ARTICLES BY ROBIN ANGUS AND ANDY ADAMS FOR
WHOM THE BARBELL TOLLSAPRIL 2001
For whom the barbell tolls . . .
``The most notable piece of speculative [financial]
architecture of the late Twenties, and the one by which, more
than any other device, the public demand for stock was satisfied,
was the investment trust or company."
J K Galbraith, The Great Crash 1929,
Pelican Books, 1975, page 72
Attention-grabbing but slightly ominous words
from J K Galbraith in a book all investment professionals should
read. Investment trusts are, of course, still with us, and many
of them are the souls of simplicity and caution. However, the
public demand for investment trust stock in the UK is today being
satisfied partly by the issue of trusts of a novel typethe
so-called "barbell trusts". These are designed to offer
their shareholders, at issue, the growth prospects of a fashionable
investment area together with a good income. On the strictest
definition of the term, three barbell trusts were issued in 1999
and 15 in 2000, raising £2.7 billion of new money in alla
figure that would double were a looser definition of "barbell"
to be employed. Supporters of barbell trusts argue that they are
an attractive way for income investors to gain exposure to growth
sectors. We believe, however, that barbell trusts may be more
costly and riskier than investors, especially retail investors,
realise. As keen supporters of the investment trust sector our
motive in writing this aricle is to increase investor awareness
of the issues and risks involved.
What are barbell trusts?
Most investment trusts hold one single portfolio
of investments as backing for all their classes of capital. By
contrast, barbell trusts hold two distinct portfolios of investmentsa
growth portfolio and an income portfolio. In pictorial form this
structure can look like a "barbell" such as is used
in weightlifting.1 The diagram illustrates the structure of one
common type of barbell trust. On the assets side of the balance
sheet are two portfolios, one of growth-oriented investments such
as Japanese equities or technology stocks, and the other of UK
bonds and high yielding investment trust securities. On the liabilities
side of the balance sheet are bank debt and two classes of share
capital. The zero dividend preference shares (zeros) are entitled
to a fixed annual capital increment. The ordinary shares (often
called ``income & residual capital shares'' in this type of
structure) are entitled to all the trust's distributable income
together with whatever capital remains after the bank debt and
the zero dividend preference shares have been repaid.
The structure shown in the diagram happens to
be that of a split capital trust (split ie. a trust with more
than one main class of share captial. However, barbell trusts
and splits must not be confused. While a good number of barbells
are splits, most splits are not barbells. It is the division of
the assets side of the balance sheet into two separate portfolios,
together with a high level of gearing, which makes a barbell2.
How barbell trusts emerged
The recent boom is new issues of split capital
trusts is nevertheless bound up with the emergence of barbells.
Why? One reason is that both splits an barbells typically offer
at least one class of high yielding paper. While the yields available
in both equity and fixed-interest markets have fallen sharply
in recent years, investors continue to clamour for the high yields
they have been used to for decades. So high-yielding equity securities
such as those offered by barbells have found a ready market. Furthermore,
investors and fund managers alike have grown accustomed to remarkably
high annual total returns from equities17.7 per cent per
annum nominal and 10.7 per cent per annum real from the FTSE All-Share
Index over the past 25 years (source: Datastream).
This has had two effects:
Investors in and promoters of new
trusts assume continuing high returns. While the past is not necessarily
a guide to the future, 25 years is a lot of "past"!
Investors in high yield securities,
anxious not to miss out on what they see a freely available capital
growth, understandably are attracted to investments offering "high
yield with a touch of equity", such as the ordinary shares
of split captial trusts and barbells.
Lastly, banks have of late been eager to lend
to the investment trust sector, whereas until recent years it
was uncommon for trusts to have significant amounts of bank debt.
The bank know their lending is safe because there is ample collateral.
It is very unlikely that a bank loan to a trust would be defaulted
on, because the bank ranks first in order of priority and the
bank checks the convenanted cover on the debt each month. Therefore,
although bank lending inevitably increases the risk of capital
loss to holders of lower-ranking classes of capital, the banks
cannot be blamed for making large sums of money available. The
Boards, not the banks, are responsible for trusts' borrowing decisions
and how these affect shareholders.
Barbell trusts may be more costly and riskier
than investors, especially retail investors, realise.
