Transcript The credit risk premia - Swiss Finance Institute

Senior Management Programme in Banking
Module IV: Asset Management
Professor Andrew Clare
Cass Business School
October 2012
Overview
 Strategic asset allocation
 Tactical asset allocation
 The tactical asset allocation game
 Alternative investments – how alternative are they?
 Liability driven investment
 Alpha – what value do active fund managers add?
 Choosing a fund manager
 Investment strategies – simple strategies for generating alpha
Strategic asset allocation
Professor Andrew Clare
Overview
 Asset allocation: what’s it all about
 Long-term expected returns
 Risk premia
 Expected risk & risk aversion
 Appendix: Yale university's endowment fund
Asset allocation


Emphasis on broad asset categories:

Equities, Bonds, Property, Currencies etc

US v UK equities etc
Main Practitioners:

Life Companies

Pension Funds

Funds of funds

Family offices
Page 4
A historic perspective on asset allocation
Real indices, logarithmic scale
100000
10000
UK equity
Gilts
T-Bills
1000
100
10
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Its simple: just buy equities !!!
Page 5
Long-term asset holdings of UK’s DB industry
100
% of total holdings
80
60
40
20
0
2003
2004
2005
2006
Equities
2007
Bonds
2008
2009
2010
2011
Other
But most institutional investors do not hold only equities.
Page 6
Strategic asset allocation

Strategic refers to longer-term outlook, bedrock of investment goals

Defining a benchmark for tactical asset allocation

Getting it wrong can be very costly

Should the aim be to

maximise expected return, or

maximise expected return, while simultaneously seeking to minimise expected
risk ?
Page 7
Using the MVE framework
A mean-variance frontier for asset classes
Expected return
Efficient frontier
Individual asset
classes
Standard deviation, risk

Often asset allocators make use of the MV framework

But to do so we need to know: expected returns, variances and covariances
to construct the frontier
Page 8
Long-term expected returns
Determining expected returns

Historic returns could be misleading – over the last ten years the FTSE-100
has fallen !!!

So asset allocators try to take a forward-looking view. We will try to do the
same and apply this view to:

Cash

Government bonds

Corporate bonds

Equity
Page 10
Long-term expected return components


There are three components of expected return on all assets

Ex ante real return

Compensation for future inflation

Compensation for risk
Let’s begin by determining the “neutral rate”, which comprises the first two
components
Page 11
The ex ante real return

In a world with no inflation and no risk, investors would still require a return
from their investments, but how much ?

It would depend upon the ‘opportunity cost’ of foregone consumption

It’s closely related to the potential growth rate of the real economy
Page 12
Average real GDP growth since 1970
Average annual, real GDP growth since 1970
3.50%
Average real GDP growth, %pa
3.00%
2.50%
2.00%
1.50%
1.00%
0.50%
0.00%
Australia
Canada
France
Japan
UK
USA
Long-run economic growth low: ex ante real return should be low too
Page 13
The ex ante real return

Despite many new inventions - railways, telephones, microchip, the internet
etc - economic growth has actually been remarkably stable

Perhaps then historic GDP growth will be a good guide to long term future
real GDP growth

On the other hand, is the credit crunch a paradigm shifting event … the end
of capitalism as we know it ?

Such estimates probably a good proxy for the long term ex ante real return

Yields on long-dated index-linked gilt market can give us a clue to what the
market thinks about trend growth
Page 14
Yields on long-dated index-linked gilt
Real, long term govt bond yield (UK)
6.0
6.0
5.0
5.0
4.0
4.0
3.0
3.0
2.0
2.0
1.0
1.0
0.0
0.0
-1.0
-1.0
-2.0
-2.0
-3.0
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011
-3.0
Yield on index-linked gilt, %pa
Yield on index-linked gilt, %pa
UK recessions
UK’s real long-term economic growth (was) similar to index-linked bond yield
Page 15
Yields on long-dated index-linked bonds
Short term real yields (pre-crisis)
Short term real yields (post-crisis)
4.0%
3.0%
3.5%
Short-term, real yields, %pa
Short-term, real yields, %pa
2.0%
3.0%
2.5%
2.0%
1.5%
1.0%
1.0%
0.0%
-1.0%
0.5%
0.0%
-2.0%
Australia
France
Italy
Japan
Sweden
UK
USA
Australia
France
Italy
Japan
Sweden
UK
USA
There’s clearly more to default-free real yields
Page 16
Compensation for future inflation

