Tải bản đầy đủ
1 The general usefulness and 'industrial organization' of active portfolio management

1 The general usefulness and 'industrial organization' of active portfolio management

Tải bản đầy đủ

24

Bindseil, U.

debate on the usefulness of active management may in fact appear surprising, since already Grossman and Stiglitz had shown convincingly in
their seminal paper of 1980 that the question of the general usefulness of
active management is misplaced. Instead, active management needs to be
part of a competitive equilibrium itself:
If competitive equilibrium is defined as a situation in which prices are such that all
arbitrage profits are eliminated, is it possible that a competitive economy always be
in equilibrium? Clearly not, for then those who arbitrage make no (private) return
from their (privately) costly activity. . .We propose here a model in which there is
an equilibrium degree of disequilibrium: prices reflect the information of informed
individuals (arbitrageurs) but only partially, so that those who expend resources to
obtain information do receive compensation. (Grossman and Stiglitz 1980, 393)

Taking some complementary assumptions, Grossman and Stiglitz concretely model the associated equilibrium, being characterized by an amount
of resources invested into informational activities, and a degree of efficiency
of market prices.6 In equilibrium, a population of active portfolio managers
with comparative advantages in this profession will emerge, in which the
least productive active manager will just be at the margin in terms of earning
the costs associated to him. Berk and Green (2002) develop an equilibrium
model in which the flows of funds towards successful active managers
explain why in equilibrium, the different qualities of managers do not imply
that excess returns of actively managed funds are predictable. They assume
that returns to active management are decreasing with the volume of funds
managed, and the equality of marginal returns is then simply ensured by
the higher volumes of new funds flowing to the successful managers.
In a noisy real world equilibrium with risk, there will always be a significant share of active managers who will have generated a loss, ex post. In
equilibrium, anyway, active portfolio management will not be an arbitrage,
i.e. the decision to invest money in a passively or in an actively managed
fund will be similar to the decision to invest in two different stocks: it will be
a matter of diversification, and in a world with transaction costs and thus
imperfect diversification, probably also of personal knowledge and risk

6

In contrast to this view, Sharpe (1991) argues that necessarily, ‘(1) before costs, the return on the average actively
managed dollar will equal the return on the average passively managed dollar and (2) after costs, the return on the
average actively managed dollar will be less than the return on the average passively managed dollar’. He proves his
assertion by defining passive management as strict index tracking, and active management as all the rest. The two
views can probably be reconciled when introducing some kind of noise traders into the model, as it is done frequently
in micro-structure market models with insider information (see e.g. Kyle 1985).

25

Central banks and public institutions as investors

aversion. It appears plausible that any large investor should, for the sake
of diversification, at least partially invest in actively managed portfolios.7
This however does not imply that all large investors should do active
management themselves. For analysing whether public institutions should
be involved in active portfolio management, it is obviously relevant to
understand in general how in equilibrium the portfolio management
industry should look like. Some factors will favour specialization of the asset
management industry into active and passive management. At the extreme,
one may imagine an industry structure made up only of two distinct types
of funds: pure hedge funds and passive funds. This may be due to the fact
that different management styles require a different technology, different
people, and different administration. The two activities would not be mixed
within one company exactly as a car maker does not horizontally integrate
into e.g. consumer electronics (e.g. Coase 1937; Williamson 1985). It would
just not be organizationally efficient to pack into one company such diverse
activities as passive management and active management.
Other factors may favour non-specialization, i.e. that each investment
portfolio is complemented by some degree of active management. Indeed, the
general aim of diversification could argue to always add at least a bit of
active management, as limited amounts add only marginal risk, especially
since the returns of active management tend to be uncorrelated to returns of
other assets. In the case of a hedge fund, in contrast, there is little of such
diversification, as the risks from active management are not pooled with the
general market risks. It could also be argued that active management is
preferably done by pooling lots of bets (views), instead of basing all the
views on a few bets. One might thus argue that by letting each portfolio
manager think about views/bets, more comes out than if one just asks a few,
even if those are, on a one-by-one comparison basis, the better ones. Creativity in discovering arbitrages may be a resource too decentralized over
the population of all portfolio managers to narrow down the use of this
resource just to a small subset of them. Expressed differently, the marginal
returns of active management by individuals may be quickly decreasing,
such that specialization would have its limits.
7

Interestingly, the literature tends to conclude that index funds tend to outperform most actively managed funds, after
costs (e.g. Elton et al. 2003). This might itself be explained as an equilibrium result in some CAPM like world (because
returns of actively managed funds are so weakly correlated to returns of the market portfolio). This extension of the
CAPM of course raises a series of issues, in particular: The CAPM assumes homogenous expectations – how can this be
reconciled with active management? Probably, an actively managed portfolio is not too different from any other
company who earns its money through informational activities, and the speciality that an actively managed portfolio
deals with assets which are themselves in the market portfolio should after all not matter.

