Finance Help

T i

G Q

a

A R R A A

J G G N N N N

K B C F

1

f i a 2 2 m d i c o l s 8 o o i

(

h 1

Journal of Financial Stability 27 (2016) 74–94

Contents lists available at ScienceDirect

Journal of Financial Stability

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j f s t a b i l

his time is different: Causes and consequences of British banking nstability over the long run

areth Campbell, Christopher Coyle ∗, John D. Turner ueen’s Management School, Queen’s University Belfast, Riddel Hall, 185 Stranmillis Road, BT9 5EE Northern Ireland, UK

r t i c l e i n f o

rticle history: eceived 13 January 2016 eceived in revised form 24 June 2016 ccepted 27 September 2016 vailable online 30 September 2016

EL classification: 10 21 13 14

a b s t r a c t

This paper addresses three questions: (1) How severe were the episodes of banking instability experi- enced by the UK over the past two centuries? (2) What have been the macroeconomic indicators of UK banking instability? and (3) What have been the consequences of UK banking instability for the cost of credit? Using a unique dataset of bank share prices from 1830 to 2010 to assess the stability of the UK banking system, we find that banking instability has grown more severe since the 1970s. We also find that interest rates, inflation, lending growth, and equity prices are consistent macroeconomic indicators of UK banking instability over the long run. Furthermore, utilising a unique dataset of corporate-bond yields for the period 1860 to 2010, we find that there is a significant long-run relationship between banking instability and the credit-risk premium faced by businesses.

© 2016 Elsevier B.V. All rights reserved.

23 24

eywords:

anking crises ost of credit inancial instability

. Introduction

The severity of the banking crisis of 2007–8, combined with the act that they are infrequent events, has led scholars to look increas- ngly to history in order to understand better the potential causes nd consequences of banking instability (Reinhart and Rogoff, 009; Grossman, 2010; Jordà et al., 2011; Schularick and Taylor, 012). However, such studies use a narrative approach in deter- ining whether a crisis has occurred or not, making it difficult to

ifferentiate between different levels of instability. This approach s also subjective in that there is divergence of opinion over what onstitutes a banking crisis. As a consequence, time-series analysis f the causes and consequences of banking instability is somewhat imited. One way of overcoming these difficulties is to use bank tock prices to measure instability (Reinhart and Rogoff, 2009, p. ). In this paper, we adopt this approach by constructing a dataset f monthly British bank share prices covering a 181-year period, in

rder to develop a precise and continuous measure of UK banking nstability over the long run.

∗ Corresponding author. E-mail addresses: [email protected] (G. Campbell), [email protected]

C. Coyle), [email protected] (J.D. Turner).

ttp://dx.doi.org/10.1016/j.jfs.2016.09.007 572-3089/© 2016 Elsevier B.V. All rights reserved.

Our share-price dataset is used to address two key questions: (1) What are the macroeconomic indicators of instability over the long run? (2) What effect does banking instability have on the cost of credit? We examine potential causes and consequences of banking instability using Granger causality tests based on vector autoregres- sion (VAR) models. Firstly, we test which macroeconomic variables consistently act as good predictors of banking instability over our sample period. Secondly, we examine if there is a long-run his- torical relationship between banking instability and the credit-risk premium faced by firms. In order to measure the credit-risk pre- mium for the UK, a new proxy is developed, using a hand-collected dataset of UK corporate bond yields. This, to the best of our knowl- edge, is the first long-run time-series test of the hypothesis that banking instability affects the cost of credit.

Using our bank-share-price dataset, we find that although the British banking system has faced episodic bouts of instability at var- ious points throughout the past two centuries, instability increases significantly in the final quarter of the twentieth century. This increase in instability culminated in the 2007-8 banking crisis, which according to our share-price measure is on a totally dif- ferent scale to all previous episodes of banking instability in the

UK. Notably, several of the UK banking crises identified as such by Reinhart and Rogoff (2009) do not manifest themselves in our data. Furthermore, using our measure, the classic nineteenth-century crises do not appear to be as severe for the banking system as stan-

Financ

d c

r p o s

d p s i a c A t m t c

i t e R a a r t b

n a 1 2 T i c ( i m t

l b b h e 2 a o fi U t

s a o S o m c c t b c

G. Campbell et al. / Journal of

ard narratives would suggest. However, this is not to say that these rises did not affect other parts of the financial system.

In terms of leading macroeconomic indicators of crises, our esults imply that interest rates, inflation, credit growth, and equity rices have significant predictive power for UK banking instability ver the long run. This suggests that the stability of these factors hould be an important policy goal.

Our study also provides the first evidence of a significant pre- ictive relationship between banking instability and the credit risk remium over the long run. Furthermore, our long-run evidence uggests that the increase in the cost of credit following banking nstability appears to be more severe and prolonged for smaller nd more risky firms. This highlights the importance of maintaining hannels of credit to small businesses following banking instability.

caveat to these results is that we use bond yields as a proxy for he cost of credit. In most cases, credit obtained through the bond

arket or from banks should be close substitutes, but it is possible hat this is not always the situation and our measure of the cost of redit may be imperfect.

The findings of this paper contribute to the literature on bank- ng stability in several ways. First, it augments the literature on he frequency, severity and measurement of financial crises (Bordo t al., 2001; IMF, 1998; Caprio and Klingebiel, 1996; Reinhart and ogoff, 2009; Schularick and Taylor, 2012). It does so by developing

new measure of banking instability, which helps provide more of n idea of the scale of banking crises than the commonly-used nar- ative approach. This approach suggests, that at least in the case of he UK, the extant literature overestimates the occurrence of severe anking crises.

Second, the paper contributes to the literature on the macroeco- omic factors that make banking crises more likely to occur (Hardy nd Pazarbasioglu, 1998; Demirgüç -Kunt and Detragiache, 1998a, 998b, 1999; Kaminsky and Reinhart, 1999; Bergman and Hansen, 002; Van den End, 2006; Davis and Karim, 2008; Schularick and aylor, 2012; Bordo et al., 2003). In particular, it does so by show- ng that over the long run, credit growth has been an important ontributor to UK banking instability. However, unlike Jordà et al. 2014), we find little evidence that house prices contribute to bank- ng instability over the long run. Although the housing boom was

ajor contributor to the 2008 financial crisis, it was not an impor- ant contributor to most previous crises.

Finally, this paper augments the literature which highlights the ink between banking instability and the credit-risk premium faced y firms. It is the first long-run analysis of how banking insta- ility affects the cost of credit, whereas most previous literature as focused on particular crisis episodes (Bernanke, 1983; Ding t al., 1998; Borensztein and Lee, 2000; Hall, 2010; Krishnamurthy, 010). The only other long-run view of crises and credit is Bordo nd Haubrich (2010), who trace the effect of credit distress events n output in the United States from 1875 onwards. Similar to the ndings of this extant literature, the long-run evidence from the K suggests that banking instability is followed by an increase in

he cost of credit intermediation. This paper is structured as follows. In section two, our hypothe-

es on the leading macroeconomic indicators of banking instability nd the effect of banking instability on the cost of credit are devel- ped. Section three discusses our data sources and methodology. ection four examines the evolution of British banking instability ver the last two centuries. Section five analyses fluctuations in key acroeconomic variables around specific and noteworthy banking

rises and then presents Granger causality tests for the macroe- onomic indicators of UK banking instability. Section six analyses

he consequences of banking instability for the cost of credit to usiness. Section seven then summarises the main findings and oncludes.

ial Stability 27 (2016) 74–94 75

2. Banking instability: theory and hypotheses

2.1. Macroeconomic indicators of banking instability

Banking instability can develop from the decisions made by banks in the peaks and troughs of the business cycle. Dur- ing economic upswings, banks may underestimate the problems associated with asymmetric information, which may lead to over-lending and increased lending to risky projects. In such an environment, banks are susceptible to a sudden macroeconomic shock, which may decrease the ability of borrowers to repay. Thus, this potential ‘boom and bust’ nature of bank activities may lead to increased instability (Minsky, 1977; Gavin and Hausmann, 1996, p. 13; Grossman, 2010, p. 74; Reinhart and Rogoff, 2009, p. xxvii). Notably, many studies have found that rapid credit growth appears to be consistently associated with systemic bank- ing crises (McKinnon and Pill, 1997; Kaminsky and Reinhart, 1999; Eichengreen and Arteta, 2000; Allen and Gale, 2000; Logan, 2001; Hardy and Pazarbasioglu, 1998; Gourinchas et al., 2001; Wong et al., 2010). In particular, cycles of over lending that drive or are accom- panied by booms in equity and housing markets have been found to increase banking instability (Borio and Lowe, 2002; González- Hermosillo, 1999; Jordà et al., 2011; Reinhart and Rogoff, 2009, pp.158–160; Reinhart and Rogoff, 2008; Schularick and Taylor, 2012).

In this paper, we hypothesise that credit growth and asset prices are associated with banking instability. In addition, changes in nominal interest rates and inflation can stimulate credit booms and sharp changes in asset prices (Alchian and Kessel, 1962), and may therefore be associated with credit booms. Flamini and Milas (2015) identify a positive relationship between interest rate volatility and financial instability in the UK, US and Sweden. As the UK banking system has never heavily relied on foreign-currency-denominated debt, exchange rate movements should not result in increased costs of servicing such debt (Kaufman, 1999, pp.15–16). Notably, Eichengreen and Arteta (2000) find that currency problems are not an important cause of banking crises.

It is possible that macroeconomic triggers are not endogenous to the banking system. For example, Gavin and Hausmann (1996, pp. 27–28) use the analogy of a chain to emphasise the macroeconomic roots of banking crises. When macroeconomic forces place strain on the banking system, the weakest banks are the ones most likely to fail, but it is the macroeconomic tension, as much as the weakness of individual banks, that causes the failures. In other words, macroe- conomic shocks which are exogenous to the banking system make it unstable. For example, increases in the nominal interest rate may cause problems as borrowers may face difficulties in servicing their debt, which potentially increases the number of loan defaults and increases the probability of the bank being in financial distress.

2.2. Banking instability and the cost of credit

The economic consequences of instability in the banking system are potentially severe and far reaching (Bernanke, 1983; Demirgüç - Kunt et al., 2006; Hoggarth et al., 2002; Laeven and Valencia, 2010; Laeven, 2011; Deaton, 2012). Although there is little dispute about the economic costs associated with banking crises, the channels through which problems in the banking sector affect economic output have long been queried. Friedman and Schwartz (1963) argue that banking difficulties can exacerbate the economic situa- tion through the monetary channel due to a rapid decrease in the money supply and money multiplier. Another channel is the extent

to which output is constrained by the effect that banking instability has on the availability and cost of credit.

Disruption of the banking system reduces the ability of banks to alleviate the asymmetric information problem effec-

7 Financ

t 1 C a c e t n t o b a

t t o c a a 2 n d

i m f c a

3

r o 2 L ( t fi s t r f a t R t t c o

l d t p t t

v t c g U

d

the corporate bond market was an important alternative source of capital for companies of all sizes. Equally, banks stepped in to offer finance following the decline in the UK corporate bond market in

1 Since joint-stock banks could not establish in London until 1833 due to the Bank of England’s charter, the earliest share price for a London bank is August 1834.

2 For example, if we have a new peak in the index of bank returns in 2000 at 100, and the index closed at 80 in 2001, 60 in 2002 and 105 in 2003, the drawdown each year would be −20%, −40% and 5% respectively.

3 We collect data on annual basis, as a number of the variables we use are not available at a greater frequency than annual for the full sample, including a number

6 G. Campbell et al. / Journal of

ively and efficiently (Mishkin, 1991; Bernanke and Gertler, 1989, 990;Holmstrom and Tirole, 1997; Kiyotaki and Moore, 1997; hristiano et al., 2010). In addition, periods of distress may be ccompanied by shocks to asset prices, which erode a borrower’s ollateral-to-debt burden (Bernanke 1983). Banks, operating in an nvironment of increased uncertainty, face a dilemma of whether o reject those seeking credit, charge higher premiums, or a combi- ation of both, as they attempt to recapitalise or reduce exposure o risky assets. This results in a higher cost of credit intermediation r credit-risk premium, particularly for small- and medium-sized orrowers (Bernanke, 1983; Bernanke and Gertler, 1995; Calomiris nd Mason, 2003).

As banks have an advantage in assessment of credit risk through heir developed expertise in screening and monitoring, alterna- ive channels of credit are likely to have a similar increased cost f credit intermediation (Bernanke 1983, p.264). In bond markets, redit-related components of yield spreads will increase, reflecting ssessments of default risk, but also non-credit components, such s increases in illiquidity of credit markets (Churm and Webber, 007). Notably, Altman et al. (2010) highlight the causal link run- ing from bank loan returns to bond returns during periods of istress.

In this paper, we hypothesise that banking instability results n an increase in the credit-risk premium on the corporate bond

arket faced by firms and that this increase is particularly severe or smaller firms. Although our focus in this paper is on the credit hannel, we also test for the presence of a monetary channel by nalysing the effect of banking instability on the money supply.

