2009 equity gilt study

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.


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


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/


4/95



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


5/95


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.


6/95



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


7/95



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.


8/95



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


9/95



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


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


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%.


10/95



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


11/95



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.


12/95



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


13/95



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


14/95



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/204768/0


15/95



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.


16/95



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.


17/95



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


Mining -17% 6% 6%

Manufacturing -11% -1% 10%

Transp -14% -1% 10%

Retail -11% -3% 14%

Finance -20% 0% 11%

Services -20% 2% 14%


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


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



18/95



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


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


19/95



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


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.


20/95



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%


Industrial production

0

100

200

300

400

500

600

700

800


0

100

200

300

400

500

600

700

800

S&P500

Credit spread (RHS

inverted)

-40%

-20%

0%

20%

40%

60%

80%


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.


21/95



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


0

50

100

150

200

250

300

350

S&P500

Credit spread (RHS

inverted)

-15%

-10%

-5%

0%

5%

10%

15%


Industrial Production

10,000

12,000

14,000

16,000

18,000

20,000

22,000

24,000

26,000

28,000


0

50

100

150

200

250

300

350

S&P500

Credit spread (RHS

inverted)

-15%

-10%

-5%

0%

5%

10%

15%


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


22/95



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.


23/95



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


24/95



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


25/95



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.


26/95



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%.


27/95



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


28/95



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


Model

-40

-20

0

20

40

60

80

100

Q2 90 Q2 93 Q2 96 Q2 99 Q2 02 Q2 05 Q2 08


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


29/95



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


30/95


31/95



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


32/95



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

2009 equity gilt study

Documents