8/8/2019 2009 Equity Gilt Study
1/95
Gilt StudyEquity
2009Please see analyst certification(s) and important disclosures starting on the inside back cover.
8/8/2019 2009 Equity Gilt Study
2/95
"Government's view of the economy could besummed up in a few short phrases: If itmoves, tax it. If it keeps moving, regulate it.
And if it stops moving, subsidise it"
Ronald Reagan
"Blessed are the young, for they shall inheritthe national debt"
Herbert Hoover
"A budget tells us what we can't afford, but itdoesn't keep us from buying it"
William Feather
Borrowing dulls the edge of husbandry
William Shakespeare, Hamlet
"The significant problems we face cannot besolved at the same level of thinking we wereat when we created them"
Albert Einstein
Published by Barclays Capital
12 February 2009
8/8/2019 2009 Equity Gilt Study
3/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 1
Equity Gilt Study 2009
54th Edition
The Equity Gilt Study has been published continuously since 1956, providing data,
analysis and commentary on long-term returns from financial assets in the UK and US.
Our UK database begins in 1899, while the US database starts in 1925. The US data are
kindly provided by the Centre for Research in Security Prices at the University of
Chicago Graduate School of Business. Our purpose in publishing this data is to provide
investors with a perspective on long-term asset returns. In the 2009 Study, we examine
the causes of the very poor equity returns of the past decade, we discuss asset returns
under varying classes of inflationary and deflationary conditions and we analyse the
permutations and causes of the credit cycle. Our colleagues at Barclays Global Investors
contribute a chapter highlighting the advantages offered by ETFs and examining how
they can be used in a variety of strategies. As always, we also provide a thoroughanalysis of historical returns from a wide variety of asset classes in both the UK and US,
together with the relevant data.
We hope you enjoy the essays and find the data useful.
Tim Bond Deborah Fuhr
Sree Kochugovindan Shane Kelly
Barclays Capital Barclays Global Investors
Website: www.equitygiltstudy.com
E-mail: [email protected]
http://www.equitygiltstudy.com/mailto:[email protected]:[email protected]://www.equitygiltstudy.com/8/8/2019 2009 Equity Gilt Study
4/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 3
Table of contentsExecutive summary
Chapter 1 The lost decade 5Chapter 2 Deflated markets 15Chapter 3 Viral economics 22Chapter 4 Back to beta with exchange traded-funds (ETFs) 34Chapter 5 UK asset returns since 1899 54Chapter 6 US asset returns 60Chapter 7 Barclays indices 64Chapter 8 Total investment returns 88
8/8/2019 2009 Equity Gilt Study
5/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
4 Global Asset Allocation Strategy Barclays Capital
Chapter 1 The lost decade
Equity investors have been on a wild and ultimately disappointing ride over the past
decade. Equities have been the worst-performing asset class since 1997, sharply
underperforming all other asset classes. We examine the causes of this relative
weakness, and find that the utility of simple valuation measures has been thoroughly
vindicated by the dreadful recent returns from equities. We show how future long-termreturns from equities the equity risk premium can be forecast. We also describe the
factors that cause equity valuations to fluctuate over time. Finally, we compare the
outlook for stock and bond returns over the next 10 years.
Chapter 2 Deflated markets
The financial turbulence of 2008 has led to a fascination with the Depression era.
Concerns over deflation have replaced the inflation scare which prevailed in the first
half of 2008. This raises the question over which state presents the greater evil,
deflation or high inflation? In this article, we compare the performance of a range ofassets and equity sectors across different inflation regimes since the 1920s. We
differentiate between phases of good and bad deflation to gain further insight and find
that, in fact, credit conditions may be a more important factor to consider in
determining trends in asset performance. Perhaps the focus on deflation or stagflation
has been a diversion in comparison to the importance of credit regimes.
Chapter 3 Viral economics
Both the occurrence and the economic impact of the credit crunch caught policy-
makers, regulators, bankers, analysts and investors largely unprepared. We examinewhy this might be so, given the empirical evidence that credit cycles are inherently
predictable and of considerable importance to the path of economic growth. Our
conclusions highlight the endemic instability of a pure free market system.
Chapter 4 Back to beta with ETFs
In 1975 Charles Ellis highlighted the shortfall of active managers in his often referenced
article The Losers Game published in the Financial Analysts Journal, July/August 1975.
He reported that over the prior decade 85% of all institutional investors who tried to beat
the stock market underperformed the S&P 500 index. In 1976, the first indexed fund was
launched in the US. Since then ETFs have become popular and widely used investment
vehicles. This article discusses the many advantages offered by ETFs and examines how
they can be used in a variety of portfolio strategies.
Chapter 5 and 6
We publish last years US and UK asset Returns, placing them within a historical context.
Equities had a terrible year globally. The FTSE All-share real total returns were the
weakest since the 1970s. The US equity returns were the weakest since the Great
Depression. Government bonds were the main beneficiary of the financial turbulence of
2008. In both the UK and the US, government bonds were the best-performing asset of
the year. They also produced the best average annual returns over 20 years. Bonds
rarely outperform equities over such a long holding period.
8/8/2019 2009 Equity Gilt Study
6/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 5
Chapter 1 The lost decadeTim Bond
Equity investors have been on a wild and ultimately disappointing ride over the past
decade. Equities have been the worst performing asset class since 1997, sharply
underperforming all other asset classes. We examine the causes of this relative weakness,and find that the utility of simple valuation measures has been thoroughly vindicated by
the dreadful recent returns from equities. We show how future long-term returns from
equities the equity risk premium can be forecast. We also describe the factors that
cause equity valuations to fluctuate over time. Finally, we compare the outlook for stock
and bond returns over the next 10 years.
Equity returns over the past decade have been among the worst on record. In nominal
terms, the -0.3% annualised return from US equities since 1998 is the fourth-worst 10-
year return of the past 83 years. Only those 10-year periods ending in 1937, 1938 and
1939 have delivered lower returns. Similarly, over the past 109 years, only the decade
ending in 1974 saw a weaker 10-year nominal return from UK equities. For the sake of
record, the 1964-74 UK equity return was 1.02%, while the 1998-2008 return was
1.05%. In both the US and UK, the real total return from equities over the past decade
has been negative.
Figure 1: 10-year rolling total returns, nominal and real, US and UK equities
-10
-5
0
5
10
15
20
25
1935 1945 1955 1965 1975 1985 1995 2005
rolling nominal 10 year return, US equity
rolling real 10 year return USequity
-10
-5
0
5
1015
20
25
30
35
1909 1929 1949 1969 1989 2009
rolling 10 yr nominal
returns, UK equity
rolling 10 year real
return, UK equity
Source: CRSP, Barclays Capital
As a natural reaction to this long phase of poor returns, there has been much talk of the
death of the equity cult. While such talk may accurately represent investors
disenchantment with equities as an asset class, it is most likely a poor forecast for
future equity returns. Prospective returns from equities are at the most attractive levels
seen for some 20 years in the US and over 25 years in Europe and the UK, even if ex-
post returns have been feeble.
The weak returns from equities over the past decade are not due to some intrinsic
problem with the asset class. Rather, they are attributable to the extreme overvaluation
of equities at the start of the decade. Although the growth in corporate profits has been
robust over the period in question, investors were paying a very high premium to
access these profits at the start of the decade. This premium has hampered, not to say
eradicated, positive returns. From 1997 through to 2002, equities were valued at
unusually expensive levels relative to earnings and corporate net worth. The collapse inequities after 2001 partially corrected this overvaluation, as equity prices declined by
more than earnings during the 2001-03 global slowdown. The subsequent economic
8/8/2019 2009 Equity Gilt Study
7/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
6 Global Asset Allocation Strategy Barclays Capital
boom from 2003-08 generated a strong trend in profitability and in turn generated a
strong rally in stock markets. However, throughout this period, valuations declined and
equity prices generally underperformed profits. When the surge in growth ended
abruptly in 2008, equity prices fell in line with the actual and expected decline in
profits. Expensive valuations therefore caused equity returns to underperform profits
following the 2001 slowdown and then did the same during the ensuing boom, while
finally failing to provide a cushion when the business cycle turned down. Over theentire period, equities behaved like an expensive and eternally out-of-the-money call
option on corporate earnings.
Put bluntly, the past decade has provided investors with an object lesson on the critical
importance of long-term valuation metrics. In Figure 2 we display a history of the two
most important of these measures for equities. The right hand panel presents the
history of Tobins Q the market value of equity/corporate net worth. The series was
constructed using data generated by Stephen Wright of the School of Economics at
Birkbeck College. The left hand panel displays a trailing real PE ratio, using a 10-year
moving average of earnings, drawn from data compiled originally by Robert Shiller.