. Data and methodology

Studies on banking crises over the long run generally use a nar- ative and binary approach in determining whether a crisis has ccurred or not (e.g., Goodhart and Delargy, 1998; Bordo et al., 001; Grossman, 2010; Reinhart and Rogoff, 2009; Valencia and aeven, 2012; Schularick and Taylor, 2012). Reinhart and Rogoff 2009), for example, define a banking crisis as bank runs that lead o closure, merging or a public-sector takeover of one or more nancial institutions or the closure, merging, takeover or public- ector assistance of an important financial institution. Following his approach, we could measure banking stability over the long un using bank failure data. We eschew this approach because bank ailures do not mean that a banking system is unstable and may ctually promote banking stability, as was the case in the UK in he nineteenth century (Baker and Collins, 1999). In addition, the einhart and Rogoff (2009) definition includes financial institutions hat may not be creating deposits or intermediating credit. In addi- ion, in the case of the UK, there are long periods of time whereby we annot be sure if Bank of England assistance was secretly provided r if distressed mergers occurred or were arranged.

Consequently, in this paper, in order to assess quantitatively the ong-run time-series variation in British banking instability, a new ata-set of monthly share prices of British commercial banks for he period 1830 to 2010 is constructed. The reasoning here is that roblems or uncertainty surrounding the banking sector, no matter he source, are likely to be reflected in bank share prices, assuming hat markets are relatively efficient.

The data-set starts in 1830 because prior to this there were ery few joint-stock commercial banks since banking incorpora- ion law had only been liberalised in the mid-1820s. Foreign or olonial banks, which are registered in London, are excluded as they

enerally conducted the majority of their business outside of the K.

Monthly share price data were hand-collected for all Lon- on banks from 1834 onwards, for all Scottish banks from 1830

ial Stability 27 (2016) 74–94

onwards, and for all Irish banks for 1830–1921.1 English provin- cial banks (i.e., banks located outside a 65-mile radius of London) were relatively small and their shares traded infrequently until the 1870s. We have share price data for the provincial banks traded on the London Stock Exchange up until 1868 and share price data for all provincial banks from 1869 onwards. Appendix Table A1 provides details on our data sources. Bank returns are equally-weighted and adjusted for events such as changes in capitalisation, stock splits and reverse stock splits. Our preference for equally-weighted returns arises from the concept that problems at smaller banks may highlight systemic issues. However, to ensure that our results are not being driven by our weighting methodology, we collected data on market capitalisation (see Appendix Table A1 for sources) to construct a value-weighted series. As can be seen from Fig. 1, there is little difference between the equally- and value-weighted series – the correlation between the two series is 96.1%. Our results are robust to using the value-weighted series.

The absolute value of average bank returns, a non-parametric measure of volatility, is used as the baseline measure of bank- ing instability. Absolute returns have been shown to be a simple and accurate measure of volatility in time series data (Ding et al., 1993; Granger and Sin, 2000; Forsberg and Ghysels, 2007). Abso- lute returns are also more robust than other measures of volatility in the presence of large movements, such as booms and crashes (Davidian and Carroll, 1987; Cotter, 2004). Robustness tests are run using drawdown of bank returns, measured by the change in an index of bank returns from its previous historical peak, nomi- nal bank returns, and volatility of bank returns, measured by the non-overlapping annual standard deviation of average monthly returns.2

As a key objective of this paper is to test potential macroe- conomic indicators of banking instability as well as the effect of banking stability on the cost of credit for business over the long run, data for several other variables were collected.3 First, potential macroeconomic indicators were collected i.e., UK equity returns, GDP, dollar exchange rates, interest rates, inflation rates, commod- ity prices (proxied by wheat prices), net public debt, M3 money supply data, house price data, and bank lending data (see Appendix Table A1 for data sources).4 Second, in order to test the relation- ship between banking instability and the cost of credit, we use the UK corporate bond yield differential over government bonds as a proxy for the cost of credit intermediation or the premium borrow- ers must pay for credit. The yield spread between government and corporate debt is a commonly used measure of changes in the cost of credit during specific periods of banking instability (Bernanke, 1983; Hall, 2010; Bordo and Haubrich, 2010), and is recognised by the Bank of England as a proxy for the spread on bank borrowing for larger businesses (Butt and Pugh, 2014). In the UK, bank and bond markets have acted as good substitutes for loan capital. Histori- cally, when banks were reluctant to advance money to businesses,

of hand collected variables. 4 While changes in bank capital ratios may be an important predictor of changes

in bank instability, this is difficult to measure accurately over the long run due to the presence of unlimited liability, reserve liability, uncalled capital, and hidden reserves. Therefore, the capitalisation of banks is not a focus of this paper.

G. Campbell et al. / Journal of Financial Stability 27 (2016) 74–94 77

F N d. Fail S

t c r a t w d

d l f i b w w o o f d

b c M n a

a B 2

p t c

w y A o b ( I

ig. 1. British bank returns, 1830–2010.gr1 otes: The banking crises documented in Reinhart and Rogoff (2009) are highlighte ources: See Appendix Table A1.

he 1970s (Coyle and Turner, 2013). Therefore, any increase in the ost of credit as the result of banking difficulties should also be eflected in the bond market. This is supported by the fact that the verage annual corporate bond yield in our sample has a correla- ion of 90 per cent with the bank bill rate between 1870 and 2005, hich is the rate on short-term debt instruments in which banks ealt.5

Our corporate bond dataset consists of hand-collected annual ata, running from 1860, the first date for which we can find pub-

ished data on UK corporate bonds, to 2010 (see Appendix Table A1 or data sources).6 Corporate bond yields are calculated from bonds ssued by domestic UK companies. A company was considered to e a UK company if its corporate headquarters and the main market here its securities traded were in the UK and its main activities ere based in the UK.7 Our corporate bond dataset contains 49,804

bservations. The yields on the government debt market are based n Consol prices for the period 1860 to 1955, and from 1955, a port- olio of high-coupon, long-dated bonds (see Appendix Table A1 for ata sources).

Several studies have specifically used the yield differential etween Baa corporate bonds and U.S government bonds when onstructing a measure of the cost of credit (e.g., Bernanke, 1983;

ishkin, 2009; Bordo and Haubrich, 2010; Hall, 2010). As there is

o equivalent ‘Baa’ rating system for our UK sample, we use the verage annual yield differential for our full sample of corporate

5 The bank bill rate refers to the prime bank bill rate for 1870–1982 from Capie nd Webber (2010), and the eligible bank bill rate for 1983–2005, available from the ank of England Interactive Database. Banks ceased accepting eligible bank bills in 005. 6 All financial companies that issued bonds are excluded as these firms, for a large

art of the sample period, were mainly financial trusts which issued bonds with he aim of investing the bulk of the proceeds in the equities and bonds of foreign ompanies (Jefferys, 1977, p.262).

7 The criteria used to determine whether a firm was based in the UK or overseas as based on (a) the section in which it appears in the relevant stock-exchange

earbook (e.g., Foreign Railways), (b) the company’s name (e.g., New Zealand and ustralia Land Co. Ltd.), or (c) if the company had a head office in a foreign country f operation as well as one in the UK. This information was obtained from a com- ination of the Investor’s Monthly Manual (1864–1930); Burdett’s Official Intelligence 1882-98), the Stock Exchange Official Intelligence (1899-[Stock Exchange Official ntelligence, 1899]1933), and the Stock Exchange Yearbook (1934–2010).

ed/nationalised banks are removed from sample at point of failure/nationalisation.

bonds over UK government bonds. However, we also subdivide our corporate bond risk premium into two equal subsets for each year in the sample – one containing firms with above median yields and one containing firms with below median yields. This enables us to assess whether banking instability has a different effect on the credit-risk premium of firms in different risk grades.

As can be seen for Table 1, which contains summary statistics of our key variables, bank returns are volatile, with a standard deviation of 20.32% and a large range in annual returns. Several other variables appear similarly volatile. Money supply, commod- ity prices, public debt growth, and equity returns all have standard deviations of 13% or greater. Notably, the volatilities of each of the three risk premium variables are similar over the full sample period.

In order to test our hypotheses, multiple-dimension vector autoregression (VAR) models and Granger causality tests are used in order to determine if there is a statistically significant relation- ship between macroeconomic variables and banking instability and between banking instability and the credit-risk premium. As the variables concerned may simultaneously influence one another over the sample period and the exact structure of the underlying relationship of the multiple time series is unknown, VAR models are used to account for the dynamics of all variables. The multivariate VAR model is of the form:

Y t = ˇ0 + ˇ1Y t-1 + . . .. + ˇmYt-m = ut (1) where Yt is a vector of all the macroeconomic and banking stability variables included in the system, �0 is a vector of constants, �1. . .� m are matrices of coefficients of all the lagged variables, m represents the number of lags of each variable, and ui is a set of error terms. The exact specification of the VAR model is discussed in greater detail in Sections 5 and 6.

4. British banking stability over the long run

4.1. Nineteenth-century crises

From Figs. 1–3 which contain our series of British bank returns

and also highlight episodes defined by Reinhart and Rogoff (2009) as banking crises, we see that several years in which crises are said to have taken place during the nineteenth century show negative stock returns. In particular, the crises in 1847, 1857, 1866 and 1878

78 G. Campbell et al. / Journal of Financial Stability 27 (2016) 74–94

Table 1 Summary Statistics.

Mean Median Standard Deviation Minimum Maximum % of Years <0 No. of Obs.

Average Bank Returns (%) 5.32 2.60 20.32 −76.50 154.64 34.81 181 Absolute Bank Returns (%) 11.92 6.24 17.26 0.01 154.64 n/a 181 Annual SD of Bank Returns (%) 2.99 1.70 3.09 0.30 18.86 n/a 181 Corporate Bond Risk Premium (%) 0.94 1.07 0.96 −2.28 2.86 12.58 151 Drawdown (%) −3.82 −2.19 19.78 −82.81 51.76 58.56 181 High Risk Premium 1/2 (%) 1.81 1.60 0.82 0.19 4.29 0.00 151 Low Risk Premium 1/2 (%) 0.08 0.60 1.34 −5.14 1.49 23.18 151 Real Interest (%) 1.70 2.28 5.31 −20.83 18.43 24.86 181 GDP Growth (Nominal, %) 4.73 5.02 6.10 −17.87 26.27 18.33 180 GDP Growth (Real, %) 2.00 2.34 2.93 −9.80 9.92 18.89 180 M3 Money Supply Growth (%) 8.89 4.90 14.94 −7.54 140.82 13.07 176 Inflation (%) 2.76 1.80 6.05 −14.00 25.20 27.62 181 Equity Returns (Nominal, %) 10.86 10.16 17.73 −48.80 145.60 20.44 181 Equity Returns (Real, %) 8.10 7.60 18.04 −64.80 121.40 26.52 181 Wheat Price Growth (%) 2.90 1.52 19.22 −34.86 100.00 44.75 181 Exchange Rate 3.98 4.79 1.50 1.30 9.97 n/a 181 Public Net Debt Growth (%) 4.47 0.00 13.06 −10.96 95.37 29.83 181 House Price Growth (%) 5.65 3.45 7.63 −15.79 34.83 10.00 110 Bank Lending Growth (%) 7.36 5.00 10.73 −14.00 65.00 19.21 151

Notes: The corporate bond risk premium is the excess of current yield on debentures over and above Consols or high-coupon, long-dated government bonds. The corporate bond risk premium over government bonds is further categorised into two subsets: High Risk Premium 1/2 (highest 50% of yields) and Low Risk Premium 1/2 (lowest 50% of yields).

Table 2 Performance of bank stocks during banking crises documented by Reinhart and Rogoff (2009).

Crisis Banks – Market Return (%)

Per cent of banks to lose 10–50%+ of value Total banks Worst performing bank

Return of worst performing bank (%)

>10% loss >20% loss >30% loss >40% loss >50% loss

1836–39 1836 −3.96 0% 0% 0% 0% 0% 14 Provincial Bank −6.67 1837 −1.32 0% 0% 0% 0% 0% 16 Provincial Bank −4.76 1838 3.96 0% 0% 0% 0% 0% 17 Royal Bank of Scotland −0.62 1839 −0.16 0% 0% 0% 0% 0% 21 London Joint Stock Bank −9.26 1847 3.55 50% 9% 0% 0% 0% 22 North of Scotland −22.54 1857 −1.90 35% 4% 4% 0% 0% 23 Glasgow Union/Union Bank −30.58 1866 −4.53 27% 10% 7% 0% 0% 30 Alliance of London & Liverpool −38.14 1878 −6.13 39% 18% 2% 1% 0% 106 Clydesdale −41.24 1890 0.88 1% 0% 0% 0% 0% 97 Yorkshire Banking Company −38.82 1914 7.28 0% 0% 0% 0% 0% 45 Union of London & Smiths −6.02 1974 −8.00 100% 100% 100% 100% 100% 7 Nat. West. Bank −69.85 1984 25.70 14% 0% 0% 0% 0% 7 Midland −16.78 1991 13.33 0% 0% 0% 0% 0% 7 Barclays −3.97 1995 22.83 0% 0% 0% 0% 0% 7 Nat. West. Bank 28.79 2007–8 2007 −39.93 88% 75% 38% 38% 0% 8 Northern Rock −92.87 2008 −35.83 100% 100% 80% 80% 80% 5 HBOS −90.36

Notes: As our focus is the state of the overall banking sector in any given period, and we do not have share price data for several of the banks that failed during our sample period, which is further complicated by unlimited liability and nationalisation, failed/nationalised banks are removed from sample at the point of failure/nationalisation and n banks m n acro a

a T t s i v

n s s c t p t c f

ot included as 100% loss. However, including 100% losses for failed/nationalised entioned above and due to the large difference in the number of banks in operatio

s years in which banking crises occurred.

re associated with negative returns. However, as can be seen from able 2, relative to the overall market, the fall in bank stocks during hese four episodes were not particularly severe and very few bank tocks lost more than 20% of their value. Notably, the banking crises n 1836–9 and 1890 are not associated with negative returns or olatility in the market for bank shares.

