the credit-anstalt myth - London School of Economics and Political

Flora Macher
London School of Economics
[email protected]
THE CREDIT-ANSTALT MYTH
R EVISITING THE CAUSES OF THE H UNGARIAN F INANCIAL CRISIS OF 1931
The mainstream view on the Central European crises of 1931 is that they resulted from the
flight of foreign financiers who became wary of these countries’ ability to repay their large debt
pile. When fear became reality with the fall of the Austrian Credit-Anstalt, foreign creditors and
depositors started withdrawing their capital from the whole region. By using a rich
macroeconomic dataset, a bank-by-bank database and relying on an early-warning indicator
analysis, this paper puts this widely-held view under the microscope for the case of Hungary. I
find that, in contrast with the prevailing view, domestic structural problems and the flight of
domestic financiers brought about the calamities of 1931. The Hungarian crisis of 1931 was
caused by a currency crisis in late 1928 and an agricultural crisis of 1930. The 1928 currency
crisis motivated the central bank to pursue a restrictive monetary policy which squeezed
liquidity out of the economy at a time when foreign financing was drying out and the economy
was sinking into recession. The agricultural crisis of 1930 reduced incomes and consumption
and induced a wave of defaults in the economy. Due to these domestic structural problems, the
banking sector was increasingly vulnerable to shocks. The shock came in October 1930 in the
form of political events, and in response, domestic depositors started slowly retreating from
Budapest banks. The banking crisis became nationwide in August 1931 when the further
decline of grain prices created a new wave of defaults. The banking crisis then activated a
currency crisis and hence, the Hungarian crisis turned into a twin crisis. The collapse of the
Credit-Anstalt and foreign depositors’ retreat thus had a contribution of second order
importance to this episode.
1
Flora Macher
London School of Economics
[email protected]
TABLE OF CONTENTS
Table of figures............................................................................................................................................................................ 3
Introduction..................................................................................................................................................................................4
Section I - Literature review ................................................................................................................................................8
I.1 – Three generations of financial crisis modeling...........................................................................................8
I.2 – The mainstream interpretation of the Hungarian crisis of 1931 ......................................................9
Section II – A new interpretation of the Hungarian crisis of 1931 ............................................................... 14
II.1 - Defining financial, banking and currency crises..................................................................................... 14
II.2 - What do banking and currency crisis measures indicate?................................................................ 15
II.3 - The currency crisis of late 1928...................................................................................................................... 17
II.4 - The agricultural crisis of 1930 ......................................................................................................................... 22
II.5 - The twin crisis of 1931 ........................................................................................................................................ 27
Section III - Early-warning indicator analysis ......................................................................................................... 31
III.1 - Identifying financial crises in Hungary between 1927 and 1931 ................................................ 31
III.2 - Defining early-warning indicators ............................................................................................................... 35
III.3 - Interpreting the signals ..................................................................................................................................... 37
III.3.i - Financial sector indicators ....................................................................................................................... 37
III.3.ii - Balance-of-payments indicators .......................................................................................................... 38
III.3.iii - Indicators of the real economy ............................................................................................................ 39
III.3.iv - Fiscal policy indicator............................................................................................................................... 39
III.3.v – The main findings of the early-warning indicator analysis................................................... 39
Section IV - Crisis triggers .................................................................................................................................................. 41
Section V - Conclusion .......................................................................................................................................................... 44
References and sources ....................................................................................................................................................... 45
References ............................................................................................................................................................................. 45
Archival sources ................................................................................................................................................................. 47
Sources of figures .............................................................................................................................................................. 48
Sources of early-warning indicators........................................................................................................................ 51
Appendix 1 ................................................................................................................................................................................. 52
Appendix 2 ................................................................................................................................................................................. 53
2
Flora Macher
London School of Economics
[email protected]
TABLE OF FIGURES
Figure 1 The framework of the Hungarian economy ........................................................................................... 11
Figure 2 The mechanism behind the arguments of the Hungarian literature ........................................ 12
Figure 3 The Exchange Market Pressure (EMP) index........................................................................................ 15
Figure 4 Total assets under insolvency as a share of banks' deposits ........................................................ 16
Figure 5 The monthly change in deposits for Budapest and non-Budapest banks .............................. 16
Figure 6 Hungary's balance-of-payments .................................................................................................................. 18
Figure 7 Hungary's trade account .................................................................................................................................. 18
Figure 8 The central bank's rediscount and the gold cover ............................................................................. 19
Figure 9 The mechanism behind the currency crisis of late 1928 ................................................................ 20
Figure 10 Rearranged balance-of-payments to demonstrate country's financing need ................... 20
Figure 11 The share of agriculture in the central bank's rediscount........................................................... 22
Figure 12 Domestic National Income (DNI).............................................................................................................. 23
Figure 13 The decomposition of the change in agricultural DNI ................................................................... 23
Figure 14 The mechanism of the agricultural crisis of 1930 ........................................................................... 24
Figure 15 The financing sources of the banking sector ...................................................................................... 25
Figure 16 Lending by the banking sector................................................................................................................... 26
Figure 17 The share of agricultural lending in total lending ........................................................................... 26
Figure 18 The mechanism behind the 1931 crisis ................................................................................................ 27
Figure 19 The effects of the banking crisis on the central bank..................................................................... 29
Figure 20 Decomposing the changes in the gold cover....................................................................................... 30
Figure 21 The triggering mechanism of the Budapest banking crisis......................................................... 41
Figure 22 The triggering mechanism of the nationwide banking crisis .................................................... 42
Table 1 The currency crisis index (EMP) ................................................................................................................... 32
Table 2 The banking crisis index .................................................................................................................................... 34
Table 3 The signals of macroeconomic indicators ................................................................................................ 37
3
Flora Macher
London School of Economics
[email protected]
INTRODUCTION
The Great Depression comprised the most severe economic and financial turmoil that the world
has ever experienced.1 It was also a global phenomenon which culminated in a steep decline in
industrial and agricultural production, persistent unemployment, and a fall in world trade. This
was also the period with the highest frequency of financial crises in the past 150 years.2 In the
abundance of interwar financial crises, those of 1931 were the most prominent: their
emergence contributed to turning what was previously an economic recession into a prolonged
depression.3 The 1931 financial crises emerged in Central Europe and their start is signified by
the failure of the Austrian Credit-Anstalt on May 11, 1931. Subsequently, a number of other
regional countries, such as Germany and Hungary, experienced bank runs and could only
maintain the stability of their currency by introducing exchange controls.
The widely held view is that the panic in Central Europe was brought about by the flight of
foreign creditors. The countries of the region financed their post-war reconstruction through
foreign loans and by the late 1920s they reached high levels of indebtedness. Hungary was so
heavily indebted that by 1929 all of the new loans it obtained went towards debt service.4 When
international lending suddenly declined, the debtors of Central Europe had to turn to their
domestic banks to accommodate their financing need. The problem was that even local banking
systems were highly dependent on foreign capital. As a result, foreign creditors were becoming
increasingly doubtful of whether regional central banks would be able to defend their own
currency. After the failure of the Credit-Anstalt, foreign creditors’ doubts transpired into their
massive withdrawal from the region. Since financiers lacked sufficient information to make a
distinction between the various regional countries’ level of distress, they simply pulled their
funds from the whole region. Therefore, what started off in Austria, quickly spread to other
nations.5
This mainstream view relies upon national historiographies. For this paper I have reviewed the
literature of one Central European country, Hungary, and I find that the international view is
fully reflected in the national studies.6 The Hungarian literature similarly emphasizes that high
I would like to thank Professor Max-Stephan Schulze and Dr. Tamás Vonyó, my two supervisors for their
guidance on this work. I have also consulted Dr. Olivier Accominotti at several stages in the preparation of
this paper. Dr. Ágnes Pogány, Professor Peter Eigner, Dr. Clemes Jobst, Dr. Nathan Marcus, Dr. Matthias
Morys, Professor Albrecht Ritchl, and Dr. Tobias Straumann have all provided valuable insights to my work. I
also relied on help from Veronika Katz Kálniczkyné, archivist at the Hungarian National Archive. I am
grateful for their help.
My research is funded by the Economic and Social Research Council.
1 Although in the wake of the current Great Recession a few authors argued that what we are presently
experiencing is comparable to the Great Depression in depth and bredth, by now it is clear that the
current slump has not come close to the devastation of the late 1920s, early 1930s. Barry Eichengreen
and Kevin O’Rourke, ’A Tale of Two Depressions’, VoxEU, http://www.voxeu.org/article/tale-twodepressions-redux, 2009, 2010 and 2012
2 Michael Bordo et al., 'Is the Crisis Problem Growing More Severe?', Economic Policy, 16/32 (2001), 5182.
3 Charles Poor Kindleberger, The World in Depression, 1929-1939 (Harmondsworth: Penguin Books,
1987).
4 T. Ivá N Berend, Decades of Crisis : Central and Eastern Europe before World War Ii (Berkeley, Calif.:
Berkeley, Calif. : University of California Press, 1998).
5 Barry Eichengreen, Golden Fetters (Oxford University Press, 1996), Kindleberger, The World in
Depression, 1929-1939, C. H. Feinstein, Banking, Currency, and Finance in Europe between the Wars
(Oxford: Clarendon Press [u.a.], 1995) xviii.
6 T. Iván Berend, Válságos Évtizedek: Közép- És Kelet-Európa a Két Világháború Között (Budapest:
Gondolat Könyvkiadó, 1982b), T. Ivan Berend, 'A Világgazdasági Válság (1929-1933) Sajátos Hatásai
Közép-Kelet Európában', Történelmi Szemle, 25/1 (1982a), T. Iván Berend, Válságos Évtizedek: A 20.
4
Flora Macher
London School of Economics
[email protected]
levels of debt and foreign capital flight were behind the Hungarian crisis. In particular, I have
identified four important claims in the Hungarian literature’s interpretation of the crisis. First,
authors argue that the financial system’s demand for liquidity emerged in 1931 because their
foreign creditors became wary of the stability of the Hungarian currency and not because
depositors or creditors started questioning banks’ financial health. The financial system’s
increased liquidity needs hence, only worsened already existing balance-of-payments
difficulties but did not cause them. Second, the literature pinpoints policy failure - in the form of
deficit spending and overborrowing - as one of the two factors behind the crisis. As a result of
high government spending and indebtedness, the budget was not balanced and the level of debt
service was high. This put pressure on central bank reserves once foreign capital became scarce,
and thus made the currency vulnerable to an attack. Third, the studies propose that the other
factor behind the crisis was declining export revenues due to deteriorating terms of trade.
Through this, the literature implies that the central bank’s reserves were declining due to the
drop in export income which, in turn, damaged the stability of the currency and contributed to
the currency crisis. Fourth, studies point to crisis propagation from Austria as the trigger event
of the actual crisis. They suggest that after the fall of the Credit-Anstalt, foreign creditors started
quickly retreating not only from Austria but also from Hungary.
The purpose of this paper is to put the widely-held view on the Hungarian crisis under close
scrutiny. Relying on contemporary statistical publications, press records and archival evidence,
I have developed a new interpretation of this crisis which in certain elements departs from the
key positions of the current historiography. I argue that initially and predominantly, domestic
depositors’ flight triggered the events of 1931 and the financial system had a prominent role in
bringing about the calamities. The episode constituted a twin crisis which originated in the
banking sector and was subsequently transmitted to the monetary system, affecting the stability
of the currency. The causes of the banking crisis can be traced back to domestic structural
problems: a currency crisis in late 1928 and to an agricultural crisis in 1930.
The late 1928 event was a currency crisis which had serious reverberations on how the crisis of
1931 enfolded. It motivated a restrictive monetary policy which squeezed credit out of the
banking system and hence of the ailing real economy. This policy stance was a wall of defense
which protected the gold cover and kept the currency stable up until late May 1931. Holes only
appeared on this wall when the banking crisis became nationwide. At that point, the central
bank was forced to provide emergency liquidity financing to the banking system.
The agricultural crisis of 1930 was the result of falling domestic agricultural prices which
closely followed the declining trend of global prices. As prices fell, the domestic income of the
agricultural sector also dropped. Since agriculture employed over 50% of the labor force, their
falling income led to a contraction of domestic consumption.7 This in turn, induced a recession
in non-agricultural sectors as well. As the whole of the real economy sank into a slump from
1930, the proportion of non-performing loans increased, and the banking sector was becoming
insolvent.
Század Első Fele Közép- És Kelet-Európai Történetek Interpretációja (Budapest: Magvető Könyvkiadó,
1987b), T. Iván Berend, 'A 20. Század Nagy Gazdasági Válságai a Történelem Folyamataiban
(Hasonlóságok És Különbségek Az 1930-as És Az 1970-1980-as Évek Között)', in T. Ivan Berend and Knut
Borchard (eds.), Válság, Recesszió, Társadalom : Az 1930-as És Az 1970-1980-as Évek Összehasonlítása :
Válogatott Tanulmányok (Budapest: Kö zgazdasá gi é s Jogi Kö nyvkiadó , 1987a), T. Ivá N Berend et al.,
MagyarorszáG GazdasáGa Az Elsö ViláGháBorú UtáN, 1919-1929 (4; Budapest: Akadé miai Kiadó , 1966),
Michael Kaser and Rudolf Nö tel, 'Kelet-Euró pa Gazdasá gai a Ké t Vilá gvá lsá gban (Eastern European
Economies in Two World Crises)', in T. Ivá N Berend and Knut Borchard (eds.), Válság, Recesszió,
Társadalom : Az 1930-as És Az 1970-1980-as Évek Összehasonlítása : Válogatott Tanulmányok (Budapest:
Közgazdasági és Jogi Könyvkiadó, 1987).
7 In 1920, 58% of the population was employed in agriculture, according to the League of Nations
Statistical Yearbooks.
5
Flora Macher
London School of Economics
[email protected]
Based on the above, I propose that the agricultural crisis and the lack of liquidity support from
the central bank in the aftermath of the late 1928 currency crisis together made the banking
sector vulnerable to shocks and this caused the 1931 crisis. The shock came in waves from
October 1930 and was initiated by domestic depositors. Foreign capital owners’ contribution to
the panic was negligible. The capital flight became nationwide from August 1931, shook the
whole banking system and financial institutions’ increased demand for central bank money
undermined the stability of the currency as well, bringing about a twin crisis.
With reference to the four main tenets of the mainstream view, my findings are the following.
First, I find that the banking system had an active role in generating the crisis. The currency
crisis of late 1928 and the agricultural crisis of 1930 made the financial system vulnerable to
shocks and when this shock came, banks’ demand for liquidity support from the central bank
substantially increased. I argue that this was the only channel through which the currency
became involved in the crisis. That is, the connection between the banking system and the
monetary system, through banks’ increased liquidity demand, became active before the
monetary system was in crisis-mode.
With regards to the factors that the literature has pinpointed as the causes behind the crisis, my
findings reveal the following. It is clear that policy-makers failed in a sense that the country
became highly indebted. This, nevertheless, applied only until 1928 and was an important
contributing factor to the currency crisis in late 1928. After this episode, however, the country
was forced to significantly cut back on its borrowing and balance its trade account because the
heydays of large foreign capital inflows of 1927-28 never returned. Therefore, policy failures in
this sense did not put pressure on central bank reserves after this early currency crisis. With
regards to the other cause mentioned by the literature - declining export revenues contributing
to the a fall in monetary reserves - I find that until the 1931 crisis, agricultural exports per se
played into the country’s misfortune only to a limited extent. It was primarily the domestic
economy that suffered from declining agricultural prices. Consequently, the reserves of the
central bank and hence, the gold cover and the stability of the currency were largely unaffected
by the agricultural crisis. The only channel through which the agricultural crisis could have
trickled into the monetary system would have been the increased liquidity demand of the
financial system. Banks did feel the impact of the agricultural crisis through the real economy’s
generally degrading loan quality and this did increase their demand for more liquidity from the
central bank. This, nevertheless, remained a latent and unserved demand due to restrictive
monetary policy. Therefore, the currency was entirely sealed off from the impact of the
agricultural crisis.
Finally, in connection with the triggering event, I am doubtful about the literature’s position that
the collapse of the Credit-Anstalt and the subsequent flight of foreign deposit-holders had a
primary role in generating the crisis. My findings reveal that the fall in deposits was
predominantly attributable to domestic currency deposits. In addition, if foreign creditors had
been fleeing the country or if domestic financiers had been converting their domestic currency
deposits into foreign currency deposits, these should be observable through the decline of
central bank reserves. Reserves, nonetheless, remained almost flat even immediately before the
crisis. This implies that the crisis was domestic in nature and was at its origins directed against
the banking system. The banking system’s increased demand for liquidity, and hence the
increase in banknotes in circulation was what injured the gold cover and hence damaged
currency stability.