26

Bindseil, U.

Finally, one needs to recognize that portfolio management firms do not
tend to manage only one portfolio, but several ones, such that one firm may
have both actively and passively managed portfolios. It can then also pack
active and passive portfolios together into mixed portfolios, following e.g.
a so-called ‘core-satellite’ approach. Having actively and passively managed
portfolios in one firm may have the disadvantage mentioned above of
putting together two different production processes (like manufacturing
cars and consumer electronics), but at the same time it has the advantage to
allow for coordination within a hierarchical organization (e.g. in an optimized core-satellite approach).
As the fund management industry is made up of hedge funds, mixed
funds (i.e. tracking an index with some intermediate risk budget to deviate
from it, being organized or not in a core-satellite way), and passively
managed funds, it seems that neither of the two factors working in different
directions completely dominates the other. It is in any case important to
retain that for every investor, the decision to have some funds dedicated
to active management is at least to some extent independent of whether it
should do active management itself. In other words: once an investor has
decided to dedicate funds to active management, he still faces the ‘make or
buy’ decision determining the efficient boundaries of a firm. While the former decision is one which is to be modelled mainly with the tools of portfolio theory, the latter is one in which tools from the industrial organization
literature would need to be applied.
5.2 Public institutions and central banks as active investors
The fact that public institutions tend to have more complex and rigid
decision-making procedures, and less leeway in the selection and compensation of portfolio managers due to rules governing the employment of
public servants, could be seen as argument against genuine active portfolio
management. Being in competition with less constrained players also looking
for mispriced assets, active management could thus end up in losses, at least
if the fixed costs are correctly accounted for. Alternatively, it could be
argued that the private sector should not be overestimated either and that
there are enough financial market inefficiencies to allow also the central
bank to make additional money by position taking. Eventually, a good
performance attribution model should be in place to decide on whether
active management contributes positively to performance (see Chapter 7 of
this book). The conclusion may be different for different types of positions

27

Central banks and public institutions as investors

taken. What is important is to come eventually to a net cost–benefit
assessment, i.e. one in which the full costs of active management enter the
analysis. Also, possible positive externalities (point 9 in Section 2) would
have to be considered in this analysis. Also the fact that central banks are
insiders inter alia on interest rates could be seen to argue against active
management. It may however be possible to remedy this issue partially
with a Chinese wall, behavioural rules, or by precluding the kind of positions which are most subject to potential use of insider information, namely
yield curve and duration position in the central bank’s own currency. These
measures could also be combined, on the basis of some assumptions on
the likely relevance of insider information for different types of positions.
The specificity that public institutions do not have the task to maximize their
income, but social welfare, would also argue against active management of
central banks, as it is at least partially a re-distributive, zero-sum activity
(the argument of Hirshleifer (1971)). One could argue that active portfolio
management, being based on private information, is by definition not
compatible with transparency and accountability standards that should
apply for public institutions, and thus are not a natural activity for public institutions. Related to that, one may argue that active management
unavoidably represents a source of reputation risk. Central banks have special
reasons to develop market intelligence, since they need to implement
monetary policy in an efficient way, and need to stand ready to operate as
lender of last resort. Active management could be an instrument contributing to the central bank’s best possible understanding of financial markets,
which is useful for other core central bank tasks, such as monetary policy
implementation or the contribution to financial stability. A counterargument could be that intelligent passive portfolio management, using a
variety of different instruments, also force the staff to understand.
These specificities are affected by an outsourcing of active management
to private investment companies to a different extent. Depending on which
weight is given to the different pro- and con-active management specificities, one thus may or may not find outsourcing attractive. It has also been
argued that a partial outsourcing is attractive, as it provides a further reference for assessing the performance of active managers in general. On the
other side one may argue that outsourcing is itself labour intensive, and
would thus be worth it only if the outsourced amounts are substantial. It
has been estimated that at least two-thirds of central banks use external
managers for some portion or even all of their reserves.