One possible reason as to why we do not find that these ineteenth-century crises are associated with large falls in bank tocks is that a lot of commercial banking was still performed by mall private banks, particularly in the first half of the nineteenth entury. However, the failure rate of private banks was never higher han 2.3% during any of these crises (Turner, 2014, p. 53). Another ossible reason is that there were not systemic problems during

his era; rather the banking problems experienced in the nineteenth entury were more idiosyncratic. Below we provide some context or each crisis which illustrates this point.

in our sample only makes the decline in 2007-8 more standout due to the points ss the sample. Dates included are those highlighted by Reinhart and Rogoff (2009)

The banking problems of 1836–9 were chiefly manifested in the prominent failures of the Agricultural and Commercial Bank of Ireland and Northern and Central Bank of England. Unlike most of their counterparts, these two banks had expanded rapidly (particu- larly their branch network) following their establishment in 1834. They were also both riddled with unique governance problems.

During the 1847 crisis, several medium-sized banks in the north-east of England failed and several other English banks tem- porarily suspended payments. The 1847 crisis was in reality a commercial crisis accompanied by a money-market crisis, which was precipitated by the fear that the newly-established Bank Char- ter Act curtailed the ability of the Bank of England to act as a lender of last resort.

Following the arrival of news from the U.S. in the autumn of 1857 that railroad securities had fallen substantially and that there were numerous bank suspensions on the Atlantic coast, many

Financ

B p s N f a (

O m h s b f f t a

o l i m T m w q E R o o f f

o i a r g m s s f 9 b s

4

F c s i W

t m r a

R r k n F 1

G. Campbell et al. / Journal of

ritish firms involved in trading with the U.S. failed, which placed ressure on their banks. Three important banks failed in this cri- is – Liverpool Borough Bank, Western Bank of Scotland, and the orthumberland and Durham District Bank. Each of these three

ailures was idiosyncratic, with aggressive risk-taking, long-term ccumulation (and hiding) of bad debts, and concentrated lending Turner, 2014, pp. 77–78).

The 1866 crisis was largely precipitated by the failure of verend, Gurney and Company, a discount house, which after accu- ulating large (but hidden) bad debts from the late 1850s onwards

ad converted to a limited-liability public company in 1865. Several mall banks as well as the Birmingham Banking Company, a large ank which was the most senior bank in the West Midlands, failed ollowing the Overend Gurney collapse. In each of these cases of ailure, there was clear mismanagement and in several instances here was concentrated lending, insider lending, and fraudulent ccounting (Turner, 2014, pp. 82–84).

The central event of the 1878 crisis was the failure of the City f Glasgow Bank − Glasgow’s premier bank, which had the third argest branch network in the UK. The £5 million deficit between ts assets and liabilities was met by calls on its shareholders, the

ajority of whom were bankrupted by the action (Acheson and urner, 2008). The City of Glasgow Bank had a very risky asset anagement strategy, concentrated loan portfolio, and directors ho had engaged in fraud (for which several of them were subse-

uently incarcerated). Other banks, and in particular the Bank of ngland, had never trusted this bank (Clapham, 1944, vol. 2, p.309; ait, 1930). The other major bank to fail in this crisis was the West f England and South Wales District Bank, and similar to the City f Glasgow Bank, it had concentrated loans, falsified accounts, and raudulent attempts were made to conceal the bank’s real position rom its shareholders (Turner, 2014, p. 87).

At the time of its failure in 1890, Barings Brother and Co. was one f the most powerful merchant banks in Europe. During the 1880s, t had invested heavily in the emerging economies of Argentina nd Uruguay. However, economic difficulties in these countries esulted in Barings suffering such large losses that it required a uarantee fund co-ordinated by the Bank of England and funded by ajor commercial banks to prevent its collapse. It is entirely pos-

ible that this action taken by the Bank of England, which was fully upported by the UK Treasury, prevented commercial banks from ailing or their stock prices from falling (Collins and Baker, 2003, p. 9). However, on closer inspection of Fig. 3, we do not see a fall in ank stocks in the month or months before the Bank of England’s upport arrangement was announced.

.2. Twentieth-century crises

When we move into the twentieth century, we observe from igs. 1–3 that bank stocks became more volatile in the twentieth entury compared to the nineteenth century, particularly in the econd half of the twentieth century. This is consistent with studies ndicating that financial instability has increased in the post Bretton

oods era (Bordo et al., 2001). Notably, although bank stocks fell by nearly 25 per cent during

he Great Depression (Fig. 1), they fell by less than the overall stock arket. This is consistent with the view that Britain did not expe-

ience a banking crisis during this period (Grossman, 1994; Capie nd Wood, 2012, p. 333).

Of the five twentieth-century crises identified by Reinhart and ogoff (2009), only 1914 and 1974 are associated with negative eturns on bank stocks. The breakdown of European capital mar-

ets and the foreign-exchange crisis in the summer of 1914 did ot cause major instability in the banking sector as indicated by ig. 1 and Table 2. Although capital markets were closed from July 914 through the rest of 1914, bank stocks did not decline when

ial Stability 27 (2016) 74–94 79

the market reopened. However, the extension of the usual August Bank Holiday in 1914 by four days plus the issue of Treasury notes contributed to the easing of liquidity pressures faced by banks fol- lowing the breakdown of the European markets (Sayers, 1976, vol. 1, pp.74–76).

The negative returns on bank stocks in 1974 clearly stand out in Fig. 1. Secondary banks, which had raised funds on the newly- liberalised money markets and advanced them in the form of property and consumer finance loans, got into difficulties in late 1973 (Capie, 2010, pp.524–586; Reid, 1982). Commercial banks, coordinated by the Bank of England, provided liquidity support, and, as a result, the difficulties almost passed unnoticed (Reid, 1982, p. 200). Fig. 1 shows significant volatility during this period with large negative returns of −27.79 per cent in 1973 and −62.85 per cent in 1974, indicating much greater instability than at any earlier stage in the sample. However, the large fall in bank stocks is quickly followed by a huge rebound in returns the following year (Fig. 1). While this may reflect the difficulties faced by the secondary banks in the UK, this pattern is also found in the overall equity market during this period, suggesting that bank shares were really influ- enced by other factors, including the 1973 Oil Crisis and extremely high inflation. When the returns of the overall equity market are taken into account, the excess returns on bank stocks are only −8.0% (Table 2).

As can be seen from Fig. 1 and Table 2, there is little evidence of negative returns in 1984, 1991 and 1995. In fact, banks’ stocks outperform the overall stock market in each of these years. 1984 is classified as a banking crisis by Reinhart and Rogoff (2009) because of the failure of Johnson Matthey Bankers, a bullion dealer which had expanded into lending in the early 1980s. This bullion dealer was taken over by the Bank of England and the Bank co-ordinated a private-sector indemnity fund. These actions may have prevented a fall in bank stocks, but from Fig. 3, we see that in the month before the Bank’s action, bank stock prices actually rose. The 1991 crisis is designated as such by Reinhart and Rogoff due to the collapse of the Bank of Credit and Commerce International, an international bank with operational headquarters in London. The 1995 crisis is designated as such following the collapse of Barings Brothers after one of its traders amassed huge losses on Nikkei 225 Index futures. However, neither of these failures appear to have had a negative effect on the commercial banking system.

4.3. The twenty-first-century crisis

As can be clearly seen from Fig. 1 and Table 2, the collapse of the banking system in 2007–8 is of a totally different order of magni- tude than all previous crises. In terms of negative returns, this crisis is unparalleled, with returns of −26.87% and −76.50% in 2007 and 2008 respectively. In total, six of the nine big British banks were insolvent, with over half of the domestic banking system in terms of total assets requiring a bailout. The scale of the 2007–8 instability is apparent in Table 2, with the stock of the worst performing bank in either year falling by more than 90%, far beyond any fall in any other year over the 181-year sample period. Indeed, the fall in bank stock prices was only arrested by the extraordinary support mea- sures taken by the Bank of England and UK Treasury. Consequently, we can truly say of the 2007–8 crisis that ‘this time is different’!

4.4. Summary

Several of the banking crises identified by Reinhart and Rogoff (2009) do not manifest themselves in our data. This arises because

of the definition used by Reinhart and Rogoff (2009), which means that the failure of one financial institution constitutes a crisis. Quite clearly, the idiosyncratic failures of the City of Glasgow Bank in 1878, Barings in 1890, Johnson Matthey Bankers in 1984, BCCI in

80 G. Campbell et al. / Journal of Financial Stability 27 (2016) 74–94

Fig. 2. Annual volatility of British bank returns, 1830–2010. Notes: Annual volatility is calculated using the standard deviation of monthly bank returns. The banking crises documented in Reinhart and Rogoff (2009) are highlighted. Sources: See Appendix Table A1.

F N d. Fail S

1 c d o

5

t t i w t r a a y a

ig. 3. Monthly returns of British banks, 1830–2010. otes: The banking crises documented in Reinhart and Rogoff (2009) are highlighte ources: See Appendix Table A1.

991, and Barings in 1995 were not viewed as banking crises by ontemporaries and therefore do no manifest themselves in our ata. This demonstrates the limitations of the qualitative definition f banking crises used in many studies of banking crises.

. Leading indicators of banking instability

In this section, we examine macroeconomic variables to see if hey prove to be leading predictors of UK banking instability over he long run. Panel A of Table 3 presents changes in these variables n the years before, during and after the twelve UK banking crises

hich have been documented by Reinhart and Rogoff (2009). Five hings are worthy of note. First, there are on average large positive eal equity returns two years before crises followed by large neg-

tive returns the year of banking crises. Second, real interest rates nd inflation are higher than their historical averages in the two ears before these crises, as is real GDP growth, which shows an cceleration of economic activity in the year before crises. Third,

ed/nationalised banks are removed from sample at point of failure/nationalisation.

broad money supply growth is also consistently higher than the sample average in the years leading up to banking crises. Fourth, changes in wheat prices, a proxy for commodities, are on average negative two years before a banking crisis; however, price growth outstrips its historical average by many times in the year before and the year of these crises. Fifth, bank lending and house price growth show large above average increases in the years leading up to banking crises, which would support the view that significant credit growth and house price growth fuelled by easy credit may trigger financial instability. Panel B of Table 3 presents figures for the most significant individual periods of banking instability in our sample, namely those events in which 10% or more of all banks lost more than 20% of their market value. Several of the trends visible in Panel A are again evident in individual nineteenth and

twentieth century crises, highlighting similarities in the origin and evolution of banking instability. However, there are also interesting differences in variable changes around the crises in Panel B. Specif- ically, in the years before the 1878 crisis, which had its origins in an

G .

C a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

8 1

Table 3 Average variable changes during British banking crises documented by Reinhart and Rogoff (2009).