Inflation expectations affect the nominal expected return on assets

How does one go about forecasting inflation ?
Page 17
The recent low inflation environment
Inflation in a selection of developed economies since 1960
30.0
25.0
US
Japan
Germany
UK
Canada
Italy
Annual inflation %
20.0
15.0
Source: Thomson Financial
10.0
5.0
0.0
1960
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
-5.0
Will the low inflation environment stick this time?
Page 18
Inflation targeting
Inflation targeting: the 1990’s epedemic
Number of inflation targeting central banks
60
50
40
30
20
10
0
1970s 1980s
1990
1991
1992
1993
1994
1995
1996
1997
1998
Source: Bank of England
Inflation targeting has had a big impact upon the inflation environment
Page 19
Inflation targeting
Inflation targets in a selection of developed economies
Country/region
Euro-area
UK
Australia
Canada
New Zealand
Sweden
USA
Target/inflation goal
ECB aims to keep CPI inflation below ceiling of 2.0%
MPC aims to keep CPI inflation within 1.0% of 2.0% target
Australia’s FRB target inflation between 2.0% to 3.0%
Bank of Canada aims to keep CPI inflation within 1.0% of 2.0% target
Reserve Bank of New Zealand aims to keep CPI between 1.0% to 3.0%
Riksbank aims to keep CPI inflation within 1.0% of 2.0% target
Indications from Fed officials that 2.0% for core PCE inflation is “preferred”

Most seem to target between 2 to 3%

Why not target 10% or 0% ?
Page 20
Market “inflation expectations”
Are market inflation expectations consistent with targets ?
Break evens over time
Ten-year break evens
4.0%
10.0
Pre-crisis
USA
Australia
8.0
6.0
4.0
2.0
Dec-11
3.5%
Break even inflation rates, %pa
Ten-year break evens, %pa
UK
3.0%
2.5%
2.0%
1.5%
1.0%
0.5%
0.0
1985
0.0%
1988
1991
1994
1997
2000
2003
2006
2009
Australia
Canada
France
Italy
Japan
UK
USA
Page 21
Compensation for future inflation

In the UK it seemed reasonable in the past to assume inflation of around
2.0% (CPI), that is, 2.5% (RPI). But what about now ?

In Europe 2.0%

In USA – the Fed have just launched QE3 – an indefinite commitment to
expand the money supply

Today, arguably, the inflation picture hasn’t been this uncertain for some time
Page 22
Putting it all together: an example

Putting together an estimate of trend growth and expected inflation gives a
neutral policy rate for an economy

Neutral rate will be close to expected return on cash

For the UK prior to the credit crunch it might have been:


2.25% for growth

2.5% (RPI) for inflation

Giving a grand total of 4.75%
But what about now ?
Page 23
The ‘neutral rate’

Policy rates will cycle around their ‘neutral rates’

The return on cash will be closely related

These neutral rates can change themselves if:

trend growth changes (productivity improvements, labour migration, credit
crunch)

monetary policy regime changes

The return on cash is the basis for future expected returns on all assets

The risk premium is what distinguishes them
Page 24
Pre and post crisis policy rates
Policy rates in Jun ’07 and Dec ‘11
14.0
12.0
Jun-07
Dec-11
8.0
6.0
4.0
2.0
U
SA
K
U
ia
Ja
pa
n
M
ex
ic
o
Po
la
nd
R
us
So
si
a
ut
h
Af
So
ric
a
ut
h
Ko
re
a
Ta
iw
an
In
d
EU
na
C
hi
a
ad
C
an
az
il
0.0
Br
Policy rate, %pa
10.0
Page 25
Risk premia
Risk premia

Why do we want to be compensated for bearing risk ?

