Great Expectations:

Regime-Based Asset
Allocation Seeks
Higher Return,
Lower Drawdowns

By BNY Mellon Investment Strategy EXECUTIVE SUMMARY

and Solutions Group1

Research from BNY Mellon’s Investment Strategy and Solutions

Group (ISSG) has found that dynamically adjusting asset class
exposures as growth and inflation expectations shift has the
potential to significantly improve risk adjusted returns for
asset allocation strategies. Analysis of capital market returns
from 1988 to 2013 showed that the ISSG’s regime-based
portfolio would have achieved a 50% improvement to the Sharpe
ratio compared with that of a typical institutional portfolio.2
Moreover, the ISSG’s regime-based approach has the potential
to provide meaningful downside protection during periods of
extreme market stress, such as the bursting of the technology
bubble from 2000-2002 and the global financial crisis from
2007-2009, suggesting its potential risk management utility.

The group’s work highlights the potential benefits of moving away from static
strategic asset allocation strategies to more opportunistic approaches that
incorporate macroeconomic indicators into asset class weightings. Unlike previous
research on regime-based or risk-based asset allocation, the ISSG has broken new
ground on three levels. First, they developed a more granular understanding of
complicated patterns of macroeconomic regimes and their effects on asset prices,
especially during transition periods. More significantly, they have pointed to the
importance of shifts in growth and inflation expectations rather than just levels for
signaling regime changes. Finally, they used these insights to develop a probabilistic
model to analyze growth and inflation expectations data with a view toward
predicting the probability of regime changes and adjusting exposures accordingly.3

1 The Investment Strategy and Solutions Group is part of The Bank of New York Mellon, a principal
banking subsidiary of BNY Mellon.
2 The typical institutional portfolio is based on Greenwich Associates data, as explained in the
disclosure section. The reason for the time period chosen is explained in footnote 4. See p. 16
for performance comparison.
3 No investment strategy can predict or guarantee performance.
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 2

The following discussion specifies how the group defined macroeconomic regimes
and their effects on asset class performance by analyzing 40 years of market and
economic data. Against this more detailed understanding of regimes and their
transitions, the team describes how it used inflation and growth expectations data
to develop their model. They compare the performance of a typical institutional
portfolio against that of their regime-based portfolio over the last 25 years and
document the improved risk and return results for the regime-based portfolio.4
Stress-testing their model, they look at how the regime-based portfolio would
TABLE OF CONTENTS have performed in two periods of extreme market duress.

Let No Crisis Go to Waste: Having described how the ISSG model works, the team addresses different ways
Rethinking Asset Allocation investors might consider implementing a regime-based asset allocation approach.
Approaches 3 These include a full-fledged implementation of an asset allocation structure that
would dynamically weight asset classes based on macroeconomic views. By contrast,
Mapping Regimes and Their a partial implementation would maintain strategic portfolio weights across
Effects on Asset Prices 5 traditional asset classes but make shifts within specific asset classes to reflect
macroeconomic views. They also weigh up the costs and benefits of adjusting
Building a Probabilistic asset class exposures by rebalancing or using synthetic overlays.
Model to Predict Regimes 9
While the importance of asset allocation decisions on investment returns has
Better Performance with long been documented,5 the ISSG believes the current environment of modest
Regime-Based Asset expected market returns and heightened volatility requires a fresh look at asset
Allocation 13 allocation approaches. The financial crisis taught painful lessons about the limits
of traditional diversification and the need to achieve a deeper understanding of the
Stress-Testing the Model 15 macroeconomic influences on asset class performance and correlations. The ISSG
believes an asset allocation approach that is mindful of and responsive to portfolio
Implementation Considerations 18 risk factors across regimes has the potential to achieve investors’ long-term return
objectives while better protecting portfolios against devastating drawdowns.
Conclusion 20

Appendix 21

4 To mitigate small-sample bias that could arise from too narrow a data field, the ISSG allowed for the
maximum sample for estimation prior to conducting the out-of-sample exercise for prediction. This
resulted in an initial in-sample estimation period from February 1973 to February 1988 and out-of-sample
period of February 29, 1988 to June 30, 2013. Data adequacy and test size and power properties were also
additional parameters that dictated their sample selection process.
5 Gary P. Brinson, L. Randolph Hood and Gilbert L. Beebower, “Determinants of Portfolio Performance,”
The Financial Analysts Journal, July/August 1986.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 3

LET NO CRISIS GO TO WASTE:

RETHINKING ASSET ALLOCATION APPROACHES
Heavy losses incurred by institutional investors during the global financial crisis
of 2007-2009 have prompted a rethinking of traditional asset allocation practices.
More recent market turmoil, driven by concerns over high levels of sovereign debt
and flagging GDP growth, once again highlighted the intimate connection between
macroeconomic conditions and asset class performance. As investors revisited
assumptions about traditional asset allocation practices, diversification and asset
class correlations, our goal was to help them integrate macroeconomic influences
Our goal was to help
on asset class behaviors into their asset allocation strategies. Our belief was that an investors integrate
asset allocation structure that could dynamically overweight assets that behaved macroeconomic influences
well in certain environments and underweight those that performed badly might on asset class behaviors
contain greater upside potential, while protecting against significant drawdowns.
into their asset allocation
To understand the latest asset allocation challenges, we think it is helpful to strategies.
remember how investor thinking has evolved. For many years, investors tended
to hold equities, fixed income and cash according to their return requirements
and risk tolerances. However, during the multi-decade bull market that began in
the early 1980s, many investors began abandoning cash allocations as a “drag on
performance.” Cash allocations were increasingly replaced by a new category of
uncorrelated assets lumped together as “alternatives,” whether it was real estate,
private equity, or hedge funds. Investors were drawn to alternatives’ potential
to deliver a higher expected rate of return within the same volatility target for
the overall portfolio. This putative “free lunch” was based on the historical
low correlation of alternatives to traditional asset classes. Notions of optimal
diversification changed, as more investors turned to alternatives in lieu of cash
allocations. A new asset allocation framework emerged based on three standard
buckets of stocks, bonds, and alternatives. The illiquidity of many alternative
asset classes was regarded as an acceptable risk for institutional investors
with long-term investment horizons.

The global financial crisis changed that view, as many investors learned painful
lessons about liquidity and the limits of diversification when it is needed most.
The crisis has engendered a new respect for tail risk and prompted widespread
soul-searching about liquidity, diversification and asset class correlations. It
has not, however, significantly dampened the return expectations of many
institutional investors confronted with ongoing pension fund deficits and
other investment challenges.
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 4

Instead, investors are increasingly looking for ways to improve their asset allocation
approaches to address tail risk and the instability of asset class correlations, without
sacrificing return expectations. The goal is to understand the underlying forces
that drive asset class performance and risk in order to enhance return, minimize
drawdown risk and avoid reverting to low-yielding cash allocations. This has led
to a number of asset allocation frameworks that define regimes in different ways.
One approach is to categorize asset classes according to their behavior across
different growth and inflation regimes.
Our research into over 40
years of U.S. macroeconomic According to this taxonomy, assets are organized into risk buckets consisting of growth
conditions and asset class assets, inflation-sensitive assets, and deflation-sensitive assets. Subsets of traditional
asset classes can fall into multiple risk buckets depending on the underlying instrument’s
behavior reveals a complex
sensitivity to growth and inflation. For example, some types of fixed income can
picture of how macroeconomic be categorized as growth (high yield bonds), inflation-sensitive (Treasury Inflation
regimes unfold and the Protected Securities), and deflation-sensitive (U.S. Treasuries) assets. Generally
transitions between speaking, these three risk buckets correspond to macroeconomic regimes that can
be described much like Goldilocks’ three bowls of porridge: Too Hot ( inflation),
those regimes.
Too Cold (deflation), and Just Right (growth).