Year (t) Average Bank returns (%)

Public net debt growth (%)

Equity returns (real, %)

Interest rate (real, %)

GDP growth (real, %)

Inflation (%) Money supply growth (%)

Wheat price growth (%)

Exchange rate Bank lending growth (%)*

House price growth (%)*

Panel A: All Documented UK Banking Crises, 1830–2010 −2 16.83 2.99 14.48 1.90 2.60 3.24 15.4 −1.20 4.08 14.9 7.88 −1 1.29 3.62 1.98 1.54 3.64 3.66 19.74 12.69 3.85 9.75 11.6 0 −8.12 3.38 −3.98 1.77 0.68 3.70 9.36 10.10 3.63 8.05 5.33 1 17.92 11.49 22.82 2.52 1.19 2.93 13.23 2.18 3.69 3.19 3.19 2 10.08 14.04 11.17 1.47 2.80 3.81 9.03 2.84 3.66 5.51 7.52 Panel B: Significant UK Banking Instability: 1866, 1878, 1974, 2007–8 −2 12.39 −1.25 15.29 4.08 1.68 −0.90 13.86 −10.20 9.97 42.17 −1 −0.81 0.00 9.26 2.30 4.23 0.90 12.46 4.08 7.69 2.25 1866 −7.96 −1.27 −3.52 −3.24 1.15 6.50 15.45 19.33 6.88 2.27 1 −8.93 0.00 6.15 −3.02 −0.94 6.10 2.30 29.11 6.75 −4.75 2 −0.139 −3.846 21.25 4.76 3.049 −1.70 −0.741 −1.06 6.83 4.65 −2 1.25 1.39 6.06 3.32 0.72 −0.30 0.48 2.21 5.42 1.05 −1 4.62 0.00 6.72 3.71 0.63 −0.70 −2.88 22.88 5.08 0.77 1878 −9.67 0.00 −0.79 5.21 0.33 −2.20 −3.45 −18.19 4.89 −2.16 1 −4.11 1.37 28.29 7.34 −1.90 −4.40 −3.92 −5.6 4.85 −12.01 2 3.66 0.00 8.16 −0.09 7.07 3.00 0.18 1.19 4.84 0.47 −2 48.98 7.18 11.50 1.00 3.66 7.10 23.93 18.88 2.50 28.56 30.93 −1 −27.79 2.90 −33.50 0.54 7.20 9.20 26.88 72.97 2.45 28.32 34.83 1974 −62.85 9.71 −64.8 −2.66 −1.31 16.00 18.26 21.88 2.34 14.95 10.54 1 154.64 14.68 121.4 −11.2 −0.62 24.20 9.51 −5.13 2.22 4.49 7.25 2 −17.99 21.96 −13.6 −3.46 2.63 16.50 10.48 29.39 1.80 16.93 7.78 −2 5.14 10.76 19.96 2.99 2.17 1.34 12.20 4.50 1.82 9.04 5.83 −1 16.10 9.20 14.35 2.12 2.85 2.05 13.90 29.56 1.84 13.02 7.37 2007 −26.87 7.99 2.77 2.23 2.56 2.33 13.40 76.63 2.00 12.92 9.08 2008 −76.5 5.00 −33.42 2.34 0.55 2.32 10.10 −32.73 1.85 13.10 1.95 1 7.83 17.50 28.19 0.84 −4.92 3.61 5.40 −1.12 1.57 2.81 −0.75 2 9.94 23.12 9.00 2.300 1.91 2.17 −1.40 53.53 1.55 −1.17 11.31 Full sample average

5.38 4.47 8.10 1.70 2.00 2.76 8.89 2.90 3.98 7.36 5.65

Notes: Panel A includes all banking crises discussed in Reinhart and Rogoff (2009). Panel B includes the years 1866, 1878, 1974 and 2007-8; periods in which 10 per cent or more of all banks lost more than 20 per cent of their market value (see Table 2). *Bank lending growth is from 1860 onwards and house price growth is from 1901 onwards. Sources: See Appendix Table A1.

8 Financ

i g i y ( G

p s b i a i t

v f l a l t r t e r d

f p 1 w v i t c n t r

a G t b G t t d t f G a g L o d

p m p L p

2 G. Campbell et al. / Journal of

diosyncratic bank collapse, real interest rates are high and lending rowth, inflation and real equity returns are relatively low. It is also nteresting to note that, on average, GDP growth is positive in the ears following crises documented by Reinhart and Rogoff (2009) Panel A). However, for the four most severe crises (see Panel B), DP growth is negative in the year following the crisis.

While Table 3 highlights these relationships during specific eriods of banking crises, we now turn to examine if there is a tatistically significant relationship between these variables and anking instability over the long run using our full sample. Specif-

cally, multiple dimension VAR models and Granger causality tests re used in order to determine if there is any statistically signif- cant relationship between various macroeconomic variables and he level of British banking instability.

Prior to the analysis of the linkages between macroeconomic ariables and financial instability, the specification of the functional orm of the system, such as lag order was determined. Various ag order selection criteria tests, including the Akaike, Schwarz, nd Hannan-Quinn information criterion, indicate the optimum ag length to be two. In addition, Dickey-Fuller tests were used to est for stationarity. All variables, with the exception of exchange ates, were stationary; the exchange rate variable was therefore ransformed to ensure stationarity. Due to the presence of het- roskedasticity and autocorrelation in some variables, the model is un using heteroskedasticity- and autocorrelation-consistent stan- ard errors.

Tests are carried out for three main periods. First, tests are run or 1860–2010, the period for which we have bank lending and risk remium data. Second, the full dataset of bank share prices from 835 to 2010 is tested.8 Third, tests are also run on the period for hich we have UK house-price data i.e., 1901–2010. The absolute

alue of average bank returns, a non-parametric measure of volatil- ty, is used as the baseline measure of banking instability. These ests are rerun using drawdown of bank returns, measured by the hange in an index of bank returns from its previous historical peak, ominal bank returns, and volatility of bank returns, measured by he non-overlapping annual standard deviation of average monthly eturns.

The results of Granger causality tests for our baseline model re shown in Panel A of Table 4. For this section, we only report ranger causality results following the VAR which depict the rela-

ionship between all variable lags and our variable of interest, anking instability, for our three time periods (the full matrices of ranger causality results following all VAR models are reported in

he Appendix Tables). These tests examine the null hypothesis that he lags of all the variables included in each model (Columns 3–14) o not Granger cause changes in the variable in Column 2. From he first row of Panel A Table 4, we can reject the null hypothesis or several indicators. It appears that indicators which consistently ranger cause changes in UK banking stability from 1860 to 2010 re: real equity return growth, bank lending growth, money supply rowth, real rates of interest, inflation, and net public debt growth. ags of lending, interest rates, and inflation are significant at the ne per cent level and equity returns, money supply, and public ebt are significant at the five per cent level.

Row 2 of Panel A Table 4 shows the results for the full sample eriod (1835–2010). These results confirm that real equity returns, oney supply, real rates of interest, and inflation are significant

redictors of changes in UK banking stability over the long-run.

ags of interest rates and inflation are again significant at the one er cent level.

8 The years 1830-34 are dropped due to volatility caused by too few observations.

ial Stability 27 (2016) 74–94

The above results are consistent with our predictions and with findings from most indicator models of financial crisis in the extant literature. Exchange rates do not appear to explain changes in bank- ing stability during this period in any test, supporting the findings of Eichengreen and Arteta (2000). Our findings also suggest that commodity prices (proxied by wheat prices) and output also do not significantly explain changes in banking stability.

The final row of Panel A Table 4 presents the results for the period 1901–2010. Granger causality tests are largely consistent with the results from the previous two periods. Real interest rates and inflation again prove to be indicators of changes in banking instability. There is also further evidence that changes in bank lending growth, real equity returns, money supply growth and net public debt growth are significant predictors of instability. It appears that house price growth is not a significant predictor of banking instability during this sample period. While house price bubbles have been associated with banking instability in the recent past, this may not have always been the case. Notably, there have been several periods of strong house price growth during the sam- ple period, which were not associated with banking instability.

Several of these results are robust across alternative measures of banking instability. Panel B of Table 4 includes results of Granger causality tests when the above specifications are rerun, replacing absolute bank returns with drawdown, nominal bank returns and standard deviation of bank returns as the measure of instability. Using drawdown of bank returns as a measure of instability, results suggests that real equity returns, lending growth, inflation and real rates of interest are significant indicators of changes in banking instability over the long-run, which is consistent with the results above.

Using nominal bank returns and standard deviation of bank returns as measures of instability also show that lags of real interest rates, inflation, and bank lending growth Granger cause changes in bank instability. However, real equity returns is not a significant predictor of banking instability at the 10% level. Interestingly, for the standard deviation of bank returns, the addition of house prices indicates that they have explanatory power for changes in bank share price volatility at the 10% level.

The correlation matrix between VAR variable residuals, which captures contemporaneous relationships between variables, is pre- sented in Table 5. Notably, there is a positive correlation between our banking instability variable and the risk premium, real equity market returns, lending growth and inflation, all of which are statis- tically significant. There is also a negative correlation with the real interest rate, significant at the five per cent level, which could be attributed to increased bank earnings over the short run, following interest rate rises.

Fig. 4 displays the impulse responses of banking instability to a one standard error shock to each indicator variable, with 90 per cent confidence intervals. These functions give a clearer picture of the direction of the relationship between variables. From Fig. 4, we see that instability rises in response to a positive shock to money supply and inflation, with the peak response occurring after one to two years. A shock to bank lending growth is also associated with increased instability. However, instability decreases one year after a positive shock to real equity returns and a positive shock to real interest rates only slightly increases instability after one year. In order to determine how much the ordering of our variables in the VAR matters, Fig. 5 uses the reverse of the ordering used in Fig. 4. The results in Fig. 5 are largely similar to those in Fig. 4, with the main difference being the response of banking instability to a shock to real interest rates and inflation, with the effect of inflation

being decreased and the effect of real interest rates becoming more pronounced. In this analysis, inflation is allowed to affect interest rates contemporaneously, where the reverse was true in Fig. 4. A positive shock to the real interest rate variable increases banking

G .

C a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

8 3

Table 4 Granger causality tests of macroeconomic indicators of UK banking instability.

Dependant Variable

Bank Instability (P-Value)

Risk premium (P-Value)

Equity returns (real, %) (P-Value)

Bank lending growth (%) (P-Value)

GDP growth (real, %) (P-Value)

Money supply growth (%) (P-Value)

Inflation (%) (P-Value)

Interest rate (real, %) (P-Value)

Exchange Rate (P-Value)

Public net debt growth (%) (P-Value)

Wheat price growth (%) (P-Value)

House price growth (%) (P-Value)

Panel A: Absolute returns 1860–2010 Bank Instability 0.047 0.581 0.023 0.010 0.899 0.021 0.002 0.000 0.298 0.029 0.146 1835–2010 Bank Instability 0.041 0.013 0.832 0.083 0.009 0.000 0.945 0.192 0.240 1901–2010 Bank Instability 0.039 0.845 0.035 0.017 0.732 0.088 0.009 0.000 0.754 0.036 0.149 0.168

Panel B: Drawdown 1860–2010 Bank Instability 0.001 0.770 0.083 0.016 0.638 0.919 0.008 0.072 0.043 0.681 0.376 1835–2010 Bank Instability 0.000 0.118 0.639 0.160 0.049 0.069 0.266 0.914 0.382 1901–2010 Bank Instability 0.006 0.651 0.066 0.241 0.241 0.729 0.000 0.015 0.197 0.621 0.360 0.623 Standard deviation of returns 1860–2010 Bank Instability 0.000 0.100 0.470 0.039 0.574 0.044 0.003 0.000 0.016 0.336 0.086 1835–2010 Bank Instability 0.000 0.486 0.512 0.752 0.031 0.000 0.174 0.093 0.063 1901–2010 Bank Instability 0.000 0.335 0.475 0.152 0.451 0.002 0.005 0.001 0.695 0.222 0.098 0.094 Nominal returns 1860–2010 Bank Instability 0.473 0.196 0.167 0.017 0.659 0.888 0.000 0.001 0.075 0.139 0.151 1835–2010 Bank Instability 0.591 0.165 0.874 0.358 0.000 0.001 0.174 0.494 0.253 1901–2010 Bank Instability 0.567 0.043 0.146 0.049 0.416 0.997 0.000 0.000 0.645 0.124 0.086 0.556

Notes: Each row represents the p-values from Granger causality tests following a vector autoregression. Only the results relevant to the banking instability variable are presented here. The exchange rate is differenced to ensure stationarity.

Table 5 Contemporaneous correlations between variables.

Bank Instability Risk Premium Equity returns (real, %)

Bank lending growth (%)

GDP growth (real, %)

Money supply growth (%)

Inflation (%) Interest rate (real, %)

Exchange Rate Public net debt growth (%)

Wheat price growth (%)

Bank Instability 1.000 Risk Premium 0.142* 1.000 Equity returns (real, %) 0.362*** −0.102 1.000 Bank lending growth (%) 0.153* −0.173** 0.130 1.000 GDP growth (real, %) 0.074 −0.049 0.162** 0.053 1.000 Money supply growth (%) 0.060 −0.067 0.048 0.171** 0.350*** 1.000 Inflation (%) 0.153* −0.015 −0.157* −0.002 0.167** 0.070 1.000 Interest rate (real, %) −0.177** −0.043 0.075 0.008 −0.192** −0.068 −0.990*** 1.000 Exchange Rate −0.026 −0.083 0.078 0.392*** 0.008 0.094 −0.188** 0.191** 1.000 Public net debt growth (%) −0.068 −0.108 −0.080 −0.080 0.056 0.177** 0.345*** −0.327*** −0.007 1.000 Wheat price growth (%) −0.107 −0.181** −0.159* 0.028 0.302*** 0.259*** 0.339*** −0.318*** −0.0990 0.201** 1.000

Notes: ***, **, *, denote significance at the 1%, 5%, and 10% respectively.

84 G. Campbell et al. / Journal of Financial Stability 27 (2016) 74–94

-3 -2.5

-2 -1.5

-1 -0.5

0 0.5

1 1.5

2

0 2 4 6 8 10

P er

ce nt

Years

PublicNetDebt Growth(%)

-2

-1

0

1

2

3

4

5

0 2 4 6 8 10

P er

ce nt

Years

Money Supply Growth (%)

-5

-4

-3

-2

-1

0

1

2

3

0 2 4 6 8 10

P er

ce nt

Years

Interest Rate (real, %)

-2.5 -2

-1.5 -1

-0.5 0

0.5 1

1.5 2

2.5

0 2 4 6 8 10

P er

ce nt

Years

Exchange Rate

-3

-2

-1

0

1

2

3

4

0 2 4 6 8 10

P er

ce nt

Years

Bank Lending Growth (%)

-10 -8 -6 -4 -2 0 2 4 6 8

0 2 4 6 8 10

P er

ce nt

Years

Equity Returns(real, %)

-2 -1.5 -1

-0.5 0

0.5 1

1.5 2

2.5 3

0 2 4 6 8 10

P er

ce nt

Years

GDP Growth (real, %)

-4 -3 -2 -1 0 1 2 3 4 5

0 2 4 6 8 10

P er

ce nt

Years

Wheat Price Growth (%)

-2

-1

0

1

2

3

4

5

0 2 4 6 8 10

P er

ce nt

Years

Inflation (%)

-2

0

2

4

6

8

10

12

0 2 4 6 8 10

P er

ce nt

Years

Absolute Bank Returns

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

0 2 4 6 8 10

P er

ce nt

Years

Risk Premium (%)

Fig. 4. Impulse response functions of banking instability (order 1). Notes: These VAR Cholesky orthogonalized impulse response functions are estimated using annual data from 1860 to 2010, the ordering of the variables is as shown above. Each function displays the response to a one standard deviation shock in each indicator variable, with bootstrap 90% confidence intervals.