Figure 2: Trailing real PE ratio, real, based on 10-year average earnings, Tobins Q ratio
(corporate equity market value/net worth at replacement cost), 1900-2009
0
5
10
15
20
25
30
35
40
45
50
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
real PE
average
plus 1 standard deviation
minus 1 standard deviation
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
Tobin's q, Total Assets
average
plis 1 s tandard deviation
minus 1 s tandard deviation
Source: Shiller, Wright, Federal Reserve, EcoWin
The charts should serve to illustrate the extraordinary overvaluation of equity markets in
the second half of the 1990s. From the end of 1996 onwards, the US was consistently
valued at well over 1 standard deviation expensive to the long run average of both these
yardsticks. The overvaluation was the most extreme of the past century and indeed of
recorded stock market history. Perhaps unsurprisingly, subsequent returns from buyingequities at such prices were poor, despite the 2003-08 period recording the strongest and
most synchronised phase of global economic growth since the 1960s.
In essence, investors were paying too much to access corporate earnings and corporate
assets during the stock market bubble years. S&P 500 operating earnings per share rose
more than 50% between the end of 1998 and the end of 2008. European profits rose by
considerably more. Had equity prices kept pace with earnings, we estimate that the
annualised US equity real return would have been in the region of 3-4%, not minus 2.7%.
A 3-4% real return is certainly well below the 6.4% long run average for US equities and it
would have been below the 5.7% real return from government bonds, but at least it
would have beaten cash, which delivered an annual real 0.7% over the period in question.
As it was, equities were the single worst-performing asset class during the 1998-2008
decade for the sole reason that they were the most overvalued asset class at the start of
that period.
8/8/2019 2009 Equity Gilt Study
8/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 7
The rather brutal lesson we can glean from the past 10 years is that valuations, rather
than macroeconomic conditions, and the progress of corporate profits, are the core
determinant of equity market returns. The investment industry as a whole devotes
enormous resources to the analysis of each quarters corporate profits and considerable
effort is expended on forecasting the economic cycle. Yet very little attention is paid to
aggregate valuation. Unfortunately, this balance of attention is flawed. Over the long
run, equity valuations appear to be the primary driver of equity returns, with economicconditions and profit trends contributing little, if anything, to the overall total return
from an investment in equities. Profits and growth explain the more minor fluctuations
of equity prices from quarter to quarter and year to year, but they are incapable of
explaining multi-year returns. Admittedly the converse also applies, in the sense that
valuations tend not to be able to explain shorter-run fluctuations in the stock market.
At time scales of much under five years, valuation becomes less relevant, with the
economic and profit cycle becoming the key explanatory variables. However, since
equities are typically held for the long run and are predominately owned by institutions
or individuals with long liability structures, it would seem reasonable to suggest that
more attention should be paid to the predictive capability of valuations. In short, if the
equity premium is forecastable, it makes sense for us to avail ourselves of the forecast.
To illustrate this point, consider the two valuation metrics mentioned earlier Tobins
Q ratio and Shillers PE ratio based on 10-year real earnings. Both these metrics have a
solid empirical record of successfully forecasting equity returns. In Figure 3 we illustrate
this relationship. The charts compare rolling 10-year annualised nominal returns from
US equities to the Q ratio and to the real PE ratio at the start of each of the 10-year
periods. The returns are plotted annually and the sample period begins in 1925. The
charts demonstrate the strong negative correlation between these two valuation
metrics and subsequent returns, showing how expensive valuations are associated with
low returns and vice versa.
Figure 3: Correlation, 1935-2008, 10-year rolling annualised returns from US equities, real PE
ratio and Tobins Q ratio at the start of each 10-year period
Q ratio
R2 = 0.6743
-10
-5
0
5
10
15
20
25
0 0.5 1 1.5 2
Q ratio
10
ye
arreturn
(annualised)
Shiller trailing PE ratio and 10 year returns
R2 = 0.6249
-5
0
5
10
15
20
25
0 10 20 30 40 50
Source: Shiller, Wright, Barclays Capital
The relationships are sufficiently robust to allow the possibility of forecasting. Figure 4
illustrates the results of a regression exercise, where the Q ratio and the real PE ratio
are used independently to forecast 10-year equity returns. Both variables produce very
similar forecasts. In Figure 5 we combine both metrics in a single model and also
perform an out-of-sample test on the methodology, stopping the regression in 1990 to
allow the model to forecast thereafter. This test shows both that the relationships are
reasonably stable over time, and that the model was effective in forecasting the general
trend in equity returns after 1990. In particular, the model correctly forecast that equity
8/8/2019 2009 Equity Gilt Study
9/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
8 Global Asset Allocation Strategy Barclays Capital
returns would be negative over the past 10 years. The model failed to predict the 1997-
2002 stock market bubble and therefore under-estimated returns in the periods ending
in those years. However, any gains that an investor could have made from riding the
bubble would have been temporary and reliant on an exceptional and perhaps
improbable ability to time the market.
Figure 4: 10-year rolling annualised nominal returns from US equities, 1935-2008, actual and individually regressed from Q ratio and real PE ratio
-10
-5
0
5
10
15
20
25
1935 1945 1955 1965 1975 1985 1995 2005
return over 10years from buying stocks in this year
forecast from Shiller real PE
forecast from Q ratio
Source: Shiller, Wright, Barclays Capital
Figure 5: 10-year rolling annualised nominal returns from US equities,
1935-2008, actual and regressed from combined Q ratio and real PE
ratio, with out-of-sample test from 1990
-10
-5
0
5
10
15
20
25
1935 1945 1955 1965 1975 1985 1995 2005
return over 10years from buying stocks in this yearforecastreturn over 10years from buying stocks in this year
out of sample from 1990 onwards
Source: Shiller, Wright, Barclays Capital
Currently, the model suggests that a purchase of US equities at the close of 2008 will
deliver a nominal annualised return of between 12.4% and 13.4% over the next 10
years. This forecast is corroborated by a similar exercise with another long-standing
valuation yardstick, the price/dividend ratio. We find this metric less efficient in
forecasting equity returns, but nonetheless it produces statistically significant results, as
illustrated in Figure 6. The price/dividend ratio, which was similarly over-valued at the
1998-2001 peak, suggests the future 10-year return (end-2008 to 2018) will be an
annualised 11.2%. Averaging these various results produces a forecast for the future
10-year nominal equity risk premium of 12.3%.
8/8/2019 2009 Equity Gilt Study
10/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 9
Figure 6: US price/dividend ratio and 10yr rolling nominal returns, correlation and regression model
R2 = 0.4212
-10
-5
0
5
10
15
20
25
0 20 40 60 80 100
price/dividend ratio
nominal10yearrollingreturn
-10
-5
0
5
10
15
20
25
1935 1945 1955 1965 1975 1985 1995 2005
return over 10years from buying stocks in this year
forecast from price / dividend ratio
Source: Barclays Capital
If the methodologies described above accurately forecast the past decade of poor
equity returns, the question arises as to why these signals were broadly ignored byinvestors. It was certainly not due to a lack of information. By way of illustration, Robert
Shillers book Irrational Exuberance, which espoused the 10-year real PE ratio as a core
yardstick for prospective equity returns, was published in 2000. Similarly, work by
independent analysts such as Andrew Smithers highlighted the relative expense of
equities signalled by Tobins Q ratio as early as 1998. If the warning signals were
available, they were not generally acknowledged by market participants. To be sure, the
equity bubble of 1997-2001 was widely seen as such. At the time, proponents of a New
Era in valuations were at least partly counterbalanced by a vociferous minority, who
accurately defined the trend as an unsustainable boom. However, from 2004 onwards,
the strong growth in profits generated by a buoyant global economy tended to obscure
the point that equity returns were continuing to be dampened by a persistent trendtowards lower valuations. Coincident returns were certainly strong, partially reflecting
the strength in profits; however, once profits turned down, equity prices fell back in
lock-step with the drop in earnings. Stock markets therefore underperformed earnings
during the expansion, but performed in line with earnings during the contraction. At
the start of 2009, equity prices are slightly lower than they were at the end of 1998,
even though prospective profits for the impending year are likely to be considerably
higher than they were in 1998, a deep global recession notwithstanding.