Figure 1: The Goldilocks Model of Economic Regimes

Just Right

Too Cold Too Hot

Source: ISSG

However, we believe this basic temperature scale masks important gradations

between these three points, which have important implications for asset class
performance. Even more misleading, in our view, is the implication, from this scale
shown above, that economies heat up and cool down in a sequential,orderly process.
On the contrary, our research into over 40 years of U.S. macroeconomic conditions
and asset class behavior reveals Rising Growth
a far more complex picture of how macroeconomic
regimes unfold and the transitions between those regimes.

Perfection Warming

Falling Inflation Rising Inflation

Cooling

Too Hot
Too Cold

Falling Growth

Too Hot

Warming Cooling
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 5

MAPPING REGIMES AND THEIR EFFECTS ON ASSET PRICES

We believe the Goldilocks scenario of Too Hot (rising inflation choking off growth),
Just Right
Too Cold (falling inflation and falling growth), and Just Right (positive growth and
low inflation, which encompass Warming, Cooling and Perfection subsections)
Too Cold Too Hot
regimes does not adequately capture all of the possible permutations given the
two macroeconomic variables of inflation and growth. In our view, there should be
a minimum of four regimes to represent the possible combinations of growth and
inflation scenarios (rising and falling growth, rising and falling inflation). While four
regimes depict the four possible scenarios, we think a fifth scenario, a Too Cold
A more nuanced five-
regime, represents a special case of the falling inflation and falling growth regime, bucket framework has
when growth contracts sharply as in the case of economic recessions. Admittedly, profound implications
it would be possible to introduce ever more dissections, but this has to be balanced for understanding asset
with a practical need to decipher and identify regimes meaningfully.
class behavior.
Figure 2: Three Regimes to Five

30%
Rising Growth
20%
10%

Return
0%
-10%
-20%
Perfection Warming
-30%
-40% ity ity
qu qu u
E
t'l
E Eq
Falling Inflation Rising Inflation S. EM
U. In

Cooling

Too Hot ■ Warming

Too Cold

Falling Growth

Source: ISSG

A more nuanced five-bucket framework has profound implications for understanding

Too
asset class behavior. By contrast, investors Hotthe basic three bucket framework
using 100%

might be inclined to allocate away from equities and other growth assets as GDP 80%

begins to decline. But historical data show that growth sensitive assets can still 60%

have positive real returns even as GDP is declining (or Cooling) on average. A rules- 40%

based, three-bucket system might halt investing in growth assets as GDP begins 20%
Warming Cooling
to decline, despite the fact that there is still positive return potential for them 0%

during a Cooling period. -20%

88
89
90
91
92
93
19
19
19
19
19
19
1
While growth and inflation have clear implications for asset class performance,
Perfection
investors recognize that changes in the price of an asset are driven by expectations ■ Warming
about these factors, not simply the changes in level. The current price of an asset
reflects an expectation of the inflation rate, real growth rate and risk premium.
Changes in the price of an asset are a function of changes in expected inflation,
real growth and the risk premium. These changes in expectations can be measured

Too Cold

8% 100%
e Point Revision

6%
80%
4%
60%
2%

0% 40%
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 6

in aggregate through the use of forecasted inflation and growth rates (holding the risk
Rising Growth
premium constant). CPI and real GDP data from the Survey of Professional Forecasters
compiled and maintained by the Federal Reserve Bank of Philadelphia provide a long
history of how real growth and inflation expectations have changed over time. As we
developed our historical view of economic regimes and their transitions, we focused
Perfection Warming
on changes in inflation and growth expectations as opposed to level changes in order
to better align changes in asset prices with macroeconomic shifts.
Falling Inflation Rising Inflation
It is important for investors In addition to tracking macroeconomic regimes according to changes in inflation
Cooling
to understand the non- and growth expectations, we believe it is also important for investors to understand
Too Hot
sequential movements the non-sequential movements across different regimes over time. Investors often
Too Cold
think of the economy as ebbing and flowing in a neat, sequential pattern of heating
across different regimes
and cooling. The typical picture is that of an economy Warming up, getting Too Hot,
over time. and then Cooling until the point of Too Cold. While this image is easy to understand,
Falling Growth
it does not correspond to actual experience in most macroeconomic cycles. Instead,
we found a far more complex pattern of regime transitions.

Figure 3: Complex Transitioning Across Macroeconomic Regimes

Too Hot

Warming Cooling

Perfection

Too Cold
Source: ISSG

In fact, our research shows that a transition from Too Hot to Cooling has not
8%
happened in the past 40 years. The Too Hot regime has been succeeded by Perfection
Percentage Point Revision

6%
(rising
4%
growth and falling inflation) two of the four times it was experienced in
the 2%
last 40 years. This more complex pattern of transitions presents a significant
hurdle
0%
for investors, as it complicates the challenge of trying to predict the order of
macroeconomic
-2%
regimes. However, it does provide a richer understanding of how the
economy
-4% can transition through time. Figure 4 shows the interaction of the year-
over-year
-6% revisions 6
to expectations of growth and inflation since 1970. This helps
capture
-8% a trend in investors’ expectations of the macroeconomic environment.
-10%
73

76

77

79

80

82

83

85

86

88

89

91

92

95

97

98

00

01

03

06

07

09

10

12
74

94

04
19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

20

20

20

20

20

20

20

20
19

19

20

■ 12M Revision in Expected Inflation ■ 12M Revision in Expected Real GDP

■ Perfection ■ Warming ■ Cooling ■ Too Hot ■ Too Cold

6 See appendix for additional information.

40

35
 Too Cold GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 7

Figure 4: Revisions to Expectations of Growth and Inflation since 1970

8% 100%
Percentage Point Revision

6%
80%
4%
60%
2%

0% 40%
-2% 20%
-4%
Regime lengths can
0%
-6% vary substantially.

88
89
90
91
92
93
94
9
19
19
19
19
19
19
19
19
-8%

-10%
73

76

77

79

80

82

83

85

86

88

89

91

92

95

97

98

00

01

03

06

07

09

10

12
74

94

04
19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

19

20

20

20

20

20

20

20

20
19

19

20
■ 12M Revision in Expected Inflation ■ 12M Revision in Expected Real GDP

■ Perfection ■ Warming ■ Cooling ■ Too Hot ■ Too Cold

Source: ISSG & Philadelphia Federal Reserve as of 6/30/2013.

While transitions between regimes appear to be quite arbitrary, there are some
discernible patterns over the last 40 years. For example, the three Too Cold
100%
regimes
40 have been succeeded by Warming. While this experience may not hold
true
35
indefinitely, we can understand why this pattern exists. Too Cold regimes are 80%
characterized by a drastic decline in growth expectations and decreasing inflation 60%
30
expectations. We know that the U.S. Federal Reserve has historically combated
Number of Months

40%
dramatic
25 growth declines by adding substantial stimulus to the economy in the
20%
form of lower interest rates. However, thatAverage
stimulus often comes with rising real
length
20
prices, a tailwind for increasing inflation expectations and thus setting the stage 0%
for15a Warming regime.

88
89
90
91
92
93
94
19
19
19
19
19
19
19
19
10
The second interesting pattern is the propensity of the Warming and Cooling regimes
5
to rotate back and forth. Again, this is intuitive as there can be extended periods of
relatively
0 benign economic activity. A third insight from our drill down into regime
transitions isWarming
that the Perfection Perfection Cooling
regime generally follows periodsToo
of Hot Too Cold
high inflation
expectations. As such, if inflation expectations are not at a relatively high level, we
are unlikely to experience the best scenario for equity-like assets. Investors might
use this insight to dampen the return expectations on equity-like instruments and/
or allocate capital to assets that perform well in benign or increasing inflation
expectation environments.

