Business School
ACTL4303 AND ACTL5303
ASSET LIABILITY MANAGEMENT
Week 12
Currency and International Diversification
Performance Measurement and Attribution
• Market efficiency, anomalies, and behavioral finance
• Fundamental law and transfer coefficient
• The three hurdles for active management:
1. Are there anomalies to market efficiency?
2. Can some managers translate anomalies into consistent
outperformance?
3. Can funds reliably appoint and retain successful managers?
• Secured Loans – how secure? Spread duration. Secondary market
liquidity. The cycle of default rates.
• FCFF versus DDM. CAPM beta adjustments.
• Hedge funds. Alpha reliability net of fees versus market risk premiums.
Revision (1)
2
• Economic indicators. What leads, what lags? How foes the market move
relative to the economy?
• APRA SPG-530 and asset allocation, constant mix rebalancing
• Black-Litterman
• The asset-liability modeling framework and interest rate sensitivity
Revision (2)
3
Mercer Survey – Rolling 3 Year Quartiles and Median
Australian Equity Manager Performance
Past performance is not a reliable indicator of future performance. You should not rely on
past performance to make investment decisions
4
Australian Equity Manager versus Fund Performance
- periods ending June 2015
Past performance is not a reliable indicator of future performance. You should not rely on
past performance to make investment decisions
Sources: S&P/ASX, Mercer, FTSE/ASFA, SuperRatings
Active Management contribution is Mercer survey median minus S&P/ASX 200
Tax Effect is FTSE ASFA Superannuation Index less non-tax adjusted index
Super Fund Fees are estimated from a survey of Fund Product Disclosure Statements
Median Super Fund is from SuperRatings (SR50) Australian Share Options Survey.
5
Practical Beta Estimation (3)
• Thin trading can lead to low beta estimates – trade to
trade estimation, Dimson and Marsh (1983)
• Direct Leverage Adjustment
if a stock has the same unlevered beta as the market its
beta can still be different to one due to leverage
bL = bU ´ 1+ (1- t) D /E( )( )
6
Multifactor Model Example
Ri = E(Ri) + BetaGDP (GDP) + BetaIR (IR) + ei
Ri = Excess return for security i
GDP = GDP deviation from expected
BetaGDP= Factor sensitivity to GDP deviation
IR = Interest rate deviation from expected
BetaIR = Factor sensitivity for Interest Rate deviation
ei = return due to firm specific events
7
US Leading Economic Indicator
- Conference Board
Note the equity market is a leading economic indicator. Economic
indicators don’t lead the market
8
US Coincident, Lagging Indices
9
The market cycle leads the business cycle
• The market cycle is more ahead of the business cycle at the peak than the
trough
• The market cycle transpires in phases, often starting abruptly
10
Firm Valuation Model
ValueofFirm = tFCFF
1+WACC( )t
t=1
t=¥
å
• The current value of equity is obtained by deducting the
value of outstanding debt from the value of the firm
• The FCFF is a pre debt cash flow and is independent of
the current capital structure
• However some assumption about an ‘optimal’ capital
structure needs to made to determine the WACC
• As with the DDM the cash flows can be summarised in
phases
11
FCFF = EBIT x (1 – Tax rate)
+ Depreciation
- Capital Expenditure
- Δ(Working Capital)
If statements comply with US GAAP (Generally Accepted
Accounting Principles):
FCFF = Cash flow from operations
+ Interest expense x (1 – Tax rate)
- Capital Expenditure
FCFE = FCFF – Interest expense x (1 – Tax rate)
Free Cash Flow to the Firm (FCFF)
12
ROE = Net Profit
Pretax Profit
x
Pretax Profit
EBIT
x EBIT
Sales
Sales
Assets x x
Assets
Equity
(1) x (2) x (3) x (4) x (5)
Margin Tax
Burden
Interest
Burden
Decomposition of ROE
x
Turnover Leverage x x x
When forecasting ROE or ROA and trending towards stable
assumptions, a margin/turnover decomposition can be useful
13
• Perold and Sharpe, Dynamic Strategies for Asset Allocation, FAJ Jan-Feb
1988
• How to respond as market movements change the asset allocation
• Constant mix rebalancing (CMR) sells the outperforming asset class and
buys the underperforming asset class
• CMR is effective, and enhances performance, when markets oscillate
without trend
• Usually operates after a threshold of drift is breached
• However if an asset class trends (equities in a bull market) CRM will
underperform buy and hold
Managing asset allocation around the Strategic
Asset Allocation (1)
Constant Mix Rebalancing
This week’s coverage
Bodie et al
Chapter 24 Portfolio performance evaluation
Chapter 25 International diversification
Revision
16