-3

-2

-1

0

1

2

3

4

5

0 2 4 6 8 10

P er

ce nt

Years

Risk Premium (%)

-6 -4 -2 0 2 4 6 8

10 12 14

0 2 4 6 8 10

P er

ce nt

Years

Absolute Bank Returns

-3

-2

-1

0

1

2

3

0 2 4 6 8 10

P er

ce nt

Years

Inflation (%)

-2

-1

0

1

2

3

4

5

6

0 2 4 6 8 10

P er

ce nt

Years

Wheat Price Growth (%)

-4

-3

-2

-1

0

1

2

3

4

0 2 4 6 8 10

P er

ce nt

Years

GDP Growth (%)

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2

0 2 4 6 8 10

P er

ce nt

Years

Equity Returns(real, %)

-3

-2

-1

0

1

2

3

4

0 2 4 6 8 10

P er

ce nt

Years

Bank Lending Growth (%)

-2.5 -2

-1.5 -1

-0.5 0

0.5 1

1.5 2

2.5

0 2 4 6 8 10

P er

ce nt

Years

Exchange Rate

-0.5 0

0.5 1

1.5 2

2.5 3

3.5 4

4.5 5

0 2 4 6 8 10

P er

ce nt

Years

Interest Rate (real, %)

-1.5 -1

-0.5 0

0.5 1

1.5 2

2.5 3

3.5

0 2 4 6 8 10

P er

ce nt

Years

Money Supply Growth (%)

-3.5 -3

-2.5 -2

-1.5 -1

-0.5 0

0.5 1

1.5

0 2 4 6 8 10

P er

ce nt

Years

PublicNetDebt Growth (%)

F N ed usi E ator v

i s d

t

ig. 5. Impulse response functions of banking instability (order 2). otes: These VAR Cholesky orthogonalized impulse response functions are estimat ach function displays the response to a one standard deviation shock in each indic

nstability sharply in the first year in this ordering, while a positive hock to inflation results in a small increase in instability, which

isappears after two years.

The question arises to how the regulatory environment and he safety net affect bank behaviour in a particular macroeco-

ng annual data from 1860 to 2010, the ordering of the variables is as shown above. ariable, with bootstrap 90% confidence intervals.

nomic environment. For example, low interest rates and high equity returns may not stimulate a credit expansion if it is constrained by

regulation. On the other hand, the presence of a safety net may incentivise banks to expand credit and increase the risk of their portfolio. Interestingly, in the case of the UK, there were substantial

G. Campbell et al. / Journal of Financial Stability 27 (2016) 74–94 85

Fig. 6. UK corporate bond risk premium, 1860–2010. N s ove b of Hig y ). S

r 2 B o s t H t a

6

U a e a

a s p p a a g

o b p s t d b d 4 r c o a t

otes: The corporate bond risk premium is the excess of current yield on debenture ond risk premium over government bonds is further categorised into two subsets ields). Highlighted banking crises are those discussed in Reinhart and Rogoff (2009 ources: See Appendix Table A1.

estraints on credit from World War II until the early 1970s (Turner, 014, pp. 181–186). This corresponds to a very stable period in ritish banking. Correspondingly, the credit growth prior to 2008, ccurred in a very lax regulatory environment with a substantial afety net. The institutional setting will therefore also be an impor- ant determinant of how banks behave and whether credit grows. owever, over the long-run, the key results in this section suggest

hat credit growth, equity returns, inflation and real rates of interest re important leading indicators of banking instability.

. Banking instability and credit

In this section, we evaluate whether banking instability in the K is associated with an increase in the credit-risk premium. The verage yield spread of UK corporate bonds over long-dated gov- rnment bonds is used as a measure of the credit risk premium and

proxy for the cost of credit intermediation (Fig. 6). As not all crises documented by Reinhart and Rogoff (2009)

ppear to be associated with significant instability in the banking ector, we begin by analysing changes in the corporate bond risk remium in two specific subsamples of banking instability: (1) all eriods highlighted as banking crises in Reinhart and Rogoff (2009) nd (2) four individual periods of the most severe instability, which re defined as events in which 10 per cent or more of all banks lost reater than 20 per cent of their market value.

Panel A of Table 6 reports average values of bank returns and ur risk premium variables in the two years before and after all UK anking crisis documented by Reinhart and Rogoff (2009) for the eriod 1830–2010. We can see that on average, bank returns fall harply in the year of these episodes, falling 9.24%. Furthermore, here is a significant increase in the corporate bond risk premium uring these episodes, suggesting a strong relationship between anking instability and higher credit costs for companies. The yield ifferential increases significantly, with an average increase of over 0% from year t−1 to t0. When we limit the sample to just the higher isk firms, which are more likely to be small, the increase in the

ost of credit is much greater in absolute terms. The risk premium f the lower risk firms increases to a much smaller degree, on aver- ge, during banking difficulties. From Panel A, it is also apparent hat for the two years following the episode, yield spreads remain

r and above Consols or high-coupon, long-dated government bonds. The corporate h Risk Premium 1/2 (highest 50% of yields) and Low Risk Premium 1/2 (lowest 50% of

persistently higher, on average, for the high-risk firms than in the years immediately before the crisis.

This evidence may corroborate the view that it is the small- est borrowers that are disproportionally hit by severe instability (Bernanke, 1983, pp. 264–265; Calomiris and Mason, 2003; Frederiksen, 1931, p.131). While these risk premium changes undoubtedly reflect normal changes in default risk across the busi- ness cycle, the relationship between large increases in this risk premium during significant banking difficulties is clear.

Panel B of Table 6 analyses the four most severe periods of insta- bility in our sample individually in order to ascertain whether these events evolve similarly, or if there is variation in how the cost of credit reacts. Panel B shows that across all periods, the corporate bond risk premium increases for the higher risk sample; however the increases are much more severe during the most recent crises. The risk premium for lower risk firms increases to a much lesser extent during these events. Interestingly, during the 1974 banking instability, while we see large increases in the risk premium of our higher risk firms, the premium of the lower risk sample decreases significantly, indicating a ‘flight to quality’. This is then reversed, to a degree, as the premium for higher risk firms falls in the years following 1974.

If we look specifically at changes in the yield differential around the recent 2007-8 banking crisis, we can see that this is by far the most severe episode of banking instability in our sample. First, the decline in bank returns is the largest of any sub-sample. From 2006 to 2008, the average corporate bond risk premium in our sample increased from 0.77% to 2.76%. This represents an increase of 258% in the risk premium firms had to pay to raise money on the bond market in the UK. This is by far the largest annual increase through- out the entire sample period. The British government and the Bank of England claimed to have opened channels of credit to individ- uals and small businesses affected by this banking crisis. Panel B indicates that these policies may have been successful in that the average cost of credit to all businesses had returned to a more normal level by 2009.

The results in Table 6 support the view that significant banking instability is accompanied by large increases in the cost of credit to businesses, and it is the riskiest/smallest firms which suffer the most and/or longest. In the remainder of this section, we examine

86 G. Campbell et al. / Journal of Financial Stability 27 (2016) 74–94

Table 6 Average variable changes during periods of banking instability.

Year (t) Average bank returns (%) Corporate bond risk premium (%) High risk premium 1/2 (%) Low risk premium 1/2 (%)

Panel A: All Documented UK Banking Crises, 1860–2010 −2 16.01 0.74 1.67 −0.19 −1 0.86 0.57 1.57 −0.44 0 −9.24 0.80 1.96 −0.34 1 22.08 0.65 1.76 −0.45 2 12.02 0.78 1.90 −0.34 Panel B: Significant UK Banking Instability: 1866, 1878, 1974, 2007–8 −2 12.39 0.97 1.18 0.75 −1 −0.81 0.92 1.14 0.71 1866 −7.96 1.14 1.44 0.85 1 −8.93 1.30 1.63 0.98 2 −0.139 1.18 1.47 0.90 −2 1.25 0.75 0.86 0.63 −1 4.62 0.92 1.11 0.72 1878 −9.67 0.93 1.13 0.73 1 −4.11 0.95 1.19 0.73 2 3.66 0.84 1.10 0.58

−2 48.98 −0.32 0.97 −1.60 −1 −27.79 −0.67 1.17 −2.51 1974 −62.85 −1.51 2.11 −5.14 1 154.64 −2.06 0.81 −4.91 2 −17.99 −1.82 0.85 −4.50 −2 5.14 0.66 1.30 0.02 −1 16.10 0.77 1.51 0.04 2007 −26.87 0.96 1.70 0.22 2008 −76.5 2.76 4.20 1.36 1 7.83 0.67 1.70 −0.36 2 9.94 0.19 1.35 −0.96

N Panel m e cor s owest S

i i t G t f (

i a w s a r a f c f

m m T m m H ( m c r s p n b

otes: Panel A includes all banking crises discussed in Reinhart and Rogoff (2009). ore of all banks lost more than 20 per cent of their market value (see Table 2). Th

ubsets of High risk premium 1/2 (highest 50% of yields) and Low risk premium 1/2 (l ources: See Appendix Table A1.

f there is a statistically significant link between overall banking nstability and the corporate bond risk premium. In order to do his, we present further results from the earlier VAR models and ranger causality tests, now focusing on how lags of all variables in

he system affect the risk premium variable, with additional results ocusing on our subsamples of low-risk firms and high-risk firms Table 7).

The null hypothesis for these tests is that the lags of variables n Columns 3–12 of Table 7 have no significant effect on the vari- ble in the ‘Dependent Variable’ column. From Row 1 of Panel A, e see that there is strong evidence that instability in the banking

ector Granger causes changes in the corporate bond risk premium t around the one to two per cent significance level (Column 3). This esult adds weight to the belief that banking instability is associ- ted with a breakdown in the ability of markets to channel funds rom lenders to borrowers. Banking instability is also a predictor of hanges in the risk premium at the one per cent significance level or both risk premium sub-groups.

It is interesting to note that lags of the returns on the equity arket have significant predictive power for the credit-risk pre- ium at the one per cent level for low-risk firms (Row 3, Panel A).

his would suggest that the general condition of corporations in the arket plays a significant role in explaining the cost of credit pre- ium that these firms face over the long run, as one would expect. owever, this effect is not present for the most risky subset of firms

Row 2, Panel A), which may suggest that changes in the risk pre- ium of this subset are actually less related to the past condition of

orporations. Similar results are evident in Panel B which presents esults using alternative measures of instability. This, again, may upport the belief that it is the smallest borrowers that are dispro-

ortionally affected by banking instability, with projects which in ormal times would be funded, frozen from credit markets due to anking difficulties (Bernanke, 1983, pp. 264–265).

B includes the years 1866, 1878, 1974 and 2007-8; periods in which 10 per cent or porate bond risk premium over government bonds is further categorised into two

50% of yields).

While banking instability itself is consistently a significant pre- dictor of changes in GDP (Appendix Table A2), we do not find evidence of significance at the 10 per cent level that changes in lags of the risk premium Granger cause direct changes in real GDP growth over the full sample. This may be because such a link is only significant during periods of extreme instability and large increases in the risk premium. An alternative explanation may be that changes in the risk premium affects GDP through a transmission mechanism via changes in the equity market, as there is a significant relationship running from risk premium to equity returns and equity returns to GDP in this model (Appendix Table A2). It also appears that any direct effect decreases further as we move to the more risky subset of firms, suggesting an impor- tance of large (less risky firms) for economic growth.

There is also evidence in Appendix Table A2 that banking insta- bility Granger causes changes in the money supply in the UK over the long run. In addition, lags of money supply have consistent predictive power for output changes, as would be suggested by the monetary transmission mechanism. This finding offers some support to the Friedman and Schwartz (1963) view that banking difficulties may exacerbate the economic condition of a country through the monetary channel.

7. Conclusions

This paper uses 181 years of bank share prices to develop a new measure of British banking instability over the long run. We find that the British banking system has faced several episodes of instability throughout this period. However, our evidence suggests that banking instability in the UK has grown more severe in the

final quarter of the twentieth century, culminating in by far the largest banking crisis in our sample in 2007–8. The ultimate roots of the 2007–8 banking difficulties, as with previous periods of banking instability, are likely to be found in regulatory failures and

G .

C a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

8 7

Table 7 Granger causality tests using risk premium sub-samples, 1860–2010.