Fees and expenses
Initial and recurring costs to ordinary shareholders
of barbells are much higher than may appear at first sight. First
come issue expenses, which include a significant element of marketing
costs. Issue expenses can be anything from 2 per cent to as high
as 4 per cent of the gross capital sum raised by a new trust,
including bank debt. So expenses expressed as a percentage of
the starting net asset value attributable to the trust's ordinary
shares will be every much higher.
If ordinary shares reprsented 50 per cent of
the gross capital subscribed incuding bank debt (as is the case
for the aggregate figures shown in Table 1) and issue expenses
were 4 per cent of the gross captial subscribed, the expenses
would approach 9 per cent of the ordinary shares' starting NAV.
If the ordinary shares were 33 per cent of gross
capital and issue expenses gain were 4 per cent of the gross capital
subscribed, expenses would be nearly 14 per cent of starting NAV.
There are startlingly high percentages. Perhaps it is not surprising
that such trusts often do not put a starting NAV of their ordinary
shares "up front" in the Prospectus. Then there are
management fees. These are typically 1 per cent of gross assets
per annum. At first glance, the figure looks reasonable by today's
investment trust sector standards and modest compared to the management
fees for many unit trusts and OEICs. However, the key point is
that again, management fees are charged as a proportion of gross
assets rather than net assets. This includes assets financed by
bank debt and zero dividend preference capital as well as by ordinary
share capital. So the holders of the ordinary share capital may
be paying management fees at the rate of 2 per cent or 3 per cent
per annum of the assets actually attributable to them. By contrast
with unit trusts, total expenses may be significantly higher than
management fees alone. Fitzrovia, a company that specialises in
monitoring expenses of funds, estimates that on average two-thirds
of total expenses for investment trusts are represented by the
Hence, other expenses could boost the figures
quoted above to an annual 3 per cent or 4 per cent of the assets
attributable to the holders of the ordinary share captiala
massive hurdle to clear before one even reaches the starting line.
Finally, most barbell trusts hold shares in other investment trusts.
So there are management expenses upon management expenses.
Accountingthe new flexibility
The investment trust Statement of Recommended
Practice (SORP) allows management expenses and debt interest to
be allocated between income and capital in different ways. Hasn't
this made the impact of costs and expenses less onerous?
Not quite. In fact, the SORP raises as many
questions as it answers. The SORP sets out the principle that
mangement expenses and interest costs should be allocated in line
with expected returns.
So if the Board expects 25 per cent of a trust's total return
to come from captial, one would expect that it would charge 25
per cent of management expenses and 25 per cent of interest costs
to capital. What, then, are we to make of those trusts, and there
are many, which charge no less than 75 per cent of such expenses
and costs to capital? It is clearly unrealistic to expect that
75 per cent of a trust's returns will come from capital if (as
is not unusual) it has a total portfolio yield more than double
that of the FTSE All-Share Index. We believe that the directors,
whose responsibility it is to sign the Directors' Responsibility
Statement in the Report and Accounts, should give this matter
careful consideration. If some Boards go against the spirit of
the SORP in charging to much to capital, others are not subject
to the SORP at all. 13 of the 18 barbells referred to in Table
1 are not investment companies under Section 842 of the Companies
Act and for them the SORP's writ does not run. Such companies
tend to charge even more to capital than do Section 842 trustssometimes
as much as 100 per cent.
Risks from bank debt
Universal to barbell trusts, whether conventional
or split captial in structure, is a sizeable layer of bank debt.
This is quite new in the trust sector, where gearing was traditionally
in the form of debentures or preference capital. The inclusion
of bank debt in trusts' capital structures means that a major
subscriber of capital to a trust now has the right to blow the
whistle and demand either repayment or changes to the portfolio
before the end of the game, regardless of the effect on the other
subscribers of capital (the various classes of shareholder). This
has already happened in the case of two barbells, European Technology
& Income and Framlington NetNet Income. It could happen to
"We remain nervous about some of the
barbell portfolios underlying these funds. The growth portofolio
is often highly volatile, while the income portfolio has more
capital risk than many investors think. Not only could the proportion
invested in other splits fall sharply if the underlying hurdle
rates are not met, but the high yielding (née junk) bonds
that are popular in some structures are quasi equity. The worst
case scenario for investors is therefore a gowth portfolio that
does not grow and an income portfolio that suffers defaults and
capital loss. In this instance high headline yields do little
to mitigate the overall losses that will be suffered."