My interpretation departs from the story of the Credit-Anstalt contagion and I argue that the
triggers were domestic. Budapest institutions had been experiencing waves of crisis from the
last quarter of 1930. The reason behind the late 1930 depositor-wariness was the public’s
delicate trust in the banking system. This trust suffered a blow in October 1930 due to political
6
Flora Macher
London School of Economics
[email protected]
events. This triggered a silent bank run from late 1930 which was directed at financial
institutions in Budapest. The bank run exacerbated in July and turned into a heavy, nationwide
banking crisis in August 1931. The trigger behind this was a post-harvest shock and subsequent
defaults.8
The paper will introduce the above interpretation of the Hungarian crisis of 1931. I first
describe it through an analytical narrative and afterwards, I apply an early-warning indicator
(EWI) analysis to test the narrative.9 The EWI investigation evaluates the behavior of a number
macroeconomic factors prior to the crisis and determines which EWIs were foreseeing the
calamities. The results of the EWI analysis are supportive of my analytical narrative.
In the analytical narrative part as well as in the EWI analysis, I am relying on historical sources. I
have collected a wealth of macroeconomic data from the quarterly publication of the Institute of
Hungarian Economic Research, the monthly Statistical Review, the Monthly Statistical Report as
well as from the archives of the Hungarian National Bank’s department for economics, statistics
and research.10 I have also built an annual bank-by-bank database of balance sheets and profit
and loss statements based on the Big Hungarian Compass.11 All the data were manually collected
and this is the first time these sources have been applied to the analysis of the particular case
under observation. I have also reviewed the minutes of the board meetings of the Hungarian
National Bank and the Central Commission for Financial Institutions12 between 1925 and 1931
as well as a number of contemporary newspapers for 1928-1931 to better understand the
events taking place during the period.13
The paper has two contributions. On the one hand, it relies on a large database which has not
been used for the analysis of this case before. Using this database I have been able to establish a
more nuanced overview of the financial position of Hungary in the years leading to 1931. On the
other hand, this is the first study which applies rigorous analysis to the investigation of the
causes of the Hungarian crisis of 1931.
The structure of the paper is as follows. Section I introduces the literature that my analyses rely
on. Then Section II presents my interpretation of the Hungarian crisis of 1931. Subsequently,
Section III puts this interpretation to the test through an early-warning indicator analysis.
Section IV describes the triggering events behind the crisis and Section V concludes.
This very last bit of my argument is tentative.
Graciela L. Kaminsky and Carmen M. Reinhart, 'The Twin Crises: The Causes of Banking and Balance-ofPayments Problems', The American Economic Review, 89/3 (1999), 473-500.
10 The Institute of Hungarian Economic Research is Magyar Gazdaságkutató Intézet in Hungarian. Its
publication is available at the major Budapest libraries. The Statistical Review is Statisztikai Szemle in
Hungarian which is available online through the website of the Central Statistical Office. The Monthly
Statistical Report is Statisztikai Havi Közlemények in Hungarian which is also available at the major
Budapest libraries as well as at the library of the London School of Economics. The archive of the
Hungarian National Bank is available at the Hungarian National Archive (HNA) under various files but I
have been primarily using files Z6 and Z12.
11 The Big Hungarian Compass is Nagy Magyar Compass in Hungarian and is available at all major
Budapest libraries and for some years even at the library of the London School of Economics.
12 In Hungarian: „Pénzintézeti Központ”. It was established in the 19th century for the oversight of
foreign-owned assets then its operations were suspended. It was re-established in 1916 with a new
purpose: to supervise banks. It was owned partially by the state but the majority of joint-stock financial
institutions were also its quota holders (around 500 of them, including all large institutions). Its purpose
during the interwar period was to conduct annual audits of all of its member institutions, except for the
largest ones, and provide bailout financing to ailing banks. The archival records of the Central
Commission are at the Hungarian National Archive under files Z90-95 and Z1497, Z1505, Z1599.
13 The newspapers I reviewed are Magyar Pénzügy (English: Hungarian Finance) and A pénzvilág (English:
Financial World).
8
9
7
Flora Macher
London School of Economics
[email protected]
SECTION I - LITERATURE REVIEW
This section offers a backdrop to the analysis of the 1931 Hungarian crisis. First, I discuss the
theoretical approaches to why a financial crisis emerges and describe the analytical models that
have been developed on the basis of these theoretical tenets. The second sub-section offers an
overview of the existing literature related particularly to the 1931 financial crisis in Hungary.
I.1 – THREE GENERATIONS OF FINANCIAL CRISIS MODELING
There are two contentious views in the literature on why financial crises occur: the monetarist
approach, developed by Friedman and Schwartz14 and the view opposing it, advocated by
Kindleberger.15 Monetarists propose that financial crises occur as a result of the contraction of
the money supply which then leads to a decline in aggregate economic activity. Based on this
approach, a financial crisis necessarily involves a banking crisis because it is through the
financial system that the contraction of the money supply is being distributed in the economy.
Nevertheless, the financial system does not actively cause the crisis, it only has a propagator
role. The underlying assumption of this view is that the decline of the money supply is induced
by some sort of policy failure: a misguided decision by policy-makers leads to the contraction of
the money supply which then produces a decline in aggregate economic activity. The alternative
view adopts a much broader definition and argues that financial crises could arise from a
number of origins and may involve sharp declines in asset prices, failures of large financial and
non-financial firms, disruptions in foreign exchange markets, or a combination of all of these. A
financial crisis based on this definition is not necessarily monetary in its origins but may very
well generate from the real economy and may not be the result of policy failure.
The theoretical analysis of financial crises has been built around these two opposing views on
the origins of financial crises and the analyses have developed through three generations. The
models of the first two generations adopt the monetarist view on the causes of financial crises
and work with the assumption that financial crises have monetary origins and their ultimate
cause is policy failure.16 Because of this assumption, the models focus exclusively on balance-ofpayment crises. The banking system does not have an active crisis-trigger role in these models,
only a crisis-distributor function. First-generation models, whose emergence is usually signified
by Krugman’s paper, apply the basic assumption that the government’s “uncontrollable need for
seigniorage income” to finance the budget deficit undermines the fixed parity of the currency
and causes a panic. 17 However, the credibility of first-generation models has been gradually
questioned because the underlying policy conflict has not been observable in all cases.18 This
gave rise to second-generation models which assume that policy does not actually have to fail
for a currency crisis to occur - it is sufficient if the public’s expectations are pessimistic about
the path policy-makers may take.19 If, for instance, unemployment is high then the public may
reasonably expect an increase in the budget deficit. This may induce a currency crisis even
before the government would actually choose to enforce an expansionary monetary policy.
Milton Friedman and Anna J. Schwartz, A Monetary History of the United States 1867-1960 (12.;
Princeton: Princeton University Press, 1963).
15 Charles Poor Kindleberger, Manias, Panics and Crashes (Basingstoke: Palgrave, 2001).
16 Paul Krugman, 'Balance Sheets, the Transfer Problem, and Financial Crises', International tax and public
finance, 6/4 (1999), 459-72.
17 Paul Krugman, 'A Model of Balance-of-Payments Crises', Journal of Money, Credit and Banking, 11/3
(1979), 311-25.
18 Michael P. Dooley, 'A Model of Crises in Emerging Markets', The Economic Journal, 110/460 (2000),
256-72.
19 Reference in: Krugman, 'Balance Sheets, the Transfer Problem, and Financial Crises',
14
8
Flora Macher
London School of Economics
[email protected]
Then in the mid-1990s the monetarist view and first- and second-generation models built
around it lost some of their allure. The reason for this was that these models could not plausibly
explain the causes of the East Asian crises which did not demonstrate apparent policy failure
and still, a number of countries experienced a protracted period of financial distress. In
addition, from the 1980s, advocates of financial liberalization gained a strong platform in policymaking which gave rise to more lenient regulation on capital flows and banking.20 The
consequence was that - whereas balance-of-payment crises dominated the previous decades from the late 1980s twin crises, those involving the financial sector as well as the currency,
started to become increasingly frequent.21 These experiences gave rise to third-generation
models in the analysis of financial crises. These incorporated Kindleberger’s broader definition
of financial crises which admits that a crisis may not only have a monetary but also a real
economy cause and may not necessarily be the result of policy failure. This new approach has
assigned a much greater role to the private sector and in particular to the banking system. The
signature work in this area is that of Kaminsky and Reinhart who have recognized that currency
and banking crises occur at the same time. 22 They identified these events as twin crises. Based
on their empirical investigation of financial crises in industrial and developing countries
between 1970 and 1995, they find that banking sector problems in most cases precede balanceof-payment problems. They argue that as the crisis of the currency unfolds, it further deepens
the banking panic, activating a “vicious spiral” in which the two types of crisis reinforce each
other. Two tenets of the Kaminsky-Reinhart approach are a departure from first- and secondgeneration models: it allows the real economy to be a source of distress and it proposes that the
banking system can have an active role in bringing about the crisis.
This new approach to financial crisis analysis has been empirically tested on recent episodes
but seldom on panics of the more distant past. One exception is Schnabel’s work, which applies
the framework of third-generation models to the German crisis of 1931.23 The goal of this paper
is to do the same for the Hungarian crisis of 1931.
I.2 – THE MAINSTREAM INTERPRETATION OF THE HUNGARIAN CRISIS OF 1931
The Great Depression was a global phenomenon which culminated in a steep decline of
industrial and agricultural production, persistent unemployment, and a fall in world trade. This
was also the period with the highest frequency of financial crises in the past 150 years. 24 In the
abundance of interwar financial crises, those of 1931 which originated in Central Europe stand
out: their emergence contributed to turning what previously was an economic recession into a
prolonged depression.25
Around May-July 1931, three Central European countries, Austria, Germany and Hungary
experienced a financial collapse in close succession. The calamity started off in May with the fall
of the Austrian Credit-Anstalt, the largest universal bank of the country. Then in July the German
Danatbank experienced liquidity problems and this produced a crisis in Germany. Hungary also
had to deal with bank runs from July, and in response, authorities introduced a number of
measures: a bank holiday, the closure of the stock exchange, the de-listing of banks from the
stock exchange, restrictions on deposit withdrawals and capital controls.
Charles W. Calomiris, Fragile by Design : The Political Origins of Banking Crises and Scarce Credit, ed.
Stephen H. Haber (Princeton : Princeton University Press, 2014).
21 Michael D. Bordo and Antu P. Murshid, 'Are Financial Crises Becoming Increasingly More Contagious?
What Is the Historical Evidence on Contagion?', (National Bureau of Economic Research, 2000).
22 Kaminsky and Reinhart, 'The Twin Crises: The Causes of Banking and Balance-of-Payments Problems',
23 Isabel Schnabel, 'The German Twin Crisis of 1931', The Journal of Economic History, 64/3 (2004a), 82271.
24 Bordo et al., 'Is the Crisis Problem Growing More Severe?',
25 Kindleberger, The World in Depression, 1929-1939.
20
9
Flora Macher
London School of Economics
[email protected]
Of these Central European crises, the events in Germany have already been investigated from a
number of perspectives.26 Furthermore, the recent third-generation approach to financial crisis
analysis has been applied to the German case and that study has motivated a lively debate on
the origins of the panic.27 The literature on Austria and Hungary, on the other hand, somewhat
lags behind when it comes to applying new analytical models to the examination of these two
countries’ events. This paper investigates the Hungarian crisis based on a new approach, as a
first step in catching up our understanding of Austria and Hungary with that of Germany.
The mainstream view of the Hungarian crisis of 1931 is that it was a currency crisis at its origins
and was triggered by the flight of foreign financiers following the fall of the Credit-Anstalt.
Although it is not explicitly stated or analyzed, the prevalent view implies that the Hungarian
episode is describable through first- or second generational models. 28 The international
literature’s view on the episode is fully reflected in the Hungarian historiography.29 I have
constructed a system of diagrams to illustrate the complex interrelations between different
mechanisms in the economy. I will use these diagrams throughout the paper to pinpoint the
arguments of the literature and my own. The main points of the Hungarian literature can be best
understood with the help of Figure 1 and Figure 2.
Figure 1 schematically describes the three parts of the Hungarian economy - the real economy,
the banking system and the monetary system – and points out those links between the three
parts and those exogenous factors which were relevant during the interwar period. Since the
country was predominantly agricultural, the real economy was influenced by two external
factors: the success of the harvest and the changes in agricultural prices. Since more than 50%
of the workforce was employed in agriculture, the state of agriculture influenced their
purchasing power and the latter affected the rest of the real economy. The health of the real
economy in turn, had an influence on the banking sector through the quality of credit. If the real
economy was in a recession, the proportion of non-performing loans, i.e. defaults increased and
this had a negative impact on the banking system. The direct connection between the real
economy and the monetary system was export revenues. Since income from exports was
realized in foreign currency, the changes in external trade affected the reserve levels of the
central bank.
Moving on to the monetary system: overall stability was highly dependent on one exogenous
factor, the availability of capital in the world economy. Since the country was on the gold
exchange standard, monetary policy-makers’ hands were tied when it came to the stimulation of
the economy. Being on the gold standard meant that the exchange rate was fixed and the
minimum gold cover requirement was set by legislation. The gold cover is the ratio of the
foreign currency and the gold reserves of the central bank and the total banknotes in
Harold James, 'The Causes of the German Banking Crisis of 1931', The Economic History Review, 37/1
(1984), 68-87, Theo Balderston, 'The Banks and the Gold Standard in the German Financial Crisis of
1931', Financial history review, 1/1 (1994), 43-68, Peter Temin, Lessons from the Great Depression (1989;
Cambridge, Mass: MIT Press, 1991).
27 Schnabel, 'The German Twin Crisis of 1931', (, Peter Temin, 'The German Crisis of 1931: Evidence and
Tradition', Cliometrica, 2/1 (2008), 5-17, Thomas Ferguson and Peter Temin, Made in Germany: The
German Currency Crisis of July 1931 (2003) 1-53, Isabel Schnabel, 'Reply to Thomas Ferguson and Peter
Temin's "Comment on the German Twin Crisis of 1931"', The Journal of Economic History, 64/3 (2004b),
877-78.
28 Harold James, The End of Globalization: Lessons from the Great Depression (Cambridge, Mass: Harvard
University Press, 2002) vi-vi, Eichengreen, Golden Fetters, Schnabel, 'The German Twin Crisis of 1931',
29 Berend, Válságos Évtizedek: Közép- És Kelet-Európa a Két Világháború Között., Miklós Szuhay and T. Iván
Berend, A Tőkés Gazdaság Története Magyarországon: 1848-1944 (Budapest: Közgazdasági és Jogi
Könyvkiadó, 1978)., Kaser and Nötel, 'Kelet-Európa Gazdaságai a Két Világválságban (Eastern European
Economies in Two World Crises)'.
26
10
Flora Macher
London School of Economics
[email protected]
circulation. The minimum legal requirement for Hungary’s gold cover was 24%, set at the
establishment of the independent central bank on June 30, 1925.30 The gold cover was the
indicator of the stability of the currency: if in high territory, it reflected a strong currency; if
declining or low, it demonstrated an unstable currency. The monetary authority could not
resort to printing money (increasing the banknotes in circulation), or to reducing its reserves to
increase liquidity in the economy (i.e. to stimulate the economy). These would have
deteriorated the gold cover. The monetary system was hence highly dependent on the
availability of foreign capital. If more foreign capital arrived to the country, reserves increased,
the gold cover rose and banknotes in circulation could potentially be raised. The nature of this
monetary regime naturally had a strong influence on the banking system since the additional
banknotes in circulation were distributed into the economy through the financial system. If the
gold cover improved due to the inflow of foreign loans, the discount window of the central bank
could potentially be widened towards financial institutions, banks would become more liquid
and they could in turn increase the liquidity (by lending more) of the real economy.
Figure 1 The framework of the Hungarian economy
General framework
Harvest
Agricultural
prices
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Gold cover
Inflow of
foreign
capital
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
The health of the third part of the economy, the banking system, was conditional upon one
external factor: domestic trust in financial institutions. Hungary - similarly to Germany and
Austria - experienced a hyperinflationary period in the early 1920s. Stabilization took place
from mid-1924 through mid-1926 based on the reconstruction scheme of the League of Nations
(just like in Austria).31 In the early period of the stabilization, in 1925-26, the financial system
was highly unstable: in 1925 27% and in 1926 an additional 20% of joint stock banks were in a
state of distress.32 From the end of 1925 by the end of 1927 the number of joint-stock banks
declined from 903 to 692.33 Those institutions that remained standing were starved for deposits
League of Nations files on the reconstruction of Hungary, LSE Archive
League of Nations files on the reconstruction of Hungary, LSE Archive
32 I define a bank in distress if it was under liquidation or bankruptcy procedure, conducted a merger,
terminated its (financial) operation in the given year or did not issue financial statements for two
consecutive years.