Another vital consideration for investors is that regime lengths can vary. We found
that the Warming and Cooling environments (typically fairly benign) tend to last
longer on average, but also exhibit a higher variance in length, while the extremes
(Perfection, Too Hot and Too Cold) tend to be shorter and have more consistent
duration. For example, Warming regimes averaged a length of 25 months but ranged
from 11 months to 37 months. In contrast, the Too Cold regimes had an average
length of 10 months and had a tighter range of 7 months to 15 months. The Too Cold
regime length is the shortest on average, another example of the Fed interceding
to counter the market’s dramatically decreasing growth expectations.
 ■ Perfection ■ Warming ■ Cooling ■ Too Hot ■ Too Cold

GREAT EXPECTATIONS: REGIME-

BASED ASSET ALLOCATION // 8

Figure 5: Regimes Vary in Length

40

35

30

Number of Months
25
Average length
20
Asset prices behave
differently according to 15

investor perceptions of the 10

coming economic regimes. 5

0
Warming Perfection Cooling Too Hot Too Cold

Source: ISSG

Asset prices behave differently according to investor perceptions of the coming

economic regimes. U.S. equities have earned a real return of 5.8% since 1973 (see
Figure 6). More importantly, the anticipated economic regime has had a profound
effect on when that 5.8% was earned and lost. For example, equities gained a real
return of 14.5% in periods of rising expectations for growth coupled with falling
expectations for inflation (Perfection). Contrast this with another scenario, that of
falling growth expectations and falling inflation expectations (Too Cold), in which
equities returned a negative 26.9%.

Figure 6: Equity Returns Across Regimes (Real, 4/30/1973 - 6/30/2013)

Contribution
Regime Inflation Growth Frequency Real Return
to Return
Steady/ Steady/
Warming 42% 8.1% 3.4%
Rising Rising

Perfection Falling Rising 16% 14.5% 2.2%

Cooling Falling Falling 24% 12.7% 2.9%

Too Hot Rising Falling 11% -5.9% -0.7%

Sharply
Too Cold Falling 6% -26.9% -2.0%
Falling

All Regimes 100% 5.8% 5.8%

Source: ISSG, Ibbotson & Bloomberg as of 6/30/2013. Returns calculated using ISSG’s
regimes. Please see appendix for index descriptions.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 9

In addition to equities, we found that nearly all assets performed in a similarly

intuitive fashion. For example, TIPS outperformed nominal bonds in the two
scenarios of rising inflation expectations (Warming and Too Hot). Inflation sensitive
assets (such as commodities) performed best in Too Hot regimes and they performed
better in Warming regimes than Cooling. The major exception was emerging markets
(EM) equities during Perfection regimes. Intuitively, emerging market equities should
benefit from rising growth and falling inflation expectations; however, this was
not the case in the data set analyzed. This is most likely due to the short history
of emerging market indices and the fact that the Asian currency crisis occurred We apply our insights into
during one of the two Perfection regimes for which we have high-quality data revisions to inflation and real
for EM equities. GDP to identify regimes in
Figure 7: Asset Class Performance Across Regimes
real time, so that investors
might leverage this approach
30% on a prospective
60% basis.
20%
10% 50%
Return

0%
-10% 40%
-20%
-30% 30%
ty e s igns s ds S es
-40% ty ui ty t nd rie
n P ti
ui ui ta re Bo RI S di sh
Eq Eq Eq Es Bo ve su d ST TIP o Ca
S. t'l al p. So ea iel Yr m 20%
U. In EM Re Co
r
ba
l Tr Y
5+ Co
m
o igh 1
Gl H
10%

■ Warming ■ Perfection ■ Cooling ■ Too Hot ■ Too Cold 0%

Warming Perfection
Source: Please see appendix for index descriptions.

BUILDING A PROBABILISTIC MODEL TO PREDICT REGIMES

Our detailed investigation into regime durations and transitions over the last 40 years
using historical series of revisions to inflation and real GDP expectations provided
valuable insights into how macroeconomic regimes unfold overtime. We believed we
could
100%apply these insights to identify regimes in real time, so that investors might 40%
leverage
80% this approach on a prospective basis.
60% 20%
Our40%
goal was to create a model using multinomial logistic regression7 to help predict
regime
20% probabilities by processing new information about changing real GDP and 0%
inflation
0% expectations and mapping that to what we already knew about the current
economic regime. This would allow us to test for the probability of a certain regime, -20%
-20%
based on a set of possible variables for growth and inflation data.
88
89
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10

12
13
11

-40%
20
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20

20
20

■ Warming ■ Perfection ■ Cooling ■ Too Hot ■ Too Cold -60%

1988 1989 1990 1991 1992 1993 1994 1995 1996 1

■ Equity ■

100%

80%
7 Multinomial logistic regression models are typically used to predict the probabilities of different
60%
possible outcomes of a predefined, dependent variable, given a set of independent variables.
40%

20%

0%
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 10

30%
The model incorporates the most recent levels and rates of change for the real GDP
20%
and10%
inflation expectations revisions series, as well as the regime the economy was

Return
experiencing
0% six months ago. It then generates a set of probabilities about which
regime
-10% the economy will “choose” over the next month. We tested the probability
predictions
-20% of our model against the actual regimes that occurred based on
-30%
historical data from 1988 to the end of June 2013.
ns ds s S
-40% ity ity ty at
e ds eig ies on IP iti
e
qu qu ui st on er ur sh
dB TR S od
E
t'l
E Eq E
p.
B
So
v
ea
s Ca
el S TIP
Rising Inflation S. M al r l r Yi Yr m
In U.
order
to use
In thisEmodel
Re in real
time
Co a for predicting
T gh
regime
+ m
probabilities,
Co we
Using our model we can ob Hi 15
developed a procedure for Gl
re-estimating our model at each monthly time period
produce a set of probabilities using only the data known up to that time period. Using the model that is estimated
■ Warming ■ Perfection Cooling ■ Too Hot Too Coldregime
for the regime that the at any given point in time, we can produce ■a set of probabilities for
■ the
that the market will “choose” over the next month. It was important to avoid
market will “choose”
introducing look-ahead bias into the calculations. Using this “expanding window”
over the next month. approach to model estimation, the model would likely get better over time at
assigning odds to the current state of the economy as it gathered more and
more historical data points over which to estimate.

Figure 8: Regime Probabilities Through Time

100%
80%
60%
40%
20%
Cooling 0%
-20%
88
89
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10

12
13
11
20
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20

20
20
■ Warming ■ Perfection ■ Cooling ■ Too Hot ■ Too Cold

Source: ISSG

Figure 8 shows the times series of the monthly estimations of the regime
probabilities on the top half, as well as the actual regime that was assigned to
the month based on knowledge of the full time period (on the bottom half). The
probability of a Warming regime is much higher on average than that of the other
four regimes. Warming regimes prevailed in 44% of the months in our testing
period from 1988-present (see Figure 9).
100%

80%

60%

40%

20%

0%
11
88
89
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10

12
13
20
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20

20
20
1
03

06

07

09

10

12
04
20

20

20

20

20

20
20

n in Expected Real GDP

■ Too Cold

100%

80%
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 11

Figure 9: Warming Regimes: Probability vs Actual

100%

80%
100%
60%
80%
40%
60%
20%
40%
0%
20%

11
88
89
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10

12
13
0%

20
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20

20
20
11
88
89
90
91
92
93
94
95
96
97
98
99
00
01
02
03
04
05
06
07
08
09
10

12
13
20
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20
20

20
20
Source: ISSG

The probabilities of Too Hot and Too Cold regimes are quite close to zero most of
the time, though they periodically spike in response to large moves in revisions
to the real GDP and CPI expectations data (see Figure 10).