Cazenove, Investment Trust Companies Annual Review, 10 January
A number of barbells are splits, most splits are
These words, from one of the most respected
brokers in the sector, are reminder of the portfolio risks run
by barbell trusts. The most obvious risk is that of investing
in a fashionable specialist area like technology. Whereas traditional
splits tended to invest in a broad UK portfolio with an income
flavour, follows of TMT over 2000 know only too well how risky
it is investing in a single ``gowth'' area. However, high-yielding
split securities and bonds can be risky too. It is therefore important
not to be misled by the illusion of conservatism offered by the
two portfolios of barbell trustsand legally, of course,
there is only one portfolio anyway. Assessment of risk for investment
trust securities often involves looking at hurdle rates, which
can be defined as the required annual growth rate of total assets
to repay a given class of shareholder at the wind-up date. The
hurdle rate does allow for the fact that the interest cost of
part, and in some cases all, of the bank debt (and possibly other
prior ranking capital too) is being met out of capital. However,
a significant proportion of a barbell trust's assets will be held
to generate income. Such assets may not preserve their capital
value. Because promoters' and investors' expectations of rates
of return from equities have been shaped by the experience of
the bull market of the last quarter of a century, the hurdle rates
for the ordinary shares of barbells may be perceived as being
less demanding than they actually are. Yes, there will have been
``wealth warnings'' galore. But how many investors in the ordinary
shares of barbells really expect to receive back significantly
less than the captial they have subscribed?
The "Magic Circle"
"As reverse leverage did its work, investment
trust managements were much more concerned over the collapse in
the value of their own stock than over the adverse movements in
the stock list as a whole. The investment trusts had invested
heavily in each other. As a result the fall in Blue Ridge hit
Shenandoah, and the resulting collapse in Shenandoah was even
more horrible for the GoldmanSachs Trading Corporation."'
J K Galbraith, The Great Crash 1929, Pelican
books, 1975, page 145.
Translate today's so-called "Magic Circle"
of split capital trust mangers (ie those whose trusts hold shares
in one another) back to the Wall Street of 1929 about which J
K Galbraith wrote, and it is easy to see what the worries are.
To begin with, cross-holdings make it hard to
disentangle the ownership of these trusts. It is also very difficult,
if not impossible, to pin down where unit trusts managed by fund
mangement groups within the "Magic Circle" hold shares
in them. So there is a problem of accountability and transparency.
Who owns What? But of even greater importance is what the inter-related
structure of cross-holdings could lead to in a falling market.
The risks created by geared trusts investing in other geared trusts
are very real. Substantial price declines in the ordinary shares
of some individual barbell trusts might all too easily become
a self-feeding downward spiral as the net asset values of the
ordinary shares of other trusts that held them fell in their turn.
Confidence in barbell trusts in general could thus ebb away, causing
still further price declines in their ordinary shares. If this
happened, many retail investors who were direct holders of such
shares would be disadvantaged and it is inevitable that their
confidence in the investment trust sector as a whole would suffer.
If barbell trusts started to unravel, confidence,
in the investment trust sector as a whole would be affected by
the adverse publicity. The ordinary shares of barbell trusts are
also more costly than some investors assume, in terms of fees
and expenses deducted from net asset value. Because barbells are
complicated and difficult to understand, they should put more
emphasis on communicating their investment characteristics to
investors. In particular, there is an urgent need for the significant
risks and expenses involved to be spelt out more clearly in their
Prospectuses and Report and Accounts.
Andy Adams is Director of the Centre for
Financial Markets Research, University of Edinburgh and Robin
Angus is an Adviser to the Centre.1 The term "barbell",
in a financial context, orignated in the bond market. "A
spread or portfolio position made up of short-maturity and long-maturity
fixed income securities with nothing in the middle." P Moles
and N Terry, The Handbook of International Financial Terms,
Oxford 1997.2 For a lucid exposition of the split capital prinicple
see "Dual Purpose Funds" by John Newlands, September
2000, Professional Investor (pp 14-17). For unit trusts
on average there is roughly a 90 per cent/10 per cent split for
management fees vs other expenses.4 Other (non-mangement) expenses
in most cases are charged to revenue account. The uplift in zeros
each year is taken to capital account.
"How many investors in the ordinary
shares of barbells really expect to receive back significantly
less than the capital they have subscribed?"
26 Merrill Lynch, 2001, Zero Dividend Preference Shares:
Understanding the risks and how to price them. Back