Global asset management: costs, productivity, and operational risk

In recent years the costs and productivity of operations in global asset management have come more and more under scrutiny. Over the last couple of years this sector has become significantly more cost conscious and many efforts are underway to drive down costs. However, increasing productivity and reducing costs affect the firms in many other ways as well; in particular, there is an increase in operational risk. It has become clear that running an asset management firm optimally requires a balance between costs and operational risk. In this article we discuss in detail the trade-offs between costs, productivity and operational risk in global asset management.

by Adjunct Professor Marcelo Cruz and Professor Michael Pinedo, Stern School of Business, New York University

Download article as pdf

The global asset management industry can be divided into two broad categories, namely, (i) institutional asset management and (ii) retail asset management. The two categories have fairly different cost characteristics, productivity metrics and risk metrics. The institutional asset management groups include the traditional investment advisory firms (for instance, UBS Asset Management), hedge funds and private equity funds (which may be either buyout or venture capital). This type of asset management company deals directly with large institutions and does not have to deal with the complexities of the retail business. The retail category includes all the mutual fund companies (for example Fidelity and Vanguard) and the pension funds (such as TIAA-CREF). The retail category must maintain individual accounts for individual people. The retail category requires a significantly higher level of technology (such as websites and call centres) than the institutional category. This is in order to do individual account registration, maintenance, accounting and a host of other activities. Asset management in general has undergone some major changes over the last two decades. These changes took place in two phases. In the first phase, which took place from the early 90s on until 2008, assets under management have skyrocketed because of the high liquidity in the global financial markets, which was caused by abundant credit and ever-increasing personal asset prices (for instance, house prices). This phase also witnessed the formation of hedge funds, in other words, asset managers who typically have a smaller number of clients and who can take very risky positions and bets that may yield substantial returns. These funds continue to play an important role in the financial market by exploring arbitrage opportunities.

Asset Management Phase 2

The industry is now experiencing a second phase of major changes. After the credit crunch crisis the sector has been suffering tremendously as world markets tumbled across the globe and liquidity, once abundant, became either very restricted or non-existent. Even when the markets recover it is not to be expected that the same level of liquidity and wealth will be in place. Many money managers in the US for the first time ever ‘broke the buck’, in other words, their funds’ quoted stock prices fell below one US dollar. Hedge funds that were created all over the world have been closing in droves. A sector that a few months ago thrived in a vast sea of liquidity faces today a very different reality.

Before the crisis most asset managers were not too worried about the costs or the risks involved since the ever-increasing personal wealth made their assets under management grow at a steady pace. Errors and high costs were buried under increased revenues and profit which came from a larger asset base and high returns in the world markets.

Today the story is very different. The largest asset managers have in many cases seen their assets under management fall by 30% or 40%, not only because of the drop in asset prices that impacted their funds but also because clients withdrew funds either out of necessity or out of fear of the financial conditions of their asset managers. The crisis also brought regulatory failures to light, such as the Bernard Madoff case which was one of the largest operational risk events in history. Many investors close to retirement lost their pensions not only because of market conditions but also because of a lack of caution and risk management from pension fund managers.

This new situation has forced asset managers to develop much more careful discipline around costs, productivity and risk management, concepts usually associated with manufacturing but that are starting to play a role within asset management. Each of these factors has received a lot of attention in academic journals and the trade press. However, the interdependencies and trade-offs between costs, productivity and operational risk have not been analysed in depth yet. For example, it is very likely that a reduction in costs, made without proper planning, can increase the level of operational risk exposure. In this article, we study these factors and their interdependencies in more detail.

Asset managers are susceptible to all forms of risks, namely market, credit and operational risks. These risks would manifest themselves in two ways: impact on client funds (in other words, indirect to the asset manager) or direct impact to the asset manager. The client funds are subject mainly to market and credit risk. Market risks are due to the daily fluctuation of asset prices and credit risks are due to the possibility that some counterparties, with whom the funds do business, can go bankrupt and make financial assets worthless. Such losses would have an indirect impact on the asset manager’s revenue as losses in funds entail a smaller commission; however, most of the losses are to the fund holder. Asset managers themselves are particularly vulnerable to operational risk. Errors in processing transactions or a system failure can cause severe damages and impact the balance sheet of the asset manager. Regular failures with regard to compliance with local regulations or very basic business ethics may generate large operational losses and subsequent reputational impacts. A sample of recent losses is listed in table 1 below.

