 |
Home | Previous Page | |
|
||
 |
|
|
|
||
 |
|
|
|
|
|||||||||||||||
|
|
|||||||||||||||
|
|
Speech by SEC Staff:
|
This was prepared for the Keynote Address at the "Hedge Fund Regulation and Compliance Conference" on May 12, 2005 in New York City. The Securities and Exchange Commission disclaims responsibility for any private publication or statement of any SEC employee or Commissioner. This presentation expresses the author's views and does not necessarily reflect those of the Commission, the Commissioners, or other members of the staff. |
The subject of my presentation today is conflicts of interest in asset management. I want to focus upon how the economic interests of a decision-maker who serves as an agent for an underlying investor can diverge from that of the investor (principal). This theme potentially applies to many of the types of decisions that are relevant in the overall asset management process.
At the onset of my remarks I should emphasize, that of course, the views and perspectives that I am expressing today are my own and not those of the Commission or my colleagues on the staff. The approach and perspective that I will provide is based upon the "principal-agent" or agency paradigm, which is one of the central frameworks in modern economics. I view this framework as addressing many important themes related to asset management and reflecting many aspects of commercial relationships more broadly. While I will point to a range of contexts in which the incentives of the principal and the agent who acts on his behalf diverge, only in certain instances does the resulting behavior violate acceptable norms and become problematic. I also feel that I should reinforce both this point and my earlier disclaimer by noting that my perspective and training is that of a financial economist rather than as an attorney. My approach will point to a variety of situations in which there are incentive conflicts in an economic sense, but not in every instance behavior that would be deemed problematic in a legal sense.
Before turning to a number of applications concerning asset management, I think it is helpful to lay out some basic features of the "Principal-Agent" framework. The Agent acts on behalf of the Principal because of the agent's comparative advantage in some activities, which may derive from scale or scope economies. The delegation by an agent can be to another agent, i.e., the agency relationship can be nested and reflect a hierarchical structure. The incentives of the "principal" and the "agent" making decisions on his behalf differ for a variety of reasons.
The decisions of the agent are often viewed as expensive to monitor, even sometimes characterized as "unobservable." For example, the level of managerial "effort" would be difficult for the investor to monitor. Another way to summarize this is that an agent, such as a portfolio manager, often possesses considerable discretion, especially because the relevant information is not disclosed or is disclosed only infrequently or imperfectly, and relative performance is extremely difficult to judge (given the amount of noise in the market). Of course, the agent needs to be compensated for his services and the nature of the underlying compensation structure must be consistent with the agent being willing to perform the intended tasks. In other words, the overall level of compensation must be consistent with the agent's "participation constraint," so that the agent's utility is as least as large as he would derive from his best alternative.
Of course, the risk preferences of the investor and the agent who invests on his behalf differ. As an illustration consider the typical portfolio manager and his investing client. The wealth of the portfolio manager is typically limited compared to his overall set of clients and especially relative to the capital market as a whole. Consequently, the portfolio manager is not able to assume personally the full set of investment risks-i.e., he is relatively risk averse compared to investors, so that the risk preferences of the portfolio managers and underlying investors diverge.
The risk preference example illustrates the very general point that the business goals of the principal and agent differ. For example, the typical Investment Adviser obviously cares a lot about the fees he receives, while the investor (or the "principal" in my terminology) cares about the payoffs from the investments made on his behalf net of fees. Given the Investment Adviser's interest in the fees that he might receive over time, many Advisers are quite naturally very interested in growing their businesses, possibly beyond the size that their investment ideas might support.
This notion is closely related to the mechanism concerning how the Investment Adviser gets paid. Financial economists have often observed that the relationship between investment flows and therefore, both the size of a mutual fund and the total fees paid by a fund, are convex in the investment performance (e.g., Chevalier and Ellison (1997) and Sirri and Tufano (1998)). In other words, the marginal benefit to a mutual fund adviser of improved performance is especially great when the base level of performance is already strong. This means that the payoffs to a fund adviser are especially great when the manager's performance is particularly strong.