33 Nagy Magyar Compass (Big Hungarian Compass)
30
31
11
Flora Macher
London School of Economics
[email protected]
and hence kept interest rates on deposits high. From 1926-27 savers were slowly returning to
the banking system. This renewed trust, however, was delicate and banks’ fear of runs and the
public’s fear of bank collapses persisted through the period.34
Figure 2 highlights the key points of the Hungarian literature within the general framework
introduced above. The studies argue that two exogenous factors brought about the Hungarian
crisis of 1931. On the one hand, the decline of agricultural prices had a strong impact on export
revenues. Since export revenues directly influenced foreign exchange reserves, their decline
caused reserves to fall and weakened the stability of the currency. The other factor was the
availability of foreign capital which became limited prior to the crisis and this, again through
lower reserve levels, also undermined the gold cover. Authors argue that these two factors
together made the currency highly vulnerable.
Figure 2 The mechanism behind the arguments of the Hungarian literature
The mechanism behind the argument of the Hungarian
literature
Harvest
Agricultural
prices
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Gold cover
Inflow of
foreign
capital
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
According to the literature, the trigger event that brought under this vulnerable monetary
system was the collapse of the Austrian Credit-Anstalt. This induced a flight of foreign depositholders in Austria. Since Austrian banks were closely linked to Hungarian institutions, foreign
financiers started fleeing Hungarian banks as well. As financial institutions were quickly losing
their foreign deposits, they had to turn to the central bank for liquidity support. The central
bank had to increase its discount window and thereby increase the banknotes in circulation.
This was one source of pressure on the gold cover. In addition, since - according to the
mainstream literature - foreign and not domestic currency depositors were leaving the country,
this put an additional burden on the gold cover through the depletion of central bank reserves.
The gold cover which was already vulnerable due to the fall in the availability of foreign capital
and the export decline, could not bear this pressure coming through the banking system.
Therefore, the already unstable currency was buried under the liquidity demands of the
financial system.
34
HNA, Z6, 2. doboz (File Z6, box 2) - Minutes of the Board of Governors of the Hungarian National Bank
12
Flora Macher
London School of Economics
[email protected]
Although it is not specifically stated or analyzed, Hungarian historiography implies that the
events of 1931 fit first-generation models. This underlying assumption is detectable through
four positions that the studies take in their interpretation of the crisis. The first is connected to
the role authors assign to the banking system in bringing about the crisis. They propose that the
financial system only had a crisis-distributor role but it did not actively cause the crisis.
According to this approach, banks’ demand for liquidity financing emerged only because their
foreign creditors became wary of the stability of the Hungarian currency and sought not only to
withdraw their deposits but also to convert deposits into foreign currency. Thus the banking
system started to experience a crisis when the monetary system was already in panic-mode.
Afterwards, the connection between the banking system and the monetary system, through
banks’ increased liquidity demand, further aggravated the currency crisis.
The second and third points of the literature describe which factors generated currency
vulnerability (proxied by a low gold cover). On the one hand, authors pinpoint policy failure - in
the form of deficit spending and overborrowing - as one of the two factors behind the crisis. As a
result of high government spending and indebtedness, the budget was not balanced and the
country’s debt service levels were high. When foreign capital became less abundant and the
government could not obtain new loans to service the accumulated foreign debt, it had to rely
on the reserves of the central bank. Depleting reserves reduced the gold cover and thus the
currency became vulnerable to an attack. The studies propose that the other factor behind the
crisis was declining export revenues due to deteriorating terms of trade in agriculture. The
trade account was unbalanced and hence the central bank’s reserves were falling. Thus the
struggles of the real economy aggravated the stability of the currency through declining export
revenues.
It is important to note that the studies imply that the causes of the crisis exerted their impact
through channels that entirely avoided the financial system. Authors do propose that
deteriorating terms of trade (i.e. falling agricultural prices) - besides causing exports to drop also contributed to the decline in domestic incomes. However, they do not take their point
further and do not assume that the troubles of the real economy could have actually impacted
the banking system through degrading credit quality. Further, authors do not assume that
degrading credit quality could have made the banking sector vulnerable and increased banks’
reliance on the central bank for liquidity support. Based on the mainstream view, the real
economy influenced the monetary system only through declining export revenues and the
banking system entered the picture only when the currency crisis was already under way.
The final important position that the Hungarian literature takes is with regards to the trigger
behind the actual panic. Studies point to crisis propagation from Austria as the event that
generated the financial disorder. Authors suggest that after the fall of the Credit-Anstalt, foreign
creditors started quickly retreating not only from Austria but also from Hungary. Thereby
authors propose that the immediate currency crisis was brought about by the decline in central
bank reserves.
This paper’s interpretation of the Hungarian financial crisis differs from the main claims of the
existing literature. I will argue that the crisis can be more accurately described through the
third-generation model of financial crises. The two main tenets of my interpretation are that the
crisis was caused by structural problems in the domestic real economy and the banking system
had a central role in the events. Section II will advance this argument and propose that the
Hungarian crisis was in fact a twin crisis brought about by domestic factors: it originated in the
banking sector, which was vulnerable due to problems in the domestic real economy, it spread
to the currency through banks’ increased demand for central bank liquidity support, and it
subsequently brought down the currency as well.
13
Flora Macher
London School of Economics
[email protected]
SECTION II – A NEW INTERPRETATION OF THE HUNGARIAN CRISIS
OF 1931
In this section I am presenting my interpretation of the Hungarian crisis of 1931. First, I explain
what definition and indicators I am applying to financial, banking and currency crises and then I
present the evidence for Hungary based on these. Afterwards, I discuss the two causes behind
the 1931 crisis: the currency crisis of late 1928 and the agricultural crisis of 1930. The next subsection brings together my findings and puts forth the proposal that 1931 was a twin crisis
caused by domestic factors.
II.1 - DEFINING FINANCIAL, BANKING AND CURRENCY CRISES
Reinhart and Rogoff provide a comprehensive typology of financial crises.35 Their definition of
financial crises incorporates inflationary, currency, banking and sovereign debt crises. The
authors differentiate between crises which are defined by quantitative thresholds and those
which are defined by events. Based on this characterization, they classify currency crises among
those defined by quantitative thresholds. A currency crash occurs when the annual depreciation
of the currency versus the US dollar is 15% or more. The authors describe banking crises by
quantitative measures as well as events. The decline in the price of bank stocks relative to the
market index, the fall in deposits, the collapse of asset prices, the rise in bankruptcies, or the
increasing proportion of non-performing loans may all serve as indicators of a banking crisis. At
the same time, since these data are seldom available at a high frequency, the authors also define
banking crises through events: bank runs that lead to closures, merging or takeover or largescale government assistance to an important (or a group of important) financial institution(s).
In this paper I analyze currency and banking crises and I refer only to these when I use the term
“financial crises”. I rely on the Reinhart-Rogoff definition when it comes to banking crises and
use quantitative as well as event-based measures. I have monthly data for deposits and
insolvencies from the quarterly publication Gazdasági Helyzetjelentés, published by Magyar
Gazdaságkutató Intézet.36 I am supplementing this dataset with a chronology of events
constructed from my review of the two major contemporary weekly financial newspapers, the
Magyar Pénzügy37 and A pénzvilág38 as well as from my review of the minutes of the board
meetings of the Hungarian National Bank and the same for the Central Commission for Financial
Institutions between 1925 and 1931.
For the definition of the currency crisis, instead of relying on Reinhart and Rogoff, I apply the
Eichengreen-Wyplosz-Rose exchange market pressure index (EMP) because I find it more
relevant to the particular case than the Reinhart and Rogoff measure.39 Since Hungary’s
currency was fixed to gold during the period under observation, currency pressures are not
observable through the exchange rate of the domestic currency to a major foreign currency (the
parity against the dollar or the pound sterling was kept fixed during the whole period).
Nevertheless, changes in central bank reserves and in interest rates may indicate when balanceCarmen M. Reinhart, Kenneth S. Rogoff, and Inc Books24x, This Time Is Different: Eight Centuries of
Financial Folly (Princeton: Princeton Univ. Press, 2009) xlv.
36 The title of the publication in English: „Economic Status Report”. The name of the publisher in English:
„Institute for Hungarian Economic Research”. The institute was a contemporary think tank, established in
1929 and sponsored by the Hungarian National Bank.
37 In English: „Hungarian Finance”
38 In English: “Financial World”
39 Barry Eichengreen, Andrew Rose, and Charles Wyplosz, 'Contagious Currency Crises: First Tests', The
Scandinavian Journal of Economics, 98/4 (1996), 463-84.
35
14
Flora Macher
London School of Economics
[email protected]
of-payment pressures were intensifying. Since the EMP index incorporates changes in reserves,
interest rates as well as the exchange rate, it is more relevant to the historical context than the
use of the exchange rate by itself.
For the construction of the EMP, I use the weekly changes in the foreign exchange and gold
reserves of the central bank, the weekly changes in the gold cover (which is my proxy for the
exchange rate) and the weekly changes in the base rate of the central bank. I fully rely on
Eichengreen-Wyplosz-Rose to construct the index from these three data series. These three
elements of the index are weighted by their standard deviation and the EMP indicates a crisis if
it swings above or below the mean by at least 1.5 standard deviations. I have collected the data
for the EMP from the reports of Magyar Gazdaságkutató Intézet and from the archival records of
the Hungarian National Bank.
II.2 - WHAT DO BANKING AND CURRENCY CRISIS MEASURES INDICATE?
Figure 3 depicts the results of the EMP index for the period of 1927-33. The index shows that
Hungary experienced a currency crisis already in late 1928. From October 31, 1928 the EMP
gives a signal through six consecutive periods (i.e. six weeks), indicating a prolonged crisis of
the currency. The EMP also gives a few signals afterwards in 1929 through 1931. However,
these are non-persistent changes that are corrected within a maximum of a two-week period
but in the majority of the cases, within one week. These are therefore, not regarded as currency
crises in this analysis.
What should also be noticed on Figure 3 is that the EMP is kept well within its bands in the
period of the 1931 crisis, i.e. in May-July, 1931. This implies that no currency crisis occurred in
the period immediately preceding the panic of 1931. And it also highlights the fact that capital
controls were introduced before a full-blown currency crisis could have been detectable
through the EMP.
Figure 3 The Exchange Market Pressure (EMP) index
Hungary experienced a currency crisis already in late 1928
The exchange market pressure index (EMP)
Crisis: protracted
period of swings
8.0%
6.0%
4.0%
2.0%
0.0%
-2.0%
-4.0%
-6.0%
-8.0%
07-Jan-27
07-Jan-28
07-Jan-29
07-Jan-30
07-Jan-31
07-Jan-32
07-Jan-33
Source: Magyar Gazdaságkutató Intézet, Eichengreen-Wylosz-Rose, 1996
15
Flora Macher
London School of Economics
[email protected]
Figure 4 and Figure 5 show the trends in banking crisis indicators. Figure 4 depicts the volume
of assets under insolvency, expressed as a proportion of the banking sectors’ deposits. There
was a continuous increase in this ratio and by late 1930, assets under insolvency amounted to
20% of banks’ deposits and 8.5% of their total assets. Although this is an indirect indicator of
banking troubles, it shows that the proportion of non-performing loans was on the increase
throughout the period under observation.
Figure 4 Total assets under insolvency as a share of banks' deposits
By Q4 1930, total assets under insolvency reached 20% of
banks’ deposits and 8.5% of banks’ total lending
Total assets under insolvency as a percentage of banks’ deposits
60%
50%
Oct 1930:
Volume of assets under
insolvency is:
• 20% of banks’ deposits
and
• 8.5% of banks’ total
lending
40%
30%
20%
10%
0
0%
Source: Magyar Gazdaságkutató Intézet, Compass
Figure 5 The monthly change in deposits for Budapest and non-Budapest banks
There was a Budapest banking crisis from Q4 1930 and it was
followed by a nationwide crisis in July 1931
Monthly change in total deposits
Q3 1930: silent bank run
Budapest banks
Non-Budapest banks
15%
0.15
10%
0.1
5%
0.05
0%
0
-5%
-0.05
-10%
-0.1
July 1931: real bank run
Source: Magyar Gazdaságkutató Intézet, Statisztikai Szemle, Statisztikai Havi Közlemények
16
Flora Macher
London School of Economics
[email protected]
Figure 5 introduces a more direct indicator of banking difficulties: the monthly change in
deposits for Budapest and non-Budapest banks.40 The diagrams imply that deposit growth was
limited from early 1929, and from late 1930 growth was in negative territory with a number of
sharp declines. In the case of Budapest banks there was a sharp decline in October 1930. This is
not observable for non-Budapest banks. Subsequently, April is the next month when another
decline takes place before the largest fall in July-August. After this, from July 17 authorities
introduced a 3-day bank holiday, implemented restrictions on deposit withdrawals and the flow
of foreign exchange.
Based on the above, there are four observations that need to be addressed. First, the late 1928
currency crisis requires an in-depth investigation. This is the purpose of the next sub-section.
Second, the limited growth of deposits from 1929 should also be explained. My argument is that
this was due to the recession that the economy was slowly sinking into. The economic
contraction was caused by an agricultural crisis which is the topic of the fourth sub-section.
Third, the jittery behavior of depositors of Budapest banks from the last quarter of 1930 should
also be examined in detail because it might reveal information on crisis triggers. Finally, it must
be noted that while the late 1930 episode was a Budapest event, the one in mid-1931 affected
financial institutions all over the country. This fact may again offer clues on the actual triggers of
the panic. I deal with these last two issues in Section IV in connection with crisis triggers.
II.3 - THE CURRENCY CRISIS IN LATE 1928
The currency crisis in late 1928, observable through the EMP index, was a typical emerging
market, post-stabilization crisis, as it is very well described by Reinhart-Végh.41 Through the
analysis of recent emerging market examples, the authors point out that post-stabilization
currency crises are highly typical and they follow the same pattern. First, a hyperinflationary
period is overcome by foreign financial support which lends credibility to and hence stabilizes
the currency. Stabilization is then followed by a large boom in the economy due mainly to an
increase in consumer spending. The increase in consumption and the decline in private savings
lead to imbalances in the trade account. This is financed through foreign loans and the countries
generally undergo “binge-borrowing”. However, when the availability of foreign capital
suddenly comes to an end, a currency crisis takes place. This is exactly the pattern that
characterized the Hungarian crisis in late 1928.
Hungary’s stabilization took place from mid-1924 through mid-1926 and was considered a big
success. The reconstruction was overseen by the League of Nations through a large foreign loan
and the primary objective of the scheme was to achieve a balanced government budget. This
was accomplished within a few months into the program. Therefore, the loan, originally to be
spent on filling the budget gap, was invested into the economy. 42 The success of the
reconstruction brought with it a large inflow of additional foreign loans. Figure 6 shows the
balance-of-payments for the period under review. The data demonstrate that after 1926 there
was a substantial increase in the inflow of foreign capital and during 1927 and 1928 a total of
1.226bn pengős of foreign loans entered the country, equivalent to 8% of the domestic national
income (DNI) each year.
Unfortunately, for non-Budapest banks, this data is only available from 1930.
Carmen Reinhart and Carlos Vegh, 'Do Exchange Rate-Based Stabilizations Carry the Seeds of Their
Own Destruction?', in Mpra Paper (ed.), (1999).
42 Files OV9/436-439 and OV9/234 at the Bank of England Archive
40
41
17
Flora Macher
London School of Economics
[email protected]
Figure 6 Hungary's balance-of-payments
The inflow of foreign capital substantially declined from 1929
1926
1927
1928
1929
1930
1931
1932
950
897
987
1238
1111
733
397
206
364
446
319
377
120
10
0
230
187
58
12
515
8
206
593
633
376
389
634
18
1156
1490
1621
1614
1500
1368
415
Inflow
Current account
Total
Capital account
Medium- and long-term capital
Short-term capital
Total
Total
Outflow
Current account
Total
1098
1376
1489
1451
1249
958
422
Medium- and long-term capital
38
86
131
100
179
100
9
Short-term capital
21
28
0
68
0
247
3
Total
59
114
131
168
179
347
12
1157
1490
1620
1619
1428
1305
433
1926
1927
1928
1929
1930
1931
1932
-148
-479
-502
-213
-138
-225
-25
Long- and medium-term capital transactions
167
278
315
218
198
20
1
Short-term capital transactions
-21
202
187
-10
12
268
Balance
147
480
502
208
210
287
7
Balance as a % of DNI
2%
8%
8%
3%
3%
5%
0%
Capital account
Total
Balance of payments
Current account
Balance
Capital account
5
Source: Statisztikai Szemle
The inflow of foreign capital during the two years after the stabilization enabled the country to
finance the imbalances of its trade account. Figure 7 shows that the trade account was in deficit
throughout the whole of 1927 and 1928. The total deficit in these two years was 718m pengős,
i.e. app. 60% of the total foreign capital inflow.