The most important cost factors and the related risks in the two categories of global asset management are:

(i) the costs and risks with regard to human resources (that is, the employees – portfolio managers, administrators and so on);

(ii) the costs and risks with regard to system development and transaction processing;

(iii) the costs and risks with regard to customer contact centres and distribution channels (such as physical assets).

In what follows we elaborate on each one of the cost and risk factors mentioned above. It is clear that the first factor is important for both categories of asset management firms. The second and third factors are more important for retail firms than for institutional firms.

Human Resources: Costs and Risks

As a service sector firm, any type of asset manager needs to hire top-talent in order to provide the best return and service to the clients. Human resource talent is needed for:

(i) general management (for instance, portfolio managers);

(ii) front office (sales force and others);

(iii) administrative and support personnel (for example, internal auditors and iT support);

(iv) research (like equity, bond and currency analysts as well as risk analysts).

As is the case with many other financial firms, asset managers have to make sure that they are able to attract and retain, above all, portfolio managers with established track records and a potential to bring in clients and provide adequate returns to their funds. Such people are the face of the firm to the outside world and a basis for attracting new clients. Compensation of such personnel is probably one of the highest costs of an asset manager. Losing top talent is very costly and increases the susceptibility to operational risk as well. There is a learning curve for new people and during this period, the probabilities of errors are higher. Particularly in the US but also in other countries, funds are often named after their portfolio managers. Typically these portfolio managers have developed such a track record and reputation that clients want to invest with them. These funds linked to a name can hold many billions of dollars in investments and the asset manager starts to become very dependent on this particular person. The risk of losing such a portfolio manager may represent a loss of revenue of many millions per year in administration and performance fees. Asset managers are, therefore, exposed in a major way to key personnel risk.

In the front office, sales people need to follow procedures and local regulations when selling pension and other types of funds. Several pension mis-selling cases have occurred in various different countries. Probably the most infamous case of pension mis-selling was the situation that arose in Britain between 1988 and 1994, after British regulators decided to allow individuals to buy pensions from private-sector providers. The regulators determined at that time that pension investors should have a choice in the entity that would provide their pension and that it should not necessarily be their employer. They should be allowed to invest, in effect, in a retail pension fund. many who decided, or who were persuaded, to buy in a retail fund should not have done so. High-pressure tactics by commission-based salespeople led to tens of thousands of people buying products that proved to be entirely unsuitable. High fees combined with poor investment returns helped shrink the retirement savings of these investors and many found themselves locked in and unable to switch to more appropriate products without incurring very high exit fees. The result was a nightmare for investors, pension providers and the government. After a long legal process the funds were told to reimburse the investor for mis-selling these pensions. Until 2008 an estimated £11.5 billion (near US$20 billion) had been paid in compensation for mis-selling across many asset managers that operated in this market. This experience serves to illustrate what can go wrong when, even with the best intentions, a choice is given to people who are unprepared for it. It also shows how greedy sales people can exploit unsuspecting consumers, and how something that starts out as a good idea can turn into a major financial liability to asset managers if not properly conducted.

The back office is also subject to operational risk. For example, risk managers, auditors and accountants play an important role in any institution since they have to guard the firm against for instance rogue traders, accounting frauds and Ponzi schemes (like the aforementioned Madoff case). It is important that the proper due diligence is done with regard to any counterparty the firm is dealing with and it is also important that in the organisational structure of the institution the reporting lines of the traders and risk managers do not overlap.

The research arm is a very important part of any asset manager. Research analysts typically are highly compensated and are supposed to produce reports with sound recommendations. However, even the research division is subject to a significant amount of operational risk. The analysts often depend in their judgments on complicated models that few people understand. It has often been the case that the assumptions underlying such models do not make sense or that the evaluation of the models has been erroneous, resulting in substantial losses.