Figure 10: Too Hot and Too Cold Regimes: Probability vs Actual

100% 20%
80%
100% 18%
20%
16%
18%
60%
80%

Return
14%
16%
40%
60% 12%

Return
14%
20%
40% 10%

Annualized
12%
0%
20%
8%
10%

Annualized
6%
8%
11
99
00
88
89
90
91
92
93
94
95
96
97
98

01
02
03
04
05
06
07
08
09
10

12
13

0%
20
19

19
20
19

19
19
19
19
19
19
19
19
19

20
20
20
20
20
20
20
20
20
20

20
20

4%
6% 1.8%
11
00

10

12
13
88
89
90
91
92
93
94
95
96
97
98
99

01
02
03
04
05
06
07
08
09

2% 0.2%
20
20

20
20
19
19
19
19
19
19
19
19
19
19
19
19
20
20
20
20
20
20
20
20
20
20

Source: ISSG 4% 1.8%

0%
2% 0.2%
-2%
0%
In general, our model correctly predicted the Too Cold regime as the most probable -2% n
ing tio
one at the appropriate times (Figure 10). However, as shown, it is sometimes difficult a rm r f ec
for the model to distinguish between Too Hot and Too Cold regimes, as these are both
W ing Pe tio
n
a rm r f ec
characterized by falling growth expectations. W Pe

Figure 11 shows the model’s track record in assigning a high probability to the actual
successor regime as defined by historical experience. We find that roughly 54% of the
time, the regime with the highest model probability matched with the actual regime
experienced. If we consider the two most probable regimes, the actual regime was
captured about 72% of the time, and 89% of the time the actual regime experienced
was one of the three most probable regimes. We believe that this ability to narrow
the scope of the probabilities of regimes might allow us to construct portfolios that
outperform traditional strategic asset allocations by adjusting to these regime shifts.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 12

Figure 11: Model Predictions vs Actual Regimes Experienced

Hit Rate Cumulative

Highest Probability 54.1% 54.1%

2nd Highest 18.0% 72.1%

3rd Highest 16.4% 88.5%

We translated regime 4th Highest 7.2% 95.7%
probabilities from Lowest 4.3% 100.0%
our model into asset
allocation decisions.
Source: ISSG

Figure 12 shows how often we predicted each of the five regimes to be the most
probable regime, versus how often each regime actually occurred during the
testing period. This was an important diagnostic for testing the model’s probability
estimation, since the model should be responsive to incoming data and should
assign high probabilities to less likely or “tail” scenarios when warranted. We found
our model performed well in this regard. It slightly overestimated the likelihoods of
Warming and Perfection regimes, while underestimating the Cooling regime, but the
model assigned high probabilities to current regimes in line with their experienced
frequency. This suggested that on average the model would lead to decisions in line
with the historical experience of macroeconomic regimes.

Figure 12: Highest Probability Regime vs Actual Sample Frequency

60%

50%

40%

30%
nds S es
Bo RI
P ti
S di sh
iel
d ST TIP m
o Ca 20%
Yr m
15+ Co
10%

■ Too Hot ■ Too Cold 0%

Warming Perfection Cooling Too Hot Too Cold

■ Predicted ■ Actual

Source: ISSG

40%

20%

0%

-20%
03
04
05
06
07
08
09
10

12
13
11

-40%
20
20
20
20
20
20
20
20

20
20

■ Too Hot ■ Too Cold -60%

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

■ Equity ■ Fixed Income ■ TIPS ■ Commodities

 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 13

BETTER PERFORMANCE WITH REGIME-BASED ASSET ALLOCATION

Once we were satisfied with the reliability of our model, we could apply the regime-
based asset allocation approach in real time to hypothetical portfolios and compare
the performance results with those of typical institutional portfolios. To that end, we
translated regime probabilities from our model into our asset allocation decisions.

The goal was to show how the model would dynamically weight six asset classes
through time to seek better performance. The asset classes in the simulation were
based on data from Greenwich Associates regarding the composition of typical
institutional portfolios. We calculated implied weights for the typical institutional
portfolio through time for four broad asset classes: Equity, Fixed Income, TIPS, and
Commodities. Within the Equity and Fixed Income categories we allocate to a
representative list of sub-asset classes, which can be found in the appendix. At
each portfolio formation date, we re-estimated the model and used the resultant
probabilities to develop a set of expected returns and a covariance matrix that acted
as inputs into our optimization process. We then formed portfolios, based on a set of
minimal constraints listed below, that were held for the subsequent month.

We weighted the historical average returns of our available assets for each of the
five regimes by the current regime probabilities. In a similar way, we also formed
a covariance matrix for optimization by regime probability-weighting historical
covariance data. To form expected returns for optimization, we blended these
historical returns with reverse-optimized “market-implied” returns using a Black-
Litterman-style Bayesian averaging process.8 These two building blocks formed the
basis for our mean-variance optimization. When solving the optimization problem,
we applied a minimal set of constraints that we believed were reasonable for a typical
institutional investor. We sought to maximize expected returns such that:

– Portfolio weights summed to 100%

– Positions were long-only

– The Equity weighting was constrained to be less than 75%, the Commodities
weighting less than 10%, the TIPS weighting less than 10%, and Fixed Income
and TIPS together had to be less than 75%.

8 
The Black–Litterman model, developed by Fischer Black and Robert Litterman, starts with market
equilibrium expected returns, and then modifies them to take into account the “views” of the investor
in a systematic way. Bayesian Inference is a method of statistical inference in which data are used to
update prior beliefs about a probability distribution to form a posterior probability estimate.
 50%

40%
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 14
30%
s
o nd IPS ties
dB TR PS o di sh
el S TI m Ca We20%found that our Regime-Based Asset Allocation model (RBAA) portfolio
Yi Yr m
15
+ Co outperformed the typical institutional portfolio, having both higher annualized
returns
10% (9.5% vs 8.0%) and lower volatility (8.6% vs 9.5%) for the period from
February 1988 to June 2013. The evolving regime probabilities allowed for timely
■ Too Hot ■ Too Cold 0%
changes in asset allocation that produced better returns during the testing period,
Warming Perfection Cooling Too Hot Too Cold
both on an absolute and risk-adjusted basis. Our regime-based portfolio had a
much higher Sharpe ratio compared with the institutional
■ Predicted ■ Actual
portfolio (0.64 vs 0.42)
(see Figure 16). Averaged over the time period, the RBAA portfolio produced better
RBAA portfolio outperformed risk-adjusted performance with a slight underweight to equities versus the
the typical institutional institutional benchmark.
portfolio, having both
Figure 13: RBAA Over and Underweights vs Institutional Portfolio
higher annualized returns
and lower volatility. 40%

20%

0%

-20%
3
04
05
06
07
08
09
10

12
13
11
00

-40%
20
20
20
20
20
20
20
20

20
20

■ Too Hot ■ Too Cold -60%

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

■ Equity ■ Fixed Income ■ TIPS ■ Commodities

Source: ISSG & Greenwich Associates.

As can be seen in Figure 13, the monthly allocations among the four broad asset
classes could vary widely from these averages. The RBAA portfolio favored equities
during Warming, Cooling, and Perfection periods, and dramatically de-risked with
Fixed Income when the Too Hot and Too Cold probabilities rose.

Figure 14: Average Weights

RBAA Portfolio Inst. Portfolio

Equities 58% 60%

Fixed Income 33% 34%

11
12
13

TIPS
04
05
06
07
08
09
10

5% 6%
20
20
20
20
20
20
20
20

20
20

Commodities 5% 0%

Source: ISSG & Greenwich Associates.

