Passive management strategies have increased in popularity since its beginnings in the early 1970s, and have performed favorably relative to actively managed investment strategies, owing to indexing's low cost, broad diversification and minimal cash drag. For the past two decades, indexing has gained traction globally.
At ANPHIKO Asset Management we are specialized in the passive management of bond portfolios. 95% of our actual managed portfolios contain only government and government sponsored securities in euros and dollars.
Fixed income passive or indexed portfolios provide diversified exposure to the market which makes them attractive as a core investment. Historically, these passively managed investments provide returns that compare favorably to non-leveraged, actively managed funds. As a result, investment volatility is mitigated.
ANPHIKO Asset Management passive management approach is based on a well-defined and proven methodology.
The secret to index investing is that by minimizing both operating and transaction costs through a strategy of buying and holding all (or a representative sample) of the securities in a market index, the index investor earns something very close to the market return.
How do we do it?
In most cases institutional clients define their investment policy and corresponding strategic and tactical asset allocation. For individual mandates, they have chosen a representative benchmark index. However, as these benchmarks do evaluate over time (composition and average parameters as yield, duration, life etc.) and they could well not serve any more completely the purpose for which they were chosen, we actively assist the client in monitoring these benchmarks, providing specific data up to individual security level. We are able to propose alternatives, composed – or even tailor made benchmarks in different currencies.
Once the benchmark defined, a detailed analysis is made in terms of the relevant risk factors: overall statistics like average maturity or coupon will not suffice when trying to replicate the index.
The benchmark index will be partitioned in segments or cells with respect to term structure (duration and convexity), quality, sectors, currency, issue structure, cash flow distribution, etc.. Since mid-2000, the number of segments has nearly doubled, illustrating the evolution of our analysis and its accuracy, but also of financial markets and their behavior.
The country repartition is also considered. Having experienced the inter-country yield spread evolution (often combined with hefty currency movements) in the early nineties and yield-convergence towards the EMU (together with the adherence to the Maastricht criteria), the country segmentation has been considered as one of the principal drivers of performance and thus tracking-error. More recently we have seen evidence (and not only by the credit crisis) that individual markets can behave out of synchronization with others and that the factors influencing these phenomena are not always founded on rational behavior.
Once a portfolio is initially constructed, on future rebalancing dates, the portfolio is being dissected in the same way as described for the benchmark. This enables to visualize and calculate the differences between both and to point out those which the optimization will consider to bring back in line with the benchmark.
We distinguish between two portfolios: the first initial portfolio and the ongoing portfolio.
When constructing the first initial portfolio, the starting point is the full benchmark universe. By the previous described partitioning into segments, we will have to match the allocation per segment between the portfolio and the benchmark. In this way, the aggregate risk/return profile can be broken down into the contribution of each segment. This is the matrix from which the actual portfolio will be constructed.
The methodology used to construct the initial portfolio is based on a linear programming method. Linear programming is a more quantitative extension of the stratified sampling approach which is a simple and flexible procedure but yet labour-intensive and relying on the manager's expertise to choose the correct sampling intervals and select the appropriate portfolio issues. Linear programming applies a constrained optimisation algorithm over a specified universe from which issues can be selected. The definition of the universe is critical, especially with respect to pricing and liquidity: the method may prove non-optimal if a selected issue cannot be traded (found) at the given price. The objective function can be of various nature: a yield measure, transaction cost, expected total return... Its extremum value should correspond to a minimal tracking error, but it is clear that the outcome depends quite heavily on the choice of the objective function. The constraints fulfill multiple purposes: first, they implement the cell partitioning and the aggregate characteristics of the benchmark. Secondly, additional constraints can arise from diversification or other concerns that would enhance the replication. It should be noted that partial rebalancing (only a subset of the benchmark is re-optimised) is almost unfeasible with this method.
When repeated multiple times, we obtain multiple portfolios all matching the criteria in each segment but also on an overall basis. In all cases, the number of issues in the portfolio will be (much) smaller than the number of issues in the benchmark universe.
We thus come the second task: to quantify the contribution of a segment mismatch to the overall return and risk. These are not necessarily equal: the largest mismatches do not implicitly lead to the largest return difference nor do they necessarily carry the most risk. There might be correlations among the mismatches in the various risk factors. Often used are historical data to assess these risk/return correlations, or compare risk/return profiles among a number of scenarios involving yield curve movement, credit downgrades, changes in volatility or exchange rates... The goal is to identify a number of "key" risk factors, such as duration matching. We use scenario analysis and sensitivity testing to select the optimal portfolio(s).