Systems Development and Transaction Processing: Costs and Risks

Scale plays an important role in asset management. The larger the portfolio, the lower the cost per transaction. However, the optimal size of a managed fund is often a balance of various tradeoffs. For example, a larger scale is preferred because of economies of scale; on the other hand, a smaller scale is preferred because a fund is then more nimble and will have an easier time meeting its benchmarks and outperforming the competition. Another aspect that has an impact on the optimal size of a fund is the error rates (operational risk), which typically is a function of the transaction frequency. It is to be expected that the probability of error increases with an increasing frequency in the rate of transactions. A larger fund, in order to meet its benchmarks, will have to take bigger bets. So, for each type of fund there is an optimal size and an optimal focus. Historically, several funds have reached sizes that apparently had been larger than optimal and therefore had to close entry for new customers (for instance, Fidelity’s Magellan).

Financial institutions in general, and asset managers in particular, traditionally have never been as careful with costs as other industries have been. In several industries, such as car manufacturing, the error rate is extremely low and very well controlled by sophisticated quality control departments, usually the most sophisticated area within an organisation after research or product development.

On the other hand, in the financial services industry, the most sophisticated departments are located either in the front office or on the revenue side. Financial derivatives are priced taking only market opportunity costs into consideration and rarely transaction costs; even if transaction costs are taken into account the analysis is not very deep. In the portfolio aggregation of these products, the final effects of processing are never considered. In this section, we try to briefly depict how a more sophisticated cost analysis can be developed for financial products based on a traditional microeconomics analysis.

Economic theory postulates that for a firm to maximise its results it is necessary that it produces such a quantity that allows equilibrium between the variation of the total cost and the variation of the total revenue. In economic terms, there are three types of revenues: total, average and marginal. The total or gross revenue is simply the result of multiplying the price p of a certain product by the quantity q negotiated. It can be represented by

R_GROSS = p.q

The average revenue is defined as the result of the division between the total revenue and the quantity negotiated. it is represented analytically by

or

The third representation of revenue, the marginal, corresponds to the variation of the total revenue in relation to the quantity sold. It is represented by:

Assuming that the variation of the quantity and the gross revenue can be infinitesimal (nice in theory but hard to imagine in business practice), the marginal revenue can be determined by the first derivative of the gross revenue in relation to the quantity sold:

or, for the sale of q units, we have the following relation:

In asset management the increased number of transactions (the ‘production’) will bring about an unexpected variable cost that is an increase in operational errors, in other words human and system factors would not perform the same when subject to a higher volume of transactions. The relationship between the number of operational errors and the transaction volume can be estimated through multifactor models. The entire analysis of revenues, production and costs based on (micro-) economic theory is complex and there is a vast literature on the subject. We will not delve into more details in this section but strongly suggest the understanding of these relationships when developing any growth strategy. It is worth noting that perhaps the most important conclusion from these relationships is that the profit will be maximised when:

C_Mg = R_Mg

The relation above says that the profit will be maximised when the marginal cost and the marginal revenue are exactly the same. This relation will hold for all cases and should be the objective of the strategy of the firm.

In what follows we present a simple example, which may help illustrate the theory above. Suppose a particular fund trades a single product with a very tight margin that is stable at 0.006% per unit trade (one unit = $100,000 notional) as seen in table 2. We can, therefore, simulate the revenue, which we do from 200,000 to 700,000 transactions processed per day. In general, the fund trader would only see the trades from the revenue side and is happy to see the revenue grow to $4,200,000 when 700,000 units are traded. This is a very general view, but revenue-generators will be very happy and would not care about the costs incurred to reach that.

Let us now analyse the costs. We divide the costs into two components; the processing costs and the error costs. The processing cost is expected to be stable at $5 per transaction. The error cost is $1.81 with a standard deviation of 3.89. Developing a 95% confidence interval for the error cost, we find that it would be $9.43. Therefore, on average it would cost $5 to process a transaction correctly and $12.43 to do a reprocessing because of errors.

We can also find a loss ratio. For this exercise we assume a simple linear model to relate the loss ratio to the number of transactions processed. The model is given by:

Loss Ratio = 0.0094957
+ 1.155737 x Transactions
R2 = 89.12%

By using the model above, we can verify that if the loss ratio is estimated to be 3.26% when the number of transactions is 200,000, the error rate climbs to 9.04% when the number of transaction grows to 700,000.