Figure 7 Hungary's trade account
The trade account was in high deficit until mid-1929 but shifted
into surplus or to an annual state of balance afterwards
Hungary’s trade account
m pengős
50
1925
-45.9
1926
-82.3
1927
-346.3
1928
-370.3
1929
-22.6
1930
+77.5
1931
+17.5
1932
-4.9
1933
+79.7
40
30
20
10
0
-10
-20
-30
-40
0
-50
Source: Statisztikai Havi Közlemények
18
Flora Macher
London School of Economics
[email protected]
However, the volume of foreign capital inflow significantly dropped in 1929. Whereas in 1928
the total inflow was 633m pengős, by the end of 1929 it fell to 376m pengős. (Figure 6) This
sudden slow-down created a liquidity crunch in the economy. Previous high levels of imports
could not be further sustained and the country was forced to sharply reduce the volume of
goods it imported. By the second half of 1929 these actions translated into a trade account
surplus. (Figure 7) Nevertheless, previous import arrangements still had to be met in the first
half of 1929. Since foreign capital was available to a more limited extent than before, the
economy had to resort to utilizing the reserves of the central bank to meet these foreign
currency obligations.
Figure 8 reveals how this pressure affected the gold cover at the central bank. The diagram
shows the gold cover (red line) and the total rediscount (blue bars) of the central bank. The gold
cover was 47% in mid-1928 but dropped 7 percentage points by the end of the year and a
further 2 percentage points by mid-1929. Around this point, in May 1929 the Governor of the
Hungarian National Bank visited the Governor of the Bank of England to resolve the crisis. After
this meeting, the Hungarian central bank received a bridge loan facility from the Bank of
England in the amount of GBP 500,000.43 The larger, USD 20m loan needed a bit more time to
seal and was eventually agreed upon in August 1929 and was provided by a group of central
banks.44 These steps were sufficient to stabilize the currency.45
Figure 8 The central bank's rediscount and the gold cover
The gold cover declined from 47% in mid-1928 to 38% in mid1929
The central bank’s rediscount and the gold cover
Gold cover
Total discount
May 1929
GBP 500k
forex facility
from BofE
Rediscount: 700
m pengős
Aug 1929
USD 20m (app.
GBP4m) from
central banks
70.0% Gold cover:
%
600
60.0%
500
50.0%
400
40.0%
300
30.0%
200
20.0%
100
10.0%
0
30-Jun-25
0.0%
30-Jun-26
30-Jun-27
30-Jun-28
30-Jun-29
30-Jun-30
30-Jun-31
30-Jun-32
Source: HNA, Z12, 128 and 129 csomó, Magyar Gazdaságkutató Intézet
HNA, Z6, 2. doboz (File Z6, box 2) - Minutes of the Board of Governors of the Hungarian National Bank,
May 22, 1929
44 HNA, Z6, 2. doboz (File Z6, box 2) - Minutes of the Board of Governors of the Hungarian National Bank,
Aug 30, 1929
45 Although this is not tangible support to my case on the currency crisis, I venture to emphasize the fact
that the very period when the central bank decided not to report its gold cover was the one I am
describing as a currency crisis. I propose that this just underscores the fact that there was a currency
crisis in late 1928. My experience is the same with financial institutions: when they stopped reporting
their financial statements, in the majority of the cases this was a sign of their financial difficulties.
43
19
Flora Macher
London School of Economics
[email protected]
The root causes of the late 1928 currency crisis were overborrowing and the imbalances of the
trade account. The trigger event, as depicted on Figure 9, was an exogenous shock: the sudden
decline in the availability of foreign capital. This reduced the inflow of foreign loans which then
put a pressure on the reserves of the central bank and induced a currency crisis. This is exactly
what the Hungarian literature has described as one of the two main causes of the 1931 crisis.
This nevertheless, was a cause not to the 1931 but to this earlier event, the currency crisis of
late 1928.
Figure 9 The mechanism behind the currency crisis of late 1928
The mechanism behind the currency crisis of late 1928
Agricultural
prices
Harvest
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Inflow of
foreign
capital
Gold cover
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
Figure 10 Rearranged balance-of-payments to demonstrate country's financing need
The rearranged balance-of-payments shows that the country’s
financing need was halved in 1929 and maintained at that level
1926
878
954
-76
63
47
16
-60
1927
801
1149
-348
74
88
-14
-362
1928
819
1189
-370
148
137
11
-359
1929
1066
1107
-41
157
173
-16
-57
1930
945
883
62
153
171
-18
44
1931
596
582
14
123
158
-35
-21
1932
343
352
-9
53
50
3
-6
Capital inflow from interest/dividend
Capital outflow from interest/dividend
Net earnings from financial activities
Capital inflow from loan repayment
Capital outflow for loan repayment
Net balance of repayments
Net balance of debt service
9
97
-87
102
38
64
-24
22
140
-117
102
86
16
-101
20
163
-143
98
121
-23
-165
15
171
-157
52
99
-47
-204
13
195
-182
40
177
-137
-319
15
218
-203
15
99
-85
-288
1
19
-19
7
9
-2
-20
Total financing need
-84
-463
-524
-260
-275
-309
-26
Capital inflow for short-term financing for the state
Capital outflow for short-term financing for the state
Net balance of short-term financing for the state
Capital inflow for short-term financing for other actors
Capital outflow for short-term financing for other actors
Net balance of short-term financing for other actors
Net balance of short-term financing
0
0
0
0
21
-21
-21
122
0
122
108
28
80
202
99
0
99
88
0
88
187
58
0
58
0
68
-68
-10
6
0
6
6
0
6
12
515
147
368
0
100
-100
268
8
0
8
0
3
-3
5
Remaining financing need after short-term financing
-105
-262
-337
-271
-264
-42
-21
Capital inflow from new loans/issues
Capital outflow for new loans/issues
Net balance of new loans/issues
Other capital inflow
Other capital outflow
Net balance of other capital flows
Net balance of investment financing
103
0
103
1
0
1
104
261
0
261
1
0
1
262
347
10
336
1
0
1
338
266
1
265
0
0
0
265
336
1
335
1
1
0
335
104
0
104
0
1
0
104
3
0
3
0
0
0
3
Capital inflow from goods sold
Capital outflow from goods purchased
Net earnings from Ex-Im
Other capital inflow
Other capital outflow
Net earnings from other activities
Total net earnings (current account balance)
20
Flora Macher
London School of Economics
[email protected]
After this early currency crisis, the country was forced to reduce its binge-borrowing and
balance its trade account. From 1929, the economy was adapted to the reduced volumes of
foreign capital. Figure 10 shows the balance-of-payments in a re-arranged format that seeks to
emphasize the financing need of the economy arising from the deficits of trade and debt service.
The row titled “total financing need” indicates the amount of financing the country needed after
the trade balance (the difference of export and import revenues) and after the balance of the
debt service (the difference between the debt service of loans received and loans provided). The
data indicate that the economy halved its financing need after 1928. Whereas in 1927 and in
1928 the total financing need was 463m and 524m pengős, respectively, by 1929 it was reduced
to 260m pengős and was maintained in this range in 1930 as well. This implies that after the
late 1928 currency crisis, the policy failures of an unbalanced trade account and bingeborrowing were corrected and policy adapted to the new circumstances. These policy failures
were thus, not at the root of the 1931 crisis.
Nonetheless, the currency crisis event did not pass without long-term consequences. The light
color of the arrow between the monetary system and the financial system on Figure 9 aims to
indicate that even though the currency crisis did not have observable, immediate effects on the
banking system (i.e. it did not bring about a banking crisis), it did have long-term repercussions
on the health of the financial sector and hence had an influence on how the events of 1931
enfolded.
The long-term consequences of the late 1928 currency event were twofold. First, the experience
made the central bank extremely cautious and protective of the parity. As a result, the currency
was kept strong, with the gold cover hovering around 50% going into the 1931 crisis.
Nevertheless, to achieve this, the central bank became ever more restrictive when it came to
rediscounting bills, i.e. providing liquidity for the banking sector. Figure 8 shows not only the
gold cover, discussed earlier, but also the volume of rediscount the central bank provided to the
financial system (blue bars). The diagram indicates that the rise of the gold cover back to over
50% was achieved in conjunction with the reduction of rediscount levels. While the average
volume of rediscount was app. 325m pengős in 1928 and 1929, by 1930 the central bank
reduced it to 218m pengős. The emergency loans that the central bank received in 1929 were
only sufficient to stop the immediate crisis. The rise of the gold cover from the mid-1929 low of
38% to around 55% by March 1931 was accomplished almost entirely by restricting the
banknotes in circulation.
This clearly had a strong impact on the financial system: it deprived banks of an important
source of financing. Already during the 1928 currency crisis, there were requests from the
financial sector that the central bank widen the discount window. However, the management of
the bank resisted these entreaties and after the 1928 crisis, in mid-1929 it went on to actually
tightening its rediscount policy.46 This was in response to the fact that financial institutions
relied on central bank funding during 1928 more than during previous years. The central bank
issued a warning to financial institutions that they should refrain from using the discount
window and they should more diligently evaluate the bills they accept. In addition, the national
bank also informed banks that bills with certain types of collateral would not be accepted for
rediscount.47 The implementation of the new policy kindled a lively debate in the board of the
central bank. Some members of the board raised concerns that perhaps the restriction of the
central bank’s discount window was a burden on financial institutions. Nevertheless, the
management of the monetary authority was unmoved and followed through with the tightening.
HNA, Z6, 2. doboz (File Z6, box 2) - Minutes of the Board of Governors of the Hungarian National Bank,
Nov 28, 1928
47 HNA, Z6, 2. doboz (File Z6, box 2) - Minutes of the Board of Governors of the Hungarian National Bank,
June 26, 1929
46
21
Flora Macher
London School of Economics
[email protected]
The other long-term consequence of the currency episode was that not only had the central
bank become more restrictive but it also developed a strong bias towards agriculture when it
came to liquidity provision. Figure 11 depicts the central bank’s rediscount practices. Red bars
show the amount of total agricultural bills rediscounted, while blue bars are the total rediscount
facility for the whole economy. The red line is the share of agricultural rediscount in the total.
Even though the total volume of rediscount substantially declined following the currency event,
the volume of agricultural rediscount stayed at the same level. This means that agricultural
rediscount actually crowded out the liquidity available for other sectors. This contributed to the
aggravating recession of non-agricultural sectors: while in 1930 the DNI decline of these sectors
was only 3%, in 1931 it was 8%.
Figure 11 The share of agriculture in the central bank's rediscount
After the 1928 pre-crisis episode, the central bank became
restrictive and biased towards agriculture
The share of agriculture in the central bank’s rediscount
Total rediscount
Agricultural rediscount
Share of agriculture in rediscount
Rediscount: 700
m pengős
0.7
600
0.6
500
0.5
400
0.4
300
0.3
200
0.2
100
0.1
0
30-Jun-25
Share of agr.
Rediscount:
%
0
30-Jun-26
30-Jun-27
30-Jun-28
30-Jun-29
30-Jun-30
30-Jun-31
30-Jun-32
Source: HNA, Z12, 60. csomó, Magyar Gazdaságkutató Intézet
The central bank’s tightening and its bias towards the agricultural sector were two
consequences of the late 1928 currency crisis which contributed to the events in 1931. These
two measures reduced the liquidity available in the economy and thus brought about the
degradation of credit quality which directly affected the health of the financial system. What
added to banks’ misfortune was that the restriction in monetary policy happened in a period
when the demand for rediscount would not have been low at all. The country was already in a
recession from 1930 with the DNI declining by 6%. The central bank therefore, narrowed its
discount facility just when the banking sector would have needed its liquidity support the most.
This monetary policy change further exacerbated banks’ difficulties which arose as a result of
the crisis of the real economy. This is the focus of the next sub-section.
II.4 - THE AGRICULTURAL CRISIS IN 1930
In 1930 another crisis, an agricultural crisis hit the Hungarian economy. Figure 12 shows the
country’s DNI from 1925 until 1934. The diagram on the left demonstrates that in 1930 the
agricultural sector contracted by 10% (color blue) and non-agricultural sectors’ income shrank
by 3% (color green). The diagram on the right shows that the agricultural sector was the
primary responsible for the recession in 1930.
22
Flora Macher
London School of Economics
[email protected]
Figure 12 Domestic National Income (DNI)
The economy was in recession from 1930 due primarily to the
contraction of the agricultural sector
Domestic national income in absolute terms and annual change, m pengős
Agriculture
Non-agriculture
DNI in absolute terms
DNI annual change
Growth rate =
8%
-4%
1%
6%
-10%
-20%
-20%
-17%
-8%
Growth rate =
12%
8%
10%
3%
-3%
-8%
-9%
-4%
-50%
8000
1000.0
7000
500.0
6000
0.0
5000
4000
-500.0
3000
-1000.0
2000
-1500.0
1000
0
-2000.0
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1925
1926
1927
1928
1929
1930
1931 1932 1933 1934
Source: Eckstein 1956, Statisztikai Szemle
Figure 13 The decomposition of the change in agricultural DNI
The agricultural contraction primarily affected domestic
agricultural revenues (not exports) and was due to price effects
The decomposition of the change in agricultural DNI, m pengős
Change disaggregated into export and
domestic agricultural revenues
Total change in agricultural DNI
Change disaggregated into price effect and
volume effect
200.0
200
300.0
100.0
100
200.0
0.0
0
-100.0
-100
100.0
0.0
-100.0
-200.0
-200
-300.0
-300
-400.0
-400
-500.0
-500
1929
1930
1931
-200.0
-300.0
-400.0
-500.0
1929
1930
1931
1929
Domestic agricultural revenues
Price effect
Export agricultural revenues
Volume effect
1930
1931
Source: Eckstein 1956, Statisztikai Szemle
Next, Figure 13 goes more in-depth into trying to understand what happened in the agricultural
sector: why it contracted from 1930. The first diagram shows the total change of agricultural
income which is data brought forward from Figure 12. Then the second diagram in the middle
decomposes the change in agricultural income into the change in domestic agricultural
revenues and export revenues. This data clearly identifies the domestic economy as the culprit
behind the fall in agricultural incomes. App. 80% of the fall in agricultural DNI was realized in
23
Flora Macher
London School of Economics
[email protected]
the drop of domestic revenues and only 20% through the fall of export revenues. Finally, the last
diagram on Figure 13 decomposes the change in agricultural income into price and volume
effects. The data pinpoint agricultural prices as the sole cause behind the decline in agricultural
incomes in 1930.
Relying on the above analysis, the agricultural crisis of 1930 can now be followed through the
framework of the Hungarian economy. (Figure 14) Based on the above analyses, the cause of the
agricultural crisis in 1930 was the decline in prices. As the prices of agricultural goods
underwent a global decline, Hungarian domestic prices, which closely followed global prices,
similarly fell. The fall in domestic agricultural prices automatically reduced the income of a large
proportion of the Hungarian economy, since over 50% of the population was employed in
agriculture. Through their declining consumption due to their falling income, other, nonagricultural sectors also experienced a contraction. In 1930 the whole economy shrank by 6%.
Based on this, I argue that the reason why deposits were not increasing as fast from mid-1929
through 1930 (Figure 5) was because the public simply had less money to place on their bank
accounts.
Figure 14 The mechanism of the agricultural crisis of 1930
The mechanism of the agricultural crisis of 1930
Harvest
Agricultural
prices
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Gold cover
Inflow of
foreign
capital
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
The banking sector was influenced by this recession through an increase in non-performing
loans. Since financial institutions did not account for degrading loan quality in their financial
statements, there is no exact figure for the proportion of non-performing loans. One source
suggests that of all the agricultural loans of the sector at least 25% were in default in 1930.48
Financial institutions could, nonetheless, not turn to the central bank for liquidity support due
to the restrictive monetary policy-stance. Therefore, from this point on the banking sector
substantially restrained its lending to the real economy. This naturally reinforced the recession
of the real economy and contributed to even more defaults.
Tivadar Dr. Surányi-Unger, Magyar Nemzetgazdaság És Pénzügy (Pécs: Dunántúli Pécsi Egyetemi
Könyvkiadó és Nyomda, 1936).
