The Role of Optimisation in Portfolio and Risk Management
Optimisation technologies have been used in portfolio management since the dawn of modern finance. Since the 1960s, fund managers have relied on optimisation technology effectively to determine what money to put into which stock to gain maximum returns and limited variability.
Optimisation technology came into its own during World War Two, and it was only a few years after this that operational researchers began to look at the use of optimisation in managing financial assets. Harry Markowitz first utilised this technology to improve returns on investments (ROI) in the stock market. Markowitz realised that being able to manage multiple assets that reacted differently in a variety of scenarios would increase performance of the portfolio and help reduce risk. Being among the first to use a computer for this purpose, Markowitz was able to take advantage of different stock prices to find instances where the prices were out of line. Due to the substantial profits that resulted from this finding, it wasn’t long before other economists caught on to this idea and formalised the technique of using portfolios to increase returns and reduce risk.
However, with information within the financial markets transferred very rapidly, this method of portfolio management became standard very quickly. Consequently, the ability to generate substantial profits declined, causing something of an arms race to increase the sophistication of technology and increase profit, which continues to this day.
As a result of this competition, portfolio management has become increasingly sophisticated and the restrictions and content of financial portfolios have evolved significantly. It goes without saying that investing in several stocks within the same sector is very risky, since most markets are highly correlated. Consequently, constraints have been put on the content of portfolios regarding the share of stocks in any given sector and, following the Asian financial crisis, in any one country.
Strong portfolios tend to reflect the market. Investing in stocks across the entire market, for example the Singapore 500, would be untenable since the large number of transactions and associated transaction costs reduces profit. Consequently, portfolio managers need to minimise the number of transactions but invest in a variety of stocks that mirror the market. This is designed to ensure profitable yields combined with low exposure to risk. Optimisation technology enables portfolio managers to do this by providing recommendations, within given parameters, on how best to allocate funds.
Many fund managers thought that the sophisticated analytics for measuring risk employed to date gave them a good handle on the risks they were facing. The near-collapse of several major investment houses and the subsequent fall in asset values in the past year has proved the contrary and many attribute the financial crisis to the supposition that these models were inherently flawed.
When the markets collapsed, portfolios were left exposed. However, this was largely due to what Donald Rumsfeld, Secretary of Defense under George W. Bush, famously termed ‘unknown unknowns’. Principles of modern finance divide the risk of an individual asset into systematic components, which broadly affect the entire market, and non-systematic components, that are specific to the asset. The latter can be mitigated by diversifying assets while the former is inherent. Optimisation can only help control known, non-systematic risk, for example that associated with company-specific factors. However, it was not possible for fund managers to protect their portfolios against inherent market risk since it is impossible to set, or for that matter imagine, the parameters required to protect against this type of risk.
The financial turbulence has turned many traditional assumptions upside-down. Among those being re-examined is the independence of risks. As discussed above, portfolio theory says that non-systematic risks can be mitigated though diversification, leaving only the systematic components. It has been assumed that this division is fairly stable over time, so that analysis of historic price movements can accurately gauge the magnitudes of these risk components. The recent crisis has revealed that market-led non-diversifiable risks can increase very rapidly, breaking historical trends and leaving even well-diversified funds unexpectedly vulnerable.
In order to address this, the industry will look to stress test risk parameters more broadly in order to increase their ability to withstand potentially unforeseen circumstances. Consequently, there is now a much larger focus on widening the scope of risk management models to account for a more extensive range of potential scenarios.
One of the key ways in which financial institutions can test the robustness of their risk models is through the adoption of Monte Carlo simulations. These originated from gambling and are mathematical models that generate data randomly, which you can apply to formulas. It allows banks to test risk models against a large set of scenarios, helping to validate the underlying assumptions. In terms of measuring risk in a portfolio of assets, it allows the manager to build a set of assumptions within a model that they can vary according to market conditions such as consumer spending and GDP. This allows portfolio managers to work out the impact of events against a portfolio and strengthen the model’s parameters accordingly, thus reducing risk and helping to build trust that the models will fully and effectively capture risks.
While it is impossible to account for every possible ‘unknown unknown’ scenario impacting upon a given portfolio, optimisation is key to reducing risk exposure. Having reacted to the recent volatility, fund managers must ensure that extensive stress testing is carried out on their portfolios in order to set parameters that will reduce the ongoing vulnerability to risk.