Dependant Variable

Bank Instability (P-Value)

Risk premium (P-Value)

Equity returns (real, %) (P-Value)

Bank lending growth (%) (P-Value)

GDP growth (real, %) (P-Value)

Money supply growth (%) (P-Value)

Inflation (%) (P-Value)

Interest rate (real, %) (P-Value)

Exchange Rate (P-Value)

Public net debt growth (%) (P-Value)

Wheat price growth (%) (P-Value)

Panel A: Absolute returns Risk Premium (Full Sample)

Risk Premium 0.018 0.000 0.104 0.340 0.345 0.556 0.242 0.249 0.639 0.540 0.137

Risk Premium (High Risk 1/2)

Risk Premium 0.007 0.004 0.350 0.842 0.636 0.825 0.465 0.091 0.813 0.531 0.048

Risk Premium (Low Risk 1/2)

Risk Premium 0.002 0.007 0.004 0.169 0.326 0.133 0.000 0.000 0.784 0.058 0.199

Panel B: Drawdown Risk Premium (Full Sample)

Risk Premium 0.020 0.000 0.777 0.344 0.507 0.197 0.189 0.334 0.946 0.150 0.085

Risk Premium (High Risk 1/2)

Risk Premium 0.067 0.040 0.461 0.828 0.798 0.618 0.783 0.512 0.995 0.271 0.001

Risk Premium (Low Risk 1/2)

Risk Premium 0.022 0.000 0.028 0.657 0.507 0.056 0.009 0.019 0.998 0.297 0.095

Standard deviation of returns Risk Premium (Full Sample)

Risk Premium 0.153 0.000 0.459 0.348 0.565 0.360 0.166 0.181 0.749 0.529 0.100

Risk Premium (High Risk 1/2)

Risk Premium 0.096 0.036 0.412 0.842 0.873 0.872 0.458 0.115 0.964 0.481 0.000

Risk Premium (Low Risk 1/2)

Risk Premium 0.005 0.000 0.070 0.500 0.4351 0.068 0.000 0.0001 0.917 0.469 0.072

Nominal returns Risk Premium (Full Sample)

Risk Premium 0.250 0.000 0.539 0.523 0.486 0.320 0.016 0.202 0.962 0.117 0.105

Risk Premium (High Risk 1/2)

Risk Premium 0.440 0.104 0.900 0.762 0.801 0.759 0.687 0.663 0.998 0.164 0.000

Risk Premium (Low Risk 1/2)

Risk Premium 0.274 0.001 0.034 0.631 0.388 0.158 0.006 0.010 0.933 0.31 0.110

Notes: Each row represents the p-values from Granger causality tests following a vector autoregression. Only the results relevant to the risk premium variable are presented here. The exchange rate and both high and low risk premium variables are differenced to ensure stationarity.

8 Financ

i H n a

b o s i r i t o t o t c h

cation and Learning.

T D

N M w

8 G. Campbell et al. / Journal of

n the political economy of banking (Rajan, 2010; Calomiris and aber, 2014; Turner, 2014). While macroeconomic conditions may ot have been the main or ultimate cause, they may have played

role in triggering or exacerbating banking difficulties. In order to investigate potential macroeconomic causes of

anking instability, we use a series of vector autoregressions and ur unique dataset to establish which variables consistently act as ignificant indicators of instability over the long run. We find that nterest rates, inflation, bank lending growth, and equity prices outinely act as indicators of changes to overall banking instability n the UK over the last two centuries. This suggests that stability of hese factors should be an important policy goal. Finally, in terms f consequences of banking instability, we find long-run evidence hat banking instability consistently acts as a significant predictor

f changes in the credit-risk premium that all businesses face, par- icularly small firms. This implies that the policy response during rises should be designed with this in mind. More fundamentally, owever, given these costs of banking instability to small firms and

able A1 ata sources.

Data Year Variab

Bank Share Prices and Market Capitalisation 1830–2010 Bank r

Money Supply (M3) 1834–1869 M3 gro 1870–1982 1983–1985 1986–2010

Interest Rate 1830–2010 Real in

Dollar Exchange Rate 1830–2010 Exchan

GDP 1830–2010 Real G

CPI 1830–2003 Inflatio 2004–2010

Equity Returns 1825–1870 Real eq 1871–1899 1900–2009 2010

Consols/Government Bonds 1830–1840 Risk pr 1840–1871 1871–1913

1914–2009 2010

Corporate Bond Yield 1860–1863 Risk pr 1864–1929 1930–2002 2003–2010

Wheat Price 1820–1980 Wheat 1980–2010

Net Public Debt 1830–1950 Net pu 1951–2010

Housing Prices 1900–1945 Housin 1946–2010

Bank Lending 1860–1880 Bank l 1881–1962 1963–2010

otes: *Average deposit growth in London and Westminster, London and County and Midl anual is made available for the years 1869–1930 by the International Center for Finance hich balance sheet information is available is used as a proxy for bank lending growth 1

ial Stability 27 (2016) 74–94

the economy at large, an appropriate regulatory structure needs to be devised which lessens the probability of future crises occurring.

Acknowledgments

We are grateful to two anonymous referees for valuable com- ments. Thanks to Graeme Acheson, Mark Billings, Jagjit Chadha, David Chambers, William Janeway, Joost Jonker, Anthony Hotson, Donal McKillop, Ranald Michie, Duncan Needham and to seminar and conference participants at Cambridge University, Queen’s Uni- versity Belfast, the Scottish Economic Society conference in Perth, the Money, Macro and Finance conference in Durham and the Eco- nomic History Society conference in Warwick, for their comments. Coyle acknowledges financial support from the Department of Edu-

Appendix A

le Source

eturns Course of the Exchange (1830–1868) The Scotsman (1834-1867, 1930-1938) Belfast Newsletter (1830-1868) Investor’s Monthly Manual (1869-1929) Banking Almanac and Yearbook (1845-1930) The Times and Global Financial Data (1930–1964) Thomson Reuters Datastream (1965-2010)

wth Gregory (1936) and Holmes and Greene (1986)* Capie and Webber (1985) BIS Annual Reports (1983–1986) Bank of England Interactive Database (2011)

terest rate Officer (2011b)

ge rate Officer (2011a)

DP growth Officer (2011c)

n O’Donoghue (2004) ONS

uity returns Acheson and Turner (2008) Grossman (2002) Dimson et al. (2011) Barclays Equity Gilt Study (2011)

emium Mitchell (1988) Course of the Exchange (1840-1871) Annual Statistical Abstract for the United Kingdom (1871-1913) Dimson et al. (2011) Barclays Equity Gilt Study (2011)

emium Course of the Exchange (1860-63) Investor’s Monthly Manual (1864–1929)** Stock Exchange Daily Official List (1930–2002) Bloomberg (2003)

price growth Mitchell (1988) London Gazette (1980-2010)

blic debt growth Mitchell (1988) National Income and Expenditure (1956–1983), Public ExpenditureStatistical Analyses (1998–2010)

g price growth Liesner (1989) Department for Communities and Local Govt. (2011)

ending growth The Bankers Magazine (1859-1881)*** Sheppard (1971) Bank of England Interactive Database (2011)

and banks used as proxy for money supply growth 1834–1969. ** Investor’s Monthly (ICF) at Yale University. ***Average lending growth of the twenty largest banks for 860–1881.

G .

C a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

8 9

Table A2 Granger causality tests of macroeconomic indicators of UK banking instability measured using absolute returns.

Dependant Variable

Bank Instability (p-value)

Risk premium (p-value)

Equity returns (real, %) (p-value)

Bank lending growth (%) (p-value)

GDP growth (real, %) (p-value)

Money supply growth (%) (p-value)

Inflation (%) (p-value)

Interest rate (real, %) (p-value)

Exchange Rate (p-value)

Public net debt growth (%) (p-value)

Wheat price growth (%) (p-value)

House price growth (%) (p-value)

Panel A: 1860–2010 Bank Instability 0.047 0.581 0.023 0.010 0.899 0.021 0.002 0.000 0.298 0.029 0.146 Risk Premium 0.018 0.000 0.104 0.340 0.345 0.556 0.242 0.249 0.639 0.540 0.137 GDP 0.029 0.903 0.052 0.934 0.002 0.072 0.145 0.216 0.397 0.041 0.042 M3 0.059 0.399 0.252 0.018 0.000 0.000 0.241 0.398 0.000 0.003 0.059 Inflation 0.020 0.027 0.006 0.003 0.001 0.151 0.000 0.001 0.000 0.000 0.094 Interest 0.060 0.032 0.006 0.008 0.002 0.131 0.000 0.000 0.000 0.000 0.078 Exchange Rate 0.160 0.371 0.331 0.257 0.171 0.198 0.971 0.623 0.068 0.485 0.225 Public Debt 0.157 0.193 0.050 0.004 0.456 0.280 0.337 0.188 0.005 0.000 0.129 Wheat Price 0.030 0.429 0.278 0.288 0.057 0.130 0.213 0.026 0.000 0.183 0.013 Equity 0.862 0.088 0.118 0.000 0.073 0.997 0.312 0.137 0.342 0.263 0.231 Lending 0.131 0.028 0.402 0.939 0.119 0.031 0.277 0.458 0.201 0.221 0.565

Panel B: 1835–2010 Bank Instability 0.041 0.013 0.832 0.083 0.009 0.000 0.945 0.192 0.240 Risk Premium GDP 0.074 0.017 0.095 0.309 0.230 0.299 0.293 0.044 0.132 M3 0.094 0.617 0.055 0.000 0.281 0.130 0.031 0.000 0.104 Inflation 0.740 0.750 0.012 0.355 0.000 0.000 0.059 0.000 0.333 Interest 0.763 0.908 0.020 0.345 0.001 0.000 0.060 0.000 0.370 Exchange Rate 0.521 0.550 0.274 0.031 0.986 0.748 0.000 0.882 0.234 Public Debt 0.166 0.048 0.744 0.484 0.230 0.093 0.701 0.000 0.230 Wheat Price 0.179 0.911 0.024 0.260 0.420 0.028 0.048 0.366 0.044 Equity 0.968 0.328 0.218 0.937 0.023 0.019 0.152 0.832 0.162

Panel C: 1901–2010 Bank Instability 0.039 0.845 0.035 0.017 0.732 0.088 0.009 0.000 0.754 0.036 0.149 0.168 Risk Premium 0.048 0.000 0.255 0.300 0.524 0.331 0.486 0.436 0.991 0.561 0.133 0.278 GDP 0.010 0.651 0.100 0.958 0.001 0.096 0.027 0.053 0.761 0.071 0.076 0.062 M3 0.023 0.184 0.176 0.000 0.001 0.000 0.086 0.115 0.059 0.005 0.231 0.291 Inflation 0.087 0.030 0.005 0.000 0.000 0.135 0.000 0.005 0.254 0.000 0.097 0.784 Interest 0.114 0.036 0.002 0.000 0.001 0.066 0.001 0.000 0.262 0.000 0.076 0.881 Exchange Rate 0.090 0.450 0.052 0.001 0.073 0.259 0.761 0.188 0.000 0.690 0.276 0.044 Public Debt 0.223 0.295 0.117 0.012 0.949 0.306 0.748 0.640 0.712 0.000 0.191 0.516 Wheat Price 0.025 0.262 0.193 0.231 0.109 0.367 0.046 0.001 0.722 0.159 0.012 0.448 Equity 0.574 0.079 0.018 0.002 0.067 1.000 0.073 0.031 0.546 0.103 0.269 0.229 Lending 0.260 0.045 0.693 0.795 0.120 0.069 0.053 0.115 0.010 0.151 0.481 0.110 House Prices 0.314 0.678 0.599 0.000 0.134 0.018 0.011 0.016 0.467 0.380 0.011 0.016

Notes: Each row represents the p-values from Granger causality tests. The exchange rate is differenced to ensure stationarity.

9 0

G

. C

a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

Table A3 Granger causality tests of macroeconomic indicators of UK banking instability measured using drawdown.