48
24
Flora Macher
London School of Economics
[email protected]
The monetary system was entirely sealed off from the impacts of the agricultural crisis. As
demonstrated on Figure 13, and in contrary to what the Hungarian literature has proposed,
export revenues did not play an important role in the fall of agricultural revenues. Thus the
direct channel between the real economy and the monetary system did not threaten the
stability of the currency. The other, indirect channel between the real economy and the
monetary system, the financial system itself, was sterilized by the tight monetary policy. Even
though banks may have had an increased demand for liquidity support due to real economy
problems and increasing defaults, this demand remained latent because of the central bank’s
wall of defense around the currency. Therefore, in 1930 the agricultural crisis stopped at the
real economy and the banking sector, and it did not leak out into the monetary system. This also
implies that the banking system became a buffer zone between the crisis-ridden real economy
and the currency. Banks swallowed the impact of the agricultural crisis and put a halt on new
lending from their own resources. I highlighted the banking system in red on Figure 14
indicating that after this point, the sector was vulnerable to shocks.
Figure 15 The financing sources of the banking sector
From 1930 the main financing source for the financial sector
was state-guaranteed loans
The annual change in the financing
sources of the banking sector
Private institutions
All financial institutions
State-owned institutions
1927
1928
1929
1930
1931
1932
Deposits
ST creditors
Rediscount
Guarantees
LT creditors
Equity
1933
1927
1928
1929
1930
1931
1932
1933
Source: Compass
Whereas the monetary authority turned a blind eye on banks’ suffering as a result of the
agricultural crisis, the government was alarmed and sprang to action. A government policy
which was introduced in 1929 gained increasing importance as the lending of the financial
sector declined. From 1929 the state started providing guarantees to the financial sector to
boost lending. These state-guaranteed loans could only be used for agricultural financing. The
magnitude and the significance of these state-guaranteed loans can be understood through
Figure 15. The diagrams all show the change in financial institutions’ financing resources (i.e.
the change in the equities and liabilities items of the balance sheet) from one year to the next.
The diagram on the left covers the entire financial system, while the two on the right cover
private (top) and state-owned (bottom) banks. The year 1930 for the whole of the financial
system indicates that the majority of new resources came from these state guarantees (color
light blue). In the case of private banks app. 50% of new financing arrived from this source
while for state banks the ratio was 100%.
25
Flora Macher
London School of Economics
[email protected]
Figure 16 shows the other side of banks’ balance sheet and explains how these resources were
used by banks (i.e. the asset side of the balance sheet). The diagram for the whole financial
system shows that the majority (specifically 82%) of new lending was sourced from state
guarantees in 1930. The ratio was 64% for private institutions and close to 100% for state-run
banks. These assets were entirely dedicated to providing new loans to the agricultural sector.
That is, 82% of new loans in 1930 went into the agricultural sector.
Figure 16 Lending by the banking sector
New lending, other than through state-guaranteed sources was
limited from 1930
The annual change in various types
of lending by the banking sector
Private institutions
All financial institutions
State-owned institutions
1927
1928
1929
1930
ST loans
Mortgages
Guarantees
LT loans
1931
1932
1933
1927
1928
1929
1930
1931
1932
1933
Source: Compass
Figure 17 The share of agricultural lending in total lending
51% of lending of the financial sector went into agriculture in
1930
Share of agriculture in total lending
Private institutions
All financial institutions
30%
8%
13%
1926
21%
1927
32%
1928
41%
1929
51%
52%
49%
54%
1930
1931
1932
1933
35%
44%
40%
83%
79%
78%
1930
1931
1932
40%
State-owned institutions
70%
Source: Compass
40%
18%
45%
46%
51%
1926
1927
1928
1929
85%
1933
26
Flora Macher
London School of Economics
[email protected]
This policy intervention had two important consequences. First, non-agricultural sectors were
entirely squeezed of credit. Not only was the monetary authority impartial towards agricultural
rediscount but the banking sector’s new lending was also almost fully re-directed towards the
agricultural sector. These steps left only bits for the non-agricultural parts of the economy to
support their growth. This two-fold crowding out effect reinforced the recession in nonagricultural sectors. Second, state-guaranteed loans increased the exposure of the financial
system to agriculture. Figure 17 shows that in 1930 the financial sector’s loan exposure to
agriculture reached 51%. This was a jump from 41% in the previous year and almost entirely
due to state intervention. This increase in banks’ exposure to agriculture occurred in a year
when agriculture contracted by 10%. And this was only the beginning of the sector’s
catastrophe: in 1931, the decline in agricultural DNI was 20%. The banking system hence
became highly exposed to a sector of the economy which experienced the steepest recession
from 1930 and contributed the most to the contraction of the whole of the economy.
The crisis of the real economy had a strong impact on the banking sector through deteriorating
credit quality. Moreover, the financial sector’s woes were exacerbated by policy-makers. On the
one hand, banks could not turn to the monetary authority for additional crisis-financing. On the
other hand, the fiscal authority had given banks an incentive to expose themselves to the worstperforming sector of the economy. All these made the financial system highly vulnerable to
shocks.
II.5 - THE TWIN CRISIS OF 1931
Figure 18 demonstrates the mechanism leading to the 1931 crisis. There were two fundamental
drivers behind the events.
Figure 18 The mechanism behind the 1931 crisis
The mechanism behind the crisis of 1931 – the argument
advanced in the paper
Harvest
Agricultural
prices
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Gold cover
Inflow of
foreign
capital
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
The first was the agricultural crisis of 1930, described in the previous sub-section. The
agricultural crisis affected the whole of the real economy because the price decline and the
subsequent fall in agricultural revenues reduced the purchasing power of the majority of the
population and this, in turn, negatively influenced the growth of other sectors as well. This real
27
Flora Macher
London School of Economics
[email protected]
economy crisis affected the banking system through the deterioration of the quality of credit.
What exacerbated the consequences of the agricultural crisis was that financial institutions’
exposure was increasing to this sector even though it was already in a recession. Since this
sector of the economy was contracting the most and contributed the most to the fall of the DNI,
banks were in fact increasing their investment into a ticking time bomb. The second factor
behind the banking crisis was the central bank’s unwillingness to provide new liquidity to the
financial system up until the last minute. Even though the real economy was in a recession and
new lending came predominantly through state-guaranteed loans, the central bank’s policy
remained tight.
Figure 15 and Figure 16 demonstrate that by late 1930 the banking sector was in a fragile state
of health. Figure 15 shows that institutions’ natural sources of financing entirely dried out or
decreased in 1930 and banks had to rely on state-guaranteed loans. Banks could only to a
decreasing extent rely on depositors (color dark blue): the increase of this source halved from
1929 to 1930. Short-term creditors were also abandoning Hungarian banks (color red). Finally,
another potentially important source of financing, the discount window of the central bank, did
not provide new resources from 1929 (color green). Therefore, the only source of financing that
financial institutions could rely on was state-guaranteed loans (color light blue). These,
nonetheless, came with strings attached and increased the exposure of the system to the crisisridden agriculture.
Increased exposure to the agricultural sector was a death sentence. While in 1930 51% of the
total lending of the financial system sat in agriculture (Figure 17), the sector contracted 10% in
that year (Figure 12). In addition, since misguided policy measures exacerbated the recession of
non-agricultural sectors, the remaining 49% of banks’ loan exposure was also to sectors that
were contracting. Even though it is difficult to obtain tangible evidence on the actual losses of
financial institutions since they did not write down their non-performing assets, contemporary
news reports confirm that banks by 1930 were struggling under defaults from both agricultural
and non-agricultural sectors.49 (Figure 4)
On the other hand, the monetary system was stable and the currency was strong going into May
1931. As depicted on Figure 8, in late February, early March of 1931 the gold cover was around
55% which was close to the all-time-high of 58.5% at the establishment of the central bank in
mid-1925. The big drop occurred during the week of May 23 when within seven days the gold
cover fell 7 percentage points. Afterwards, there were minor climb-backs but the fall was
irreversible. The ratio was 32% on July 23 and it reached its legal minimum of 24% on Aug 15.
While the financial system was highly vulnerable from 1930, the currency was very strong until
May 23, 1931. This is already indirect evidence that a banking crisis was likely to happen while
there was no indication of balance-of-payment instabilities. Nevertheless, it does not prove that
the financial sector initiated the panic. What needs to be understood is why deposit-holders
were pulling their money from the banking system: were they trying to escape the currency or
were they trying to escape the banking system? Depositors’ motives can be disentangled
through the analysis of the gold cover. If a currency crisis had occurred, we should see the
impact on the numerator of the gold cover: the decline in central bank reserves would show that
deposit-holders were withdrawing their money from banks and converting it into foreign
currency, in their escape from the domestic currency. If a banking crisis had occurred, we
should see the impact on the denominator of the gold cover: the increase in banknotes in
circulation would indicate that deposit-holders were withdrawing their money from the
banking system and keeping it in cash, in their fear of bank collapses.
49
Various issues of Financial World and Hungarian Finance.
28
Flora Macher
London School of Economics
[email protected]
The Hungarian crisis of 1931 was at the initial stage a banking crisis. Figure 19 demonstrates
this. The diagram in the top left corner depicts the changes in rediscount at the central bank.
The one at the bottom shows the evolution of central bank reserves. These two together clearly
imply that the crisis erupted as a banking event. In the case of a currency crisis, central bank
foreign exchange and gold reserves are heavily depleted because depositors and creditors
initiate a bank run to convert their money into foreign currency. Banks in this case turn to the
central bank and demand foreign exchange, pressuring the reserves of the central bank. This is
clearly not what happened in the case of Hungary since reserves barely changed in the critical
period (bottom left diagram). Moreover, the flatness of reserves cannot be explained by the
introduction of capital controls. The first move towards capital controls was through a
government decree on July 17, 1931.50 Nonetheless, reserves were flat relative to the rediscount
even before this date. This implies that depositors initiated the bank run to get access to their
cash and save it from banks but the majority of them did not seek to convert the money into
foreign currency. Banks required increasing volumes of rediscount from the central bank to be
able to meet the demands of a domestic bank run.
Figure 19 The effects of the banking crisis on the central bank
When the banking crisis erupter in July 1931, banks’ demand for
rediscount increased and this increased the banknotes in circ.
Rediscount, reserves and banknotes in circulation, Q4 1930-1931, m pengős
Rediscount at the central bank
700
600
500
July 17
Decree on
capital controls
Rediscount and banknotes in circulation
400
700
300
600
200
100
0
Central bank reserves
500
400
300
700
200
600
500
400
100
0
300
200
100
0
Source: Magyar Gazdaságkutató Intézet
Another important point that Figure 19 implies is that the volume of rediscount increased more
than the banknotes in circulation. This is demonstrated on the diagram on the right. The
rediscount shoots up in the critical period of the crisis and carries the banknotes in circulation
with it but only up to a certain point. This implies that not all of the liquidity provided by the
central bank was withdrawn by depositors. Banks kept some of it on their balance sheets as a
cushion of protection. This implies that financial institutions were preparing for further distress
and/or were taking advantage of the widened liquidity window, anticipating that the monetary
authority would return to its restrictive position again.
According to Ellis (1939) the foreign exchange market was monopolized in Hungary through a decree
issued on July 17, 1931. From this date, outward payments were forbidden without the permission of the
Hungarian National Bank. Howard S. Ellis, 'Exchange Control in Austria and Hungary', The Quarterly
Journal of Economics, 54/1 (1939), 1-185. Legislation on capital controls was passed on August 8, 1931.
This was the date when exchange controls were more efficiently enforced. Review of the Central
Commission for Financial Institutions, V. 8-9.
50
29
Flora Macher
London School of Economics
[email protected]
Figure 20 underscores the argument that reserve changes did not substantially contribute to the
crisis. The diagram decomposes the changes in the gold cover into changes in its numerator
(reserves, color blue) and changes in its denominator (banknotes in circulation, color red). It is
clear that the population was seeking to escape banks and not seeking to escape the currency.
Figure 20 Decomposing the changes in the gold cover
The changes in the gold cover were almost entirely caused by
changes in the banknotes in circulation
Decomposing the causes behind the changes in the gold cover, Q4 1930-1931
Banknotes in circulation
Changes in reserves
10.0%
8.0%
6.0%
4.0%
2.0%
0.0%
-2.0%
Source: Magyar Gazdaságkutató Intézet
Based on the above analysis, the Hungarian crisis of 1931 was a domestic banking crisis in its
origins. As the banking crisis enfolded, financial institutions’ demand for rediscount was a great
pressure on the central bank and contributed to the deterioration of the gold cover. Therefore,
what started off as a banking crisis subsequently influenced the currency as well and turned
into a twin crisis.
30
Flora Macher
London School of Economics
[email protected]
SECTION III - EARLY-WARNING INDICATOR ANALYSIS
In this section, I am applying an early-warning indicator analysis. The purpose of this analysis is
to review which macroeconomic factors were forecasting the crisis in Hungary and through that
test my interpretation of the Hungarian events. The approach relies on the macro-indicator
analysis of Kaminsky and Reinhart and is divided into three parts.51 The first part defines what a
financial crisis is. Based on this definition, I identify the crises that occurred in Hungary between
1928 and 1931 and also determine their exact start. In the second step, I describe the
macroeconomic indicators I am using and I also determine whether an indicator was signaling
the financial crisis episodes in Hungary. Finally, in the third part, I discuss this signaling
behavior.
III.1 - IDENTIFYING FINANCIAL CRISES IN HUNGARY BETWEEN 1927 AND 1931
In Section II, I already described my definition of currency and banking crises and here I am
presenting my database. In the case of currency crises, I am relying on the EichengreenWyplosz-Rose EMP index. Just like for the authors, my definition of a signal is when the EMP
index rises or sinks more than 1.5 standard deviations above or below the mean of the whole
period under observation. I also added one more condition to this definition: the EMP must give
a signal in four consecutive periods for a crisis to occur. I am using this restriction in order to
avoid false alarms given by the signal. The data are presented in Table 1.
Table 1 shows that 1927 was an “eventless” year, while from the second half of 1928 the EMP is
giving a number of one-off and two-period signals. However, none of the signals proceeds for
more than two consecutive periods and hence they do not fit the definition of a currency crisis.
The only time when the EMP is ticking for at least four consecutive weeks is between Oct 31 and
Dec 7, 1928. This is what I have previously referred to as the late 1928 currency crisis. After this
episode, the EMP gives a number of one-off signals in 1929 and 1930, but there are no periods
which constitute a crisis. The same applies to 1931. The first half of that year experiences a
number of one-off signals but there are no crisis episodes. This is despite the fall of the gold
cover, one of the three elements of the EMP. (Figure 8) The reason why we do not see the EMP
signaling in the weeks prior to the introduction of exchange controls (before July 17) is that
even though the gold cover was declining, this decrease was offset by the limited change in
reserves (another element of the EMP). (Figure 20) Following the introduction of exchange
controls, reserve change was further reduced through policy measures and this again offset the
fall in the gold cover. This suggests that policy measures averted a currency crisis before it could
have happened. Therefore, the EMP does not signal a currency crisis for 1931.
For the definition of banking crises, I am using the monthly change in domestic currency and
foreign currency deposits at Budapest and non-Budapest banks. My definition of a crisis is when
the monthly decline in deposits exceeds the average monthly change of the whole period under
observation by at least 1.5 standard deviations. The data are presented in Table 2. Table 2
shows an “eventless” 1928 and 1929. The first signal arose in October 1930 for domestic
currency deposits at Budapest institutions. Based on this, I am arguing that Budapest banks
started to undergo a crisis from the last quarter of 1930. Afterwards, 1931 appears relatively
hectic. The first signal came in January for foreign currency deposits for non-Budapest
institutions. The next appeared in April for domestic currency deposits for Budapest
institutions. Afterwards, in July Budapest institutions underwent an across-the-board decline of
all of their deposits. The same happened for non-Budapest institutions a month later. Moreover,
in the case of non-Budapest banks, the decline in domestic currency deposits was persistent for
51
Kaminsky and Reinhart, 'The Twin Crises: The Causes of Banking and Balance-of-Payments Problems',
31
Flora Macher
London School of Economics
[email protected]
four consecutive months, whereas for Budapest institutions the crisis was recurring in October
and then in December.