Risk inherent in investment classes distinguishes expected returns

Measuring risk premia is very problematic
Page 27
Risk premia on main asset classes

Government bonds – an ‘inflation risk premium’

Corporate bonds – a credit risk premium

Equities – the equity risk premium
Page 28
The “inflation risk premium”

Biggest risk in holding conventional, govt bonds is inflation

In past governments have arguably “inflated away” their debts – they may be
tempted to do this again

Investors demand an additional return, mainly because future inflation is
uncertain (other risks too)

It will depend upon the:

the monetary policy framework and

the credibility of monetary authorities
Page 29
Calculating an “inflation risk premium”
Yield on Conventional government bond (Gilt)
Minus
Yield on index-linked government bond (ILG)
Minus
Estimate of expected inflation (survey based)
Equals
Measure of inflation risk premium
This gives a good proxy for the risk premium on government bonds
Page 30
Inflation risk premia
Measure of the inflation risk premium for gilts
Change in BRP (2007-2011)
3.0
Australia
Canada
France
Italy
Japan
UK
USA
0.0%
Inflation risk premium
Moving average
2.0
Change in bond risk premia, %pa
Bank of England policy rate, %pa
2.5
1.5
1.0
0.5
0.0
-0.5
-1.0
1993
-0.5%
-1.0%
-1.5%
-2.0%
-2.5%
1995
1997
1999
2001
2003
2005
2007
2009
-3.0%
It’s fallen everywhere, but not because of receding fears of inflation
Page 31
Risk premia on main asset classes

Government bonds – an inflation risk premium = 0.50% to 1.00% ?

Corporate bonds – a credit risk premium

Equities – the equity risk premium
Page 32
The credit risk premium

Credit premium additional return over equivalent govt bond to compensate
for credit risk

Varies according to the type of firm (AAA, AA, A, BBB etc)

Outside US not much history to guide us as to likely future credit risk
premium

It’s also very volatile …
Page 33
Credit premium varies over time
6.0%
6.0%
5.0%
5.0%
4.0%
4.0%
3.0%
3.0%
BAA
Spread
2.0%
Credit spread, %pa
Credit spread, %pa
The credit risk premia
2.0%
1.0%
1.0%
AAA
Spread
0.0%
1926
0.0%
1936
1946
1956
1966
1976
1986
1996
2006
34
Credit premium varies by rating
25.0
25.0
Aaa
20.0
A
Baa
15.0
15.0
Spec
10.0
10.0
5.0
5.0
0.0
1973
0.0
1978
1983
1988
1993
1998
2003
Credit spread, %pa
Credit spread, %pa
20.0
2008
35
Credit premium varies by sector
8.0
7.0
Banking
Industrial
7.0
6.0
Telecoms
Utilities
6.0
5.0
5.0
4.0
4.0
3.0
3.0
2.0
2.0
1.0
1.0
0.0
1993
0.0
1995
1997
1999
2001
2003
2005
2007
2009
Sectoral credit spread, %pa
Sectoral credit spread, %pa
8.0
2011
36
Company specific factors
European high yield interest rate coverage
European high yield debt/EBITDA
5.0
5.0
4.0
4.0
3.0
3.0
2.0
2.0
1.0
1.0
0.0
0.0
2008
•
2009
2010
2011
2008
2009
2010
2011
Company fundamentals play an important part in the premium too
37
Risk neutrality and the credit premium
•
However, if investors are risk neutral then they will only asked to be
compensated for the potential additional loss compared with a default-free
investment
Expected loss = probability of loss x (1 – recovery rate)
•
What we expect to lose from investing in a credit risky entity is simply the
product of the probability of experiencing and the scale of that potential loss
38
The probability of loss (1920-2008)
80.0
70.0
Year 5
Year 10
Year 15
Year 20
60.0
Percentage
50.0
40.0
30.0
20.0
10.0
0.0
AAA
Aa
A
Baa
Ba
B
Caa-C
39
Recovery rates – (rating 5 years before default)
100.0%
Sr. Sec
Sr. Unsec
1992
1997
Sub.
Recovery rates, %
80.0%
60.0%
40.0%
20.0%
0.0%
1982
1987
2002
2007
40
The “risk neutral” credit premium
•
•
Credit premium = prob. of loss x (1 – recovery rate)
For example:
Expected loss rate for Baa over ten years = 5% x 40% = 2%
•
•
Or something like that
The degree to which the actual spread differs from the risk neutral, or ‘fair’
spread reflects the additional return required by risk averse investors
41
de
ed
at
R
eG
ra
de
Al
l
at
iv
-C
aa
C
en
t-G
ra
tm
ul
Sp
ec
ve
s
In
B
Ba
a
Ba
A
Aa
a
Aa
Loss rates
Loss rates (1982-2008)
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0.0%
42
Risk premia on main asset classes