In Figure 15 we see that during the sample period, the excess return deviations
of the RBAA portfolio from the institutional portfolio were fairly modest during
the Just Right economic regimes – Warming, Cooling, and Perfection. The RBAA
portfolio realized the majority of its outperformance during times of stress when
Too Hot and Too Cold regimes are predicted as most probable.

20%
17.3%
18%
16%
ualized Return

14%
12%
10%
8%
6% 3.9%
11
03
04
05
06
07
08
09
10

12
13
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 15

Figure 15: Excess Returns of RBAA Portfolio by Regime

20% 40%
17.3%
18% 30%
16% 20%
Annualized Return

14%
10%
12%
10% 0%
8% -10%
6% 3.9% -20%
4% 1.8%
-30%
2% 0.2%
0% -40%

08
07
-2%

-0

n-
c-
-0.9%

ov

De

Ja
N
ing on ng ot ld
rm cti oli oH Co
erf
e Co To To
o
Wa P

Source: ISSG & eVestment Alliance

Exhibit 16 shows the performance statistics of the RBAA model portfolio compared
with a typical institutional portfolio from February 1988 through to August 2011.
The RBAA portfolio had roughly 1.6% better annualized performance, with volatility
(annualized risk) that was about 3% less than that of the typical institutional
portfolio. This resulted in an almost doubling of the RBAA portfolio’s Sharpe ratio
compared with that of the institutional portfolio.

Figure 16: RBAA vs Institutional Portfolio Performance (Net of Fees)

RBAA Portfolio Inst. Portfolio

Annualized Return 10% 8%

Annualized Risk 9% 10%

Risk Free Rate 4% 4%

Sharpe Ratio 64% 42%

Source: ISSG. See appendix for information regarding fees.

STRESS-TESTING THE MODEL

To further showcase what we regard as distinctive benefits of applying a regime-based
approach to asset allocation, especially during times of market stress, we highlighted
our RBAA portfolio results for two particularly challenging periods in the markets, the
bursting of the technology bubble in the early part of the 2000s as well as the global
financial crisis of 2007-2009. We isolated those crisis periods as those are the points
at which a regime-focused investor would hope to outperform relative to a buy-and-
hold strategy.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 16

Case
30 Study 1: The Bursting of the Technology Bubble (2/28/00 to 11/30/02)
30
While the results overall were positive (-10.6% holding period return for the
25
25
institutional portfolio versus -0.5% holding period return for the RBAA portfolio),
we
20
20
believed it was important to determine what drove the higher risk-adjusted
returns historically. While the average weights across the regime-based portfolio
and
15 the institutional portfolio were quite similar, the dynamic nature of the regime
15
based portfolio was quite stark. For example, Figure 17 illustrates the allocation
10
10
swings relative to the institutional portfolio for the tech-bubble crisis period.
55
As shown, going into the crisis, the RBAA portfolio held a modest overweight to Fixed
Income
00 compared to the institutional portfolio. By June of 2000, the RBAA portfolio
increased this overweight position. A second leg down in the market was captured by

-88%

-66%

-44%

-22%

0%%

2%%

4%%

6%%

8%%
4%%

2%%

0%%

100%

122%

144%
%

%
oHot
Hot Too
TooCold
Cold

8
-114

-112

-110

1
investors’ expectations and the probability of a second Too Cold scenario increasing

-
under the RBAA model in November
■■RBAA of 2001. The model responded and de-risked
RBAA ■■Institutional
InstitutionalPortfolio
Portfolio
almost immediately, offering significant protection. Figure 18 shows the relative
performance of the RBAA portfolio to the institutional portfolio.

Figure 17:
Relative Asset Class Weights of the RBAA Portfolio vs Institutional Portfolio

40%
40%
30%
30%
20%
20%
10%
10%
0%
0%
-10%
-10%
-20%
-20%
-30%
-30%

5 2006
005 20062007
20072008
20082009
20092010
20102011
20112012
20122013
2013
-40%
-40%
May-00

May-01

May-02
Mar-00

Mar-01

Nov-01

Mar-02
Aug-00
Sep-00

Nov-00
Dec-00

Aug-01
Sep-01

Dec-01

Aug-02
Sep-02
Feb-00

Feb-01

Feb-02
Jun-00

Jun-01

Jun-02
Apr-00

Jan-01

Apr-01

Jan-02

Apr-02
Oct-00

Oct-01

Oct-02
Jul-00

Jul-01

Jul-02
May-00

May-01

May-02
Mar-00

Nov-00

Mar-01

Mar-02
Aug-00
Sep-00

Dec-00

Aug-01
Sep-01

Nov-01
Dec-01

Aug-02
Sep-02
Feb-00

Feb-01

Feb-02
Jun-00

Jun-01

Jun-02
Apr-00

Jan-01

Apr-01

Jan-02

Apr-02
Oct-00

Oct-01

Oct-02
Jul-00

Jul-01

Jul-02
Commodities
Commodities

■■Equity
Equity ■■Fixed
FixedIncome
Income ■■TIPS
TIPS ■■Commodities
Commodities

Source: ISSG. See appendix for index descriptions.

Figure 18: Monthly Excess Return of the RBAA Portfolio vs Institutional Portfolio

4%
4%

2%
2%

0%
0%

-2%
-2%

-4%
-4%
Mar-00
Apr-00
May-00
Jun-00
Jul-00
Aug-00
Sep-00
Oct-00
Nov-00
Dec-00
Jan-01
Feb-01
Mar-01
Apr-01
May-01
Jun-01
Jul-01
Aug-01
Sep-01
Oct-01
Nov-01
Dec-01
Jan-02
Feb-02
Mar-02
Apr-02
May-02
Jun-02
Jul-02
Aug-02
Sep-02
Oct-02
Nov-02
Mar-00
Apr-00
May-00
Jun-00
Jul-00
Aug-00
Sep-00
Oct-00
Nov-00
Dec-00
Jan-01
Feb-01
Mar-01
Apr-01
May-01
Jun-01
Jul-01
Aug-01
Sep-01
Oct-01
Nov-01
Dec-01
Jan-02
Feb-02
Mar-02
Apr-02
May-02
Jun-02
Jul-02
Aug-02
Sep-02
Oct-02
Nov-02

Source: ISSG

40%
40%
17.3%
17.3%
 2%
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 17
0%

Case Study 2: The Financial Crisis (11/30/2007 to 5/31/2009)

-2%
From November 2007 through May of 2009, the RBAA portfolio holding period
return was -8.0% while the typical institutional portfolio holding period return
-4%
was -22.0%. Figure 19 shows the relative asset class weights of the RBAA portfolio
Mar-00
Apr-00
May-00
Jun-00
Jul-00
Aug-00
Sep-00
Oct-00
Nov-00
Dec-00
Jan-01
Feb-01
Mar-01
Apr-01
May-01
Jun-01
Jul-01
Aug-01
Sep-01
Oct-01
Nov-01
Dec-01
Jan-02
Feb-02
Mar-02
Apr-02
May-02
Jun-02
Jul-02
Aug-02
Sep-02
Oct-02
Nov-02
during that time period. Of the eighteen months shown in Figure 20, the model had
a positive contribution to performance only half of the time. However, the protection
the model provided during the large down months for the institutional portfolio was
significant. The RBAA portfolio lagged (while still showing positive returns) in the
months subsequent to the equity market’s bottoming in March of 2009. RBAA added
value for the full period, participating in up markets while protecting in down.