The fluctuation in equity valuations over the past decade demands a closer
consideration. While it is easy in retrospect to ascribe weak returns to overvaluation,
such an explanation does not tell us why the overvaluation occurred in the first place.To write the decade off as an epic example of the madness of crowds would seem to be
too glib an analysis. Indeed, our own research has led us to the conclusion that simple
irrationality may have played a much smaller role in moulding recent stock market
returns than is popularly imagined. Rather, it seems that investors and the markets
were in the grip of powerful forces that were hard for any individual to withstand. We
can identify two particular and perhaps related fundamental trends at work.
First, it is reasonable to suppose that the original decline in the forward-looking equity
risk premium reflected the broad decline in macroeconomic volatility that occurred from
the mid-1980s onwards. As the peaks and troughs of the business cycle grew less
pronounced during the period generally termed the Great Moderation, so the volatility
of profits and the intrinsic riskiness of corporate liabilities also declined. The fall in ex-ante
equity risk premia, personified by the rise in PE ratios and the rise in equity market
capitalisation relative to corporate net worth, can therefore be seen as a straightforward
8/8/2019 2009 Equity Gilt Study
11/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
10 Global Asset Allocation Strategy Barclays Capital
extrapolation of prevailing conditions on the part of investors. As the collective memory
of the 1970s faded, so the extrapolative process became more firmly embedded in market
psychology. The limited macroeconomic volatility of the 1980s and 1990s came to be
seen as the norm, while the 1970s were held to be an aberration attributable to poor
economic structures such as over-mighty trade unions and poor economic
management such as excessively lax monetary policy. On the basis that these mistakes
were recognised and therefore unlikely to be repeated, market participantsunderstandably expected the decline in macroeconomic volatility to persist.
We believe that much of the move higher in equity valuations and indeed the
accompanying accumulation of leverage in the household, financial and corporate
sectors can be explained, if not justified, on this basis. Unfortunately, the extrapolative
process generated its own downfall. The increase in equity valuations simultaneously
increased the probability of poor future returns. In the same vein, the increase in
general leverage on the basis of low macroeconomic volatility raised the sensitivity of
the economic and financial systems to small changes in fundamentals. By 2007, a very
modest tightening of monetary policy by the standards of past business cycles was
sufficient to trigger a collapse in the over-extended US housing market, thereby
tripping the global economy into the worst recession of the past 50, if not 70, years. In
essence, the Great Moderation was inherently unstable and prone to self-dissolution
because people recognised its existence and adjusted their behaviour accordingly. The
eventual denoument was as sure and inevitable as the plot of any dramatic tragedy.
Indeed, since the process was generated by the actions of human beings, it is perhaps
unsurprising to find that the terms of analysis of tragic literature hubris, harmatia,
pathos and (it is to be hoped) catharsis can translate so easily into the economic field.
To frame the discussion in somewhat more quantitative terms, we can illustrate the
connection between equity valuation and economic volatility. A popular standard
explanation for shifts in equity valuation highlights the empirical inverse correlation
between PE ratios and inflation. While this explanation is observationally correct, it is
intellectually unsatisfying. This is because it fails to explain why the valuation attached
to an asset that correlates positively to inflation the stream of corporate earnings
should exhibit a negative correlation in practice. If we instead regard the link between
PE ratios and inflation as symptomatic of a deeper correlation between inflation and
economic volatility, the pieces of the jigsaw fall into place. An accelerating inflation rate
is an inherently unstable process because it is exponentially self-feeding. A persistent
rise in inflation therefore raises current and prospective macroeconomic volatility.
Under such conditions, the desire for a higher ex-ante equity risk premium is logical. As
rising inflation raises the riskiness of the economic cycle, so investors demand a greater
risk premium to compensate for an increase in the dispersion of future outcomes. In
Figure 7 we illustrate this effect at work. The graph compares a moving average of the
quarterly volatility of both real GDP growth and inflation to the US trailing earnings
yield. Self-evidently, an increase in this measure of volatility generates an increase in
the earnings yield and vice versa.
8/8/2019 2009 Equity Gilt Study
12/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 11
Figure 7: Five-year moving average of the sum of quarterly volatility in real GDP growth and
inflation, S&P 500 trailing earnings yield
0
1
2
3
4
5
6
7
8
Q2 50 Q2 57 Q2 64 Q2 71 Q2 78 Q2 85 Q2 92 Q2 99 Q2 06
2
4
6
8
10
12
14
16average volatility of GDP and inflation
earnings yield S&P 500R2 = 0.4346
0
5
10
15
20
25
30
35
0 2 4 6 8
minus 66% correlation
Source: Haver
The decline in equity earnings yields during the 1990s can therefore, be seen as a reactionon the part of investors to the decline in macroeconomic volatility. The more recent rise
in earnings yields is similarly reflective of an increase in the volatility of both inflation and
growth during the current business cycle. The relationship between coincident economic
volatility and financial asset risk premiums is a simple reflection of the extrapolative
process by which we create a model of the future based on the recent past. Such
mechanisms served us well when avoiding the multiple external threats of the African
savannah. They are perhaps less useful when our own worst enemy is ourselves.
There is, however, an additional and perhaps more inexorable explanation for
fluctuations in equity valuations. In past editions of the Equity Gilt Study we have
highlighted the link between trends in financial asset yields and trends in demographics.
In particular, fluctuations in the population cohorts of savers and the retired correlate
strongly with bond yields and earnings yields. Thus when we observe the long run
rhythm of financial asset yields, the ratio of the 35-55 year old population to the total
population correlates negatively, while the growth rate of the newly retired population,
which we define as 65-75 year olds, correlates positively. Indeed regression models in
which the sole variables are these demographic components appear to explain long
trends in stock and bond yields quite well. The same variables also explain changes in
the Q ratio over time. Figure 8 and Figure 9 illustrate these points.
Figure 8: US and UK long-dated government bond yields, actual and modelled from demographics
0
2
4
6
8
10
12
14
16
18
1922 1942 1962 1982 2002
long Gilt yie ld
model
0
2
4
6
8
10
12
14
16
1926 1946 1966 1986 2006 2026
long UST yield
Model
Source: Barclays Capital
8/8/2019 2009 Equity Gilt Study
13/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
12 Global Asset Allocation Strategy Barclays Capital
Figure 9: US trailing PE ratio and US Q-ratio, actual and modelled from demographics
0
5
10
15
20
25
30
35
1950 1960 1970 1980 1990 2000 2010
PE ratio
Modelled from
demographics
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1950 1960 1970 1980 1990 2000 2010
Q ratio
Q ratio modelled from
demographics
Source: Wright, Barclays Capital
Unusually, the relationship between demographics and bond and stock yields conforms
to both common sense and economic theory. A population in which the high savingsage cohort is dominant will be characterised by a strong demand for financial assets.
This demand will be reflected in higher-than-otherwise financial asset prices and lower-
than-otherwise yields. Conversely, since the retired will typically be sellers of financial
assets, a society in which the retired population is large or growing rapidly is likely to be
characterised by a weaker demand for financial assets and hence higher-than-otherwise
yields. The demonstrable correlation of financial asset yields with demographic trends
supports the notion of the Lifecycle Theory of Savings writ large across the economy.
Speculatively, we might add the following rider. Bearing in mind the clear connection
between inflation and financial asset yields, it is possible to infer a relationship between
demographics and macroeconomic volatility. Certainly there is some logic to a link
between the worker-dependent ratio and the propensity for inflation. After all, if
globalisation is held to have been a restraining force on inflation over the past 20 years
due to the expansion of the global labour force, it is reasonable to propose a similar
effect from a natural expansion of a workforce due to demographic trends. Without
wishing to belabour this point, we can surmise that growth in the baby-boomer
working age population over the past three decades may well have been one of the
factors keeping inflation in check. Similarly, as the baby-boom generation ages into
retirement over the next decade, it is plausible to believe that the wage bargaining
power of the remaining labour force will rise and that inflationary shocks might carry a
greater risk of persistence.
As far as financial assets are concerned, the growing demographic dominance of thehigh savings age cohort helps explain the rise in equity valuations during the 1990s. As
the boomer generation entered their peak savings years, the competition for financial
assets drove prices ever higher. It is particularly noteworthy that the US equity bubble
peaked in the same year that the US high savers cohort peaked as a ratio to the general
population. A similar timing was visible in early 1990s Japan. Meanwhile the subsequent
move lower in valuations can be explained both by the pick-up in the growth of retirees
and the decrease in the high savers-total population ratio.