Government bonds – an inflation risk premium = 0.25% to 0.50%

Corporate bonds – a credit risk premium, the starting point should be the
historic loss rate, let’s say = 1.50% to 2.00% for Baa

Equities – the equity risk premium
Page 43
The equity risk premium

ERP is the additional return required over long-dated government bond for
bearing equity risk

But what is equity risk ?

profitability

ongoing viability of company
Page 44
Equities are a poor hedge against recessions
UK real equity returns
Historic equity risk premia
100%
100%
80%
80%
60%
60%
8.0
Equities
40%
20%
20%
0%
0%
Real return, %pa
40%
5.0
Real return
Real return
6.0
4.0
Bonds
ERP
6.0
3.8
3.2
4.9
4.5
3.9
3.6
3.9
2.1
2.0
0.0
-20%
-20%
-40%
-40%
-60%
1900
-2.0
-60%
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Page 45
The Dividend Discount Model (DDM)

When we buy equities we purchase a future stream of dividends

All we need to do is calculate the “present value” of each of these dividends
and add them all up

But


dividends are paid over a long period

and are uncertain
However, if we assume that they grow at a constant growth rate, maths can
help us out …
Page 46
The Dividend Discount Model (DDM)
ERP + Risk free rate = Dividend Yield + Dividend growth
or
ERP = Dividend Yield + Dividend growth  Risk free rate

Dividend yield can be observed

Risk free rate can be observed (government bond yield)

Dividend growth – unobservable

If we apply some macroeconomic theory then we can arrive at a very simple
measure of the equity risk premium …
Page 47
Simplifying the DDM
ERP = Dividend Yield + Growth in dividends  Risk free rate

if dividends grow in line with real economy over long periods of time and

if real risk free rate is close to trend growth of the economy then (for the UK):
ERP = Dividend Yield + 2.25%  2.25%
ERP = Dividend Yield
Page 48
UK’s equity risk premium
A measure of the UK’s equity risk premium
Implied equity risk premium, % pa
12.0
10.0
8.0
6.0
4.0
2.0
0.0
1965
1970
1975
1980
1985
1990
1995
2000
2005
In 1970s required additional compensation was high
Page 49
A DDM matrix
Real risk free rate 0.70%
Real
Earnings
Growth
%
1.50%
1.75%
2.00%
2.25%
2.50%
2.75%
3.00%
2.5%
10,412
12,207
14,750
-
3.5%
6,556
7,224
8,045
9,077
-
Risk premium
4.55%
3.5%
4,720
6,556
5,057
7,224
5,446
8,045
5,900
9,077
6,436
10,412
12,207
-
4.5%
4,784
5,130
5,531
6,000
6,556
7,224
8,045
5.5%
3,766
3,978
4,214
4,481
4,784
5,130
5,531
Real risk free rate 2.25%
Real
Earnings
Growth

%
1.50
1.75
2.00
2.25
2.50
2.75
3.00
2.0%
6,436
7,080
7,867
-
2.5%
5,446
5,900
6,436
7,080
-
Risk premium
3.00%
3.5%
4,720
4,165
5,057
4,425
5,446
4,720
5,900
5,057
6,436
5,446
5,900
-
4.0%
3,726
3,933
4,165
4,425
4,720
5,057
5,446
4.5%
3,371
3,540
3,726
3,933
4,165
4,425
4,720
FTSE-100 = 5,900 on this day
Page 50
Issues with this simplification

What if firms increase dividends temporarily ?