Figure 19:
Relative Asset Class Weights of the RBAA Portfolio vs Institutional Portfolio

40%

30%

20%

10%

0%

-10%

-20%

-30%

-40%
8
8

9
7

8
08

09
08
08

08

09
08

08

08
08

09
07

08
-0
-0

-0
-0

-0
r-

r-
n-
b-

p-

b-
l-

g-

t-
n-

n-
c-

c-
ay
ar

ar
ov

ov
Ju
Ap

Ap
Oc
De

De
Au
Ju
Fe

Se

Fe
Ja

Ja
M

M
M
N

■ Equity ■ Fixed Income ■ TIPS ■ Commodities

Source: ISSG. See appendix for index descriptions.

Figure 20: Monthly Excess Return of the RBAA Portfolio vs Institutional Portfolio

8%

6%

4%

2%

0%

-2%

-4%

-6%
8

9
8

9
8
8

M 9
8
08

08

09
8
7

08
8

09
08
-0

-0
-0

-0
-0
0

0
0

-0
-0

-0
r-

r-
n-
b-

p-

b-
t-
n-

n-
l-
ay

ay
ar

ar
ov
g
c

c
Ju
Ap

Ap
Oc
De

De
Au
Ju
Fe

Se

Fe
Ja

Ja
M

M
M

Source: ISSG

Total Portfolio or Model Driven or

Cash or Synthetic?
Intra-Asset Sleeve? Qualitative Process?
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 18

IMPLEMENTATION CONSIDERATIONS

Figure 21: Implementation Questions

Total Portfolio or Model Driven or

Cash or Synthetic?
Intra-Asset Sleeve? Qualitative Process?

Source: ISSG

There are several possible ways to implement a regime-based framework in an

institutional portfolio. A full convert to regime-based investing could implement
an asset allocation structure that seeks to dynamically weight asset classes based
on the model’s macroeconomic regime forecast. A partial adopter might choose to
maintain strategic portfolio weights across the traditional asset classes (i.e., equity,
fixed income, alternatives) but make shifts within asset classes to reflect a view on
the macroeconomic state. For example, our research showed that within the equity
sleeve, using a sector rotation strategy based on inflation and growth might perform
quite well.

Another important implementation decision is whether to dynamically adjust the

allocations (total portfolio or intra-asset class) by rebalancing or using a synthetic
overlay strategy. In our view, there are pros and cons to each approach.

Investors with highly liquid portfolios consisting of only public securities might
consider rebalancing portfolio holdings to the target regime-dependent portfolio.
In normal markets, the benefit of rebalancing according to regime may outweigh
the explicit transaction costs. However, in non-normal, illiquid markets this might
not be possible. Additionally, an investor would need to consider other factors such
as realized gains and losses, relationships with managers, or the other implicit costs
of rebalancing with physical securities.

The primary benefits of adjusting allocations synthetically are speed and cost.
Equity and interest rate derivatives, for example, can be used to adjust a portfolio’s
equity beta or duration without disrupting the activity of underlying managers. Using
derivatives to adjust portfolio exposures may be the only option for portfolios with
significant holdings of illiquid investments.

Investors who are able to deploy derivatives for rebalancing face their own unique
challenges, though. Investors must be able to accommodate the operational,
regulatory and governance challenges of implementing a derivatives program, and
sufficient liquidity for collateral must be allocated to the positions. Further, there
may be significant basis risk between an investor’s holdings and the derivative
instruments with which he is able to adjust his portfolio exposures.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 19

For example, in anticipation of a Warming or Too Hot regime, an investor might want
to increase the portfolio’s sensitivity (i.e., beta) to inflation and inflation surprises.
The investor may already hold an allocation to private real asset investments with
an adequate inflation surprise beta, but it is unlikely that he would be able to deploy
sufficient private capital in time to raise the portfolio’s overall inflation sensitivity.
So, to increase the portfolio’s allocation to real assets, the investor could turn to
derivatives, based on commodity indices. Over a short or medium horizon, these
indices might not have the inflation surprise sensitivity required due to the tendency
of commodity-related assets to, at times, exhibit growth-like characteristics.

In our view, the efficacy of a regime framework for asset allocation will be affected
by the active managers within a portfolio. An investor contemplating when to over-
or underweight a manager relative to his strategy’s strategic weight in the asset
allocation structure should be aware of the regimes in which the strategy should
be poised to outperform.

Finally, investors should consider the relative benefits of a model-driven process

(state probability estimates combined with an optimization process, along the lines
of what we developed) versus a qualitative process. Quantitative processes have the
advantage of consistency, while avoiding behavioral biases. Qualitative processes
might allow investors to retain a larger degree of oversight. We believe there is value
from both, and think that a skilled investor with a good model framework by which to
frame the issue might offer the best option.
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 20

CONCLUSION
We believe incorporating macroeconomic changes into asset allocation structures in
a dynamic way might improve overall performance. Our RBAA portfolio demonstrated
the potential for improving risk-adjusted returns over time compared with more
static institutional approaches to strategic asset allocation. More importantly, given
the devastating losses suffered from the unexpected convergence of asset class
correlations during the financial crisis, we believe regime-based asset allocation has
the potential to become a powerful risk management tool during times of market
We believe incorporating stress. At the very least, understanding how changes in growth and inflation can affect
macroeconomic changes specific asset prices and correlations should enable investors to better recognize
into asset allocation the potential risks in their portfolios. Amid general expectations of protracted market
structures in a dynamic volatility and uncertainty as the global economy endures historic rebalancing, we
believe traditional approaches to strategic asset allocation with limited flexibility to
way might improve adjust to regime shifts might be at a disadvantage. A new era in financial markets
overall performance. seems to suggest that a more opportunistic approach should be considered.

BNY Mellon’s Investment Strategy & Solutions Group would like to thank
David Chapman, Director, Risk Management, at Catholic Healthcare Investment
Management Company for his extensive collaboration with the regime-based
asset allocation research project as well as his thoughtful review of and helpful
suggestions for this white paper.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 21

APPENDIX

MODEL DRIVER DEFINITIONS

Series name Start End

Expected 12M Inflation 2/29/1988 6/30/13

Expected 12M Real GDP 2/29/1988 6/30/13

12M Revision in Expected Inflation 2/29/1988 6/30/13

12M Revision in Expected Real GDP 2/29/1988 6/30/13

– A measure of the forward 12 month U.S. inflation forecast using data from the Survey of Professional
Forecasters and Consensus Economics.
– A measure of the forward 12 month U.S. real GDP forecast using data from the Survey of Professional
Forecasters and Consensus Economics.
– A measure of the aggregate revisions to the forward 12 month U.S. inflation forecast over a twelve
month time period.
– A measure of the aggregate revisions to the forward 12 month U.S. real GDP forecast over a twelve
month time period.