Following traditional optimisation analysis, the maximum profit condition, C_Mg = R_Mg, will be reached, given the current costs at 427,000 units traded. If the volume is higher, we have declining profits and need to adjust our processing capacity likewise. This type of modelling also offers us conditions to verify our capacity and see how an improvement in the process (for instance, system improvement, training process and hiring employees) will benefit the organisation and increase productivity.

In our example, the profit will be maximised at $190,052 when C_Mg = R_Mg =$6. Therefore, the average number of transactions was 239,815, given our current environment and capacity conditions; we would maximise our potential by trading 427,000 units per day. If the asset manager has a strategy trading more than that, it will have to take into consideration the costs as well.

We also performed a simulation with these data, assuming that we were able to cut the processing costs by 20%, which means, from $5 per transaction to $4 per transaction due to economies of scale. The modification is substantial. The maximum profit condition in this case is reached at 900,000 units per day, more than doubling our optimal capacity, as can be seen in table 3 below.

In another simulation, the loss ratio was cut proportionally to around 3% due to operational risk reduction by, for example, training of the employees and improving systems, and the error cost was reduced proportionally throughout the table. The maximum profit condition was reached at around 600,000 units per day. Therefore, the simple fact that we reduced the operational risk in a business unit made our optimal capacity increase by 40%. We had a dramatic productivity gain by managing the operational risk more effectively. See table 4 below for more information.

There are several other factors that affect the costs and the risks of transaction processing. Transaction processing can be outsourced (however, usually not off-shore, but preferably to a firm relatively close by so that any form of operational risk does not increase by too much). Another important factor is manual versus automated transaction processing (for instance, SWIFT). Automated transaction processing clearly has a higher productivity than manual transaction processing. However, automated transactions can only be done with regard to the more standard, plain vanilla transactions, not with regard to the more complicated esoteric transactions. Even though one would like to think that automated processing is more reliable and less susceptible to operational risk than manual processing, it is not clear that this is actually the case. For example, automated transactions are still subject to typographical errors; typographical errors often cause substantial costs to managed funds.

Costs, Productivity and Operational Risk in Customer Contact Centres

There are various channels through which an asset management firm can interact with its customers, namely:

(i) branch network (‘bricks and mortar’)

(ii) websites

(iii) call centres

(iv) sales force and broker/dealers

(v) correspondence and mailings of statements.

Each channel has its own capabilities and cost structure and is subject to its own set of operational risk factors. These different channels and contact centres, of course, have a strong interaction with one another. Each type of channel has its own advantages and disadvantages in its interaction with the client, whether the client is a pension fund or an individual investor.

The branch network usually entails a significant real estate cost. A network of branches is useful if the asset manager is a retail one. A branch may be useful in attracting new customers by facilitating the first face-to-face contact. Later an investor may communicate with the asset manager through one of the other channels. If the asset manager is an institutional one, then an extensive network of branches is usually not required. An asset manager may be content with having a small number of offices only in big cities. The operational risk at any given branch may depend on the likelihood of the occurrences of for instance floods, earthquakes and blackouts.

Both types of asset management firms have to make considerable investments in technology in order to provide their customers with internet access or access to operators. However, the website of a retail firm typically has to be more elaborate than that of an institutional firm. Retail firms typically also have large call centres in order to enable their customers to have access to operators. Websites as well as call centres have to be designed in such a way that they can handle peak traffic, since overload of a system may cause crashing of Internet access or excessive waiting times at call centres. It is not easy to forecast peak traffic, since traffic intensity may not only depend on the time of day, but also on random exogenous events such high volatility in the equities markets. Websites as well as call centres are, in general, subject to what is referred to as technology risk. A server going down may bring an entire system down and a website may then not be able to provide its customers with internet access. The system may also be subject to for instance hacking and phishing. The institution, in order to satisfy its clients, has to keep its website up-to-date continuously with regard to content as well as with regard to security.

The cost of a client doing a transaction via the web is clearly significantly less expensive for the company than the cost of a transaction via a call centre. On the other hand, a client who does his/her transactions via the web typically has a higher frequency of doing transactions than a client who does his/her transactions via a call centre. However, it is still in the interest of the company to direct as many transactions as possible to its website where the participant himself can enter the necessary information.