What if firms pay no dividends ?

What about share buy backs ?

What if the profits earned by the market are not derived from the underlying
economy ?

Adjustments to the model can be made to account for all these issues, but will
require considerable user discretion
Page 51
Assembling the building blocks


Once the asset allocator has come to a view about expected:

economic growth rates

inflation and

risk premia on a range of asset classes
then the expected return jigsaw puzzle can be put together …
Page 52
Putting it all together: an example
Example of “building block approach” to forecasting long-run asset class returns
Expected return/expected return component
9.0%
Long-run expected return
8.0%
7.0%
6.0%
5.0%
4.0%
Expected return component
3.0%
2.0%
1.0%
0.0%
Real
economic
growth
Expected
inflation
Inflation risk
premium
Equity risk
premium
Index-linked
gilts
Cash
Gilts
Equity
This was the orthodox view just under four years ago
Page 53
Questions about the building block approach

What might change the asset allocator’s views ?

Should we revisit the pre-crisis assumptions?

What about developing economy asset classes ?

What about the starting point ? (the tactical aspect)
Page 54
Expected risk: Measuring
volatilities and correlations
Expected returns is the first ingredient
Expected return: established
via 'building block approach'
A mean-variance efficient frontier for asset classes
Efficient frontier
Individual asset
classes
Standard deviation, risk
How do we get the other ingredients ?
Page 56
Historic measures of volatility and correlation

Having determined the expected return, most use historic measures of
volatility and correlation

A MVEF can then be constructed

But variances and co-variances change over time …
Page 57
Time-varying volatility
UK & US equity return volatility over time
Equity market volatility, standard deviation, %pa
12.0%
10.0%
USA
UK
8.0%
6.0%
4.0%
2.0%
0.0%
1970
1975
1980
1985
1990
1995
2000
2005
2010
Asset class volatility can vary substantially over time
Page 58
Time-varying correlations
Correlation between UK & US equity returns over time
Equity market correlation, US and UK equities
100.0%
90.0%
80.0%
70.0%
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
1970
1975
1980
1985
1990
1995
2000
2005
2010
Asset class correlation can vary substantially over time too
Page 59
Forecasting volatility and correlations

Time variation of volatility and correlation is a problem

Many fancy statistical techniques for forecasting future volatility and
correlations

But once again, there is no “correct” way to forecast volatilities and
correlations
Page 60
What about client risk tolerance ?

We now have:

expected returns

expected variances and correlations

an efficient frontier

But what is the client’s appetite for risk ?

May be dictated by “return needs”

Psychologists and economists now put a lot of effort in to trying to determine
this
Page 61
Measuring risk aversion

Risk aversion is very difficult to gauge

Answer depends heavily on the utility function assumed to describe investors’
risk-return trade off

Investment professionals in US use experiments of this kind to determine risk
aversion of their clients

These techniques are now in widespread use elsewhere too
Page 62
Risk aversion is the final ingredient
Expected return: established
via 'building block approach'
Choosing a position on the efficient frontier
B
E
D
Efficient frontier
A
Individual asset
classes
C
Source: Fathom
F
Standard deviation, risk: established using
historic estimates of volatilities and correlations
But remember: ALL optimizers essentially ‘tell’ us what we have ‘told’ them!!!
Page 63
Appendix:
Yale University Endowment fund
Overview

Yale University's endowment is seen as a "best of breed" multi asset class
investment fund

Has made substantial use of alternative asset classes

The fund aims to support the University's academic activities

And has managed to increase both the absolute size of the fund and the
absolute size of the annual support for these University activities
Annual spend
Fund value
1996
$170m
$676m
2010
$1.1bn
$16.6bn

(Though the fund has received substantial donations over this period too)

Source: "The Yale Endowment" 2010
65
Yale University revenue

Endowment revenue makes up over 40% of total uni revenue
66
The "liabilities"

The fund is used to support all University activities.