For bond markets, demographic models expected a phase of very low nominal yields
from the late 1990s onwards, but are now beginning to point towards a reversion to
higher yields. Roughly, the demographic models expected bond yields to trough arounda decade after equity yields. Such a time lag makes sense, since the demand for equities
would be hit first by a fall in the ratio of high savers to the general population, whereas
8/8/2019 2009 Equity Gilt Study
14/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 13
the demand for bonds would be sustained or even buoyed as retirees shift into
income-bearing assets. At a later point, the social security fiscal strains of an ageing
population, along with the impact of a shrinking workforce on inflation and tax
receipts, might be expected to be factors pushing bond yields higher. Looking forward,
demographic trends would seem to point to an era of low equity valuations,
accompanied by rising bond yields.
The apparent end of the Great Moderation in economic volatility conveys a similar
message, as the world struggles to adapt to the emergence of giant developing
economies. As Figure 10 and Figure 11 should hopefully illustrate, the relationship
between global growth and raw material inflation shifted unfavourably during the
current cycle. The primary cause was a severe demand shock, driven by the growing per
capita resource appetites of the large developing economies and the greater raw
material intensity of growth in industrialising economies. It was also clearly the case
that a constrained response from the supply side failed to accommodate the leap in
commodity demand. The weakness in the supply response is attributable to multiple
causes, ranging from endemic scarcity, through environmental considerations, to
under-investment. Our analysis of this topic is available in the past two years editions
of the Equity Gilt Study. Please see Chapter 1 of theEquity Gilt Study 2008and Chapter
1 of theEquity Gilt Study 2007
Figure 10: A shifting relationship between growth and inflation trends in
the correlation between global GDP growth and commodity price inflation
-20
-15
-10
-5
0
5
10
15
20
25
-2 -1 0 1 2 3 4 5 6 7
Global GDP y/y
CRBy
/y
1970-1979
2001-2008
1980-2001
Source: Haver
http://ecommerce.barcap.com/research/user/article/attachment/204768/0http://ecommerce.barcap.com/research/user/article/attachment/178471/0http://ecommerce.barcap.com/research/user/article/attachment/178471/0http://ecommerce.barcap.com/research/user/article/attachment/204768/08/8/2019 2009 Equity Gilt Study
15/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
14 Global Asset Allocation Strategy Barclays Capital
Figure 11: A shifting relationship between growth and inflation rolling
10-year coefficient between global manufacturing confidence and
commodity prices
0
5
10
15
20
25
30
35
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Coefficient between GSCI Commodity
index and Global PMI
Source: Haver
In the most recent business cycle, the net boost to global inflation, prompted by the
increased resource intensity of global growth, was sufficient to raise interest rates to
levels that catalysed the asset price deflation and de-leveraging trends that are visible at
present. An economic system whose levels of leverage and asset prices were predicated
on endlessly low inflation proved to be unsustainable at the first whiff of higher inflation.
In the short term, it is reasonable to believe that the de-leveraging of the private sectors
in a number of over-extended economies will keep global demand and hence inflation
weak. Over the longer run, it is difficult to evade the impression that this effect will fade
and that the problem of accommodating the resource appetites of the developing world
will re-emerge. As a consequence, the balance of probability seems tilted towards the
persistence of high in comparison to the last three decades macroeconomic volatility.
Overall, both demographic and economic factors suggest that equity valuations may
fall a little further and remain low for a while, before recovering later in the decade.
Both factors also suggest that bond yields are likely to trend higher at the same time.
However, of the two asset classes, we expect that equities are likely to reverse their long
phase of underperformance against bonds. As far as bonds are concerned, rising yields
will self-evidently damage returns. In contrast, as we have seen, lowly equity valuations
tend to confer higher-than-average future long-run returns. This is both because a
performance-damaging decline in valuations becomes less likely and because aneventual performance-enhancing rise in valuations becomes more likely, at low levels of
valuation. Or, to put it rather more simply, equities are likely to outperform bonds over
the next decade because equity yields are already high, whereas government bond
yields have yet to rise. To summarise, we are in an environment in which forward-
looking measures of equity risk premia should be high, compensating for a more risky
macroeconomic environment and a reversal of the demographic forces that have
supported asset prices in the recent past. If history is any guide, such a period should
present long-term investors with an opportunity to gain cheap access to corporate
profits and net worth.
8/8/2019 2009 Equity Gilt Study
16/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 15
Chapter 2 Deflated marketsSreekala Kochugovindan
Financial markets have endured one of the most turbulent episodes in history. At the time
of writing, the peak to trough fall in the S&P500 since the start of the credit crunch stands
at 52%. A collapse of that magnitude surpasses the bursting of the dotcom bubble, (adrop of -49%), the crash of 1987 (-33.7%) and the OPEC crisis of 1973 (-48%). We have
to go as far back as the 1930s to find a more acute equity sell off. Unsurprisingly, a
fascination with the depression era has developed and concerns over deflation have
replaced the inflation scare which prevailed in the first half of 2008. This raises the
question over which state presents the greater evil, deflation or high inflation? In this
article, we compare the performance of a range of assets and equity sectors across
different inflation regimes since the 1920s. We find that over the past six years, the
distribution of equity returns have switched from emulating one extreme episode of
history to another, from stagflation to deflation, in under a decade. We differentiate
between phases of good and bad deflation to gain further insight and find that, in fact,
credit conditions may be a precondition for bad deflation and, in turn, a more importantfactor to consider in determining trends in equities. Perhaps the focus on deflation or
stagflation has been a diversion in comparison to the importance of credit regimes.
Inflation extremes
We begin by examining the average real returns of US equities, bonds and cash during
the three inflation phases since 1929. High inflation is classified as inflation greater than
the long run average of 4%, the stable years include inflation between 0% and 4% and,
finally, deflation where the annual inflation rate is negative.
Historically, the US stock market has produced the best returns during periods of low and
stable inflation with an average real return of 11%. Although equities are considered to
provide a hedge against inflation in the long run, the short-term performance can be
quite poor. Equities can be susceptible to sharp declines in the face of unexpected
inflation spikes. For example, US equities slumped over 50% during 1973 and 1974 before
rallying 30% the following year. Figure 12 shows that equities produced a small positive
average real return during the high inflation years; however, examining the sector level
data over these years, we find that these returns are skewed toward the commodity-
related companies. The mining sector produced the best average returns of 6% pa.
During the deflationary years stocks provided the worst performance, with dire returns
across all the sectors. Instead, returns are concentrated in bonds and cash as both
traditionally perform well during periods of risk aversion. Credit spreads, on the other
hand, widen quite dramatically during deflationary episodes. The table below highlights
the annual change in the spread between long-run investment grade credit and 30-year
Treasuries. The deflationary years show an average spread change of 100bps since 1929.
This is in sharp contrast to the other two inflation regimes. During the low and stable
inflation years, spreads barely move with an average spread change of half a basis point.
The high inflation periods suggest that credit spreads only widen on average by 8bps.
Thus, in extreme inflation conditions, whether it happens to be deflation or high
inflation, portfolio diversification does not seem to be the best approach given that
returns are so heavily concentrated in either resource-based stocks in the case ofinflation, or government bonds in the case of deflation.
8/8/2019 2009 Equity Gilt Study
17/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
16 Global Asset Allocation Strategy Barclays Capital
Figure 12: Average real annual total returns
Deflation High inflation The stable years
US Equities -15% 0.34% 11%
20 year bonds 4% -3% 6%
Cash 5% -1% 1%
Change in credit spread (bps) 99.7 7.8 0.5
US Equity Performance
Deflation High inflation The stable years
Mining -17% 6% 6%
Manufacturing -11% -1% 10%
Transp -14% -1% 10%
Retail -11% -3% 14%
Finance -20% 0% 11%
Services -20% 2% 14%
Deflation High inflation The stable years
US Equities -15% 0.34% 11%
20 year bonds 4% -3% 6%
Cash 5% -1% 1%
Change in credit spread (bps) 99.7 7.8 0.5
US Equity Performance
Deflation High inflation The stable years
Mining -17% 6% 6%
Manufacturing -11% -1% 10%
Transp -14% -1% 10%
Retail -11% -3% 14%
Finance -20% 0% 11%
Services -20% 2% 14%
Source: CRSP, Ecowin, Barclays Capital
To put the more recent experience in context, we compare sector behaviour during the
past five years with the sector behaviour during the deflationary episode in the 1930s
and the stagflationary 1970s. We separate the equity returns over the past six years
into two phases: 2003 to July 2007 to capture the period of global growth and booming
equity and commodity markets, which originally led to the inflation scare, and the
second phase covering the credit crisis. The charts highlight a remarkable similarity
between the sector returns of the 1970s and between 2003 and 2007. In both cases, the
commodity-driven inflation spike led to portfolio returns being highly concentrated in
commodity-related equity sectors. There also appears to be some similarity between
sector returns during the great depression and the credit crunch. Financials were the
worst performing in both cases. Although we do not believe that the current crisis will
follow the path of the Great Depression the substantial global fiscal and monetary
stimulus currently employed is likely to prevent such a scenario it is, however, very
interesting to note that over the past decade, equity returns have switched fromemulating one extreme episode of history to the exact opposite extreme.