Table 1 The currency crisis index (EMP)
Date
07-Jan-27
15-Jan-27
23-Jan-27
31-Jan-27
07-Feb-27
15-Feb-27
23-Feb-27
28-Feb-27
07-Mar-27
15-Mar-27
23-Mar-27
31-Mar-27
07-Apr-27
15-Apr-27
23-Apr-27
30-Apr-27
07-May-27
15-May-27
23-May-27
31-May-27
07-Jun-27
15-Jun-27
23-Jun-27
30-Jun-27
07-Jul-27
15-Jul-27
23-Jul-27
31-Jul-27
07-Aug-27
15-Aug-27
23-Aug-27
31-Aug-27
07-Sep-27
15-Sep-27
23-Sep-27
30-Sep-27
07-Oct-27
15-Oct-27
23-Oct-27
31-Oct-27
07-Nov-27
15-Nov-27
23-Nov-27
30-Nov-27
07-Dec-27
15-Dec-27
23-Dec-27
31-Dec-27
EMP signal
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Date
07-Jan-28
15-Jan-28
23-Jan-28
31-Jan-28
07-Feb-28
15-Feb-28
23-Feb-28
29-Feb-28
07-Mar-28
15-Mar-28
23-Mar-28
31-Mar-28
07-Apr-28
15-Apr-28
23-Apr-28
30-Apr-28
07-May-28
15-May-28
23-May-28
31-May-28
07-Jun-28
15-Jun-28
23-Jun-28
30-Jun-28
07-Jul-28
15-Jul-28
23-Jul-28
31-Jul-28
07-Aug-28
15-Aug-28
23-Aug-28
31-Aug-28
07-Sep-28
15-Sep-28
23-Sep-28
30-Sep-28
07-Oct-28
15-Oct-28
23-Oct-28
31-Oct-28
07-Nov-28
15-Nov-28
23-Nov-28
30-Nov-28
07-Dec-28
15-Dec-28
23-Dec-28
31-Dec-28
EMP signal
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
1
0
0
0
1
1
0
0
1
1
0
0
1
1
1
1
1
1
0
0
1
Date
07-Jan-29
15-Jan-29
23-Jan-29
31-Jan-29
07-Feb-29
15-Feb-29
23-Feb-29
28-Feb-29
07-Mar-29
15-Mar-29
23-Mar-29
31-Mar-29
07-Apr-29
15-Apr-29
23-Apr-29
30-Apr-29
07-May-29
15-May-29
23-May-29
31-May-29
07-Jun-29
15-Jun-29
23-Jun-29
30-Jun-29
07-Jul-29
15-Jul-29
23-Jul-29
31-Jul-29
07-Aug-29
15-Aug-29
23-Aug-29
31-Aug-29
07-Sep-29
15-Sep-29
23-Sep-29
30-Sep-29
07-Oct-29
15-Oct-29
23-Oct-29
31-Oct-29
07-Nov-29
15-Nov-29
23-Nov-29
30-Nov-29
07-Dec-29
15-Dec-29
23-Dec-29
31-Dec-29
EMP signal
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
32
Flora Macher
Date
07-Jan-30
15-Jan-30
23-Jan-30
31-Jan-30
07-Feb-30
15-Feb-30
23-Feb-30
28-Feb-30
07-Mar-30
15-Mar-30
23-Mar-30
31-Mar-30
07-Apr-30
15-Apr-30
23-Apr-30
30-Apr-30
07-May-30
15-May-30
23-May-30
31-May-30
07-Jun-30
15-Jun-30
23-Jun-30
30-Jun-30
07-Jul-30
15-Jul-30
23-Jul-30
31-Jul-30
07-Aug-30
15-Aug-30
23-Aug-30
31-Aug-30
07-Sep-30
15-Sep-30
23-Sep-30
30-Sep-30
07-Oct-30
15-Oct-30
23-Oct-30
31-Oct-30
07-Nov-30
15-Nov-30
23-Nov-30
30-Nov-30
07-Dec-30
15-Dec-30
23-Dec-30
31-Dec-30
London School of Economics
EMP
signal
1
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
1
0
0
1
0
0
0
1
Date
07-Jan-31
15-Jan-31
23-Jan-31
31-Jan-31
07-Feb-31
15-Feb-31
23-Feb-31
28-Feb-31
07-Mar-31
15-Mar-31
23-Mar-31
31-Mar-31
07-Apr-31
15-Apr-31
23-Apr-31
30-Apr-31
07-May-31
15-May-31
23-May-31
31-May-31
07-Jun-31
15-Jun-31
23-Jun-31
30-Jun-31
07-Jul-31
15-Jul-31
23-Jul-31
31-Jul-31
07-Aug-31
15-Aug-31
23-Aug-31
31-Aug-31
07-Sep-31
15-Sep-31
23-Sep-31
30-Sep-31
07-Oct-31
15-Oct-31
23-Oct-31
31-Oct-31
07-Nov-31
15-Nov-31
23-Nov-31
30-Nov-31
07-Dec-31
15-Dec-31
23-Dec-31
31-Dec-31
EMP
signal
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Date
07-Jan-32
15-Jan-32
23-Jan-32
31-Jan-32
07-Feb-32
15-Feb-32
23-Feb-32
29-Feb-32
07-Mar-32
15-Mar-32
23-Mar-32
31-Mar-32
07-Apr-32
15-Apr-32
23-Apr-32
30-Apr-32
07-May-32
15-May-32
23-May-32
31-May-32
07-Jun-32
15-Jun-32
23-Jun-32
30-Jun-32
07-Jul-32
15-Jul-32
23-Jul-32
31-Jul-32
07-Aug-32
15-Aug-32
23-Aug-32
31-Aug-32
07-Sep-32
15-Sep-32
23-Sep-32
30-Sep-32
07-Oct-32
15-Oct-32
23-Oct-32
31-Oct-32
07-Nov-32
15-Nov-32
23-Nov-32
30-Nov-32
07-Dec-32
15-Dec-32
23-Dec-32
31-Dec-32
[email protected]
EMP
signal
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Date
07-Jan-33
15-Jan-33
23-Jan-33
31-Jan-33
07-Feb-33
15-Feb-33
23-Feb-33
28-Feb-33
07-Mar-33
15-Mar-33
23-Mar-33
31-Mar-33
07-Apr-33
15-Apr-33
23-Apr-33
30-Apr-33
07-May-33
15-May-33
23-May-33
31-May-33
EMP
signal
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Notes: An EMP signal of “1” indicates that the EMP index is at least 1.5 standard deviations above or below the mean. The
highlight indicates the currency crisis event.
33
Flora Macher
London School of Economics
[email protected]
Table 2 The banking crisis index
Date
31-Jan-28
29-Feb-28
31-Mar-28
30-Apr-28
31-May-28
30-Jun-28
31-Jul-28
31-Aug-28
30-Sep-28
31-Oct-28
30-Nov-28
31-Dec-28
Signal for
domestic
currency
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
foreign
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
total
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Date
31-Jan-29
28-Feb-29
31-Mar-29
30-Apr-29
31-May-29
30-Jun-29
31-Jul-29
31-Aug-29
30-Sep-29
31-Oct-29
30-Nov-29
31-Dec-29
Signal for
domestic
currency
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
foreign
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
total
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Date
31-Jan-30
28-Feb-30
31-Mar-30
30-Apr-30
31-May-30
30-Jun-30
31-Jul-30
31-Aug-30
30-Sep-30
31-Oct-30
30-Nov-30
31-Dec-30
Signal for
domestic
currency
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
1
0
0
Signal for
foreign
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
total
deposits at
Bp
institutions
0
0
0
0
0
0
0
0
0
1
0
0
Signal for
domestic
currency
deposits at
non-Bp
institutions
Signal for
foreign
deposits at
non-Bp
institutions
Signal for
total
deposits at
non-Bp
institutions
Signal for
domestic
currency
deposits at
non-Bp
institutions
Signal for
foreign
deposits at
non-Bp
institutions
Signal for
total
deposits at
non-Bp
institutions
Signal for
domestic
currency
deposits at
non-Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
foreign
deposits at
non-Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
Signal for
total
deposits at
non-Bp
institutions
0
0
0
0
0
0
0
0
0
0
0
0
34
Flora Macher
Date
31-Jan-31
28-Feb-31
31-Mar-31
30-Apr-31
31-May-31
30-Jun-31
31-Jul-31
31-Aug-31
30-Sep-31
31-Oct-31
30-Nov-31
31-Dec-31
London School of Economics
Signal for
domestic
currency
deposits at
Bp
institutions
0
0
0
1
0
0
1
0
0
1
0
0
Signal for
foreign
deposits at
Bp
institutions
0
0
0
0
0
0
1
0
0
0
0
1
Signal for
total
deposits at
Bp
institutions
0
0
0
0
0
0
1
0
0
1
0
0
Signal for
domestic
currency
deposits at
non-Bp
institutions
0
0
0
0
0
0
0
1
1
1
1
0
[email protected]
Signal for
foreign
deposits at
non-Bp
institutions
1
0
0
0
0
0
0
1
0
0
0
0
Signal for
total
deposits at
non-Bp
institutions
0
0
0
0
0
0
0
1
1
1
1
0
Notes: A signal of “1” indicates that the fall of deposits in the given month exceeds the average monthly change by at
least 1.5 standard deviations. The highlights indicate banking crisis events.
Based on the above definitions, I argue that there were three episodes of financial crisis in
Hungary. The first was a currency crisis which started off in October 1928. The second was a
banking crisis of Budapest financial institutions which emerged in October 1930. This event was
initiated by domestic deposit holders. The third crisis was a nationwide banking crisis from
August 1931 which affected both Budapest and non-Budapest banks. Early on, foreign currency
depositors were contributing to the nationwide banking crisis, but afterwards, only domestic
depositors’ flight was persistent. The banking events affected financial institutions in ebbs and
flows but they seem to have been more continuous for non-Budapest banks.
III.2 - DEFINING EARLY-WARNING INDICATORS
In the following, I am reviewing the behavior of 12 macroeconomic factors to understand which
of them foresaw the above defined three crisis episodes. Kaminsky and Reinhart apply 16
different indicators in their analysis and I have been relying on their list of indicators when I
designed mine. Because of data limitations and the continuous turbulence over my period, I
have 12 measures which I confidently deem reliable and adequate to the analysis of the present
case.
For the investigation of the financial sector, I am using deposits and interest rates on deposits at
the largest Budapest banks as indicators. I have deposits broken down into foreign and
domestic currency and further separated for Budapest and non-Budapest banks. Regarding
interest rates, I only have data for primary Budapest institutions.52 For the analysis of the
balance-of-payments, I am using imports and exports to proxy the current account and the gold
cover, the reserves of the central bank, banknotes in circulation and the foreign-domestic
interest rate differential to describe the capital account. The foreign-domestic interest rate
differential is the percentage point difference between the base rate of the Hungarian National
Bank and that of the Bank of England. For the analysis of the real economy, I am applying the
stock exchange index, the Albers-Uebele index53 and the activity ratio of non-agricultural
Primary Budapest institutions were the 12/13 largest Budapest banks based on the size of their total
assets.
53 Thilo Albers and Martin Uebele, 'Reassessing Interwar Business Cycles from a Global Perspective: A
New Monthly Dataset', unpublished paper, (2014).
52
35
Flora Macher
London School of Economics
[email protected]
sectors. Finally, to comment on fiscal health I have a government budget balance indicator
which is the ratio of the budget deficit or surplus and total government revenues.
I have constructed the indicators in the following way. For each indicator, I define a tranquil
period and I calculate the average monthly change of the indicator during the tranquil period.
My ideal definition of the tranquil period is the 12 months from June 1926 until June 1927
because this is the period right after the end of the League of Nations reconstruction program
and also the period under observation for which my crisis indicators confidently exclude the
possibility of any crisis events. Nonetheless, this has turned out to be a strict definition and in
order to have a sufficient number of meaningful indicators. Therefore, I slightly had to depart
from this definition of the tranquil period in a few cases. Appendix 1 describes my assumptions
on this matter. Once I had the tranquil periods defined for each indicator, I calculated the
average monthly change of each measure in its own tranquil period. For this calculation, I
defined monthly change as the ratio of the indicator’s value in the current and in the previous
month. The exceptions from this are interest rates indicators and the government budget
balance for which I defined monthly change as the difference in the indicator’s value in the
current and in the previous month. Subsequently, I reviewed the behavior of each indicator in
the 12 months preceding the three crisis episodes and calculated for these pre-crisis 12 months
an average monthly change for each indicator. Appendix 1 presents the average monthly change
for each indicator in the tranquil period as well as the same in the 12 months prior to the three
crisis episodes.
In defining when an indicator signals, I compare the average monthly change during the
tranquil period with the same during the 12 months prior to each crisis episode. The definition
of a signal is when the pre-crisis behavior of an indicator significantly differs from its tranquil
period behavior. The key here is to define what difference should be considered significant, i.e.
determine threshold values for each indicator. My original plan was to fully rely on the
Kaminsky and Reinhart threshold value analysis.54This study has analyzed over 100 currency,
banking and twin crises in industrialized and developing countries for the period of 1970-1995.
Based on this database, the authors have determined the threshold values for each of their
indicator, above (or below) which the indicator correctly identified a crisis. However, for a
number of reasons the Kaminsky-Reinhart threshold analysis is not applicable to the present
case. First, the monetary systems that the authors analyzed were very different from the one in
Hungary in the interwar period. For instance, based on the Kaminsky-Reinhart threshold
analysis, banknotes in circulation should increase prior to a crisis, implying that there was
monetary easing in the months preceding the crisis. Such monetary policy was considered
unorthodox in the interwar period and, as mentioned earlier, the Hungarian central bank was
following the exact opposite path and was tightening the banknotes in circulation. Therefore, a
threshold value which expects a substantial increase to indicate a crisis is not realistic in the
present case. Similar problems arise with the government budget deficit indicator. In modern
periods, a substantial increase of government budget deficit may have forecasted a crisis.
However, in the interwar period, deficit spending was unorthodox and seldom pursued or only
to a very limited extent by modern-day standards. In Hungary it could especially not reach the
proportions implied by the Kaminsky-Reinhart thresholds because the country’s monetary
system was stabilized under the League of Nations reconstruction scheme which required a
balanced budget. Second, there are a number of case specific features which rule out the use of a
generalized threshold mechanism. For instance, the Hungarian financial system had just
regained its stability at the beginning of the period under observation here. Because of that, as
the public was returning to the banking system, deposits of Budapest banks increased by
around an annualized 40% in the tranquil period. When an indicator shows such an outstanding
performance even in a period that is considered tranquil (for lack of a better benchmark) with
54Kaminsky and
Reinhart, 'The Twin Crises: The Causes of Banking and Balance-of-Payments Problems', (
Table 5, p489
36
Flora Macher
London School of Economics
[email protected]
regards to the emergence of financial crises, then a highly generalized threshold will not be able
to catch the relevant difference between the tranquil and the pre-crisis performance. I have
similar concerns in connection with the gold cover, reserves and the banknotes in circulation.
Since the Hungarian central bank was established on June 30, 1925, it was an immature bank of
issue whose monetary aggregates were likely to behave differently from that of a mature bank
of issue. For instance, the gold cover was the highest at the establishment of the bank and was
slowly declining in the months afterwards. This decline should be, in my view, considered
realistic in a period of stabilization but the Kaminsky-Reinhart threshold values may indicate it
to be a sign of crisis.
Because of the above concerns with the Kaminsky-Reinhart signaling mechanism, I have
decided to use threshold values which are not general but specific to the given case. Therefore,
my indicators give a signal when the average monthly change of the particular measure in the
12 months before the crisis episode is one standard deviation above or below (whichever is
relevant to the particular indicator) the average monthly change in the tranquil period.
III.3 - INTERPRETING THE SIGNALS
The results are presented in Table 3. I have also carried out a sensitivity analysis to understand
to what extent the signals of Table 3 are reliable.55 The results of the sensitivity analysis are
presented in Appendix 2. Table 3 shows those signals in highlight which are weak according to
the sensitivity analysis.
Table 3 The signals of macroeconomic indicators
Indicator
Domestic currency deposits at Bp. banks
Foreign currency deposits at Bp. banks
Total deposits at Bp. banks
Domestic currency deposits at non-Bp. banks
Foreign currency deposits at non-Bp. banks
Total deposits at non-Bp. banks
Total deposits
Interest rate on deposits paid by primary Bp. banks
Imports in value
Exports in value
Gold cover (reserves/banknotes in circulation)
Reserves
Banknotes in circulation
Foreign-domestic interest rate differential
Stock exchange index
Albers-Uebele index
Activity ratio for non-agricultural sectors
Government surplus(+)/deficit (-)as a % of revenues
Indicator ticks
for Bp. banking
crisis
1
0
1
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
Indicator ticks
for nationwide
banking crisis
1
0
1
1
0
1
1
0
0
0
1
1
0
0
1
1
0
0
Indicator ticks
for currency
crisis
0
0
0
NA
NA
NA
NA
NA
0
0
0
0
0
0
1
0
0
0
Note: Highlights indicate which measure is weak. See Appendix 2 for details on the sensitivity analysis.