Many of the donations are given with pre-defined activities that the donor
wishes to support, but for investment purposes the funds are "co-mingled"

In 2010 the fund supported 41% of the University's $2,681m operating budget
67
"Spending" policy

Conflicting goal: desire to support as much current spending as possible, but
preserving the value of assets to support future spending

Goal 1: Aim to produce stable/smooth stream of income for university

Goal 2: protect value of investments relative to inflation

Long-term spending rate, combined with "smoothing rule"

The smoothing rule ensures that income does not fall too far (if at all) in bad
years, but does not rise too much (if at all) in good years

(Life companies use similar rules)
68
"Spending" policy

Spending growth has outstripped the University's specific inflation measure,
that is, spending has increased in real terms

Rate of growth is smooth, due to smoothing rule
69
Spending rate

The crisis had a big impact on the “smoothed” spending rule
70
Fund value

Combination of high annual returns and ongoing contributions has led to a
massive increase in the fund's value over time
71
Investment policy

A combination of academic theory and "informed market judgement"

MVA is the starting point, stress tested for different return, vol and correlation
assumptions etc

Aim to invest predominantly in asset classes with "equity-like" returns

Avoid the "home bias" of investing in only domestic asset classes

The long-term horizon means that capital can be committed to illiquid asset
classes
72
Yale’s performance
73
Yale’s asset allocation
74
Asset allocation: actual & target

The assets used to support the spending aspirations (the liabilities) are
relatively diverse

This asset allocation structure is quite different from similar US University
funds, and very different from the sort of allocations made by UK life and
pension funds
75
Yale’s illiquidity ‘budget’
76
Yale’s fiscal highlights
77
Absolute returns

In 1990 first sizeable institution to invest in absolute return strategies

Identify managers that can enhance long-term real returns by exploiting
market inefficiencies. 50% Event driven, other 50% "Value driven strategies"

Expected real return: 5-6%

Expected risk: 10% volatility (event driven)

Expected risk: 15% "Value strat"

Policy:


performance-related fees

hurdle rates

clawback provisions

manager invests own net worth in fund
Performance: 11.5% pa with low correlation to bonds and equities
78
Domestic equities

Lower weighting than similar institutions (7% target)

Expected real return: 6%

Expected standard deviation: 20%

Benchmarked against Wilshire 5000 index

Policy:


commitment to active management

prefer managers with bottom-up research capabilities

acknowledgement that this will focus on small stocks
Performance: over last ten years 6.7%pa outperforming Wilshire 5000 by 7.4%pa
79
Overseas equities

Raison d'etre: exposure to global economy

One half of portfolio invested in high growth, emerging markets

Expected real return: 7%

Expected standard deviation: 22.5%

Again commitment to active fund management
80
Fixed income

Attracted by "certainty" of nominal cash flow; a hedge against "financial
accidents", but allocation of just 4%

Expected real return: 2%

Expected standard deviation: 10%

Benchmark index: Lehman Brothers US Treasury Index

Policy:

(internal) active management

avoiding market timing strategies, call options & credit risk
81
Private equity

Attraction: long-term, risk adjusted returns

Including buy-out funds and venture funds

Expected real return: 10.5%

Expected standard deviation: 27.7%

Policy:


avoid PE funds sponsored by financial institutions because of potential conflicts
of interest and staff instability
Actual returns since inception: 30.3%pa
82
Real assets

Investments in real estate, oil and gas and "timberland"

Attractions:

real assets so hedge against expected and unexpected inflation

visible cash flows

low correlation with other asset classes

illiquidity of such assets creates barrier to entry and raises long-term returns

Expected real return: 6%pa

Expected standard deviation: 15.5%pa

Over last ten years the portfolio has returned 10.9%pa
83
Investment performance

All components of portfolio have outperformed active benchmarks over the
last ten years
84