Figure 13: Nominal sector returns then and now
0.0 0.5 1.0 1.5 2.0 2.5
Retail
Services
Finance
Manufacturing
Transport
Mining
1970s, 100%
increase in CPI
-0.95 -0.9 -0.85 -0.8 -0.75
Finance
Services
Manufacturing
Retail
Mining
Transport 1929-1932, CPI
falls 21%
0.0 0.5 1.0 1.5 2.0 2.5
Retail
Services
Finance
Manufacturing
Transport
Mining
2003 to July 2007,
13% rise in CPI
-0.95 -0.75 -0.55 -0.35 -0.15 0.05
Finance
Services
Mining
Manufacturing
Transport
Retail The current crisis
July 2007 to Dec
2008
Source: CRSP, Barclays Capital
8/8/2019 2009 Equity Gilt Study
18/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 17
Dissecting deflation
The three inflation categories described in Figure 12 could be considered too simplistic.
Bordo and Filardo (2005)1 distinguish between good and bad deflation. They examine a
number of examples dating back to 1800 in order to identify the main features and
determinants of each phase of deflation. Historically, good deflation tends to be a
characterised by a mild decline in the price level accompanied by economic growth.
One example given by Bordo et al is the 1921-29 period where many countries
experienced robust growth as a result of the post-war recovery process, the growth of
consumerism and new high-tech industries, such as cars and radios. The bad deflation
promptly followed during 1929-33. Sharp price declines were accompanied by
dramatic falls in real output across many countries. In the US inflation fell over 20%
during those three years while real output fell by 7.6%.2
The chart below plots the performance of equity sectors during the good and the bad
deflationary episodes and highlights how the simple classification of deflation in Figure
12 masks some dramatic differences in the distribution of returns. Between January
1926 and August 1929, US CPI declined by a cumulative 3.4%. At the same time USequity sectors rallied between 70% (Retail sector) and 164% (Manufacturing). During
the bad deflation years CPI fell by a cumulative 30% while equities fell between 66%
(Manufacturing) and 90% (services).
Figure 14: Equity sector performance during the good and the bad
deflation years
0
50
100
150
200
250
300
Jan 1926 Jan 1928 Jan 1930 Jan 1932 Jan 1934 Jan 1936
70
75
80
85
90
95
100
105
Mining
0
50
100
150
200
250
300
Jan 1926 Jan 1928 Jan 1930 Jan 1932 Jan 1934 Jan 1936
70
75
80
85
90
95
100
105
MiningManufacturing
TranspRetailFinanceServicesCPI (RHS)
Good deflation Bad deflation Recovery process
Source: CRSP, Ecowin, Barclays Capital
The volatility of returns during the phases of deflation also differs quite dramatically.
Figure 15 plots the volatility of daily sector returns during the good and bad deflation
episodes alongside the stagflationary decade of the 1970s. The 1929-33 bad deflation
period shows that returns are consistently more volatile than in either of the other two
phases. During 1929-33 bad deflation period, volatility nearly doubled for the best-
performing sectors of the 1926-29 period, manufacturing and transport. These same
sectors were 2.5x and 3.5x more volatile during the 1930s than in the stagflationary
1970s. Part of this difference may be due to the fact that economic volatility was so
much greater during the 1930s. The rolling 10-year standard deviation of annual GDP
1 Bordo M, and Filardo, A (2005) Deflation in a historical perspective BIS Working Papers No 1862 Source: Bordo et al. and Ecowin
8/8/2019 2009 Equity Gilt Study
19/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
18 Global Asset Allocation Strategy Barclays Capital
growth was in the region of 9% in 1939, by the 1970s the volatility of growth had fallen
to 2.5%. The volatility of inflation during the 1930s was also higher, standing at 5% in
comparison to just 3.5% during the 1970s.
Figure 15: The good and bad volatility
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
Mining
Manufacturing
Transp
Retail
F
inance
Services
1926 -1929
1929-1933
1970s
Source: CRSP, Barclays Capital
This leads us to reassess the definition of good and bad deflation as outlined above, which
was primarily dependent upon the economic growth rate. A number of the episodes of
good deflation outlined in Bordo and Filardo (2005) were followed by a shock that led to
a banking crisis which in turn, was followed by a phase of bad deflation. Figure 16 returns
to phase of good and bad deflation and compares equity returns against the long run
credit spread as well as industrial production growth. Although the US experienced strong
economic growth during the good deflation period, annual industrial production growthreached over 40% in 1922, the economic environment was turbulent throughout the
decade, with a short bout of negative growth in 1924. Regardless of these fluctuations in
growth, equities continued to rally against the backdrop of tight and stable credit spreads.
The bad deflation phase was accompanied by a dramatic widening in credit spreads as
well as a decline in growth. The 1933 recovery only began when spreads began to tighten
back to more stable levels.
8/8/2019 2009 Equity Gilt Study
20/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 19
Figure 16: Equity returns and the credit cycle
0
100
200
300
400
500
600
700
800
Jan 19 Jan 22 Jan 25 Jan 28 Jan 31 Jan 34 Jan 37 Jan 40
0
100
200
300
400
500
600
700
800
S&P500
Credit spread (RHS
inverted)
-40%
-20%
0%
20%
40%
60%
80%
Jan 19 Jan 22 Jan 25 Jan 28 Jan 31 Jan 34 Jan 37 Jan 40
Industrial production
0
100
200
300
400
500
600
700
800
Jan 19 Jan 22 Jan 25 Jan 28 Jan 31 Jan 34 Jan 37 Jan 40
0
100
200
300
400
500
600
700
800
S&P500
Credit spread (RHS
inverted)
-40%
-20%
0%
20%
40%
60%
80%
Jan 19 Jan 22 Jan 25 Jan 28 Jan 31 Jan 34 Jan 37 Jan 40
Industrial production
Source: Ecowin
A similar analysis for the 1970s shows that equity returns were again very closely
correlated with credit spreads. In 1974, spreads contracted and equities rallied six
months before industrial production had troughed, so it was not necessary for growth
to get underway for equity and credit markets to stage a recovery. The chart also
suggests that equities recovered one month before the credit market. There is no clear
statistical evidence to suggest whether equities lead or lag the credit market, although
there is some evidence to suggest that credit spreads may be a causal factor of extreme
inflation episodes. The granger causality test between equities and inflation on monthly
data since 1919 to 2008 suggests that statistically causality runs from credit to
inflation. Bordo and Filardo (2005) test the impact of banking crisis on the probability
of deflation being good, bad or ugly. They find that if a banking crisis occurs, theprobability of a good deflation phase taking place, drops considerably, while the
probability of bad deflation rises. This suggests that a financial shock which leads to
tightening credit conditions and a dramatic widening in spreads maybe more important
in driving the volatility of inflation and in turn the volatility in financial assets. Perhaps
the focus on deflation or stagflationary regimes is a diversion in comparison to the
importance of credit regimes.
8/8/2019 2009 Equity Gilt Study
21/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
20 Global Asset Allocation Strategy Barclays Capital
Figure 17: Equity returns and the credit cycle in the 1970s
10,000
12,000
14,000
16,000
18,000
20,000
22,000
24,000
26,000
28,000
Jan 69 Jan 71 Jan 73 Jan 75 Jan 77 Jan 79
0
50
100
150
200
250
300
350
S&P500
Credit spread (RHS
inverted)
-15%
-10%
-5%
0%
5%
10%
15%
Jan 69 Jan 71 Jan 73 Jan 75 Jan 77 Jan 79
Industrial Production
10,000
12,000
14,000
16,000
18,000
20,000
22,000
24,000
26,000
28,000
Jan 69 Jan 71 Jan 73 Jan 75 Jan 77 Jan 79
0
50
100
150
200
250
300
350
S&P500
Credit spread (RHS
inverted)
-15%
-10%
-5%
0%
5%
10%
15%
Jan 69 Jan 71 Jan 73 Jan 75 Jan 77 Jan 79
Industrial Production
Source: Ecowin
The credit impact
To ascertain the potential impact of credit spreads, we examine the average equity return
during four stages of the credit cycle: 1) low stable spreads; 2) widening credit spreads; 3)
high and stable spreads; and 4) tightening spreads. Figure 18 illustrates the results, the
chart on the left plots the average quarterly equity returns since 1919, the chart on the
right adds in the maximum and the minimum return achieved during each phase.