To pursue this example somewhat, both mutual fund and hedge fund advisers are embedded in a "tournament"-perspective clients are looking to past performance in allocating flows and are inclined to allocate new investment funds to the most successful managers based upon their prior experience. A number of years ago one very successful adviser expressed this notion to me in the following form-when competing for new assignments they invariably competed against managers with similar track records. This happens because prospective clients learn about the skills of manager from their past track records. At the same time the nature of the learning process is difficult because there is so much cross-sectional variability or noise in fund returns that is independent of the manager's skill. However, a strong implication of this perspective is that advisers may be incentivized to substantially to add to the risks that their funds bear if those risks are not fully understood or detected in the marketplace because of the value in the competitive marketplace for new funds and assignments of being among the winners to the managerial tournament. It also illustrates the importance and value of trying to create "track records" for new products, why some advisers discard less successful ones and why the track record of products being evaluated by institutional clients often exceeds substantially the subsequent realized performance.1
One additional dimension to the agency problem that I feel I should highlight as I frame the principal-agent problem associated with fund management is that in some contexts the fund adviser, i.e., the agent, works for many principals at the same time. In particular, there often is an allocation problem when the same securities are being purchased or could be purchased for many different vehicles. In a context with separate accounts this can arise as a byproduct of the separate account structure if the accounts are not treated equivalently. However, in some contexts, such as the example of a fund manager working for a variety of products in the same fund family there can be situations in which there are different sensitivities to various accounts due to such factors as incentive compensation, differential management fee rates as in the case of a manager investing for both a hedge fund and traditional mutual fund, the effect of past performance within a particular product and the convexity of the flow for performance relationship, and spillovers from "Star" funds. In such circumstances I feel that it is very important for the manager to have an objective and well-defined as well as fair process for allocating positions, because the potential for problematic conflicts of interest is considerable.
At the same time I want to emphasize that it may be very appropriate for the manager to have multiple clients or work with multiple funds. There are natural scale economics in the generation of information and in portfolio decision-making.2 This is basic economics and business. The issue in my mind is the role of the manager's self interest in such decisions across clients. To the extent that academics have identified situations, even at times in the aggregate, in which the agent's self interest appears to drive decisions that does give one concern.
Now I would like to explore in a bit more detail various examples of conflicting incentives that arise in asset management. I should note that the examples I have in mind, a few of which I already have briefly touched upon, are quite varied and arise at various levels-emphasizing the hierarchical nature of incentive conflicts. These issues arise in such settings as product distribution (between advisers and brokers) and its influence upon the investor's decision to select particular fund products, the production of information including the roles of "soft dollars" and analysts-which in the case of analysts relates to the allocation of "hot" IPOs (Initial Public Offerings), risk and asset allocation decisions by fund advisers, the role of fund advisers in protecting uninformed investors against "market timing" and late trading and following explicit voluntary disclosures specified in their prospectuses, the allocation of trades within a fund complex and finally, the order routing decision including the influence of "payment for order flow" and the import of "Best Execution" practices.
The first example that I would like to address is the case of product distribution in mutual funds. Clearly, the economic incentives of brokers and financial advisers with respect to retail clients will reflect how the brokers and advisers are paid, which conflicts with the underlying preferences of investors who "pay" the fee! It is often argued that mutual funds are often "sold" rather than purchased. In my view the value of mutual funds as an investment vehicle is that they provide a simple mechanism for unsophisticated investors to acquire diversified positions and benefit from scale economies and even for sophisticated investors to obtain diversification. That being said, however, it tends to be the more sophisticated investors who recognize the importance of the role of expenses in influencing overall returns. As it turns out, marketing is important to the distribution of products with above average loads and/or expenses. I have noticed that more sophisticated individuals have a much greater awareness of the value of index and other forms of passive investing and the value of investing through mutual fund groups with a reputation for low expenses.
Yet my comments certainly should not be construed as criticism of the overall mutual fund industry. Based upon both historical evidence and economic theory we know that the marketplace offers an expected return premium for bearing aggregate risk, making it valuable for individual investors to bear at least some risk as theory emphasizes that investors are risk neutral about very small risks. Yet many investors do not understand the mechanism by which they can undertake such investments and brokers and advisers play an important role in facilitating product distribution. In effect, the customer may need to pay for "education" about how to invest, facilitating the distribution of the mutual fund products. The prevailing structure of mutual fund loads and other fee-based mechanisms finance this important distribution function. Of course, the difficulty with this approach is how distorted can be the adviser's incentives-for example, how often do advisers recommend low-cost mutual funds? Another example of conflict of interest in the distribution process that has recently received attention is the "shelf space" payments made by some mutual fund organizations to brokers.
A different type of example of where the retail distribution process leads to agency difficulties is illustrated by the nature of many arbitration cases against retail brokers. Many of these complaints focus upon issues involving "churning" (artificially high turnover rates) and the "suitability" of specific assets for the customer arise as a byproduct of compensating brokers based upon the transactions they generate. To the extent that these complaints have a strong basis, they offer interesting examples of "problematic" conflicts of interest often associated with the distribution of individual securities.