Sensitivity testing is a Monte Carlo simulation of hundreds of thousand possible yield curve movements between defined extreme values analyzing the difference in behavior between the portfolio and the benchmark. The same is done by scenario analysis, but historic yield curves and their movements form the input.
The optimal portfolios are selected on the basis of their ability to mimic the performance of the benchmark with the smallest tracking error. The final selection is done by the manager taking into account the cash flow ‐ and transition (between segments) profile of the portfolio. Cash flow differences raise tracking error ("cash drain") while transition differences increase transaction volume (and thus generating also unwanted tracking error).
Further on, the initial portfolio forms the basis on the next check between the portfolio and the benchmark. This happens on the dates where the profile of the benchmark changes to reflect bond auctions, supply, calls, bonds entering/leaving the benchmark etc. Usually this occurs on a monthly basis, the benchmark composition being static during the whole month.
On these dates, the rebalancing dates, the earlier described steps are repeated, but as we have already a sample portfolio, we rely on the following methods to optimize our portfolio:
Many institutional investors have liabilities they must pay in the future, such as the retirement benefits that pension funds pay. To meet these future obligations, pension funds, insurance companies and other institutions invest a pool of assets with the goal of paying their future liabilities from the returns on those assets. If returns are insufficient to cover the liabilities, the institution must contribute capital to fund the liabilities.
ANΦKO Asset Management's liability-driven investing shifts the focus of asset allocation back to the real purpose of the assets, which is to meet liabilities rather than to outperform a market benchmark or peer portfolio that has no relation to the institution's liabilities. Thus, the defining element of a liability-driven investment approach is that portfolio performance is benchmarked against the institution's liabilities, rather than a benchmark with no direct relation to the liabilities.
ANΦKO Asset Management's liability-driven investing approach is the ultimate use of passive management.
Liability-driven investing is a paradigm shift in institutional investing. Institutions have long measured the success of their investment strategies against market benchmarks with no connection to the underlying goal of the portfolio. Today, at the urging of regulators, investors, rating agencies and corporate managers, a growing number of institutional investors are seeking to reduce the risks associated with their liabilities by implementing liability-driven investment strategies, which benchmark portfolio performance to the liabilities.
Liability-driven investing is, by nature, a custom endeavor. Essentially, the aim of a liability-driven investment solution is to customize a portfolio that will hedge the fundamental risk characteristics of an institution's liabilities. Because no two institutions have the same liabilities, no two liability-driven investment solutions will be exactly the same.
The first step in ANΦKO Asset Management's LDI approach is to analyze our client's liability cash flows as if they were a security or portfolio of securities. We determine the sensitivity of the cash flows to a range of market variables, such as yield curve shape, currency, inflation and credit spreads. In analyzing liabilities, we employ the same tools and risk measures we use on a day-to-day basis to monitor pour existing portfolios. Thus, our analysis of liability cash flows is consistent with the analysis of security cash flows we employ later in the process when we construct the custom LDI benchmark to match the risk profile of the liabilities.
Once we have analyzed the risk profile of the liability cash flows, we build a benchmark of liquid fixed income securities to reflect the liability cash flow characteristics.
When constructing the benchmark, we can set parameters around risk tolerance on specific risk measures such as duration, yield curve sensitivity and convexity. Additionally, we introduce specific constraints and bounds that allow us to construct such a benchmark that exactly matches the duration/convexity profiles of the liabilities, while allowing some latitude on other risk measures, such as curve sensitivity, to allow for a benchmark with a desired credit profile or a benchmark with a slightly higher yield, according to client guidelines.
We also conduct a series of "stress-tests" to evaluate the risk characteristics of the liability cash flows across a variety of market scenarios compared to the cash flow streams of the benchmark.
Our model is optimized in such a way that we do not have the need to deploy short positions or derivative securities, such as swaps and "STRIPS". Our optimization process produces a benchmark which is flexible enough to adapt to changes in liabilities. We regularly (usually on a 3 month basis) update the benchmark to reflect changes in the payment stream and preserve the match between the benchmark and the liabilities.
We do not "enhance" the benchmark to capture eventual available "market opportunities" or to add alpha. In our opinion this does not add to the portfolio's stability in the long run.