Figure 18: Average, maximum and minimum quarterly equity returns across the credit cycle
8/8/2019 2009 Equity Gilt Study
22/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 21
On average, periods of low and stable credit spreads produce small yet positive returns,
with a relatively low dispersion, a range of +/-20% per quarter. Unsurprisingly, the worst
returns are posted during the spread-widening phase. The average return falls to -6%,
with a significant increase in the maximum downside to -40%. Wide and stable spreads,
our most likely scenario for the coming months, has a disappointing average return of
-4% per quarter; however, this regime has had greater upside potential than either low
and stable or rising spread regimes. The maximum historical return during this phasehas been 35%, with a maximum downside of -30%. The best returns occur during the
tightening phase, with an average quarterly return of 4.3%, and a maximum return as
high as 86%, which occurred during the sharp credit-tightening phase of the 1930s.
It seems that the similarities drawn between the current distribution and volatility of
equity returns with that of the 1930s are driven more by the credit cycle than any real
danger of a rerun of the Great Depression. Our analysis here suggests that allocating
with deflation in mind may be the wrong approach. A clearer understanding of the
credit cycle could be far more important in better understanding future trends in
financial assets.
8/8/2019 2009 Equity Gilt Study
23/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
22 Global Asset Allocation Strategy Barclays Capital
Chapter 3 Viral economicsTim Bond
Both the occurrence and the economic impact of the credit crunch caught policymakers,
regulators, bankers, analysts and investors largely unprepared. We examine why this
might be so, given the empirical evidence that credit cycles are inherently predictableand of considerable importance to the path of economic growth. Our conclusions
highlight the endemic instability of a pure market system.
The events of the past two years have served to underline the importance of the credit
cycle as an independent economic determinant. Too many traditional economic models
have tended to regard the credit cycle as identical to the interest rate cycle. Changes in
official interest rates occupy a core position in the formulation of economic forecasts,
whereas fluctuations in the price and availability of private credit attract less attention.
The comparatively minor importance attached to credit measures in economic
forecasting is hard to fathom. On some grounds, credit variables are better at
forecasting economic growth than are official interest rates. In particular, quantitative
measures of credit availability, such as that provided by the Federal Reserves Senior
Loan Officer Survey (SLO), appear to correlate more strongly with subsequent nominal
economic growth rates than do measures of official monetary policy settings. This
survey, which is unfortunately only available as a continuous series from 1990 onwards,
asks bankers a series of standardised questions about their lending habits on a quarterly
basis. The results are then reported in the form a diffusion index. Over the course of the
past 18 years, the Survey has provided an extremely accurate leading indicator for
major fluctuations in the US business cycle.
Figure 19 illustrates this point, comparing the Surveys questionnaire about whether
terms and conditions for Commercial and Industrial Loans are tightening or easing, with
the pace of nominal GDP growth. The highest correlation of -72% is achieved with theSLO survey leading growth by half a year. In Figure 20, we repeat the exercise for the real
Fed Funds rate, where the optimal correlation of -48% was achieved with the Funds rate
leading growth by two years. Fairly clearly, the Loan Officer Survey is a more effective
leading guide for the business cycle than the Fed Funds rate. We subjected the Loan
Officer survey to a Grainger causality test, to check that the correlation was not merely
picking up the pro-cyclical nature of bank lending. The test confirmed that the changes in
lending conditions did indeed stimulate changes in nominal economic activity.
Figure 19: Comparison of Senior Loan Officer Survey of terms and conditions for C&I loans
(lagged 2 quarters) against US nominal GDP growth
R2= 0.5155
-30
-20
-10
0
10
20
30
40
50
60
70
0 2 4 6 8
Nominal GDP % y/y
SLOs
urvey
0
1
2
3
4
5
6
7
8
Q4 90 Q4 93 Q4 96 Q4 99 Q4 02 Q4 05 Q4 08
-40
-20
0
20
40
60
80
1000
1
2
3
4
5
6
7
8
Q4 90 Q4 93 Q4 96 Q4 99 Q4 02 Q4 05 Q4 08
-40
-20
0
20
40
60
80
100
nominal GDP % yy
SLO survey: C&I terms and
conditions lagged 2 quarters
Source: Federal Reserve, Haver, BEA
8/8/2019 2009 Equity Gilt Study
24/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 23
Figure 20: Comparison of Real Fed Funds rate (lagged 8 quarters) against US nominal GDP growth
R2= 0.2338
-2
-1
0
1
2
3
4
5
6
0 2 4 6 8
Nominal GDP y/y
RealFu
ndsrate
R2= 0.2338
-2
-1
0
1
2
3
4
5
6
0 2 4 6 8
Nominal GDP y/y
RealFu
ndsrate
0
1
2
3
4
5
6
7
8
Q4 90 Q4 94 Q4 98 Q4 02 Q4 06
-2
-1
0
1
2
3
4
5
60
1
2
3
4
5
6
7
8
Q4 90 Q4 94 Q4 98 Q4 02 Q4 06
-2
-1
0
1
2
3
4
5
6
GDP nominal yy
Real Fed Funds rate lagged 8quarters
Source: Haver, BEA
Despite the empirical evidence, until very recently, credit variables have rarely been
allotted much weighting in economic forecasting models. The relatively greater
forecasting importance allotted to monetary policy settings rests on a belief that that
changes in actual and expected official interest rates are the primary determinant of
interest rates elsewhere in the economy. Yields for private sector credit are therefore
assumed to move in line with the interest rates controlled by monetary authorities.
On average, such a presumption is correct. For the sake of example, over the past 50 years,
long-dated US corporate bond yields display a correlation of 76% to short-dated
government bond yields. However, the average of any relationship can conceal plenty of
examples of very non-average behaviour. Figure 21 illustrates this point. The chart shows
the rolling 18-month correlation between annual changes in US corporate bond yields and
short-dated government yields. Typically, the correlation is strongly positive, but there are
sporadic outliers such as 1974/5 and 2007/8 when the correlation swings sharplynegative. There are also a couple of occasions 1994 and 1998 when the correlation fades
to zero. Significantly, each of these outliers is associated with a severe financial crisis, during
which the financial system which acts as an intermediary between the central bank and
the economy was subject to great stress. In 1974/5 and 2007/8 these financial crises
caused or exacerbated very deep recessions. In 1994 and 1998 the crises were more or less
confined to the financial markets government bonds in 1994 and credit in 1998.
Figure 21: Rolling 18-month correlation, annual change in US 3-year
note yields, annual change in long Baa corporate bond yields
-1.0
-0.5
0.0
0.5
1.0
1.5
Jan 70 Jan 74 Jan 78 Jan 82 Jan 86 Jan 90 Jan 94 Jan 98 Jan 02 Jan 06
rolling 18 month correlation, annual
change in 3 year gov yields, annual
change in US corporate yields
Source: Haver, Moodys
8/8/2019 2009 Equity Gilt Study
25/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
24 Global Asset Allocation Strategy Barclays Capital
The importance of credit variables during these outlier phases can be illustrated by
the current downturn. In Figure 22, we compare the actual path of US nominal GDP
growth during the current crisis, with those paths projected by regressions derived
from the SLO measure of credit conditions and the real Fed Funds rate. The forecast
generated by the lagged Fed Funds measure envisaged a gentle deceleration in
growth of a little over 1%, with no real risk of recession. The forecast generated by
the SLO survey, of a 4% deceleration, was much closer to the actual outcome,providing a solid warning of recession.
Figure 22: Actual US nominal GDP growth, 2006-08, compared to
forecasts generated by regressions against lagged SLO survey of credit
conditions and lagged real Fed Funds rate
1
2
3
4
5
6
7
8
Q1 06 Q2 06 Q3 06 Q4 06 Q1 07 Q2 07 Q3 07 Q4 07 Q1 08 Q2 08 Q3 08 Q4 08 Q1 09 Q2 09
Actual nominal GDP
Forecast from SLO
Forecast from Fed Funds
Source: Haver, Federal Reserve, BEA
Since the credit cycle follows its own path that is occasionally distinct from that
projected by official interest rates, it is useful to understand the dynamics that lie
behind fluctuations in the pricing and availability of credit. Focussing on the availability
of US bank credit measured by the aforementioned Senior Loan Officer survey, we
found two key variables that appear to explain the bulk of historic fluctuations in banks
appetite to lend: past monetary policy settings and the past appetite to borrow. To first
focus on the impact of monetary policy, we found that the real Fed Funds rate (deflated
by headline CPI) explained changes in C&I loan conditions with six quarter lag,.. This
result conforms to a common sense expectation that a tightening or easing of
monetary policy will eventually encourage or discourage banks in their lending policies.