III.3.I - FINANCIAL SECTOR INDICATORS
In connection with the financial sector, I am presenting two indicators: deposit measures
broken down in various ways and the interest rate on deposits offered by Budapest banks. In
the case of the Budapest banking crisis, domestic currency and total deposits give a signal. NonBudapest deposits do not tick for this crisis which confirms that this was a Budapest-episode.
That is: the likelihood of a signal of “1” to switch to a no-signal (zero), or a no-signal to switch to a
signal of “1”.
55
37
Flora Macher
London School of Economics
[email protected]
For the nationwide banking crisis, we see signals for domestic and total deposits for both
Budapest and non-Budapest banks, as well as for total deposits of the whole financial system.
This is strong reassurance that this crisis was country-wide. For the currency crisis, there are no
signals from the available deposit indicators underscoring the view that this was a single
currency crisis that did not influence the banking system. It is also important to note that
foreign currency deposits do not give a signal for either of the two banking crises. This
reinforces the argument expressed in Section II that these banking crises were initiated by
domestic depositors.
The other indicator of the financial system, the interest rate paid on deposits by Budapest
institutions does not give a signal in either of the crisis episodes. At the same time, it should be
noted that the sensitivity analysis has revealed that the zero signal of this indicator is weak for
the nationwide banking crisis. This implies that a less than 10% change in any of the
components of this signal would switch the signal to 1.56 Such a signal would indicate that while
the earlier Budapest banking crisis had not induced Budapest financial institutions to increase
their interest rates, the nationwide banking crisis was already so severe that in order to retain
their depositors they had to resort to this solution.
III.3.II - BALANCE-OF-PAYMENTS INDICATORS
In connection with the monetary system, I have two groups of indicators. To describe the
developments of the current account, I am using imports and exports data. To investigate the
capital account, I rely on the gold cover, which is my proxy for the exchange rate, central bank
reserves, banknotes in circulation and the foreign-domestic interest rate differential.
Current account measures do not give a signal in either of the three crisis episodes. In the case
of imports, the no-signal supports my view that the current account adjustment was
implemented in response to the currency crisis in late 1928. Therefore, in the 12 months
preceding the late 1928 currency crisis, the movements in imports were not substantially
different from the changes in the tranquil period. Structural adjustments to imports came after
the currency crisis. In the case of exports, the no-signal reinforces my argument that, despite
what the Hungarian literature posits, exports did not play an important role in bringing about
either of the three crisis events.
In connection with the capital account, the no signal of the banknotes in circulation underlines
the point that central bank chose not to ease the troubles of the banking system and the real
economy by providing more liquidity. This verifies that the national bank was exerting a strong
control over the banknotes in circulation through its restriction of the liquidity window, and
this way, it kept the gold cover within tight bands. In fact, banknotes in circulation declined in
the 12 months preceding the Budapest banking crisis and only increased by 1.2% and 0.5%
prior to the nationwide banking and currency crises, respectively. At the reserve level, where
the central bank was less able to exert control, things did get somewhat out of hand prior to the
nationwide banking crisis. We see reserves signaling for this crisis episode because this event
saw a bit of a decline in foreign currency deposits. This decline was so minor that it did not
make foreign currency deposit indicators tick for either Budapest or non-Budapest banks (Table
3), but it was sufficient to move the reserves to give a signal. The decline in foreign currency
deposits happened only at Budapest institutions and amounted to a monthly average fall of
0.867%, totaling app. 11m pengős through the 12 months which corresponded to 6% of the
average reserve level in the same period. The reserve change influenced the gold cover as well,
although this signal is considered weak, implying that the overall reserve change did not have a
strong impact. These two signals together suggest that currency problems were slowly
emerging after the Budapest banking crisis and in the wake of the nationwide banking crisis.
56
Based on the sensitivity analysis: 8%. See Appendix 2.
38
Flora Macher
London School of Economics
[email protected]
The fourth indicator of the monetary system illustrates that prior to the three crisis episodes the
Hungarian National Bank did not deviate from its tranquil period interest rate policy. For the
two banking crises, this suggests that in the 12 months preceding these events, the pressure on
the currency was not substantial enough to prompt a significant base rate change. This supports
the view that the 1931 crisis was a banking crisis in its origins and problems with the currency
emerged when the financial system passed the panic on to the monetary system. In connection
with the late 1928 currency crisis, the no-signal of this indicator might appear surprising.
However, considering that this event was caused by an unanticipated sudden decline of foreign
capital inflows which created a liquidity crunch, it is clear that the central bank could not have
reacted in the 12 months preceding the event. Naturally, right after the crisis, the board of the
bank increased the base rate by 1%point and then again by the same amount in six months.
III.3.III - INDICATORS OF THE REAL ECONOMY
I have three indicators to describe the performance of the real economy: the stock exchange
index, the Albers-Uebele index and the activity ratio for non-agricultural sectors. In connection
with the Budapest banking crisis and the nationwide banking crisis, the signals of the stock
market index and the Albers-Uebele index illustrate that prior to these events there was turmoil
in the real economy. The fact that the activity index does not tick is reassuring since this
measure describes the performance of industry and the troubles of the real economy arose in
the agricultural sector. The stock exchange index also signals for the currency crisis.
Nonetheless, the signal is very weak: only 1% change in any of the indicators’ components
would bring about a switch in the signal. The fact that the measure does tick, implies that the
capital flight of the currency crisis may have arose in the form of stock fire sales.
III.3.IV - FISCAL POLICY INDICATOR
My indicator for the influence of government spending is the ratio of government
deficit/surplus to government revenue. The measure is strong based on my sensitivity analysis
and it gives no signal in either of the three crisis episodes. Based on this, the Hungarian
literature’s argument that government spending and borrowing were critical in bringing about
the crisis is not confirmed.
III.3.V – THE MAIN FINDINGS OF THE EARLY-WARNING INDICATOR ANALYSIS
The results of the early-warning indicator analysis are supportive of the interpretation of
Hungarian events introduced in Section II. The analysis confirmed that the three episodes I have
identified were in fact crisis events, since a number of the indicators were giving a signal prior
to all three of them. Second, the results do not confirm the Hungarian literature’s position that
government deficit spending was behind the crisis. Third, the current account indicators
underscore my argument that exports did not play a role in these crisis episodes. Fourth, my
deposit indicators have shown that both the Budapest banking crisis and the nationwide
banking crisis were caused by the flight of domestic depositors. This supports my view that
depositors were fleeing from an ailing banking system and not from a weak currency, unlike the
Hungarian literature suggests. At the same time, and this is my fifth take on this analysis,
monetary indicators show that currency troubles did start to emerge after the Budapest
banking crisis. The capital account indicators do not signal prior to the Budapest banking crisis
but they do give a weak signal before the nationwide banking crisis. This suggests that there
was no currency crisis until July 1931 but we see pressure building up in the monetary system.
This implies that a minor demand for currency conversions arose between the crisis of
Budapest banks and the nationwide banking crisis. Even though this pressure was negligible, it
induced the central bank to further tighten the banknotes in circulation between January and
39
Flora Macher
London School of Economics
[email protected]
June 1931, in order to protect the gold cover. The stability of the currency was hence
maintained through these restrictive measures. Nonetheless, since the flight of domestic
depositors strengthened in July, the pressure from the banking system towards the central bank
increased. When banks’ demand for liquidity became untenable in July, the central bank gave in
and started widening its discount window. After this, as Figure 20 has shown, the increase in the
banknotes in circulation brought about the steep decline in the gold cover.
This sequence of events is fully congruent with the Kaminsky-Reinhart vicious-spiral model in
which the banking system hands on the crisis to the monetary system and receives it back from
the latter but at a higher proportion. The events started off with a Budapest banking crisis in
October 1930, initiated by domestic deposit-holders. This domestic deposit flight afterwards
influenced foreign currency depositors as well and they also sought to get a hold of their money.
Nonetheless, they demanded not only to withdraw their funds from banks but also to convert
their money into foreign currency. Even though the magnitude of these conversions was still
minor, Budapest banks’ demand for foreign currency increased the exposure of the exchange
rate to future shocks and induced the central bank to resort to further tightening to avoid a
currency crisis. Although the renewed tightening of banknotes in circulation secured the
stability of the currency, it further aggravated Budapest banks’ calamities. Since Budapest banks
were financiers of non-Budapest banks, their increasing distress started to affect financial
institutions all over the country and brought about a nationwide banking crisis. The distress of
the entire banking sector then buried the currency under itself.
40
Flora Macher
London School of Economics
[email protected]
SECTION IV - CRISIS TRIGGERS
A final question that requires examination is the trigger of the banking crisis, i.e. why depositors
actually initiated a run on the vulnerable banking sector. Moreover, since the Hungarian case
has presented two banking crises, one in Budapest from the last quarter of 1930 and a
nationwide crisis from mid-1931, I need to identify triggers for both cases.
Figure 21 The triggering mechanism of the Budapest banking crisis
The triggering mechanism of the Budapest banking crisis in Oct
1930
Harvest
Agricultural
prices
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Gold cover
Inflow of
foreign
capital
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
My argument with regards to the Budapest banking crisis is that the loss of public trust was an
important factor. Figure 21 demonstrates the mechanism through which this affected the
system. The loss of domestic trust had been triggered by a number of political events, starting
off with a speech from Prime Minister Bethlen about the country’s untenable financial situation.
In this speech, which occurred in mid-October 1930, Bethlen confessed that there was no
perspective of a foreign state loan in the foreseeable future and the economy had to do without
new lending. The anxious behavior of deposit-holders started right after the speech. Following
the speech, a rumor started circulating that the government would confiscate deposits from
financial institutions and invest them in the economy. A Member of Parliament also demanded
that interest rates be fully regulated and maximized.57 There were calls for nationalizing
financial institutions.58 The public’s anger against banks mounted and was leveraged upon by
the government: they introduced a poverty tax, payable by financial institutions and used for
the payment of unemployment benefits.59 Thus, late 1930 seems to be the point when it became
clear to the public that the country’s financial situation was not sustainable and there was no
prospect for short-term improvement. After this point, depositors’ behavior was very anxious,
as shown on Figure 5.
Magyar Pénzügy, Oct 15, 1930
Magyar Pénzügy, Oct 29, 1930
59 Magyar Pénzügy, Jan 28, 1931
57
58
41
Flora Macher
London School of Economics
[email protected]
Before the nationwide banking crisis in August 1931, the crisis indicator (Table 2) signals three
times. The first signal in October 1930 can be explained by the Bethlen speech. The second
signal in January 1931 may be connected to a demonstration of 200 agricultural producers who
went to Budapest, demanded their loans to be settled and placed the blame on financial
institutions for their current situation.60 In response to the demonstration, the government
started working on a program for the settlement of agricultural loans.61 After the program was
announced in March 1931, agricultural producers stopped servicing their debt. This further
aggravated banks’ liquidity.62 This can explain the third banking crisis signal in April 1931,
hitting Budapest banks.
Figure 22 The triggering mechanism of the nationwide banking crisis
The triggering mechanism of the nationwide banking crisis in
July 1931
Harvest
Agricultural
prices
Agricultural
output
Reserve
level
Exports
Real economy
Monetary system
Nonagricultural
output
Gold cover
Inflow of
foreign
capital
Banking
systems’
lending
potential
Domestic trust
in the banking
system
Availability of
foreign
capital
The trigger behind the nationwide banking crisis is more complex. On the one hand, since
Budapest banks were important financiers of non-Budapest banks, it is likely that the distress of
the former impacted the latter as well. This may explain the January 1931 signal for nonBudapest banks in Table 2. However, the interconnectedness of financial institutions is in itself
not sufficient to explain the nationwide banking crisis: it does not account for the fact that from
August 1931 the nationwide banking crisis persisted for a number of months. While prior to this
date the banking crisis mainly affected Budapest banks and came in ebbs and flows, from
August it influenced non-Budapest banks and was lasting. Something must have happened in
July-August which brought about this change in intensity and change in geographical focus.
Although my findings are still preliminary in this area, I am inclined to believe that the trigger of
the big banking crisis was connected to agriculture. On the one hand, non-Budapest banks were
the main target in this episode and they were the primary financiers of agriculture. On the other
hand, the timing of the crisis also suggests an agricultural event. The potential trigger may have
been the realization, in both a financial and symbolic sense, of the decline that agricultural
prices had undergone in the previous year. The mechanism is depicted on Figure 22. The reason
why this may be a plausible explanation is that the significant deposit flight in July and August,
Magyar Pénzügy, Jan 14, 1931
Magyar Pénzügy, Mar 1, 1931
62 Magyar Pénzügy, Mar 27, 1931
60
61
42
Flora Macher
London School of Economics
[email protected]
occurred right after the wheat harvest. Wheat was the most important commodity of Hungarian
agriculture and its harvest usually took place in late June. Based on this explanation, once the
volume and price of agricultural produce became known in late June, and once producers came
to realize that prices substantially declined from the previous year, a large number of defaults
occurred and the stability of the banking system was entirely undermined. Several notes and
loans came due upon harvest, making this realization an immediate issue of financial survival.
As the above account indicates, I am reluctant to share the view of the Hungarian literature that
crisis propagation from Austria was the key trigger of the Hungarian crisis. Authors propose
that right after the collapse of the Credit-Anstalt it was foreign creditors who started
withdrawing their money from the Hungarian banking system. However, this is not what the
data suggest. As has been demonstrated, the majority of withdrawals were domestic. It was
predominantly the Hungarian public who pulled their money out of the banking system.
Therefore, the literature’s claim about crisis propagation from Austria and the flight of foreign
depositors could only explain a minor portion of the withdrawals and cannot be regarded as a
trigger.63
What I find more relevant is investigating whether the collapse of the Credit-Anstalt could have
influenced domestic deposit withdrawals in Hungary. This is a possibility if Hungarian depositors had
been aware of the connections between Austrian and Hungarian banks. I do believe this is a possibility.
The Credit-Anstalt owned the largest Hungarian bank, the Hungarian General Creditbank. I find it possible
that the domestic public was aware of this and once the distress of the Credit-Anstalt became known, the
Hungarian General Creditbank also came under attack from Hungarian depositors. I further believe that
the German crisis may have had a similar impact on Hungarian banks since German banks were also
active investors in Hungarian banks. Moreover, the German crisis, which erupted in mid-July, may
actually have had a more important effect than the collapse of the Credit-Anstalt since July was the month
of the first big drop in deposits. I plan to investigate the propagation of the crisis among these three
countries in a separate paper.
63
43
Flora Macher
London School of Economics
[email protected]
SECTION V - CONCLUSION
The paper has demonstrated that - rather than contagion from Austria and the flight of foreign
creditors from the region - the Hungarian crisis of 1931 was caused by structural problems of
the domestic real economy and was triggered by the flight of domestic depositors. I have shown
through an analytical narrative and through an early-warning indicator analysis that the
Hungarian episode was a twin crisis which originated in the banking sector and was eventually
transmitted to the monetary system. The two main causes of the crisis were the currency crisis
in late 1928 and the agricultural crisis of 1930. The two together increased the vulnerability of
the banking system which eventually experienced a crisis in 1931 due to the flight of its
domestic financiers.
The late 1928 currency crisis was a testimony to how pernicious the gold exchange standard
was in the interwar period, especially in peripheral markets, heavily dependent on foreign
capital flows. Strong adherence to the gold standard was the prime reason why the Hungarian
National Bank chose to follow a restrictive monetary policy in the years following the 1928
currency episode. By reducing the rediscount volume and hence, tightening the banknotes in
circulation, the central bank squeezed all liquidity out of the economy at a time when the
economy was in a recession. The central bank’s single-minded pursuit of currency stability
undermined the lending power of the banking system and thereby sacrificed long-term growth.
This policy stance, even though widely pursued and considered orthodox at the time, strongly
contributed to the weakness of the banking sector and the events of 1931.
The agricultural crisis was an indication of an inherent structural problem of the Hungarian
economy. Having over 50% of the population who made their livelihood in this sector was a
great exposure for the economy. It is hence, not surprising that the government which, unlike
the central bank, was accountable to its electorate, decided to experiment with measures that
were regarded unorthodox at the time. The provision of state-guaranteed loans was not at all in
accordance with the fiscal austerity requirements of the gold exchange standard. They
demonstrated the struggle of a government trying to help the large part of the population who
were most hit by the agricultural crisis. However, the agricultural crisis further deteriorated in
1931 and these guarantees made no difference for the sector. They did, however, increase the
vulnerability of the banking system which was afterwards even more exposed to this
exceptionally crisis-ridden sector of the economy. In addition, by favoring the agricultural
sector, other parts of the economy were even more squeezed of liquidity and their contraction
worsened. Thereby the banking sector was exposed to a real economy, which was weak in every
corner.
