors to their portfolio’s performance. This allows them to discover any unintended exposures and truly differentiate between performance derived from systematic bets and performance derived from stock-specific contributions, which would be indicative of stock-picking skill.

The various factors that performance can be attributed to will depend on the specific factor model chosen during the analysis. Throughout this article we will be using the FactSet/Northfield Global Equity Risk Model that, while being a global risk model, contains regional fundamental and sector factors as well as country, currency, and economic factors. This provides an additional layer of granularity over the typical global factor model as factors will not necessarily behave the same across the world, at all times. For example, a UK Value portfolio may not necessarily perform the same as a Japanese Value portfolio, so splitting the factors out by region allows you to see these differences.

Brinson Attribution vs. Risk-Based Performance Attribution

To start, it’s important to understand the methodology of a more traditional approach, Brinson attribution, and of risk-based performance attribution.

Brinson attribution is a performance attribution model based on active weights. It is the most commonly used attribution model, in part due to its easy-to-understand nature. Excess returns are generally decomposed into allocation and security selection effects, as well as currency and other effects occasionally.

Risk-based performance attribution is a performance attribution model that utilizes a factor-based risk model. Excess returns are decomposed to active risk factor exposures, into a Risk Factors Effect (systematic), and a Risk Stock Specific Effect (stock selection). The Risk Factors Effect can be further decomposed across the individual systematic factors from the chosen factor model. Using the FactSet/Northfield Global Equity Risk Model, we can see how this is broken down:

pic 1

All attribution models have their strengths and weaknesses; a particular limitation of Brinson attribution is that the effects are constructed from the report grouping used. In the example below, the allocation effect is describing only the impact of allocation to sectors, as a sector grouping has been used in the report. If one was to change the grouping of the report to regions for example, you would expect both attribution effects to change in value.

This means that it is important to select a grouping that is relevant to the portfolio construction process; if not, the analysis can be meaningless or even misleading. It also means that we can only analyze the impact to performance from the allocation to a single factor at a time. This can represent an issue for a manager who has a stock selection process that is driven by multiple factors such as by sector allocation, but also a bias to high-quality stocks.

For this reason, risk-based performance attribution can be considered a multi-factor approach, allowing you to analyze the impact of exposures to multiple factors at once while not being impacted by the reporting grouping used.

Risk-Based Performance Attribution Example

In the example below, we are analyzing a sample European equity portfolio to illustrate how the methodologies differ. The Brinson model shows that excess return is predominantly due to stock selection with 289 basis points of the overall 390 basis points of outperformance coming from the selection effect. For an active manager who might pride themselves on being an expert stock picker, this is great news—but does it tell the entire story?

The risk-based performance attribution model shows something of a contrast. In this example, most of the 390 basis points are now shown to come from systematic risk exposures (292 basis points), while a smaller amount is still coming from stock-specific decisions (99 basis points). This implies that there were some factor tilts in this portfolio (intentional or otherwise) that were driving the majority outperformance, rather than pure stock-picking as suggested by Brinson.


We can then use the model to further decompose the Risk Factors Effect across the various factors in our chosen factor model. The majority of systematic outperformance is shown to be coming from fundamental/style factors as well as sector factors. The individual fundamental factors with the largest positive impact are Quality and Leverage, while Value has been a drag on performance.

If we focus in on the Quality factors, the portfolio overall was overweight Quality with a positive active exposure of 38 basis points. Since the FactSet/Northfield Global Equity Risk Model uses regional factors, we can then look at Quality at a more granular level. In the case of EU Quality, the overweight was a good decision as this factor had a positive return over the period and so it contributed 156 basis points to excess return. However, in the case of UK Quality, the overweight was not such a good decision as that factor performed poorly during this period, so the decision to have a positive tilt to UK Quality contributed a 39 basis points drag on performance. This highlights the benefit of having regional factors; we can see exactly which elements of the portfolio are driving the return at a deeper level—where Quality is behaving differently in the EU compared to the UK—than most other global factor models can.

This can also serve to highlight areas where unintended factor tilts are being made. Did this manager intentionally decide to tilt towards Quality in this portfolio (in which case we can conclude the decisions as described above)? Or is this just highlighting the fact that a “stock-picking” manager is actually making more systematic bets without necessarily realizing? The manager may believe that they are picking good stocks based on sound fundamental research or otherwise, but the model is highlighting that the stocks chosen share similar traits in the form of high ROA, ROE, and Cash Flow/Sales, which is how the Quality factors are defined in this factor model.


We can also look at this more graphically, focusing on the overall top/bottom factor contributors to excess return. In the chart below, where the size of the bubbles represents the contribution to excess return from that factor, we can see how the decisions made to over/underweight specific factors, along with their corresponding returns, have had an impact on the portfolio’s excess return.

This can be broadly split into four quadrants based on the x-axis and y-axis. The top right quadrant represents the area within which positively performing factors were overweighted, and the bottom left quadrant represents the area within which negatively performing factors were underweighted. The factors within these quadrants positively contributed to the portfolio’s outperformance.  

Conversely, the top left and bottom right quadrants represent areas where factor contribution to excess return was negative, either due to the overweighting of negatively performing factors or the underweighting of positively performing factors.


As shown, risk-based performance attribution has many strong points. When combined with more traditional methods of performance attribution such as Brinson, it can provide more comprehensive insight into the drivers of portfolio performance and can account for multiple layers of decision-making in a way that traditional methods like Brinson can’t. Active managers can control their factor exposures during portfolio construction to add value through both their opinions on systematic market drivers as all as expected stock specific returns. The ease at which risk-based performance attribution can be implemented make it a natural extension to more traditional methods.

With factor investing having become more commonplace in asset management, plus a greater awareness of factor tilts in both the financial community and their own investor clients, the question of how exposure to different factors contributes to a portfolio’s performance is an increasingly important question to be able to answer.

the performance function reshaped ebook

Ravinder Dosanjh

Specialist in Risk and Quantitative Analytics

Mr. Ravinder Dosanjh is a Specialist in Risk and Quantitative Analytics at FactSet. In this role, he is one of FactSet’s experts for Portfolio Risk and Quantitative Analytics and has spent the last four years specializing in workflows and solutions for portfolio and quantitative analytics including, but not limited to, factor research, portfolio construction, optimization, performance, and risk and factor attribution. Starting as a Consultant in 2012, he spent three years working with some of FactSet’s largest buy-side clients across the UK. He then joined the Analytics team in 2015 covering the same region, before working with clients across EMEA from 2019. Mr. Dosanjh is a CFA charterholder since 2017, and earned a Bachelor of Science in Economics from the University of Birmingham.