The result also conforms to the findings displayed in Figure 19 and Figure 20, whichshowed that the strongest correlation between the real Fed Funds rate and nominal
GDP growth was found at an eight-quarter lag, whereas the strongest correlation
between C&I credit conditions and growth was found at a two-quarter lag.
The second variable that we found to be important was also intuitively obvious.
Broadly, the non-financial corporate borrowing appetite appears to be a strong
influence on future credit conditions for corporate borrowers. This is a wholly
reasonable relationship, as changes in the demand for credit should as in any market
for any good or service influence the eventual pricing of that item. In this instance, we
found that a measure of corporate credit demand that also contained a leading
measure of corporate creditworthiness outperformed simpler gauges, such the ratio ofnon-financial corporate borrowing to GDP.
8/8/2019 2009 Equity Gilt Study
26/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 25
To explain our methodology, we should note that the core corporate borrowing
requirement reflects the financing gap, which is the excess of capital investment over
internal cash flow, after deducting dividend and tax payments. However, in recent
years, the non-financial corporate sector has spent considerable sums over $2 trillion
since the end of 2001 on purchasing equities. Indeed, by the fourth quarter of 2007,
companies were spending more on equity purchases than on capital investment. A
significant portion in the variation in the corporate sectors appetite to borrow has atleast in recent history been driven by changes in the sectors appetite to purchase
equity. The same was true of the previous expansion, from the mid-1990s through to
the end of the century (see Figure 23). As the right-hand panel of Figure 23 should
illustrate, during the most recent business cycle expansion, a much larger portion of the
overall increase in corporate borrowing was attributable to equity purchases than
capital expenditure. The boom in business borrowing was almost exclusively driven by a
leveraging of the corporate capital structure by private equity firms and other players.
Figure 23: Change in the quarterly annualised spending on fixed investment and equity purchases
by the US non-financial corporate sector 1995-2000 and 2002-08
-200
-100
0
100
200
300
400
500
Q1 95 Q1 96 Q1 97 Q1 98 Q1 99 Q1 00
Change from Q1 95,
annualised $ bill, non-
corporate capex
Change from Q1 95, $bill
annualised, non-fin
corporate equity purchases
-200
0
200
400
600
800
1,000
1,200
Q1 02 Q1 03 Q1 04 Q1 05 Q1 06 Q1 07 Q1 08
Change from 2001, $ annualised
$ bill, non-fin corporate purchases
of equities
Change from 2001, annualised $
bill, non-fin corporate capex
Source: Haver, Flow of Funds
The previous two business cycles, therefore, have been characterised by sizeable
substitutions of corporate debt for corporate equity in the liability structure of the
sector. The trend has been visible in the surge in M&A volumes, the prevalence of
leveraged buyouts generated by the private equity business and in the use of share
buybacks to ostensibly return profits to shareholders. The leveraging of the corporate
sector has a variety of underlying causes, including an agency problem with
management incentive structures, pension fund disenchantment with quoted equity
returns after the 2002-03 bear market, a confusion of the financial results attributable
to leverage and attributable to better management and last, but not least, sheer bullish
sentiment on the part of many CEOs. Regardless of the multiplicity of causes, the
macroeconomic impact is clear. The substitution of debt for equity inevitably weakens
the creditworthiness of the corporate sector. Despite a popular perception that the
non-financial corporate sector entered the current recession in good shape, the facts
of the matter are that the recent boom in debt-equity substitution has left the
corporate sector in its worst shape from a credit perspective of the entire post-war
period. As Figure 24 should illustrate, the debt/profit ratio for the US non-financial
corporate sector rose to 7 during the recent expansion, an all-time high. Thus, at the
peak of the profit cycle, each 1% change in interest rates on this $7 trillion stock of debt
moves profits by 7%.
8/8/2019 2009 Equity Gilt Study
27/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
26 Global Asset Allocation Strategy Barclays Capital
Figure 24: US non-financial corporate debt/non-financial corporate
internal funds*
0
1
2
3
4
5
6
7
8
Q1 52 Q1 56 Q1 50 Q1 64 Q1 68 Q1 72 Q1 76 Q1 80 Q1 84 Q1 88 Q1 92 Q1 96 Q1 00 Q1 04 Q1 08
Non-financial corporate debt/internal funds
Note: *Profits after tax, dividends, depreciation, inventory valuation. Source: Federal Reserve, Haver
To return to the main point, debt for equity substitution increases the appetite to
borrow and decreases the creditworthiness of the borrower. From both perspectives,
the trend will therefore be of relevance to the banking sectors willingness to lend to
businesses. As such, the measure of the corporate financing gap that also includes the
trend in corporate equity net purchases should be effective in explaining changes in C&I
loan conditions. If we express this measure as a ratio to internal corporate sector cash
flow, we are presented with a calculation that captures the corporate borrowing
requirement. The measure also serves as a gauge of creditworthiness, incorporating
information on the extent of equity retirement and on the changes in borrowing
relative to the changes in profit growth. This measure is more effective in explaining
variations in C&I loan conditions than a straightforward financing gap, leading by six
quarters at an 82% correlation, compared to four quarters at a 67% correlation. The
measure also outperforms a simple ratio of corporate borrowing to cash flow.
Figure 25: Senior Loan Officer Survey, net conditions C&I lending, non-
financial corporate financing gap plus corporate equity
purchases/corporate internal cash flow, both lagged six quarters
-40
-20
0
20
40
60
80
100
Q2 90 Q2 92 Q2 94 Q2 96 Q2 98 Q2 00 Q2 02 Q2 04 Q2 06 Q2 08
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2SLO survey of C&I loan conditions
Non-fin corporate financing gap, including net equity
purchases/internal funds lagged 6
Source: Haver, Federal Reserve, Barclays Capital
8/8/2019 2009 Equity Gilt Study
28/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 27
If the real Funds rate and the aforementioned measure of the corporate borrowing
requirement are combined as twin variables in a regression model, we have a reasonably
robust framework for explaining and indeed predicting fluctuations in the credit cycle.
The model is illustrated in Figure 26, below. The panel on the right displays the history of
the model in entirety, while the panel on the left displays a snapshot of the models
predictions as of Q1 of 2007. As this panel demonstrates, the model proved effective in
forecasting the subsequent general tightening in credit conditions, a signal that allowedthe Barclays Asset Allocation team to turn bearish on the credit asset class early in 2007.
On a different note, it is worth comparing Figure 25 and Figure 26. It should be clear that
the model that includes the Fed Funds rate as a variable is currently under forecasting the
actual tightness of credit conditions. Meanwhile, the financing gap alone is doing a better
job of prediction. The divergence offers an illustration of the extent to which past excessive
borrowing and leverage have made bank lending practices irresponsive to monetary
policy. It should also be noted that both versions of the model suggest that the natural
peak in tight lending condition is Q1 2009, conditions gradually easing thereafter.
Figure 26: Senior Loan Officer Survey, terms and conditions for commercial and industrial loans,
actual and modelled
-30
-20-10
0
10
20
30
40
50
60
70
Q2 90 Q2 93 Q2 96 Q2 99 Q2 02 Q2 05 Q1 08
SLO survey of C&I loan conditions
Model
-30
-20-10
0
10
20
30
40
50
60
70
Q2 90 Q2 93 Q2 96 Q2 99 Q2 02 Q2 05 Q1 08
SLO survey of C&I loan conditions
Model
-40
-20
0
20
40
60
80
100
Q2 90 Q2 93 Q2 96 Q2 99 Q2 02 Q2 05 Q2 08
SLO survey of C&I loan conditions
Model
Source: Haver, Federal Reserve, Barclays Capital
Hopefully we have demonstrated that fluctuations in the credit cycle are amenable to
forecasting. If we widen our analysis of the credit cycle to include default rates and
corporate bond spreads, we can show that they are similarly predictable on the basis of
some simple econometric models. To be sure, most such models require
macroeconomic inputs, such as GDP growth, unemployment and so on. However the
broad point is that credit pricing and credit fundamentals display reasonably reliableand stable responses to changes in underlying economic factors. Thus corporate default
rates rise after a phase in which corporate borrowing rises faster than corporate profits.