BENCHMARK COMPOSITION AND INDEX DEFINITIONS

Benchmark
Asset name Weight Index Name Start End
U.S. Equity 33.0% S&P 500 4/30/1973 6/30/13

Int’l Equity 18.0% MSCI EAFE 4/30/1973 6/30/13

EM Equity 6.0% MSCI EM 4/30/1973 6/30/13

Real Estate 3.0% FTSE NAREIT 4/30/1973 6/30/13

Corp. Bonds 12.0% Barcap US Corporate Agg 4/30/1973 6/30/13

Global Sovereigns 0.0% JPM Government 4/30/1973 6/30/13

Bond Index Global

Treasuries 14.0% Barcap US Treasury Agg 4/30/1973 6/30/13

High Yield Bonds 6.0% Barcap US High Yield 4/30/1973 6/30/13

15+ Yr STRIPS 0.0% Citi 15+ STRIPS 4/30/1973 6/30/13

TIPS 6.0% Barcap US TIPS 4/30/1973 6/30/13

Commodities 0.0% DJ-UBS Commodities 4/30/1973 6/30/13

Cash 2.0% Citi 3 Month Treasury (Cash) 4/30/1973 6/30/13

– The S&P 500 is an index designed to track the performance of the largest 500 U.S. companies.
– The MSCI EAFE Index (Europe, Australasia, Far East) is designed to measure the equity market
performance of global developed markets, excluding the U.S. & Canada.
– The MSCI EM index tracks the performance of Emerging Market Equities. Prior to 1987 the returns
are combined with the IFC emerging market returns and the MSCI EAFE index.
– The FTSE/NAREIT index is designed to track the performance U.S. Real Estate Investment Trusts.
– The Barcap U.S. Corporate Aggregate Index is designed to track the performance of U.S. Investment
Grade Corporate securities.
– The JPM Government Bond Index Global is designed to track the broad universe of global government
bonds. Results simulated before 1985.
– The Barcap U.S. Treasury Aggregate Index is designed to track the performance of U.S. Treasury
securities.
– The Barcap High Yield Index tracks the performance of high yield debt securities. Prior to 1983 returns
are regressed against the returns of the Barcap Baa and Russell 2000.
GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 22

– The Barcap U.S. Treasury Aggregate Index is designed to track the performance of U.S.
Treasury securities.
– The Citi 15+ STRIPS index tracks the performance of U.S. Treasury 15+ year STRIPS.
Results simulated prior to 1991.
– The Barcap U.S. TIPS index is designed to track the performance of U.S. Treasury Inflation Protected
Securities. Returns are simulated prior to 1997.
– Index designed to provide diversified commodity exposure with weightings based on the commodity’s
liquidity and economic significance. Results simulated prior to 1991.
– The Citigroup Three Month Treasury Bill index tracks the performance of 90 day U.S. Treasury bills.
We use data from the Federal Reserve before 1978.

These benchmarks are broad-based indices which are used for comparative purposes only and
have been selected as they are well known and are easily recognizable by investors. Comparisons to
benchmarks have limitations because benchmarks have volatility and other material characteristics
that may differ from the portfolio. For example, investments made for the portfolio may differ
significantly in terms of security holdings, industry weightings and asset allocation from those of the
benchmark. Accordingly, investment results and volatility of the portfolio may differ from those of the
benchmark. Also, the indices noted in this presentation are unmanaged, are not available for direct
investment, and are not subject to management fees, transaction costs or other types of expenses that
the portfolio may incur. In addition, the performance of the indices reflects reinvestment of dividends
and, where applicable, capital gain distributions. Therefore, investors should carefully consider these
limitations and differences when evaluating the comparative benchmark data performance.

DISCLOSURES
Investment Strategy and Solutions Group (“ISSG”) is part of The Bank of New York Mellon (“Bank”). ISSG offers
products and services through the Bank, including investment strategies that are developed by affiliated BNY
Mellon Asset Management investment advisory firms and managed by officers of such affiliated firms acting
in their capacities as dual officers of the Bank.

Typical institutional portfolio data is based on a survey compiled by Greenwich Associates on average asset
allocations for Corporate, Public, and Endowment & Foundation plans covering the period from 1990-2009.
The ISSG applied the 1990 weights to 1988 and 1989 as well, and used the 2010 weights for 2011.

HYPOTHETICAL OR SIMULATED PERFORMANCE RESULTS HAVE CERTAIN INHERENT LIMITATIONS. UNLIKE AN

ACTUAL PERFORMANCE RECORD, SIMULATED RESULTS DO NOT REPRESENT ACTUAL TRADING. SIMULATED
TRADING PROGRAMS IN GENERAL ARE ALSO SUBJECT TO THE FACT THAT THEY ARE DESIGNED WITH THE
BENEFIT OF HINDSIGHT. ALSO, SINCE THE TRADES HAVE NOT ACTUALLY BEEN EXECUTED, THE RESULTS
MAY HAVE UNDER OR OVER COMPENSATED FOR THE IMPACT OF CERTAIN MARKET FACTORS. IN ADDITION,
HYPOTHETICAL TRADING DOES NOT INVOLVE FINANCIAL RISK. NO HYPOTHETICAL TRADING RECORD CAN
COMPLETELY ACCOUNT FOR THE IMPACT OF FINANCIAL RISK IN ACTUAL TRADING. FOR EXAMPLE, THE
ABILITY TO WITHSTAND LOSSES OR TO ADHERE TO A PARTICULAR TRADING PROGRAM IN SPITE OF THE
TRADING LOSSES ARE MATERIAL FACTORS WHICH CAN ADVERSELY AFFECT THE ACTUAL TRADING RESULTS.
THERE ARE NUMEROUS OTHER FACTORS RELATED TO THE ECONOMY OR MARKETS IN GENERAL OR TO THE
IMPLEMENTATION OF ANY SPECIFIC TRADING PROGRAM WHICH CANNOT BE FULLY ACCOUNTED FOR IN
THE PREPARATION OF HYPOTHETICAL PERFORMANCE RESULTS, ALL OF WHICH CAN ADVERSELY AFFECT
TRADING RESULTS.
 GREAT EXPECTATIONS: REGIME-
BASED ASSET ALLOCATION // 23

OUR MISSION
The Investment Strategy and Solutions Group (ISSG) partners with clients
to develop thoughtful and actionable solutions to the broad investment
policy issues confronting corporate and public retirement plans, Michael Rausch, CFA, ASA
endowments and foundations, sovereign wealth funds and financial Investment Strategist
institutions and intermediaries. As the investment landscape grows more 412.236.8832
[email protected]
complex and uncertain, our seasoned investment professionals seek to
satisfy client’s appetite for trusted advisors who can help them think
through their most difficult investment challenges. To that end, we engage
in an ongoing dialogue with our institutional clients to achieve a deep Stephen Kolano, CFA
understanding of their concerns and needs. Investment Strategist
617.722.3995
For more information, please contact: [email protected]

Jeffrey B. Saef, CFA Al Trezza, CFA, ASA

Managing Director Investment Strategist
617.722.6956 412.236.7732
[email protected] [email protected]

Charles Dolan, Ph.D., CFA Ivo Batista, CFA

Senior Investment Strategist Portfolio Strategist
213.553.9540 +44.20.7163.5475
[email protected] [email protected]

Ralph P. Goldsticker, CFA Harsh Parikh

Senior Investment Strategist Portfolio Strategist
415.975.2383 617.722.7736
[email protected] [email protected]

Robert A. Jaeger, Ph.D. Stacy Devlin

Senior Investment Strategist Assistant Portfolio Manager
203.722.0986 617.722.7908
[email protected] [email protected]

Rumi Masih, Ph.D. Elena Goncharova, CFA

Senior Investment Strategist Investment Analyst
617.722.7859 617.722.7871
[email protected] [email protected]

Andrew Wozniak, CFA Michael W. Griswold, CFA

Senior Investment Strategist Investment Analyst
412.236.7940 617.722.7797
[email protected] [email protected]