• When an asset pays income, the price falls and the income is no longer
embedded in the price
• An asset return calculation therefore adds income to change in price
• At the portfolio level the valuation retains the income generated by
assets, because it is not typically payed out of the portfolio
• A portfolio return calculation does not add income to change in
valuation because the change in valuation already includes asset
income
Portfolio and asset returns
- income recognition
AssetRe turn = tP -t-1P +
tIt-1P
17
• Benchmark returns are time weighted so portfolio return calculations
need to be time weighted for comparison
• Time weighted returns for a period (eg 3 years) are calculated as the
product of sub-period returns (eg monthly)
• Sub-period returns should ideally also be calculated time weighted
• In practice sub-period returns are sometimes money weighted
because of legacy custodian practices
• Money weighted sub-period returns should be avoided where possible,
particularly where cash flows are significant and/or intra-period market
behaviour is uneven
Time Weighted (TWR) and Money Weighted
Returns (MWR)
18
When cash flows occur there are two choices:
1. Adjust for the cash flows using a Modified Dietz calculation
2. Revalue the portfolio on the date of cashflow
The sub-period portfolio return
MWR =V1 -V0 - CFtåV0 + Wt ×CFtå
TWR =Vt -CFt
V0
×V1
Vt
19
Example
• A portfolio is valued at $100m at the end of August
• A cash flow of $20m into the portfolio occurs on 20 September
• The underlying return is +5% for the first 20 days of September then -
10% in the final 10 days
• The value of the portfolio at c.o.b 20 September is $125m ($100m x
1.05 + $20m)
• The value of the portfolio at c.o.b 30 September is $112.5m ($125m x
(1 – 0.10))
• MWR = (112.5 – 100 – 20)/(100+ 20 x 1/3) = -7.03%
• TWR = ((125-20)/100) x (112.5/125) = -5.50%
• The cash flow has created more exposure to the negative performance
in the final ten days, lowering the MWR below the TWR
Calculation of sub-period returns (1)
20
• Recall a fund employs a custodian to maintain the records of their
portfolio, and manage custody.
• The investment management agreement (IMA) will also stipulate that
the investment manager maintain comparable records
• Custodians routinely reconcile their valuations to a manager’s portfolio
valuation at month end
• This ensures agreement about portfolio holdings and accruals
(outstanding trades, income due but not received)
• In the event of significant cashflow (eg 10% of the portfolio) the
custodian should be requested to value the portfolio so that a proper
TWR can be calculated
• The MWR in the event of cashflows is a second best option
• Where performance fees are involved accurate return calculation is
even more important
Calculation of sub-period returns (2)
21
• Investment management fees are normally payed directly from the
portfolio
• A manager’s performance is normally measured gross of fees
• Payments of fees should be recognised as a cash outflow and the
portfolio valuation should be gross of any fee accrual
• Example: a portfolio has a gross valueation of $100m at c.o.b 30 June
and a fee accrual of $150,000 for a net valuation of $99.85m
• The custodian valuation at 31 July shows a gross valuation of $105m
and a fee accrual of $50,000 for a net valuation of $104.95m
• A quarterly management fee of $150,000 is payed on 20 July
• The gross MWR = (105 – 100 – (-0.15))/(100 + (-0.15)x11/31) = 5.15%
• Note gross valuations are used, and the fee is treated as cash out
• If the fee was the only cash flow the MWR calculation would be
reasonable in this instance because the cash flow is small
Net and gross of management fees
22
• Comparing manager performance to benchmarks after tax is
complicated.
• The benchmark portfolio in theory should have a cost inventory affected
by inception date, and cash flows in the same way as the managers
portfolio.
• For example if the manager is forced to realise gains to meet a cash
outflow, the benchmark portfolio should be impacted similarly
• After tax benchmarks are available (eg FTSE/ASFA) but these are
generic and not matched to a particular portfolio’s cash flow history
• In practice funds can monitor tax inefficient behaviour (losing franking
entitlement by breaching the 45 day rule, realising gains after 11
months instead of 12 months)
• After tax return calculations are more relevant at the overall fund level,
rather than for individual managers within a fund. A sale might realise a
short gain for the manager (15% tax) but the total fund inventory may
enable it to be a long gain (10% tax) with parcel selection.