Rising household leverage, increases in debt-service burdens and increases in
unemployment will generate increases consumer loan defaults. Similarly, each
incremental rise in house price income ratios increases the probability of higher
mortgage defaults. These are logical and intuitive relationships whose existence can be
proven empirically. If we use these measures in tandem with standard forecasts for the
main macroeconomic variables, we can then derive a judgement about the
sustainability of particular trends. The following exhibits illustrate this point.
Figure 27 portrays a model for net loss rates on US commercial bank loan books. The losses
are expressed as an annual percentage of total loans. The explanatory variables include the
tweaked measure of the corporate financing gap described above, as well as corporate
profits, GDP growth, house prices, changes in unemployment and inflation. Although an
8/8/2019 2009 Equity Gilt Study
29/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
28 Global Asset Allocation Strategy Barclays Capital
accurate forecast by the model depends to an extent on accurate macroeconomic forecasts,
the model underlines the point that bank loan losses display predictable responses to
changes in economic conditions. In the right-hand panel ofFigure 26 we highlight the point
that the ratio of corporate borrowing to cash flow tends to lead bank loan losses by a year
and a half. Thus, even in the absence of any reliable economic forecasts, a reasonably well-
informed trend direction can be ascertained for bank loan losses.
Figure 27: Net loss, % of total, US commercial banks loans and leases, actual and modelled (left
hand panel) non-financial corporate borrowing % cash flows versus loss rate
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Mar 85 Mar 89 Mar 93 Mar 97 Mar 01 Mar 05 Mar 09
comm bank charge-off rate
total loans and leases
model
0.0
0.5
1.0
1.5
2.0
2.5
Mar 85 Mar 89 Mar 93 Mar 97 Mar 01 Mar 05
non-fin corporate borrowing/internal
funds lagged 6 quarters
comm bank charge-off rate total loans
and leases
Source: Haver, Barclays Capital
In Figure 28 we make a similar point. Here, we display a model designed to predict
defaults on consumer loans, the default rate expressed as an annual percentage of total
consumer loans, with key variables including household debt-to-income ratios, debt
servicing burdens, interest rate changes and unemployment. As with the business-
related default measures, variables capturing snapshots of household creditworthiness
tend to lead defaults by up to two years. Trends in defaults and loan losses can
therefore be ascertained, even in the absence of precise economic forecasting.
Figure 28: Net losses US bank consumer loans, actual and modelled
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Q1 85 Q1 87 Q1 89 Q1 91 Q1 93 Q1 95 Q1 97 Q1 99 Q1 01 Q1 03 Q1 05 Q1 07 Q1 09 Q1 11
Charge-Off rate consumer loans
Model
Source: Haver, Barclays Capital
These examples should serve to underline the simple point that cyclical fluctuations in
most aspects of the credit cycle are far from random. So, in the light of the financial
cataclysm that has unfolded over the past two years, the question that begs is why did the
tightening in credit conditions come as such a surprise to lenders, investors, regulators
8/8/2019 2009 Equity Gilt Study
30/95
8/8/2019 2009 Equity Gilt Study
31/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
30 Global Asset Allocation Strategy Barclays Capital
Although there were plenty of other causes of the US residential mortgage debacle, a
primary underlying reason was that participants drastically underestimated the impact
of falling house prices on mortgage defaults. This was due to the lack of any such
evidence in the US historical database. As a result, default models tended to
underestimate the connection between delinquency and house prices, lulling
participants into a false sense of comfort regarding the impact of any decline in
residential property prices on residential mortgage creditworthiness.Indeed, the reverse process applied. The extended phase of rising house prices had
reduced mortgage defaults by allowing potential delinquents the option of positive
equity extraction. This process in turn began to accumulate evidence that borrowers at
the lowest end of the creditworthy spectrum were perfectly capable of servicing
mortgages. As a result, the creditworthiness of lower credit scoring borrowers
apparently rose. The impact was to encourage lenders to weaken lending standards,
resulting in the now notorious surge in lending to the sub-prime category of borrower.
In essence, because the inputs to US mortgage delinquency models were missing a vital
piece of historical evidence, the models signally and woefully underestimated likely
defaults. This is a prime exponent of the old computing adage about junk in, junk out.
Two fundamental failures lie at the heart of the credit crunch; a failure to recognise the
volatility of house prices and failure to recognise the importance of house prices to
borrower creditworthiness. In essence, lenders did not pay attention to the potential
variability in the value of the collateral underpinning their loans. Precisely the same faults
might be attributed to the surge in lending to the corporate sector, a significant contributor
with a 38% share since the end of 2002 to the general rise in non-financial indebtedness
during the current cycle. In this case, lenders may very well have underestimated the
potential volatility of corporate profits. Certainly, the general acquiescence to higher
debt/EBITDA ratios and weaker covenanting can only be rationalised on the basis that
lenders shared the judgement of many policymakers that the business cycle had been
tamed, if not abolished. Undoubtedly, the relatively modest economic retrenchment barely a recession worthy of the name that followed the collapse of the 2000 equity
bubble reinforced this notion, and thus encouraged the tolerance of higher economic
leverage. Ironically enough, we can very well ascribe the shallowness of the 2001-03
recession to the boom in residential real estate borrowing. The severity of the downturn
was therefore blunted by the application of increased leverage. The better-than-expected
economic outcome subsequently encouraged the accumulation of even higher leverage.
Arguably, this self-feeding process has been underway since the start of the Great
Moderation the phase of declining macroeconomic volatility that began in 1982. One
does not require a spreadsheet to understand that a system in which higher borrowing
encourages higher borrowing is necessarily doomed to self-dissolution.
The credit crunch can therefore be seen as the inevitable denouement of a long-term
cycle of decreasing economic volatility, a trend that fostered an ever greater illusion of
future security. The basic cause of the crunch was a critical decrease of what can only be
described as macro-prudence on the part of most participants in the economy, whether
borrowers, lenders, regulators or policymakers. We might loosely define the quality of
macro-prudence as a general heightened awareness of the variability of future outcomes
and the instability of economic regimes, as well as a specific comprehension of the
hazards originating in collective behavioural patterns. Or, to put things more bluntly,
macro-prudence is a sensitivity to the damage that can be wrought by the madness of
crowds. A great deal of attention has been paid to incentive structures in the financial
markets, the weakening of credit oversight inherent in the originate and distribute
lending model, the opacity of structured and derivative products, the economically
unstable nature of profound current account imbalances and the myopic investor focus
on quarterly corporate results as proximate causes of the credit crunch. All these factors
8/8/2019 2009 Equity Gilt Study
32/95
BARCAP_RESEARCH_TAG_FONDMI2NBUR7SWED
Barclays Capital Global Asset Allocation Strategy 31
were certainly contributors to the rising tide of excess leverage in both the financial and
non-financial sectors. However, these elements were all in themselves merely symptoms
of the single ubiquitous underlying theme of a decline in macro-prudence.
The persistent lack of severe and prolonged negative economic outcomes from the
early 1980s onwards progressively reinforced scepticism that such outcomes could
ever occur. As Figure 30 should illustrate, returns from investment in credit assets over
the past 25 years have seen steadily positive results, with just the odd year of smalldeclines. Similarly, the financial sectors profitability, shown here as a share of GDP, has
grown steadily with only few moderate setbacks, at least until the past year.
Figure 30: Annual nominal returns, Barclays US Credit index, US financial sector profit share of GDP
-10
-5
0
5
10
15
20
25
30
35
40
45
1980 1985 1990 1995 2000 2005
annual total return, Barclays
US Credit index
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
4.0%
1980 1985 1990 1995 2000 2005
financial sector
profits as a
share of GDP
Source: BEA, Haver, Barclays Capital
Even after allowing for the survivorship bias inherent in such data series, these are hardly
the type of returns that would encourage an abundance of caution in market participants.
After all, the last major Western banking crises in the late 1980s and in the first few yearsof the 1990s failed to pose any globally systemic threats. So the lesson accrued over the
past quarter century was that the financial architecture was unshakeable, immune to the
downside of the credit cycle. Thus, despite a steady rise in leverage ratios across the
economy and an astonishing 100 percentage point rise in the us total debt/GDP ratio
over the past decade the expectation that the aftermath of credit booms might pose
mortal dangers declined just as steadily as the level of leverage rose.
Figure 31: US national leverage, total debt/GDP ratio, level and change
over 10 years
100
150
200
250
300
350
400
Q1 70 Q1 75 Q1 80 Q1 85 Q1 90 Q1 95 Q1 00 Q1 05
ratio
level
-20
0
20
40
60