28

Bindseil, U.

Table 1.8 Trading styles of central bank reserves managers according to a JPMorgan survey
conducted amongst forty-two reserve managers in April 2007

Buy and hold
Passive benchmark tracking
Active benchmark trading
Active total return trading
Segregated alpha trading

Describes exactly style

Describes somewhat style

12%
12%
64%
10%
2%

19%
24%
21%
12%
7%

Source: JPMorgan ‘New trends in reserve management – Central bank survey’,
February 2008.

One may try to summarize the discussion on the suitability of active portfolio
management for central banks and other public investors as follows. First,
genuine active management is based on the idea that private information,
or private analysis, allows detecting wrongly priced assets. Over- or underweighting those relative to the market portfolio then allows increasing expected
returns, without necessarily implying increased risk. There is no doubt that in
equilibrium, active management has a sensible role in financial markets. Second, while it is plausible as well that in equilibrium, large investors will hold
at least some actively managed portfolios, it is not likely that every portfolio
should be managed actively. In other words, it is important to separate the
issue of diversification of investors into active management from the industrial organization issue of which portfolio managers should take up this
business. Indeed, hedge funds, passively managed funds and mixed funds
coexist in reality. Third, a number of central bank specificities could appear
to argue against central banks being amongst the active portfolio managers.
There is, however, one potentially important argument in favour of central
banks being active managers, namely the implied incentives to develop
market intelligence. As it is difficult to weigh the different arguments, it is not
obvious to draw general conclusions. Eventually, central bank investment
practice has emerged to include some active management, mostly undertaken
by the staff of the central bank itself, and sometimes being outsourced.
Table 1.8 provides a self-assessment of forty-two central bank reserves
managers with regard to the degree of activism of their trading style, such as
collected in the JPMorgan reserve managers survey. It appears that the style
called in the survey ‘active benchmark trading’, i.e. benchmark tracking
with position taking in the framework of a relatively limited risk budget, is
predominant amongst central banks.

29

Central banks and public institutions as investors

6. Policy-related risk factors
Section 8 of this chapter will develop the idea of an integrated risk management for central banks. An integrated risk management obviously needs
to look at the entire balance sheet of a central bank, and at all major risk
factors, including the non-alienable risk factors (i.e. the risk factors relating
to policy tasks). This section discusses four key non-alienable risk factors
of central banks. While Section 3 explained how the underlying policy tasks
have made large-scale investors out of central banks, the present section looks
at them from the perspective of integrated central bank risk management.
Genuine threats to the structural profitability of central banks, which are
often linked to policy tasks, have been discussed mainly in the literature on
central bank capital. A specific model of central bank capital, namely the
one of Bindseil et al. (2004a), will be presented in the following section.
Here, we briefly review the threats to profitability that have been mentioned
in this literature. Stella (1997; 2002) was one of the first to analyse the fact
that several central banks had incurred such large losses due to policy tasks
that they had to be recapitalizd by the government. For instance in Uruguay
in the late 1980s, the central bank’s losses were equal to 3% of GDP; in
Paraguay the central bank’s losses were 4% of GDP in 1995; in Nicaragua
losses were a staggering 13.8% of GDP in 1989. By the end of 2000, the
Central Bank of Costa Rica had negative capital equal to 6% of GDP.8
Martı´nez-Resano (2004) surveys the full range of risks that a central bank’s
balance sheet is subject to. He concludes that, in the long run, central banks’
financial independence should be secure as long as demand for banknotes
is maintained. According to Dalton and Dziobek (2005, 3):
Under normal circumstances, a central bank should be able to operate at a profit with
a core level of earnings derived from seigniorage. Losses would have, however, arisen
in several central banks from a range of activities including: open market operations;
sterilization of foreign currency inflows; domestic and foreign investments, credit,
and guarantees; costs associated with financial sector restructuring; direct or implicit
interest subsidies; and non-core activities of a fiscal or quasi-fiscal nature.

In a recent comprehensive study, Schobert9 analyses 108 central banks’
financial statements over a total of 1880 years. Out of those, 43 central
8
9

See also Leone (1993), Dalton and Dziobek (2005).
See Schobert, F. 2007. ‘Risk management at central banks’, unpublished presentation given in a central banking
course at Deutsche Bundesbank.