Dependant Variable

Bank Instability (p-value)

Risk premium (p-value)

Equity returns (real, %) (p-value)

Bank lending growth (%) (p-value)

GDP growth (real, %) (p-value)

Money supply growth (%) (p-value)

Inflation (%) (p-value)

Interest rate (real, %) (p-value)

Exchange Rate (p-value)

Public net debt growth (%) (p-value)

Wheat price growth (%) (p-value)

House price growth (%) (p-value)

Panel A: 1860–2010 Bank Instability 0.001 0.770 0.083 0.016 0.638 0.919 0.008 0.072 0.043 0.681 0.376 Risk Premium 0.020 0.000 0.777 0.344 0.507 0.197 0.189 0.334 0.946 0.150 0.085 GDP 0.663 0.960 0.078 0.926 0.008 0.087 0.005 0.031 0.443 0.043 0.047 M3 0.093 0.534 0.009 0.023 0.000 0.000 0.141 0.188 0.000 0.022 0.139 Inflation 0.067 0.100 0.112 0.003 0.002 0.082 0.000 0.002 0.000 0.000 0.088 Interest 0.065 0.099 0.119 0.008 0.005 0.058 0.000 0.000 0.000 0.000 0.084 Exchange Rate 0.133 0.492 0.081 0.298 0.215 0.260 0.842 0.947 0.053 0.777 0.328 Public Debt 0.039 0.151 0.074 0.016 0.680 0.128 0.009 0.001 0.005 0.000 0.039 Wheat Price 0.219 0.869 0.448 0.312 0.160 0.156 0.968 0.104 0.002 0.117 0.016 Equity 0.012 0.616 0.117 0.001 0.090 0.309 0.031 0.006 0.131 0.177 0.144 Lending 0.236 0.087 0.742 0.915 0.093 0.031 0.554 0.788 0.244 0.245 0.583

Panel B: 1835–2010 Bank Instability 0.000 0.118 0.639 0.160 0.049 0.069 0.266 0.914 0.382 Risk Premium GDP 0.840 0.193 0.114 0.446 0.047 0.203 0.289 0.034 0.057 M3 0.048 0.641 0.034 0.000 0.414 0.037 0.025 0.001 0.182 Inflation 0.056 0.958 0.009 0.210 0.000 0.000 0.052 0.000 0.352 Interest 0.029 0.960 0.014 0.181 0.003 0.000 0.054 0.000 0.398 Exchange Rate 0.077 0.441 0.289 0.109 0.476 0.715 0.001 0.671 0.212 Public Debt 0.065 0.025 0.785 0.848 0.013 0.008 0.438 0.000 0.157 Wheat Price 0.360 0.537 0.047 0.368 0.540 0.185 0.049 0.301 0.052 Equity 0.004 0.037 0.384 0.257 0.012 0.002 0.164 0.828 0.158

Panel C: 1901–2010 Bank Instability 0.006 0.651 0.066 0.241 0.241 0.729 0.000 0.015 0.197 0.621 0.360 0.623 Risk Premium 0.000 0.000 0.852 0.382 0.619 0.079 0.353 0.493 0.830 0.282 0.061 0.008 GDP 0.658 0.729 0.081 0.788 0.003 0.080 0.037 0.077 0.638 0.066 0.102 0.023 M3 0.050 0.179 0.011 0.000 0.000 0.000 0.396 0.436 0.062 0.036 0.464 0.710 Inflation 0.289 0.113 0.042 0.000 0.000 0.094 0.001 0.011 0.126 0.000 0.120 0.429 Interest 0.274 0.145 0.007 0.000 0.001 0.043 0.002 0.000 0.118 0.000 0.109 0.870 Exchange Rate 0.163 0.610 0.010 0.001 0.413 0.181 0.739 0.375 0.000 0.309 0.153 0.048 Public Debt 0.048 0.148 0.147 0.086 0.870 0.117 0.091 0.001 0.678 0.000 0.061 0.521 Wheat Price 0.280 0.645 0.381 0.381 0.224 0.515 0.254 0.002 0.201 0.110 0.016 0.095 Equity 0.025 0.687 0.062 0.009 0.087 0.425 0.001 0.000 0.499 0.095 0.168 0.268 Lending 0.549 0.144 0.875 0.816 0.089 0.118 0.057 0.152 0.010 0.213 0.395 0.181 House Prices 0.275 0.961 0.531 0.000 0.378 0.023 0.003 0.002 0.502 0.350 0.007 0.025

Notes: Each row represents the p-values from Granger causality tests. The exchange rate is differenced to ensure stationarity.

G .

C a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

9 1

Table A4 Granger causality tests of macroeconomic indicators of UK banking instability measured using nominal returns.

Dependant Variable

Bank Instability (p-value)

Risk premium (p-value)

Equity returns (real, %) (p-value)

Bank lending growth (%) (p-value)

GDP growth (real, %) (p-value)

Money supply growth (%) (p-value)

Inflation (%) (p-value)

Interest rate (real, %) (p-value)

Exchange Rate (p-value)

Public net debt growth (%) (p-value)

Wheat price growth (%) (p-value)

House price growth (%) (p-value)

Panel A: 1860–2010 Bank Instability 0.473 0.196 0.167 0.017 0.659 0.888 0.000 0.001 0.075 0.139 0.151 Risk Premium 0.250 0.000 0.539 0.523 0.486 0.320 0.016 0.202 0.962 0.117 0.105 GDP 0.725 0.966 0.195 0.898 0.061 0.074 0.060 0.061 0.462 0.049 0.042 M3 0.035 0.152 0.062 0.011 0.000 0.000 0.633 0.499 0.000 0.004 0.202 Inflation 0.100 0.201 0.033 0.006 0.002 0.265 0.000 0.001 0.000 0.000 0.071 Interest 0.091 0.209 0.043 0.013 0.005 0.228 0.001 0.000 0.000 0.000 0.057 Exchange Rate 0.438 0.567 0.376 0.260 0.236 0.266 0.768 0.980 0.058 0.945 0.319 Public Debt 0.207 0.195 0.445 0.019 0.531 0.153 0.015 0.004 0.006 0.000 0.074 Wheat Price 0.850 0.906 0.767 0.314 0.134 0.158 0.937 0.173 0.002 0.091 0.017 Equity 0.174 0.265 0.617 0.001 0.036 0.642 0.012 0.009 0.130 0.127 0.132 Lending 0.191 0.035 0.963 0.911 0.107 0.048 0.661 0.759 0.277 0.181 0.597

Panel B: 1834–2010 Bank Instability 0.591 0.165 0.874 0.358 0.000 0.001 0.174 0.494 0.253 Risk Premium GDP 0.883 0.247 0.120 0.328 0.106 0.209 0.323 0.037 0.064 M3 0.354 0.933 0.050 0.000 0.573 0.103 0.055 0.000 0.266 Inflation 0.142 0.339 0.011 0.361 0.000 0.000 0.058 0.000 0.285 Interest 0.077 0.498 0.018 0.339 0.006 0.000 0.061 0.000 0.331 Exchange Rate 0.147 0.376 0.328 0.079 0.252 0.909 0.000 0.620 0.186 Public Debt 0.130 0.272 0.894 0.678 0.087 0.022 0.726 0.000 0.178 Wheat Price 0.894 0.863 0.049 0.228 0.370 0.287 0.040 0.292 0.062 Equity 0.144 0.845 0.191 0.706 0.001 0.002 0.154 0.719 0.216

Panel C: 1901–2010 Bank Instability 0.567 0.043 0.146 0.049 0.416 0.997 0.000 0.000 0.645 0.124 0.086 0.556 Risk Premium 0.027 0.000 0.155 0.626 0.660 0.138 0.051 0.439 0.742 0.134 0.041 0.009 GDP 0.860 0.737 0.560 0.863 0.003 0.121 0.065 0.109 0.623 0.080 0.081 0.050 M3 0.009 0.040 0.013 0.000 0.000 0.000 0.817 0.869 0.032 0.009 0.591 0.642 Inflation 0.427 0.103 0.103 0.000 0.000 0.229 0.000 0.009 0.077 0.000 0.086 0.542 Interest 0.482 0.098 0.072 0.000 0.001 0.127 0.005 0.000 0.078 0.000 0.068 0.899 Exchange Rate 0.481 0.646 0.183 0.000 0.231 0.141 0.983 0.443 0.000 0.406 0.186 0.051 Public Debt 0.219 0.266 0.400 0.110 0.957 0.209 0.556 0.180 0.871 0.000 0.127 0.702 Wheat Price 0.980 0.692 0.827 0.278 0.180 0.419 0.395 0.003 0.227 0.101 0.020 0.220 Equity 0.314 0.195 0.342 0.005 0.050 0.918 0.008 0.004 0.482 0.091 0.141 0.366 Lending 0.398 0.055 0.979 0.768 0.094 0.130 0.166 0.250 0.010 0.162 0.397 0.221 House Prices 0.409 0.683 0.492 0.000 0.330 0.031 0.003 0.004 0.401 0.237 0.008 0.053

Notes: Each row represents the p-values from Granger causality tests. The exchange rate is differenced to ensure stationarity.

9 2

G

. C

a m

p b

ell et

a l.

/ Jo

u rn

a l

o f

Fin a n

cia l

Sta b

ility 2

7 (2

0 1

6 )

7 4

– 9

4

Table A5 Granger causality tests of macroeconomic indicators of UK banking instability measured using standard deviation of returns.

Dependant Variable

Bank Instability (p-value)

Risk premium (p-value)

Equity returns (real, %) (p-value)

Bank lending growth (%) (p-value)

GDP growth (real, %) (p-value)

Money supply growth (%) (p-value)

Inflation (%) (p-value)

Interest rate (real, %) (p-value)

Exchange Rate (p-value)

Wheat price growth (%) (p-value)

House price growth (%) (p-value)

Panel A: 1860–2010 Bank Instability 0.000 0.100 0.470 0.039 0.574 0.044 0.003 0.000 0.016 0.336 0.086 Risk Premium 0.153 0.000 0.459 0.348 0.565 0.360 0.166 0.181 0.749 0.529 0.100 GDP 0.004 0.735 0.075 0.863 0.010 0.063 0.366 0.603 0.448 0.070 0.059 M3 0.903 0.527 0.066 0.035 0.001 0.000 0.486 0.530 0.000 0.020 0.092 Inflation 0.066 0.022 0.028 0.006 0.001 0.249 0.000 0.001 0.001 0.000 0.099 Interest 0.068 0.020 0.037 0.013 0.004 0.213 0.000 0.000 0.001 0.000 0.072 Exchange Rate 0.130 0.356 0.514 0.307 0.222 0.222 0.862 0.973 0.030 0.587 0.504 Public Debt 0.128 0.162 0.119 0.004 0.486 0.173 0.238 0.164 0.004 0.000 0.028 Wheat Price 0.006 0.409 0.596 0.276 0.166 0.203 0.214 0.015 0.000 0.182 0.031 Equity 0.784 0.098 0.129 0.001 0.062 0.985 0.336 0.176 0.354 0.206 0.146 Lending 0.021 0.033 0.341 0.926 0.135 0.027 0.644 0.624 0.223 0.149 0.415

Panel B: 1834–2010 Bank Instability 0.000 0.486 0.512 0.752 0.031 0.000 0.174 0.093 0.063 Risk Premium GDP 0.018 0.051 0.146 0.250 0.291 0.386 0.328 0.074 0.119 M3 0.262 0.054 0.070 0.000 0.570 0.194 0.128 0.002 0.268 Inflation 0.293 0.582 0.010 0.381 0.000 0.000 0.047 0.000 0.458 Interest 0.359 0.818 0.018 0.374 0.004 0.000 0.048 0.000 0.513 Exchange Rate 0.242 0.600 0.378 0.051 0.700 0.946 0.000 0.940 0.436 Public Debt 0.154 0.111 0.891 0.556 0.196 0.112 0.876 0.000 0.116 Wheat Price 0.010 0.628 0.036 0.260 0.439 0.038 0.006 0.318 0.071 Equity 0.335 0.205 0.200 0.954 0.076 0.042 0.073 0.775 0.143

Panel C: 1901–2010 Bank Instability 0.000 0.335 0.475 0.152 0.451 0.002 0.005 0.001 0.695 0.222 0.098 0.094 Risk Premium 0.455 0.000 0.739 0.330 0.721 0.333 0.373 0.390 0.932 0.613 0.083 0.259 GDP 0.002 0.494 0.217 0.833 0.010 0.020 0.060 0.159 0.831 0.094 0.174 0.085 M3 0.928 0.444 0.235 0.002 0.002 0.000 0.258 0.189 0.058 0.051 0.313 0.508 Inflation 0.433 0.042 0.022 0.000 0.000 0.272 0.001 0.008 0.148 0.000 0.115 0.708 Interest 0.306 0.034 0.006 0.000 0.001 0.168 0.000 0.000 0.177 0.000 0.087 0.923 Exchange Rate 0.116 0.730 0.054 0.000 0.395 0.281 0.876 0.616 0.000 0.450 0.415 0.115 Public Debt 0.157 0.292 0.146 0.011 0.980 0.346 0.376 0.156 0.746 0.000 0.060 0.575 Wheat Price 0.009 0.388 0.505 0.304 0.219 0.434 0.130 0.002 0.324 0.156 0.043 0.148 Equity 0.178 0.049 0.025 0.001 0.019 0.910 0.164 0.062 0.422 0.050 0.109 0.184 Lending 0.032 0.040 0.678 0.750 0.078 0.066 0.412 0.300 0.002 0.101 0.315 0.530 House Prices 0.041 0.625 0.637 0.000 0.536 0.014 0.002 0.005 0.604 0.477 0.013 0.153

Notes: Each row represents the p-values from Granger causality tests. The exchange rate is differenced to ensure stationarity.

Financ

R

A

A A A

B

B B B B B B

B

B

B

B

B B

B

B

B

B

B B

C

C

C

C

C

C

C C

C

C

C

C C C

D D

D

D

D

D

D

D

D

G. Campbell et al. / Journal of

eferences

cheson, G.G., Turner, J.D., 2008. The death blow to unlimited liability in victorian britain: the city of glasgow failure. Explor. Econ. Hist. 45, 235–253.

lchian, A.A., Kessel, R.A., 1962. Effects of inflation. J. Political Econ. 70, 521–537. llen, F., Gale, D., 2000. Bubbles and crises. Oxf. Rev. Econ. Policy 15, 9–18. ltman, E., Gande, A., Saunders, A., 2010. Informational efficiency of loans versus

bonds: evidence from secondary market prices. J. Money Credit Bank. 42, 755–767.

aker, M., Collins, M., 1999. Financial crises and structural change in english commercial bank assets, 1860–1913. Explor. Econ. Hist. 36, 428–444.

ank for International Settlements, BIS Annual Report, 1983–1986. ank of England Interactive Database, 2011. anking Almanac and Yearbook, 1845–1930. arclays Equity Gilt Study, 2011. elfast Newsletter, 1830–68. ergman, M.U., Hansen, J., 2002. Financial Instability and Monetary Policy: The

Swedish Evidence. Sveriges Riksbank Working Paper Series, 137. ernanke, B., Gertler, M., 1989. Agency costs net worth, and business fluctuations.