After tax performance
23
• First version 1998 by the CFA Institute
• Goal is to have managers calculate and present their investment
performance consistently
• All of a managers client portfolios should be grouped into composites
(eg Australian equity broad cap, Australian equity small cap etc)
• If a client portfolio has the same mandate specifications as the
composite it must be included in that composite
• Managers should present results for the relevant composite in
marketing presentations, rather than selecting the best performing
portfolio(s)
• Returns should be presented annually, rather than concealed in multi-
year averages
• Returns for periods less than a year should not be annualised
Global Investment Performance Standards (GIPS)
24
• The decision about market exposure (beta) normally rests with the
superannuation fund, not external managers
• An equity manager is expected to have a beta close to one and
performance is measured in excess of index typically without beta
adjustment
• It is worthwhile validating that beta is close to one for risk management
• The information ratio is therefore the common measure in practice
• In a multi-manager configuration tracking error still matters, and does not
diversify away
• The exception to the information ratio is absolute return managers (eg
hedge funds) where the Sharpe measure is more appropriate
Sharpe, Treynor, Jensen, M2 or Information Ratio?
IR=a p
s (ep )SharpeMeasure =
(rp - rf )
s p
25
• The numerator, αp , is normally taken as the simple excess of portfolio
return over the market, without explicit beta adjustment
• This is equivalent to assuming that the underlying beta is 1
• Similarly, σ(ep) , is the volatility of portfolio returns in excess of the market
(rP – rB)
• In practice calculations are made on monthly excess returns and the ratio
is annualised by sqrt(12)
• A cleaner approach mathematically is to work with ln((1+rP)/(1+rB)) so that
the sqrt(12) annualisation is mathematically appropriate
• Even then the underlying assumption is that monthly active returns are
independent
• Where there is serial correlation of active returns due to style bias the
annualised volatility via this simple approach is understated
• Bear this in mind in manager configuration (risk budgeting)
Information Ratio
26
• An investment option within a superannuation fund will usually have a
strategic asset allocation (SAA) giving benchmark weights for each
asset class
• A passive index, not always replicable (eg property), can be established
for most asset classes
• Benchmarks for private equity and absolute return strategies are less
obvious. In practice you might use public market indices and cash
respectively.
• The objective of attribution analysis is to analyse the incremental
performance relative to benchmark by the activities which contribute to it
– asset allocation and security selection
Performance Attribution (1)
RP -RB = AA+ SS
27
The framework is based on representing portfolio and benchmark returns as
the sum of products of asset class weights and asset class returns
Performance Attribution (2)
RP -RB = wPi × rPii=1
n
å - wBii=1
n
å × rBi = (wPi × rPi -wBi × rBi )t=1
n
å
= (wPi -wBii=1
n
å ) × rBi + wPi × (rPi - rBi )i=1
n
å
= (wPi -wBi ) × (rBi -RB )i=1
n
å + wPi × (rPi - rBi )i=1
n
å
Note the last step works because the sum of difference in weights must
be zero
28
Performance Attribution (3)
• The asset allocation contribution for bonds is (50%-30%)x(1.0%-1.55%)=-
0.110%
• The benchmark bond return, while positive, is less than the benchmark
portfolio return
• Allocating more to bonds makes a negative contribution
• There is no portfolio return for property but the security selection item is
multiplied by the portfolio allocation of 0%, so the table is complete
Asset
Class
Portfolio
Allocation
Portfolio
Return
Benchmark
Allocation
Benchmark
Return
Asset
Allocation
Contribution
Security
Selection
Contribution
Equities 50% 3.00% 60% 2.00% -0.045% 0.500%
Bonds 50% 0.50% 30% 1.00% -0.110% -0.250%
Property 0% NA 10% 0.50% 0.105% 0.000%
100% 1.75% 100% 1.55% -0.050% 0.250%
29
Multi-Period Performance Attribution(1)