Chris Harris, CFA John Zalewski

Investment Solutions Strategist Investment Analyst
+81.3.6756.4637 412.234.7864
[email protected] [email protected]
 BNY Mellon Investment Management is one of the world’s leading investment management organizations and one of the
top U.S. wealth managers, encompassing BNY Mellon’s affiliated investment management firms, wealth management
organization and global distribution companies. BNY Mellon is the corporate brand of The Bank of New York Mellon
Corporation and may also be used as a generic term to reference the Corporation as a whole or its various subsidiaries
generally. • The statements and opinions expressed in this document are those of the authors as of the date of the article,
are subject to change as economic and market conditions dictate, and do not necessarily represent the views of BNY
Mellon, BNY Mellon Asset Management International or any of their respective affiliates. This document is of general
nature, does not constitute legal, tax, accounting or other professional counsel or investment advice, is not predictive of
future performance, and should not be construed as an offer to sell or a solicitation to buy any security or make an offer
where otherwise unlawful. The information has been provided without taking into account the investment objective,
financial situation or needs of any particular person. BNY Mellon Asset Management International Limited and its
affiliates are not responsible for any subsequent investment advice given based on the information supplied.
Past performance is not a guide to future performance. The value of investments and the income from them is not
guaranteed and can fall as well as rise due to stock market and currency movements. When you sell your investment
you may get back less than you originally invested. • While the information in this document is not intended to be
investment advice, it may be deemed a financial promotion in non-U.S. jurisdictions. Accordingly, where this document
is used or distributed in any non-U.S. jurisdiction, the information provided is for use by professional and wholesale
investors only and not for onward distribution to, or to be relied upon by, retail investors. • Products or services
described in this document are provided by BNY Mellon, its subsidiaries, affiliates or related companies and may be
provided in various countries by one or more of these companies where authorized and regulated as required within
each jurisdiction. This document is not intended for distribution to, or use by, any person or entity in any jurisdiction
or country in which such distribution or use would be contrary to local law or regulation. This document may not be
distributed or used for the purpose of offers or solicitations in any jurisdiction or in any circumstances in which such
offers or solicitations are unlawful or not authorized, or where there would be, by virtue of such distribution,
new or additional registration requirements. Persons into whose possession this document comes are required
to inform themselves about and to observe any restrictions that apply to the distribution of this document in their
jurisdiction. The investment products and services mentioned here are not insured by the FDIC (or any other state
or federal agency), are not deposits of or guaranteed by any bank, and may lose value. • This document should
not be published in hard copy, electronic form, via the web or in any other medium accessible to the public, unless
authorized by BNY Mellon Investment Management International Limited.
In Australia, this document is issued by BNY Mellon Investment Management Australia Ltd (ABN 56 102 482 815,
AFS License No. 227865) located at Level 6, 7-15 Macquarie Place, Sydney, NSW 2000. Authorized and regulated
by the Australian Securities & Investments Commission. • In Brazil, this document is issued by BNY Mellon
Serviços Financeiros DTVM S.A., Av. Presidente Wilson, 231, 11th floor, Rio de Janeiro, RJ, Brazil, CEP 20030-905.
BNY Mellon Serviços Financeiros DTVM S.A. is a Financial Institution, duly authorized by the Brazilian Central Bank
to provide securities distribution and by the Brazilian Securities and Exchange Commission (CVM) to provide securities
portfolio managing services under Declaratory Act No. 4.620, issued on December 19, 1997. • Securities in Canada
are offered through BNY Mellon Asset Management Canada Ltd., registered as a Portfolio Manager and Exempt
Market Dealer in all provinces and territories of Canada, and as an Investment Fund Manager in Ontario. • In
Dubai, United Arab Emirates, this document is issued by the Dubai branch of The Bank of New York Mellon, which
is regulated by the Dubai Financial Services Authority. • If this document is used or distributed in Hong Kong,
it is issued by BNY Mellon Investment Management Hong Kong Limited, whose business address is Level 18,
Three Pacific Place, 1 Queen’s Road East, Hong Kong. BNY Mellon Investment Management Hong Kong Limited
is regulated by the Hong Kong Securities and Futures Commission and its registered office is at 6th floor,
Alexandra House, 18 Chater Road, Central, Hong Kong. • In Japan, this document is issued by BNY Mellon Asset
Management Japan Limited, Marunouchi Trust Tower Main Building, 1-8-3 Marunouchi Chiyoda-ku, Tokyo 100-0005,
Japan. BNY Mellon Asset Management Japan Limited is a Financial Instruments Business Operator with license
no 406 (Kinsho) at the Commissioner of Kanto Local Finance Bureau and is a Member of the Investment Trusts
Association, Japan and Japan Securities Investment Advisers Association. • In Korea, this document is issued by
The Alcentra Group BNY Mellon AM Korea Limited for presentation to professional investors. BNY Mellon AM Korea Limited, 29F One IFC,
BNY Mellon ARX 10 Gukegeumyung-ro, Yeongdeungpo-gu, Seoul, 150-945, Korea. Regulated by the Financial Supervisory Service.
• In Singapore, this document is issued by The Bank of New York Mellon, Singapore Branch for presentation to
BNY Mellon Cash Investment Strategies
professional investors. The Bank of New York Mellon, Singapore Branch, One Temasek Avenue, #02-01 Millenia
BNY Mellon Western Fund Management Tower, Singapore 039192. Regulated by the Monetary Authority of Singapore. In Singapore, this document is to be
Company Limited distributed to Institutional Investors (as defined in the Securities and Futures Act, Chapter 289 of Singapore) only. •
The Boston Company Asset Management, LLC This document is issued in the UK and in mainland Europe, by BNY Mellon Asset Management International Limited,
CenterSquare Investment Management, Inc. 160 Queen Victoria Street, London EC4V 4LA. Registered in England No. 1118580. Authorized and regulated by the
Financial Conduct Authority. • This document is issued in the United States by BNY Mellon Investment Management.
CenterSquare Investment Management Holdings, Inc.
The Dreyfus Corporation BNY Mellon owns over 95% of the parent holding company of The Alcentra Group, which is comprised of the following
affiliated investment advisers: Alcentra, Ltd and Alcentra NY, LLC. • BNY Mellon ARX is the brand used to describe the
EACM Advisors LLC
Brazilian investment capabilities of BNY Mellon ARX Investimentos Ltda. • BNY Mellon Cash Investment Strategies is
Hamon Investment Group a division of The Dreyfus Corporation. • BNY Mellon Western FMC, Insight Investment Management Limited and
Insight Investment Meriten Investment Management GmbH do not offer services in the U.S. This presentation does not constitute an offer
Mellon Capital Management Corporation to sell, or a solicitation of an offer to purchase, any of the firms’ services or funds to any U.S. investor, or where otherwise
Meriten Investment Management unlawful. • BNY Mellon Western Fund Management Company Limited is a joint venture between BNY Mellon (49%)
and China based Western Securities Company Ltd. (51%). The firm does not offer services outside of the People’s
The Newton Group
Republic of China. • BNY Mellon owns 90% of The Boston Company Asset Management, LLC and the remainder is
Siguler Guff & Company LP owned by employees of the firm. • BNY Mellon owns a 19.9% minority interest in The Hamon Investment Group Pte
Standish Mellon Asset Management Company LLC Limited, the parent company of Blackfriars Asset Management Limited and Hamon Asian Advisors Limited both of
Walter Scott & Partners Limited which offer investment services in the U.S. • Services offered in the US, Canada and Australia by Pareto Investment
Management Limited under the Insight Pareto brand. • The Newton Group (“Newton”) is comprised of the following
affiliated companies: Newton Investment Management Limited, Newton Capital Management Limited (NCM Ltd),
Newton Capital Management LLC (NCM LLC), Newton International Investment Management Limited and Newton
Fund Managers (C.I.) Limited. NCM LLC personnel are supervised persons of NCM Ltd and NCM LLC does not provide
investment advice, all of which is conducted by NCM Ltd. Only NCM LLC and NCM Ltd offer services in the U.S. • BNY
Mellon owns a 20% interest in Siguler Guff & Company, LP and certain related entities (including Siguler Guff Advisers
LLC). • BNY Mellon Asset Management International Limited and any other BNY Mellon entity mentioned above are all
ultimately owned by BNY Mellon, unless otherwise noted.

©2013 The Bank of New York Mellon Corporation. 9/13 bnymellonim.com BNY Mellon Investment Management