The most common transaction for call centres is actually an address change. On the web, the participant himself could do this but there must be an audit routine supporting it to make sure it is not a fraudulent address change.

Call centres require a great deal of sophisticated metrics and monitoring in order to be managed well. Among the metrics are such items as time waiting for call pickup and efficiency in assigning calls to multiple locations. A good system here makes it easier to balance workloads between two or more sites, monitoring the follow up of call centre people on their commitments to service. The performance characteristics, reliability and quality of service of call centres typically depend on the workload. The higher the workload, the more productive the operators are, but, on the other hand, the higher the operational risk, the lower the quality of service.

There are good reasons for diversifying call centres into various locations and even countries. Time zone variation is useful for locations offering 24/7 service. Obviously, for instance weather and earthquake risk support the use of diverse locations. A call centre can be off-shored and/or outsourced. Many financial firms in the U.S. have off-shored their call centres to India. The cost differential with regard to the salaries of the operators still seems to be significant. After making the decision whether or not to off-shore its call centres, the institution still has to make the decision whether or not to outsource its call centres. The outsourcing may be done with regard to the setup of the call centre and/or with regard to its management. The setup of a call centre is typically outsourced to companies that specialise in such operations. If the call centre is a large operation, then the management of the call centre may be kept in-house. Outsourcing and/or offshoring is done differently in the USA than in Europe, for various reasons. The number of languages the call centre of an american firm has to deal with tends to be smaller than the number of languages a European call centre has to deal with. The set of languages that have to be dealt with in North America is usually different from that in Europe.

Nowadays, the telecom technology is sufficiently advanced to allow call centre operators to work from home, with all the necessary access to company databases and have flexible hours. So a call centre manager, who suddenly sees a peak in incoming traffic, can mobilise a number of part time workers and bring them online within half an hour. The cost characteristics as well as the risk metrics are very much influenced by the training and the monitoring of the call centre employees. For instance, some call centres in the financial services world require very extensive training and licensing which can cause lots of operating problems if there is not a good compliance system in place. Also, there should be some monitoring on how difficult questions are answered or shepherded on to another operator. One factor that is very important is the turnover rate of the operators; in other words, the number of operators leaving each year. This turnover rate has a high impact on costs, productivity and operational risk. New operators have to be trained and are less productive in their initial period; their error rate is, of course, also higher than that of more experienced operators. The training period is costly as well.

The relationship between the sales force or broker/ dealers and the asset manager has to be designed very carefully. The salary and incentive structure has to be designed in such a way as to ensure that the sales force does the proper due diligence when it interacts with the clients. Two aspects of such interactions require scrutiny: first, a representative of the asset manager has to present the products and its terms and conditions of business in a correct manner and has to verify the appropriateness of the asset manager’s products and conditions with regard to any given client. He or she also has to verify the creditworthiness of the client. The salary structure of the sales force should not be based only on one up-front commission for any new client but rather on a more long-term commission structure that rewards a long term relationship with a client. Correspondence and statements, sent either by regular mail or by email, has a significant influence on the operations of the other contact centres. If all clients receive their statement on the same day, say, at the beginning of the month, then the call centres and websites will immediately be subject to a sudden peak in demand. Any peak in demand will immediately increase the error rate. It is therefore preferable that mailings to clients are, as much as possible, spread out over time. This has to be done of course with client consent.

Conclusion

It is clear that asset managers have to invest continuously in their human resources as well as in technology. Both areas represent major cost components and are subject to various forms of operational risk. As seen above, a reliable productivity/cost analysis cannot be done without a thorough analysis of the operational risk involved. Costs, productivity and operational risk are strongly intertwined. For a firm to optimise its investments and operations, all the possible factors and trade-offs have to be taken into account. Such an optimisation process may not be particularly easy.

Michael Pinedo is the chair of the Department of Information, Operations and Management Sciences at the Stern School of Business at New York University. His research interests focus primarily on operations in financial services.

Marcelo Cruz is an Adjunct Professor at the Stern School of Business at New York University. He is the author of the well-known book ‘Modelling, Measuring and Hedging of Operational Risk in Financial Services’ (published by Wiley).