AAt =i=1
n
å (wtPi -wt
Bi ) × (r tBi -Rt
B )
SS t =i=1
n
å wtPi × (rt
Pi - rt
Bi )
1+SS*
t =1+RtB + AAt +SS t
1+RBt + AAt
=1+RP
t
1+RBt + AAt
1+ AA*
t =1+RtB + AAt
1+RtB
The attribution elements need to be converted to a multiplicative
structure for multi-period attribution
30
Multi-Period Performance Attribution(2)
The attribution ratios are then multiplied through time to
calculate full period attribution
AA = ( (1+RBt )) × ( (1+ AA*
t
t=1
p
Õt=1
p
Õ )-1)
SS = ( (1+RBt
t=1
p
Õ ) × (1+ AA*
t )) × ( (1+ SS*
t
t=1
p
Õ )-1)
(1+RPt )-1= ( (1+RB
t )-1t=1
p
Õt=1
p
Õ )+ AA+SS
31
• It is more straightforward to attribute gross performance rather than
after tax/fees performance
• The impact of tax and fees can be summarised at the total portfolio
level
• A portfolio return may not exactly equal the sum of products of sector
weights and sector returns due to cashflow distortions etc
• At the calculation stage any discrepancy should be isolated in sub-
period and multi-period attribution – not obscured by arbitrarily merging
with other elements
• If the discrepancy accumulates significantly over time then there could
be something structurally wrong with the attribution framework (eg the
portfolio returns have been assumed gross of fees when in fact they
have been calculated net)
• Random unbiased discrepancies should not accumulate significantly
Performance attribution in practice
32
Australian companies investing overseas (1)
33
• ‘NAB sells off HomeSide’ (Money Management 12/12/01)
‘National Australia Bank (NAB) has rid itself of its troubled
American mortgage business, HomeSide, after offloading it to
Washington Mutual Inc for $3.7b
‘The Florida-based business , acquired by NAB in 1997 for
$1.7b , made spectacular losses of more than $3b as a result
of bungled interest-rate calculations …’
‘NAB chief executive Frank Cicutto vowed .. That NAB would
never again revisit mortgage products in foreign markets’
NAB’s US adventure
Australian companies investing overseas (2)
34
• ‘NAB committed to UK despite Clydesdale downgrade’ (AFR
29/9/11)
• ‘NAB’s Clydesdale on credit watch in Britain’ (AFR 2/6/13)
• ‘Brokers urge NAB to move on Clydesdale sale’ (AFR
21/1/14)
• ‘RBS fallout casts shadow over NAB UK’ (AFR 29/1/14)
• ‘Cameron Clyne confident NAB will “get out” of UK (AFR
10/9/14)
• ‘NAB flags profit drop on $1b UK write-downs’ (AFR 9/10/14)
NAB’s UK adventure
Australian companies investing overseas (3)
35
• ‘QBE buys into Argentina, Hong Kong’ (AFR 7/3/12)
• ‘QBE playing catch-up after aggressive expansion’ (AFR
21/1/13)
• ‘Profit warning tipped for QBE after US strife’ (AFR 9/12/13)
• ‘QBE should sell Winterthur, take write-down’ (AFR 15/4/14)
• ‘Argentinian woes cause yet another QBE downgrade’ (AFR
30/7/14)
QBE’s ill-fated global acquisition strategy
Australian companies investing overseas (4)
36
‘‘We were a company with $500 million in cash, rail, port,
wagons, the mine was already developed, the wash plant
was going and they still managed to screw it,” he said.
“And I don’t want to talk publicly about how that happened,
but that’s just an example of how these things work.
“They just think they are sending people from here to there
and think, ‘Oh yeah, how hard can that be?’
“And it’s bloody hard. You have to manage it and you have to
manage it well.” – vendor who initially sold asset to RIO
Rio takes massive loss on Mozambique sale (paid $3.9b,
sold for $50m) (AFR 31/7/14)
• Some companies have unique products that generate strong demand
overseas (eg Cochlear hearing implants, Computershare registry
services)
• But generally Australian companies tend to acquire smaller overseas
operations that are cheap often because they are in compromised
competitive positions
• At the outset there is hope that the talents of Australian management
can turn these frogs into princes (why hasn’t a local acquirer tried?)
• In practice that is often a steeper challenge than it first appears
• Overseas diversification via the overseas subsidiaries of Australian
companies is a constrained alternative
• Buying overseas companies gives access to stronger global companies
• Are these available at good prices?