30

Bindseil, U.

banks recorded at least once an annual loss, and 146 years of losses were
observed in total. She attributes 41 per cent of loss years to the need to
sterilize excess liquidity (which is typically due to large foreign exchange
flows into the central bank balance sheet), and 33 per cent to FX valuation
changes (i.e. devaluation of foreign reserves). Only 3 per cent would be
attributed to credit losses, and there is no separate category regarding losses
due to market price changes other than foreign exchange rate changes. In
other words, interest rate risks were not considered a relevant category,
probably because never was an end of year loss driven by changes of interest
rates. These findings confirm that policies, and in particular foreign
exchange rate policies, are the real threat to central bank profitability and
capital, and not interest rate and credit risks; although those are the types of
risks to which central bank risk managers devote most of their time, as these
are the risks that are controlled through financial risk management decisions, while the others are largely implied by policy considerations, which
may be seen to be outside the reach of financial risk management. However,
even if a total priority of policy considerations would be accepted, still the
lesson from the findings of Schobert and others is that when optimizing
the financial assets of a central bank from the financial risk management
perspective, one should never ignore the policy risk factors and how they
correlate with the classical financial risk factors. In the following, the four
main identified policy risk factors are discussed in more depth.
6.1 Banknotes, seignorage, and liquidation risk
The privilege to issue banknotes is a fundamental component of central
bank profitability, and hence the scenario that this privilege will lose its
relevance is one of the real long-term risk factors for central banks. For
a very long time, central bankers and academics have speculated about
a future decline in the demand for banknotes. Woodford (2001, section 2)
provides an overview of recent literature on the topic. One may summarize:
while there are a variety of reasons why improvements in information
technology (like the more systematic use of smart cards) might be expected
to reduce the demand for banknotes, it does not appear that those developments are in real competition to the main uses of banknotes, which
explain the high amounts of banknotes in circulation (of around EUR 1500
per capita in the euro area). Moreover, the actual use of e.g. smart cards has
progressed only slowly, while e.g. credit cards have been in circulation for
a long time. Goodhart (2000), for example, argues that the popularity of

31

Central banks and public institutions as investors

currency will never wane – at least in the black-market transactions that
arguably account for a large fraction of aggregate currency demand – owing
to its distinctive advantages in allowing for unrecorded transactions. The
evolution of banknotes in circulation over the last years in both USD and
EUR, has not given any indication of a decreasing trend (more the contrary). Another indicator against the hypothesis of a forthcoming disappearance of the circulation of banknotes and coins in the case of the euro
area is a look at denominations in circulations. In fact, only 83 out of EUR
620 billion of currency in circulation was denominated in banknotes with a
nominal value of less than EUR 50 or in coins, i.e. the typical transaction
balances. More than 50 per cent of the value of currency in circulation was
in denominations above EUR 50, i.e. denominations one rarely is confronted with. In line with this observation, Fischer et al. (2004) conclude
that the various methods they apply all indicate rather low levels of transaction balances used within the euro area, namely of around 25–35 per cent
of total currency.
In case one would be able to establish a probability distribution for the
evolution of banknotes in circulation over a certain, say ten-year horizon –
how to integrate exactly the risk factor ‘volume of banknotes in circulation’
into a risk management model for central banks? Two main types of risks
may be distinguished in relation to banknotes. First, a decline of banknotes
would imply a decline of seignorage, even if the assets counterbalancing
banknotes would be perfectly liquid. This is the topic of Section 7. Second,
in the short term, the decline of banknotes creates liquidation and liquidity
risk. ‘Liquidation risk’ is the risk that assets need to be liquidated before
the end of the investment horizon. In such a case, the originally assumed
investment horizon would have been wrong, and accordingly the assumed
optimum asset allocation would actually not have been optimal. ‘Liquidity
risk’ is the risk that due to the need to undertake large rapid sales, prices
obtained are influenced in a non-favourable manner. For the issue considered here (reversal of trend growth in banknotes in circulation), liquidation risk could appear more important than liquidity risk. It should be
noted that it is not only the uncertainty about the demand for banknotes
which creates liquidation risk. Similar risk factors are: (i) for domestic
financial assets, the need to build up foreign reserves; (ii) for foreign
reserves assets, the need to liquidate those assets for foreign reserve interventions; (iii) the need to buy assets for some other policy reasons, such as
emergency liquidity assistance to banks. The case with uncertainty of nonmaturing liabilities has been modelled e.g. by Kalkbrener and Willing (2004),