Of course, market participants recognize and appreciate the value of information. Nevertheless, information is not the typical commodity. Indeed, it is an unusually awkward commodity from an economist's perspective in that it can be difficult for an information provider to price his services or even to identify a viable strategy for getting paid. At the root of this problem is the traditional perspective in modern economics that information is a public good in that, like the building of a road or police and fire protections, its benefits can be consumed by many individuals simultaneously, so that it is not efficient to exclude individuals from its benefits. The pricing of information for individual users is delicate as a direct consequence of the difficulty of excluding users from the benefits of information and the limited cost associated with the distribution of information, relative to its production. Consequently, information, like other public goods, can be under-provided relative to its low marginal cost, absent an explicit pricing or institutional mechanism to overcome these issues.3
With that introduction to the pricing of information, I think it useful to briefly discuss several market approaches to the pricing of information. The examples I will highlight are the services of equity analysts both in the context of "soft dollars" and the IPO process as well as the pricing of the services of credit rating agencies. While in each of these issuances various potential conflicts of interest have been widely identified and discussed, I think it is instructive to step back and reflect upon the nature of the underlying pricing model for informational services and the delicate nature of the economic viability of such products.
Because of the difficulty in getting investors to pay for research, issuers may stimulate research on their securities by paying for such research. Examples of this include the willingness of issuers to pay credit-rating firms and the importance of analyst coverage to issuers of IPOs. One context where asset managers are willing to pay for research is when the payment uses soft dollars that are generated as a byproduct of commissions. Notice that the soft dollars are really paid directly by the investors rather than by the adviser as the commissions are directly charged to the investors unlike direct payments for research, encouraging the adviser to purchase additional research. Of course, this is related to agency or conflict of interest concerns with soft dollars and commissions.
Also, note that analysts and research can be supported by alternative models of bundling. On the one hand, the "IPO model" uses the sell side to "bundle" the distribution of securities with sell-side analysis and research. On the other hand, the "soft dollar model" bundles the payments of commissions, i.e., transactional services, with support for desired analysts. Our universities face the same issue in getting paid directly for research, so they "bundle" research and education, potentially using tuition payments to help support the research enterprise. To some degree we can view the IPO mechanism and soft dollars as alternative mechanisms to help support research and as substitutes for one another. This illustrates how we can view the mechanism used to fund research as consistent with the inadequacy of the direct payment for research model.
This portion of the example illustrates that trying to remove all potential conflicts of interest may not be optimal if there are other frictions in the environment. At the core of this is the classic "public goods" nature of research, which is not an easy problem to remedy. While bundling is an attempt to resolve the public goods problem for funding research, the bundling approach itself can distort marginal incentives and lead to inefficiency. In both the IPO and soft dollar settings, this has been the subject of considerable attention in recent years. Yet because of the underlying public goods problem associated with paying for research, any attempts to address the agency problem underlying the IPO mechanism or soft dollars must be intertwined and recognize the difficulty of funding research by alternative routes.
On a prior occasion4 I spoke about the pricing of the services of credit rating agencies. Some of the source of value from the perspectives of credit rating agencies is the use of the ratings for various regulatory purposes. However, this requires that the ratings are widely available. This is not compatible with excluding users and charging the users for ratings. Instead, the approach of the credit rating agencies is to charge companies for rating their instruments, which is a natural approach for dealing with the exclusion problem. There is considerable demand by listed companies to have their companies rated. Consequently, the rating service is able to contract with and charge companies in return for rating their issues and exclude from ratings those companies that do not pay. In some instances, the rating services provide unsolicited ratings for companies that do not purchase ratings.5
There are a variety of reasons why the fund adviser has a different attitude toward risk than the investor whose funds are being managed. As previously discussed, the fund manager is interested in building his/her asset management business. As the manager competes for funds he is sensitive to competing in a tournament against other managers (in large part based upon past performance). This can encourage the manager to take on more risk. Similarly, in those situations in which there are explicit incentive fees (e.g., supplemental fees above a specified threshold) there can be inducements to take on more risk. Mutual fund and even hedge fund clients are sensitive to short-run performance in light of agency concerns and the continuing evaluation of the portfolio manager. This can lead the institutional manager to be reluctant to assume risks which are not reflected in the portfolio benchmark and not seize some of the opportunities offered in the marketplace. For example, most managers are reluctant to stray very far from their benchmarks. This is not only a rational response to benchmarking, but it illustrates some of the costs associated with agency issues in portfolio management. Indeed, despite the huge sums invested through hedge funds, these factors may limit the ability of these funds to eliminate market mis-pricing. These observations about the benchmarking process are reinforced by reflecting upon the role of "consultants" who can severely punish managers who do not closely track their stated benchmarks by withdrawing funds from such managers.
Much of the attention upon the mutual fund industry in recent years concerns issues associated with mutual fund valuation. "Market timers" and "late traders" exploited staleness in pricing. The costs of these strategies are borne by uninformed, passive long-term investors. In some of the cases "arrangements" were entered into with "timers" due to the potential management fee received by the adviser. In other instances the timers were employees or the fund did not follow policies detailed in the prospectus so that the fund violated its announced disclosures. The incentives of the advisers and long-term passive investors differed in these mutual fund valuation matters.