Australian companies investing overseas (5)
37
Sovereign Risk
38
‘Kingsgate Consolidated Limited (ASX: KCN) wishes to advise
that business is continuing as usual at its Chatree operations in
Thailand following the move by the army to take control of the
administration of the country on Wednesday 22 May. This
followed the imposition of martial law on Tuesday 20 May 2014
which was implemented after several months of anti-government
protests in Bangkok.’
– Company Announcement 27/5/14
Social Capital and Economic Development
39
‘Social capital is the set of institutions – including the legal
framework, the financial system, the nature of corporate
governance, political practices and traditions, educational and
health levels, the structure of taxes, etc. – that determine the
way individuals are given incentives to create value with the
tools and infrastructure that they have.’
– Michael Pettis, Peking University
The Pettis thesis is that a deficiency in social capital makes it
hard for developing economies to transition to developed
economies. The development of social capital conflicts with the
interests of entrenched rent-seeking elites.
Emerging markets (1)
• Emerging markets currently represent around 11% of the MSCI All
Countries Index. Volatility of 24% compares to 18% for developed
markets
40
Emerging markets (2)
41
• For 19 developed markets since 1900 the correlation is firmly negative
(-0.39 in local currency, -0.32 in $US)
• For 22 developed markets over the period 1970-2011 the correlation is
virtually zero (-0.04 in local currency, 0.01 in $US)
• For 15 emerging markets over the period 1988-2011 the correlation is
firmly negative (-0.41 in local currency and -0.47 in $US)
• Equity returns depend on profitability per share
• Economic growth can occur through expansion of the capital base (debt
and equity funded) without improvement in profitability per share
• Scrutiny of how companies manage their capital (adequate returns on
retained capital) is an important prerequisite for equity performance
• Capital management may be better scrutinised in developed markets
Does stronger economic growth mean higher
equity returns?
42
Exposure to non-$A currencies improves
diversification (to a point)
43
• A portfolio has two layers of allocation
• Asset class weights, divisible by country (eg US equity allocation) which
add to 100%
• Currency allocation weights (eg to $A, $US, Euro, Yen etc) which add to
100%
• For example it is possible to allocate to UK equities, but cross-hedge
the currency exposure to Euro
• Separating assets and currencies allows allocation to strong markets in
local currency terms, but where the currency is weak
• Without separation a perspective on the currency will influence the
allocation to the asset (eg avoid markets where currency is weak)
• Asset returns should be considered in excess of the local interest rate,
and currency returns considered as including the local interest rate
• In that way currency allocation creates forward premiums
Asset and currency separation (1)
44
Asset and currency separation (2)
Asset % Currency %
Australian
Equities
50 $A 60
US Equities 25 $US 15
European Equities 10 Euro 20
Japanese Equities 15 Yen 5
100 100
Starting with the currency profile of the assets, this currency allocation
could be achieved by a 10% cross-hedge from Yen to Euro, and a
10% hedge from $US to $A. Currency hedging establishes a new
currency exposure and an interest rate differential (forward premium).
45
Currency and asset correlation - Australia
Asset returns in this illustration are a composite of 40% Australian
equities, 30% international shares (hedged), 15% Australian bonds
and 15% international bonds (hedged). R-squared is 21%.
46
• International investors view Australia as a global cyclical exposure
• When the global economy is strong, Australia’s terms of trade tend to be
strong (commodities) and the Australian dollar is well supported.
Overseas investors are attracted to Australian assets, especially
resource companies
• When there are concerns about the global economy overseas investors
have less appetite for the Australian dollar, and Australian assets
• In times of crisis overseas investors have a tendency to repatriate
capital, selling peripheral asset/currency exposures
• For the above reasons the Australian dollar has a tendency to be weak
when assets returns are weak
• But there is a silver lining - Australian dollar weakness buoys the returns
from overseas assets due to currency translation
Currency and asset correlation - Australia
47
• CAPM, market efficiency – for and against
• Modeling asset class returns – equity trend stationarity, Nelson Siegel,
regime switching, historical context
• Expectations of asset class returns – Black-Litterman
• Inflation and asset class returns, by type of inflation
• Equity management styles – value and growth
• Property - valuation
• Risk budgeting – manager allocation
Revision Focus
48
Thank you for your attention
Greg Vaughan
School of Risk and Actuarial Studies
University of New South Wales
49