DTU406 PORTFOLIO MANAGEMENT Dr. Nguyen Thi Hoang Anh Lecture 6: Application: Portfolio Management and Style Analysis Contents Introduction Passive portfolio management Active portfolio management Benchmarks Style analysis 2 Introduction In the last lectures, we considered how to compute the optimal portfolio of risky assets, based on estimates of expected returns, the variance-covariance matrix and the risk free rate; when all investors have the same information about these inputs, they will all reach the same optimal portfolio, and this will be the market portfolio, which comprises all risky assets in proportion to their market values When markets are efficient, the optimal strategy for any investor - whether individual or institutional - is to invest in the market portfolio; this is known as passive portfolio management In practice, however, investors know that not all markets are perfectly efficient all of the time, and that not all securities are priced correctly; there is therefore an opportunity to engage in active portfolio management by identifying stocks that are mispriced in order to achieve a better trade-off between risk and return than is provided by the market portfolio This lecture provides an introduction to the methods of portfolio management and the evaluation of portfolio performance and style 3 Passive portfolio management Passive portfolio management in principle amounts to investing in the market portfolio, but in practice, this is infeasible given its size Instead, passive fund managers normally construct a relatively narrowly based portfolio that is designed to track a broad based index (such as the FTSE 100 or S&P 500) over time The manager is assessed by how well the fund tracks the target index, or benchmark Passive fund construction techniques include Full replication: investing in all of the stocks in the target index Sampling: investing in a representative selection of stocks in the target index Quadratic programming: investing in a portfolio whose composition is designed to track the target index as closely as possible (i.e. to minimise the tracking error of the portfolio) 4 Active portfolio management 1 Active portfolio management is based on the premise that there exist, at any point in time, certain stocks that, for various reasons, are either under- or over-priced Active portfolio management attempts to identify these stocks, and include them in an otherwise passive portfolio, either with increased weight (if they are undervalued) or decreased weight (if they are overvalued) Mispriced stocks may be identified using fundamental analysis (such as cash flow valuation, screening on the basis of ratios), or using technical analysis (such as chartism and the use of moving averages trading rules) 5 Benchmarks So far, we have analysed portfolio performance in the context of the CAPM; specifically, portfolio (or individual stock) performance is measured relative to the market portfolio; the risk of the portfolio (or stock) is measured by its CAPM Beta In practice, fund managers do not typically invest in the market portfolio; instead, portfolio managers specialise in countries, industries or types of stocks (such as value stocks or small stocks) Their objective is usually to create a portfolio that tracks a passive benchmark, such as an industry index, or an index of value stocks, but to add value through stock selection Moreover, their performance is measured with respect to that same benchmark In evaluating a portfolio manager’s performance, we need to know the style of the portfolio; if it is a portfolio of value stocks, it would not make sense to evaluate it against a benchmark comprised of growth stocks 6 Style analysis 1 In order to establish the appropriate benchmark for an active fund, we use style analysis The resulting benchmark is known as the style benchmark portfolio, which is a portfolio of individual style indices, reflecting passive investments in the different styles, such as value, growth, small-cap, large-cap, European equities, emerging market equities, government bonds, etc We can then decompose the performance of the active fund into (a) the performance of the fund that is due to the fund’s style (i.e. the performance of the style benchmark portfolio) and (b) the performance that is due to the fund manager’s stock selection ability We evaluate the active fund by comparing its performance with that of the style benchmark portfolio using the performance measures, such as the Sharpe ratio, Treynor’s measure and Jensen’s alpha 7 Style analysis 2 Introduced by William Sharpe Regress fund returns on indexes representing a range of asset classes. The regression coefficient on each index measures the fund’s implicit allocation to that “style.” R –square measures return variability due to style or asset allocation. The remainder is due either to security selection or to market timing. 8 Style analysis 3 Suppose that we are analysing the style of the Allianz NFJ Small Cap Value A fund To undertake style analysis for the fund, we first choose a set of style indices. For example, we might (rather narrowly) choose style indices for Large cap/Growth Large cap/Value Small cap/Growth Small cap/Value To replicate these style indices, we can use data on exchange traded funds (ETFs), which can be obtained from http://finance.yahoo.com/etf Here we have five years of monthly data on the fund and the four style indices 9 Style analysis 4 We then construct a style benchmark portfolio from these style indices using arbitrary weights (here set to 25% in each index), and calculate the tracking difference (the difference between the fund return and the benchmark return) The tracking error is the standard deviation of the tracking difference 10 Style analysis 5 To complete the style analysis of the portfolio, we use Excel’s Solver to select the style benchmark portfolio weights to minimise the fund’s tracking error, subject to the constraints that (a) the weights sum to unity, and (b) the weights are positive (because this is a mutual fund and hence not permitted to take short positions) After implementing Solver we have 11 Style analysis 6 Comparing the results of our style analysis with the style box from Yahoo Finance, we see that we have correctly classified the fund as a Small-cap/Value fund Note, however, that fund also has a non-negligible position in Large-cap/Value stocks, and in evaluating the performance of this fund, we should take this into account 12 Style-based performance measurement We are now in a position to evaluate the performance of this fund. In particular, we have estimated the appropriate style benchmark for the fund, and we can compare the performance of the fund with the performance of this benchmark We could use any of the standard evaluation measures described earlier, but here we will focus simply on the absolute performance In particular, we can compute the annualised average return of the fund, the component of this that is due to the fund’s style (which is given by the annualised average return of the benchmark) and the component that is due to the manager’s stock selection (which is the difference between the two) This fund appears to have performed rather well, with about a third of the fund’s return coming from the manager’s ability to select stocks 13 Practice exercise You are an analyst working for a fund management company. You have been asked to evaluate the style and performance of a mutual fund. 1. Download five years of weekly data for these funds from http://finance.yahoo.com/funds, and calculate simple returns. 2. Estimate a benchmark style portfolio for the fund, using style indices for value/growth and size. As proxies for the style indices, use data on the respective exchange traded funds from http://finance.yahoo.com/etf. 3. How does your classification of the fund compare with the classification given on Yahoo! Finance? 4. Evaluate the performance of the fund relative to the benchmark portfolio. 14