A recent academic paper documents strategic cross-fund subsidization of "high family value" funds compared to lower value ones within a mutual fund family.6 This study links the differential performance within a fund complex to both preferential allocation of IPOs and the extent of cross-trading within a complex. Of course, there are other illustrations of potential types of incentive conflict in asset trading. For example, even on an ex ante basis the early trades in a program will tend to obtain more favorable executions and less price impact. Perhaps more fundamentally, it is important for managers to avoid illegally exploiting the "look back" option they possess in trade assignment and instead make contemporaneous assignment to avoid inherent conflicts of interest. Among the sources of differing incentives within a fund complex across products are incentive compensation, differing fee rates, nonlinearity in the flow-performance relationships and the effect of "Star" Funds.
A final example of a conflict of interest in asset management is the "routing" choice of individual brokers in selecting a platform to which to route an order for execution. In many instances such decision-making has not been at the individual trade level and is indirectly influenced by the agent's receipt of funds for routing orders after receiving "payment for order flow" inducements to route orders. Indeed, my understanding is that "Best Execution" standards allow the routing decision to be defined on an ex ante basis rather than at a "trade-by-trade" level based upon the prevailing relationship in the quotes among markets. The specific nature of the conflict seems somewhat analogous to the case of certain expenses being borne by the mutual fund rather than its adviser-which can break the adviser's incentive to minimize costs.
What can we learn from these examples? First, I think that clearly stated and defined policies and procedures are helpful in the investment management process. This helps limit the types of conflicts of interest that might arise. By limiting needless "discretion" and articulating to everyone in a transparent fashion-e.g., employees within the organization, customers and regulators--the "intent" and "procedures" of the firm become clearer. I see this as useful from both a marketing perspective, as customers are increasingly sensitive to conflicts, as well as from the perspective of internal management. The development and communication of such guidelines makes it much easier to assess the policy of the firm and for regulators and outside auditors to audit "compliance" with the stated policy. Some problematic issues, such as ex post trade allocation, should be managed by asset managers by formulating and enforcing sensible policies.
Agency theory offers a valuable lens for assessing many regulatory issues. It provides a framework for identifying the potential differences in incentives among the parties to an economic situation. In fact, in various contexts in which products are bundled there are important potential agency conflicts. The recommendations of analysts, the incentives to elect soft dollars rather than reduced commissions and the allocation of IPOs to valuable clients are examples of the types of issues in which there are strong conflicts of incentives in the investment banking and trading processes. Various types of intermediaries, such as pension consultants, credit rating agencies, governance experts and investment advisers are simultaneously entering into business transactions with different portions of the financial network. On the one hand, this naturally leverages and develops the expertise and comparative advantages of the parties, but does leave the system vulnerable to potential conflicts of interest.
In the asset management process itself there are strong potential conflicts that arise because of the inherent nature of the fee and compensation relationship, differences in risk preferences between the investor and manager and the differential importance to the agent / portfolio manager of the multiple principals that he faces. For example, in the investment management process the allocation of hot IPOs by a mutual fund, exploiting implicit "look back" options when trades are not completely allocated in a timely fashion and trading for some accounts ahead of others to either front run or reduce price impact are examples of the types of problems that can arise. In fact, agency conflicts extend all the way through the distribution stream as a broker's or adviser's compensation structure influences the nature of his recommendations among investment vehicles. Empirical research has been valuable in identifying a number of facets of the agency problem in asset management.7
As an economist, I also should emphasize that the existence of natural conflicts in incentives does not necessarily imply a breach of duty or responsibility by the agent or "wrongdoing," but I do think the lens of agency theory certainly augments our understanding of investment management behavior and in some situations can help identify behavior that is problematic. In my view this perspective also highlights the importance and value of well-defined policies and procedures that are followed by the agent in directly mitigating conflicts of interest and also adverse consequences from potentially problematic situations. It also is crucial for the agent to understand the potential conflicts of interest that he confronts and disclose those when warranted.
I welcome your questions.
Chevalier, J. and G. Ellison, 1997, "Risk Taking by Mutual Funds as a Response to Incentives," Journal of Political Economy 105, 1167-1200.
Massa, M., Matos, P., and J. Gaspar, 2005, "Favoritism in Mutual Fund Families? Evidence on Strategic Cross-Fund Subsidization," Journal of Finance, forthcoming.
Sirri, E. and P. Tufano, 1998, "Costly Search and Mutual Fund Flows," Journal of Finance 53, 1589-1622.
Spatt, C., "Frictions in the Bond Market," keynote address presented at the Second MTS Conference on "Financial Markets: The Organization and Performance of Fixed-Income Markets" in Vienna, Dec. 16, 2004. http://www.sec.gov/news/speech/spch121604cs.htm
http://www.sec.gov/news/speech/spch051205css.htm
| Home | Previous Page | Modified: 05/17/2005 |