"A Case for Arithmetic Attribution" , abstract of a paper published in the Winter, 2012/2013 issue of the Journal of Performance Measurement.

In the ongoing debate over the use of Arithmetic versus Geometric attribution methods, Mark David believes that short shrift has been given to the Arithmetic view. Separating the concept of Management Effect from that of a full performance attribution methodology, he conducts a thorough, point-by-point evaluation of both in light of recent contributions to the discussion. He finds, with regard to Management Effect, that both methods are useful and answer different, meaningful questions about the performance of the manager. Yet, after demonstrating an attribution method that excels under all evaluation criteria – including proportionality - he concludes without hesitation that arithmetic attribution is a more precise and powerful tool for measuring the effectiveness of the manager’s investment process.

Download the associated spreadsheet.

"Bespoke Attribution: Illustrating the Manager's Story" , abstract of a paper published in the Winter, 2009/2010 issue of the Journal of Performance Measurement.

It is a commonplace in discussions of performance attribution that the methodology should reflect the investment process it purports to measure. Yet, in practice, we find managers have far more often implemented widely available “standard” methodologies than not. The resultant attribution reveals a palpable and frustrating gap between the investment story that the manager is trying to tell and the measures being used to illustrate it. In this paper, we explore recently implemented, real-world attribution methods that demonstrate exceptions to this observation. We conclude by walking through a case study that illustrates techniques for crafting an attribution methodology tailored to a specific investment process.

"Analyzing Performance of Global Portfolios" , summary of an article published in the May/June 2010 issue of the Investments & Wealth Monitor, by the Investment Management Consultants Association.

Portfolio performance analysis – comprising both performance measurement and attribution – is an indispensable tool for the investment manager. Appropriate performance analytics give the manager direct insight into and an objective evaluation of the effectiveness of their investment process. Perhaps more importantly, they also provide a concise illustration to the client that explains what the manager did to add value.

Managers of US-only portfolios who accept these basic tenets are very likely already to have in place a methodology and system for analyzing performance. The question we address in this article is: Given a robust and operational platform for US-only performance analysis, what is required to extend that platform to global portfolios, maintaining an equivalent level of analytic effectiveness?

"Practical Pointers for Implementing Transaction-based Performance Measurement" , a presentation delivered at the "Symposium on Measuring Performance & Risk" , hosted by Financial Research Associates , April 23-24, 2009.

Transaction-based performance measurement is the ne plus ultra for accuracy in performance analytics. As many have observed, however, its implementation is rarely straightforward or trouble-free. We highlight the pitfalls that may befall the unwary, and present a framework for maximizing feasibility and minimizing risk of a transaction-based performance initiative.

• Defining a simple and intuitive methodology
• Generalizing a model for all security types
• Segmentation (drill-down) and aggregation
• Dealing with accounting data idiosycracies
• Output data design
• Prototyping and testing for accuracy
• Preparing for attribution

"Alternate Attribution" , a presentation delivered at the "Crash Course in Attribution" conference, hosted by Financial Research Associates , September 19-20, 2007.

Brinson & Fachler introduced their framework over 20 years ago; it has since become the single most implemented methodology for performance attribution. More recently, managers' efforts to differentiate their investment products and processes from the pack have motivated them to take advantage of sophisticated models and tools. This evolution has resulted in a large and diverse zoo of proprietary investment processes that frequently bear no resemblance whatsoever to "Sector Allocation" and "Stock Selection".

Attribution effects should measure the effectiveness of individual elements of the investment process. In this presentation, we examine opportunities and challenges associated with matching attribution to reflect proprietary investment processes, including:

• Factor-based attribution
• Risk-adjusted attribution
• Liability-driven benchmarks and attribution
• Strategy-based attribution
• Decision-based attribution
• Tailoring an attribution methodology

"Transaction-based Performance: A Framework for Evaluating Measurement and Attribution Methodologies"

Published in the Journal of Performance Measurement
Spring, 2007; Volume 11, Number 3

In this paper, we lay out a set of criteria for comparing and evaluating the spectrum of performance methodologies from “Monthly holdings-based” to “Capture 100% of transaction activity”. In addition to specifying the components that a complete methodology should address, we offer a standard of comparison in the form of a feasible transaction-based methodology. By doing so, we explicitly attempt to raise expectations of the value and feasibility of transaction-based performance.

Download the spreadsheet associated with this paper.

Download a presentation of this paper given at the World Research Group's 2007 Performance Attribution conference.

Download another presentation, this time with a Fixed Income slant and case study, given at the World Research Group's 2007 Fixed Income Performance Attribution Forum.

"Sector-level Attribution Effects with Compounded Notional Portfolios"

Published in the Journal of Performance Measurement
Spring, 2006; Volume 10, Number 3

One of the drawbacks most commonly cited in criticism of the arithmetic, multi-period attribution linking method known as “compounded notional portfolios” (CNP) or “exact multi-period Brinson” is its lack of allowance for linking attribution effects at the sector level. This paper details an extension to the CNP method for deriving and linking sector-level effects that is simple, intuitive, consistent with common industry practice and precisely additive to the total level effects over any range of time periods. It then proceeds to evaluate this extension in the light of criteria which have previously been applied to the CNP and other linking methods.

Download the spreadsheet associated with this paper.

Download a presentation of the findings of this paper, delivered at the 2006 Northfield Information Services client conference.

"Why Are We Still Building Investment Performance Systems?"

White Paper, January 2006

In this paper, we discuss recent developments in the theory and utilization of investment performance measurement and attribution, and why these continue to push the requirements for technological solutions beyond what vendor-supplied, off-the-shelf systems can deliver.

Major shifts in Performance requirements on several fronts are covered, including:  

"Implementing a Fixed Income Attribution Model in a Business and Technology Environment: Practical Choices and Their Implications"

Presented at World Research Group Conference - Leveraging Performance Attribution Analysis for Fixed Income Investments;
April 28th, 2005

The team charged with implementing fixed income attribution in the context of an existing business and technology environment must choose from a large number of functional and design options. Each of these choices can dramatically affect not only the size and risk of the effort, but the ultimate usability and operability of the attribution system. We present a comprehensive framework to address and resolve these choices as an input to implementation planning and design and comment upon their implications from the points of view of both the business client and the systems implementer. Focus on:

The various enterprise clients for attribution and their differing requirements

Data sources, including holdings, benchmark and analytic

Sources of returns & contributions

Flexible drill-down in an online environment

Periodicity and linking

Pre-calc vs. re-calc strategies