Am. Econ. Rev. 79, 14–31. ernanke, B., Gertler, M., 1990. Financial fragility and economic performance. Q. J.

Econ. 105, 87–114. ernanke, B., Gertler, M., 1995. Inside the Black Box: The Credit Channel of

Monetary Policy Transmission. NBER Working Papers, 5146. ernanke, B., 1983. Nonmonetary effects of the financial crisis in the propagation

of the great depression. Am. Econ. Rev. 73, 257–276. loomberg L.P., 2003–2010. ordo, M.D., Haubrich, J.G., 2010. Credit crises, money and contractions: an

historical view. J. Monet. Econ. 57 (1), 1–18. ordo, M., Eichengreen, B., Klingebiel, D., Martinez Peria, S.M., 2001. Is the crisis

problem growing more severe? Econ. Policy 24, 51–82. ordo, M.D., Dueker, M.J., Wheelock, D.C., 2003. Aggregate price shocks and

financial stability: the United Kingdom 1796–1999. Explor. Econ. Hist. 40, 143–169.

orensztein, E., Lee, J-.W., 2000. Financial Crisis and Credit Crunch in Korea-Evidence from Firm-Level Data. IMF Working Papers, 00/25.

orio, C., Lowe, P., 2002. Asset Prices, Financial and Monetary Stability: Exploring the Nexus, Bank for International Settlements Working Paper 114.

urdetts Official Intelligence, 1882–98. utt, N., Pugh, A., 2014. Credit spreads: capturing credit conditions facing

households and firms. Bank Engl. Q. Bull. 54, 137–147. alomiris, C.W., Haber, S.H., 2014. Fragile by Design: The Political Origins of

Banking Crises and Scarce Credit. Princeton University Press, Princeton. alomiris, C.W., Mason, J.R., 2003. Consequences of bank distress during the great

depression. Am. Econ. Rev. 93, 937–947. apie, F., Webber, A., 2010. A Monetary History of the United Kingdom:

1870–1982. Routledge, London. apie, F., Wood, G., 2012. Money Over Two Centuries: Selected Topics in British

Monetary History. Oxford University Press, Oxford. apie, F., 2010. The Bank of England, 1950 to 1979. Cambridge University Press,

Cambridge. aprio, G., Klingebiel, D., 1996. Bank Insolvency: Bad Luck, Bad Policy, or Bad

Banking? in Annual World Bank Conference on Development Economics. World Bank, Washington D.C.

entral Statistical Office, National Income and Expenditure (1956-1983). hristiano, L.J., Motto, R., Rostagno, M., 2010. Financial Factors in Economic

Fluctuations. European Central Bank Working Paper Series 1192. hurm, R., Webber, L., 2007. Decomposing corporate bond spreads. Bank Engl. Q.

Bull. 47, 533–541. lapham, J.H., 1944. The Bank of England: A History, 2 vol. Cambridge University

Press, Cambridge. ollins, M., Baker, M., 2003. Commercial Banks and Industrial Finance in England

and Wales, 1860–1913. Oxford University Press, Oxford. otter, J., 2004. Absolute Return Volatility Geary WP2004/15. ourse of the Exchange, 1840–71. oyle, C., Turner, J.D., 2013. Law: politics and financial development: the great

reversal of the U.K. corporate debt market. J. Econ. Hist. 73, 810–846. atastream, 1965–2010. avidian, M., Carroll, R.J., 1987. Variance function estimation. J. Am. Stat. Assoc. 82,

1079–1091. avis, E.P., Karim, D., 2008. Comparing early warning systems for banking crises. J.

Financ. Stab. 4, 89–120. eaton, A., 2012. The financial crisis and the well-being of americans. Oxf. Econ.

Pap. 64, 1–26. emirgüç -Kunt, A., Detragiache, E., 1998. Financial Liberalization and Financial

Fragility. International Monetary Fund Working Paper, WP/98/83. emirgüç -Kunt, A., Detragiache, E., 1998b. The determinants of banking crises in

developing and developed countries. IMF Staff Pap. 45 (1). emirgüç -Kunt, A., Detragiache, E., 1999. Monitoring Banking Sector Fragility: A

Multivariate Logit Approach with an Application to the 1996/97 Banking Crises. IMF Mimeo.

emirgüç -Kunt, A., Detragiache, E., Gupta, P., 2006. Inside the crises: an empirical analysis of banking systems in distress. J. Int. Money Finance 25, 702–718.

epartment for Communities and Local Government, 2011. Housing Market: House Prices from 1930, Annual House Price Inflation, United Kingdom, from 1970. London: DCLG.

ial Stability 27 (2016) 74–94 93

Dimson, E., Marsh, P., Staunton, M., 2011. Credit Suisse Global Investment Sourcebook 2010. Credit Suisse, Zurich.

Ding, Z., Granger, C., Engle, R., 1993. A long memory property of stock market returns and a new model. Empir. Finance 1, 83–106.

Ding, W., Domac, I., Ferri, G., 1998. Is there a credit crunch in East Ssia? In: Policy Research Working Paper Series. World Bank.

Eichengreen, B., Arteta, C., 2000. Banking Crises in Emerging Markets: Presumptions and Evidence, Centre for International Development Economics Research Working Paper. Haas School of Business, pp. 115.

Flamini, A., Milas, C., 2015. Distribution forecast targeting in an open-economy: macroeconomic volatility and financial implications. J. Financ. Stab. 16, 89–105.

Forsberg, L., Ghysels, E., 2007. Why do absolute returns predict volatility so well? J. Financ. Econom. 5, 31–67.

Frederiksen, D.M., 1931. Two financial roads leading out of depression. Harv. Bus. Rev. 10, 131–148.

Friedman, M., Schwartz, A.J., 1963. A Monetary History of the United States, 1867–1960. Princeton University Press for NBER, Princeton, N.J.

Gavin, M., Hausmann, R., 1996. The Roots of Banking Crises: the Macroeconomic Context. Inter-American Bank Working Paper 318, 1–20.

Global Financial Data, 1930–1964. González-Hermosillo, B., 1999. Determinants of Ex-ante Banking System Distress:

A Macro-Micro Empirical Exploration of Some Recent Episodes. IMF Working paper, 99/33.

Goodhart, C., Delargy, P.J.R., 1998. Financial Crises: Plus Ca Change, Plus Cest La Meme Chose. Int. Finance 1, 261–287.

Gourinchas, P.-O., Valdes, R., Landerretche, O., 2001. Lending booms: latin america and the world. Econ. J. Latin Am. Caribb. Econ. Assoc. 1, 47–89.

Granger, C., Sin, C., 2000. Modelling the absolute returns of different stock market indices: exploring the forecastability of an alternative measure of risk. J. Forecasting 19, 277–298.

Gregory, T.E., 1936. The Westminster Bank Through a Century, 2. Oxford University Press.

Grossman, R.S., 1994. The shoe that didn’t drop: explaining banking stability during the great depression. J. Econ. Hist. 53, 654–682.

Grossman R.S., 2002. New indices of British equity prices, 1870-1913, J. Econ. Hist. 62, 121-46.

Grossman, R.S., 2010. Unsettled Account: The Evolution of Banking in the Industrialised World Since 1800. Princeton University Press, Princeton.

Hall, R.E., 2010. Why does the economy fall to pieces after a financial crisis? J. Econ. Perspect. 24, 3–20.

Hardy, D., Pazarbasioglu, C., 1998. Leading Indicators of Banking Crises: Was Asia Different? IMF Working Paper no. WP/98/91 (June).

Hoggarth, G., Reis, R., Saporta, V., 2002. Costs of banking system instability: some empirical evidence. J. Bank. Finance 26, 825–855.

Holmes, A.R., Greene, E., 1986. Midland: 150 Years of Banking Business, London. Holmstrom, B., Tirole, J., 1997. Financial intermediation loanable funds, and the

real sector. Q. J. Econ. 112, 663–691. HMTreasury, Public Expenditure Statistical Analyses (1998-2010). International Monetary Fund, 1998. World Economic Outlook, Washington, D.C.:

IMF (May). Investors Monthly Manual, 1864–1929. Jefferys, J.B., 1977. Business Organisation in Great Britain 1856–1914. Arno Press,

New York. Jordà, Ò., Schularick, M., Taylor, A.M., 2011. Financial crises, credit booms, and

external imbalances: 140 years of lessons. IMF Econ. Rev. 59, 340–378. Jordà, Ò., Schularick, M., Taylor, A.M., 2014. Betting the house. J. Int. Econ. 96, 2–18. Kaminsky, G.L., Reinhart, C.M., 1999. The twin crises: the causes of banking and

balance-of-payments problems. Am. Econ. Rev. 89, 473–500. Kaufman, G., 1999. Banking and Currency Crises and Systemic Risk: A Taxonomy

and Review. Loyola University Chicago and Federal Reserve Bank of Chicago mimeo.

Kiyotaki, N., Moore, J., 1997. Credit cycles. J. Political Econ. 105, 211–248. Krishnamurthy, A., 2010. How debt markets have malfunctioned in the crisis. J.

Econ. Perspect. 24, 3–28. Laeven, L., Valencia, F., 2010. Resolution of Banking Crises: The Good, the Bad, and

the Ugly. IMF Working Paper WP/10/146. Laeven, L., 2011. Banking crises: a review. Annu. Rev. Financ. Econ. 3, 17–40. Liesner, T., 1989. One Hundred Years of Economic Statistics: United Kingdom,

United States of America, Australia, Canada, France, Germany, Italy, Japan, Sweden. London: Economist Publications.

Logan, A., 2001. The UK Small Banks Crisis of the Early 1990: What Were the Leading Indicators of Failure? Bank of England Working Paper.

London Gazette, 1980–2010. McKinnon, R.I., Pill, H., 1997. Credible economic liberalizations and over

borrowing. Am. Econ. Rev. 87, 189–193. Minsky, H.P., 1977. The financial instability hypothesis: an interpretation of keynes

and an alternative to standard theory. Nebraska J. Econ. Bus. 16, 5–16. Mishkin, F.S., 1991. Asymmetric information and financial crises: a historical

perspective. In: Hubbard, G.R. (Ed.), Financial Markets and Financial Crises. University of Chicago Press Chicago, pp. 69–108.

Mishkin, F.S., 2009. Is monetary policy effective during financial crises. Am. Econ.

Rev. 99, 573–577.

Mitchell, B.R., 1988. British Historical Statistics. Cambridge University Press, Cambridge.

O’Donoghue, J., Goulding, L., Allen, G., 2004. Composite price index, 1750–2003. Econ. Trends 604, 38–46.

9 Financ

O O O R

R

R

R

R

S

S

Working Papers 12/163, International Monetary Fund. Van den End, J.W., 2006. Indicator and Boundaries of Financial Stability. DNB

Working Paper, no. 97. Wong, J., Wong, T.-C., Leung, P., 2010. Predicting banking distress in the EMEAP

4 G. Campbell et al. / Journal of

fficer, L.H., 2011. Dollar-Pound Exchange Rate from 1791, MeasuringWorth. fficer, L.H., 2011. What Was the Interest Rate Then? MeasuringWorth. fficer, L.H., 2011. What Was the U.K. GDP Then? MeasuringWorth. ait, R.S., 1930. The History of the Union Bank of Scotland. John Smith and Son,

Glasgow. ajan, R.G., 2010. Fault Lines: How Hidden Fractured Still Threaten the World

Economy. Princeton University Press, Princeton. eid, M., 1982. The Secondary Banking Crisis 1973–75: Its Causes and Course.

Macmillan, London. einhart, C.M., Rogoff, K.S., 2008. Banking Crises: An Equal Opportunity Menace.

NBER Working Paper, 14587. einhart, C., Rogoff, K., 2009. This Time Is Different. Eight Centuries of Financial

Folly. University Press Princeton, NJ, US, Princeton.

ayers, R.S., 1976. The Bank of England: 1891–1944, 2 vol. Cambridge University

Press, Cambridge. chularick, M., Taylor, A.M., 2012. Credit booms gone bust: monetary policy,

leverage cycles, and financial crises, 1870–2008. Am. Econ. Rev. 102, 1029–1061. Sheppard, D.K., 1971. The Growth and Role of UK Financial

ial Stability 27 (2016) 74–94

Institutions 1880–1962. Methue, London. Stock Exchange Official Intelligence, 1899–1933. Stock Exchange Daily Official List, 1930–2002. The Annual Statistical Abstract for the United Kingdom, 1871–1913. The Bankers Magazine, 1859–1881. The Scotsman, 1834–1867, 1930-38. Turner, J., 2014. Banking in Crisis: The Rise and Fall of British Banking Stability,

1800 to the Present. Cambridge University Press, Cambridge. Valencia, F., Laeven, L., 2012. Systemic Banking Crises Database: An Update. IMF

economies. J. Financ. Stab. 6, 169–179.