Tracking Error Optimization with Corporate Social Responsibility Constraints

Summary

The document asks how to choose software or packages for minimizing a portfolio’s tracking error while adding corporate social responsibility constraints to an S&P 500 portfolio. The proposed study compares five specifications of expected tracking error: four linear forms attributed to Rudolf et al. and the conventional tracking-error variance attributed to Roll. It aims to examine achieved ex post tracking error, portfolio returns, and variance under those choices.

The author reports that a supervisor suggested Solver, but the available version cannot handle the desired number of variables, and the author is new to R. PortfolioAnalytics is mentioned, though the author does not know how to express the different tracking-error specifications in it. The document provides no optimization formulation, package recommendation, implementation steps, or empirical results; it is a request for guidance rather than a completed method. Its value is in identifying the modeling and tooling problem, while the actual comparison and conclusions remain unresolved.

Key ideas

• The proposed portfolio study adds corporate social responsibility constraints to an S&P 500 benchmark.
• It plans to compare four linear tracking-error specifications with a conventional variance-based specification.
• The intended outputs include ex post tracking error, portfolio returns, and portfolio variance.
• The author is seeking optimization software that can handle many variables and support the different formulations.

Tags

# What to use for Tracking error minimization


# What to use for Tracking error minimization












What programs/packages can one use to minimize a portfolio's tracking error? What I am trying to do is see what ex post TE, portfolio returns and variance can be achieved when adding CSR constraints to the S&P500 by running 5 different specifications of expected(ex ante) TE. 4 linear specifications from Rudolf et al (1999) and the traditional TE variance from Roll (1992).

My thesis supervisor told me to use solver but I don't have access to a version that can handle more than 100 variables and I am completely new to R. I have taken a look at some packages like PortfolioAnalytics but it is not obvious to me how I should put in the different TE specifications. In matrix notation the TE specifications are as follows:

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.