The above policy issues might ring familiar to the reader in the aftermath of the current
economic recession. We are seeing a dovish vs. hawkish monetary policy divide across the
Atlantic, with the US Federal Reserve supportive of monetary easing and the European Central
Bank advocating much tighter measures.64 And if we go back to the origins of the 2008 financial
crisis, we can identify fiscal policy measures behind it. Dating back to the days of Bill Clinton’s
presidency, US administrations had not put constraints on subprime lending. Policy-makers and
regulators overlooked egregious lending practices because they could build political capital by
being supportive of widespread home-ownership. The fact that we are returning to the same
patterns well illustrates Kindleberger’s point that financial crises are a “hardy perennial”.65
At the same time, the restrictive monetary policy stance of the European Central Bank (ECB) seems to
be passing since the election of Mario Draghi as president in the aftermath of the eurozone crisis.
Nonetheless, the ECB still remains restrictive vis-a-vis the Fed and any easing seems to be opposed by the
German Bundesbank.
65 Kindleberger, Manias, Panics and Crashes.
64
44
Flora Macher
London School of Economics
[email protected]
REFERENCES AND SOURCES
REFERENCES
Albers, Thilo and Uebele, Martin (2014), 'Reassessing Interwar Business Cycles From a Global
Perspective: a New Monthly Dataset', unpublished paper.
Balderston, Theo (1994), 'The banks and the gold standard in the German financial crisis of
1931', Financial history review, 1 (1), 43-68.
Berend, T. Ivan (1982a), 'A világgazdasági válság (1929-1933) sajátos hatásai Közép-Kelet
Európában', Történelmi Szemle, 25 (1).
Berend, T. Iván (1982b), Válságos évtizedek: Közép- és Kelet-Európa a két világháború között
(Budapest: Gondolat Könyvkiadó).
--- (1987a), 'A 20. század nagy gazdasági válságai a történelem folyamataiban (Hasonlóságok és
különbségek az 1930-as és az 1970-1980-as évek között)', in T. Ivan Berend and Knut
Borchard (eds.), Válság, recesszió, társadalom : Az 1930-as és az 1970-1980-as évek
összehasonlítása : Válogatott tanulmányok (Budapest: Közgazdasági és Jogi Könyvkiadó).
--- (1987b), Válságos évtizedek: A 20. század első fele közép- és kelet-európai történetek
interpretációja (Budapest: Magvető Könyvkiadó).
Berend, T. Ivá n (1998), Decades of crisis : Central and Eastern Europe before World War II
(Berkeley, Calif.: Berkeley, Calif. : University of California Press).
Berend, T. Ivá n, et al. (1966), Magyarország gazdasága az elsö világháború után, 1919-1929 (4;
Budapest: Akadé miai Kiadó ).
Bordo, Michael, et al. (2001), 'Is the crisis problem growing more severe?', Economic Policy, 16
(32), 51-82.
Bordo, Michael D. and Murshid, Antu P. (2000), 'Are Financial Crises Becoming Increasingly
More Contagious? What is the Historical Evidence on Contagion?', (National Bureau of
Economic Research).
Calomiris, Charles W. (2014), Fragile by design : the political origins of banking crises and scarce
credit, ed. Stephen H. Haber (Princeton : Princeton University Press).
Dooley, Michael P. (2000), 'A Model of Crises in Emerging Markets', The Economic Journal, 110
(460), 256-72.
Dr. Surányi-Unger, Tivadar (1936), Magyar Nemzetgazdaság és Pénzügy (Pécs: Dunántúli Pécsi
Egyetemi Könyvkiadó és Nyomda).
Eichengreen, Barry (1996), Golden Fetters (Oxford University Press).
Eichengreen, Barry, Rose, Andrew, and Wyplosz, Charles (1996), 'Contagious Currency Crises:
First Tests', The Scandinavian Journal of Economics, 98 (4), 463-84.
Ellis, Howard S. (1939), 'Exchange Control In Austria and Hungary', The Quarterly Journal of
Economics, 54 (1), 1-185.
Feinstein, C. H. (1995), Banking, currency, and finance in Europe between the Wars (Oxford:
Clarendon Press [u.a.]) xviii.
Ferguson, Thomas and Temin, Peter (2003), Made in Germany: The German Currency Crisis of
July 1931 1-53.
Friedman, Milton and Schwartz, Anna J. (1963), A monetary history of the United States 18671960 (12.; Princeton: Princeton University Press).
James, Harold (1984), 'The Causes of the German Banking Crisis of 1931', The Economic History
Review, 37 (1), 68-87.
--- (2002), The end of globalization: lessons from the Great Depression (Cambridge, Mass:
Harvard University Press) vi-vi.
Kaminsky, Graciela L. and Reinhart, Carmen M. (1999), 'The twin crises: the causes of banking
and balance-of-payments problems', The American Economic Review, 89 (3), 473-500.
45
Flora Macher
London School of Economics
[email protected]
Kaser, Michael and Nötel, Rudolf (1987), 'Kelet-Európa gazdaságai a két világválságban (Eastern
European Economies in Two World Crises)', in T. Ivá n Berend and Knut Borchard (eds.),
Válság, recesszió, társadalom : Az 1930-as és az 1970-1980-as évek összehasonlítása :
Válogatott tanulmányok (Budapest: Közgazdasági és Jogi Könyvkiadó).
Kindleberger, Charles Poor (1987), The world in depression, 1929-1939 (Harmondsworth:
Penguin Books).
--- (2001), Manias, panics and crashes (Basingstoke: Palgrave).
Krugman, Paul (1979), 'A Model of Balance-of-Payments Crises', Journal of Money, Credit and
Banking, 11 (3), 311-25.
--- (1999), 'Balance sheets, the transfer problem, and financial crises', International tax and
public finance, 6 (4), 459-72.
Reinhart, Carmen and Vegh, Carlos (1999), 'Do Exchange Rate-Based Stabilizations Carry the
Seeds of Their Own Destruction?', in MPRA Paper (ed.).
Reinhart, Carmen M., Rogoff, Kenneth S., and Books24x, Inc (2009), This time is different: eight
centuries of financial folly (Princeton: Princeton Univ. Press) xlv.
Schnabel, Isabel (2004a), 'The German twin crisis of 1931', The Journal of Economic History, 64
(3), 822-71.
--- (2004b), 'Reply to Thomas Ferguson and Peter Temin's "Comment on The German twin crisis
of 1931"', The Journal of Economic History, 64 (3), 877-78.
Szuhay, Miklós and Berend, T. Iván (1978), A tőkés gazdaság története Magyarországon: 18481944 (Budapest: Közgazdasági és Jogi Könyvkiadó).
Temin, Peter (1991), Lessons from the Great Depression (1989; Cambridge, Mass: MIT Press).
--- (2008), 'The German crisis of 1931: evidence and tradition', Cliometrica, 2 (1), 5-17.
46
Flora Macher
London School of Economics
[email protected]
ARCHIVAL SOURCES
Hungarian National Archive (HNA)
File Z6, box 1-2 - Minutes of the Board of Governors of the Hungarian National Bank
File Z12, binder 60, 61,119,128 – Department of Economics, Statistics and Research at the
Hungarian National Bank
File Z91 – Minutes of the Board of the Central Commission of Financial Institutions
Files Z92, Z923, Z925, Z926 – other files of the Central Commission of Financial Institutions
Bank of England Archive
Files OV9/436-439 and OV9/234, Papers of Otto Ernst Niemeyer, League of Nations
LSE Archive
League of Nations files on the reconstruction of Hungary
47
Flora Macher
London School of Economics
[email protected]
SOURCES OF FIGURES
Figure
Data
Source
Figure 1
The framework of the Hungarian economy
Own work
Figure 2
The mechanism behind the arguments of the Hungarian literature
Own work
Figure 3
The Exchange Market Pressure (EMP) index
Figure 4
Figure 5
includes: the gold cover
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 129. csomó (File Z12, binder 129)
includes: the base rate of the central bank
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z6, 1-2. doboz (File Z6, box 1-2)
includes: the reserves of the central bank
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Total assets under insolvency as a share of banks' deposits
includes: the assets involved in insolvency
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
includes: demand and term deposits for Budapest and non-Budapest institutions
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
The monthly change in deposits for Budapest and non-Budapest banks
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
includes demand and term deposits
includes foreign and domestic currency deposits
sample: the 12/13 largest Budapest banks, the 34/35 largest non-Budapest banks
sample covers app. 85% of total deposits in the country
Figure 6
Hungary's balance-of-payments
Statisztikai Szemle (Statistical Review)
48
Flora Macher
London School of Economics
Figure
Data
Source
Figure 7
Hungary's trade account
Statisztikai Havi Közlemények (Monthly Statistical Report)
Figure 8
The central bank's rediscount and the gold cover
[email protected]
includes: total rediscount
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 128. csomó (File Z12, binder 18)
includes: the gold cover
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 129. csomó (File Z12, binder 129)
Figure 9
The mechanism behind the currency crisis of late 1928
Own work
Figure 10
Rearranged balance-of-payments
Statisztikai Szemle (Statistical Review)
Figure 11
The share of agriculture in the central bank's rediscount
includes: total rediscount
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 128. csomó (File Z12, binder 18)
includes: agricultural bills rediscounted
Hungarian National Archive, Z12, 60. csomó (File Z12, binder 60)
Figure 12
Domestic National Income (DNI)
Eckstein, 1956
Statisztikai Szemle (Statistical Review)
Figure 13
The decomposition of the change in agricultural DNI
Eckstein, 1956
Statisztikai Szemle (Statistical Review)
Figure 14
The mechanism of the agricultural crisis of 1930
Own work
Figure 15
New financing sources of the banking sector
Nagy Magyar Compass (Big Hungarian Compass)
Figure 16
Lending by the banking sector
Nagy Magyar Compass (Big Hungarian Compass)
Figure 17
The share of agricultural lending in total lending
Nagy Magyar Compass (Big Hungarian Compass)
49
Flora Macher
Figure
Figure 18
Figure 19
Figure 20
London School of Economics
Data
The mechanism behind the 1931 crisis
The effects of the banking crisis on the central bank
includes: banknotes in circulation
includes: total rediscount
includes: the reserves of the central bank
Decomposing the changes in the gold cover
includes: banknotes in circulation
includes: the reserves of the central bank
[email protected]
Source
Own work
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 128. csomó (File Z12, binder 18)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
50
Flora Macher
London School of Economics
[email protected]
SOURCES OF EARLY-WARNING INDICATORS
Indicator
Domestic currency deposits at Bp. banks
Foreign currency deposits at Bp. banks
Total deposits at Bp. banks
Domestic currency deposits at non-Bp. banks
Foreign currency deposits at non-Bp. banks
Total deposits at non-Bp. banks
Total deposits
Interest rate on deposits paid by primary Bp. banks
Imports in value
Exports in value
Gold cover (reserves/banknotes in circulation)
Reserves
Banknotes in circulation
Foreign-domestic interest rate differential
Stock exchange index
Albers-Uebele index
Activity ratio for non-agricultural sectors
Government surplus(+)/deficit (-)as a % of revenues
Source
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Albers-Uebele
Albers-Uebele
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 128. csomó (File Z12, binder 18)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Hungarian National Archive, Z12, 1-2. doboz (File Z12, boxes 1-2)
Bank of England website:
(http://www.bankofengland.co.uk/statistics/Documents/rates/baserate.pdf)
Statisztikai Havi Közlemények (Monthly Statistical Report)
Albers-Uebele
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
Magyar Gazdaságkutató Intézet (Institute for Hungarian Economic Research)
51
Flora Macher
London School of Economics
[email protected]
APPENDIX 1
Definition of tranquil periods, calculation of monthly averages and standard deviations for the tranquil periods and the calculation of the
pre-crisis 12-month averages
Indicator
Domestic currency deposits at Bp. banks
Foreign currency deposits at Bp. banks
Total deposits at Bp. banks
Domestic currency deposits at non-Bp.
banks
Foreign currency deposits at non-Bp. banks
Total deposits at non-Bp. banks
Total deposits
Interest rate on deposits paid by primary
Bp. banks
Imports in value
Exports in value
Gold cover (reserves/banknotes in
circulation)
Reserves
Banknotes in circulation
Foreign-domestic interest rate differential
Stock exchange index
Albers-Uebele index
Activity ratio for non-agricultural sectors
Government surplus(+)/deficit (-)as a % of
revenues
St. dev.
of
average
monthly
change
of
tranquil
periods
0.0232
0.0359
0.0237
Average
monthly
change
in
tranquil
periods
0.0279
0.0118
0.0363
Average
monthly
change
in the
crisis
period
for Bp.
banking
crisis
0.0039
-0.0010
0.0026
Average
monthly
change in
the crisis
period for
nationwide
banking
crisis
-0.0083
-0.0087
-0.0086
Average
monthly
change
in the
crisis
period
currency
crisis
0.0265
0.0008
0.0215
Notes to Bp.
banking
crisis
average
9-month ave.
9-month ave.
9-month ave.
9-month ave.
Start date
of time
series
31-Jan-28
31-Jan-28
28-Feb-25
Start of
tranquil
period
31-Jan-28
31-Jan-28
30-Jun-26
End of
tranquil
period
31-Jan-29
31-Jan-29
30-Jun-27
Number
of
tranquil
periods
12
12
12
31-Jan-30
31-Jan-30
31-Jan-30
31-Jan-30
31-Jan-30
31-Jan-30
31-Jan-30
31-Jan-30
30-Sep-30
30-Sep-30
30-Sep-30
30-Sep-30
8
8
8
8
0.0087
0.0337
0.0083
0.0120
0.0059
0.0023
0.0058
0.0084
0.0054
-0.0028
0.0052
0.0057
-0.0049
0.0019
-0.0049
-0.0082
NA
NA
NA
NA
31-Jan-29
28-Feb-25
28-Feb-25
31-Jan-29
30-Jun-26
30-Jun-26
31-Jan-30
30-Jun-27
30-Jun-27
12
12
12
0.6250
0.1168
0.2088
0.0208
0.0391
0.0263
-0.2083
-0.0074
-0.0123
0.1458
-0.0222
-0.0129
NA
-0.0041
0.0018
28-Feb-25
28-Feb-25
28-Feb-25
28-Feb-25
31-Jan-26
28-Feb-25
31-Jan-26
30-Jun-26
30-Jun-26
30-Jun-26
30-Jun-26
30-Jun-26
30-Jun-26
30-Jun-26
30-Jun-27
30-Jun-27
30-Jun-27
30-Jun-27
30-Jun-27
30-Jun-27
30-Jun-27
12
12
12
12
12
12
12
0.0436
0.0372
0.0405
0.7500
0.0808
0.0152
0.0280
0.0063
0.0106
0.0054
-0.0417
0.0775
0.0111
0.0115
0.0060
-0.0046
-0.0070
0.0833
-0.0084
-0.0078
-0.0063
-0.0374
-0.0369
0.0123
0.2500
-0.0147
-0.0073
-0.0041
-0.0138
-0.0122
0.0053
0
-0.0038
- 0.0022
0.0003
28-Feb-27
28-Feb-27
29-Feb-28
12
0.1231
0.0095
-0.0149
-0.0284
-0.0111
Notes to
nationwide
banking
crisis
average
Notes to
currency
crisis
average
9-month ave.
9-month ave.
No data
No data
No data
No data
No data
8-month ave.
8-month ave.
52
Flora Macher
London School of Economics
[email protected]
APPENDIX 2
Sensitivity analysis
I have checked the sensitivity of my indicators to the following: i) changes in the tranquil period average; ii) changes in the standard deviation of the tranquil period; iii) changes in the
threshold of one standard deviation. I have calculated by what percentage these three measures need to change in order for the indicator’s signal to switch from its original sign (i.e. to
switch from zero to one or vice versa). If the change to cause the indicator to switch is less than 10%, then I consider the indicator’s signal weak.
Domestic currency deposits at Bp. banks
Foreign currency deposits at Bp. banks
Total deposits at Bp. banks
Domestic currency deposits at non-Bp. banks
Foreign currency deposits at non-Bp. banks
Total deposits at non-Bp. banks
Total deposits
Interest rate on deposits paid by primary Bp. banks
Imports in value
Exports in value
Gold cover (reserves/banknotes in circulation)
Reserves
Banknotes in circulation
Foreign-domestic interest rate differential
Stock exchange index
Albers-Uebele index
Activity ratio for non-agricultural sectors
Government surplus(+)/deficit (-)as a % of revenues
Sensitivity to the change in
tranquil period standard deviation
Indicato
Indicator
Indicato
r ticks
ticks for
r ticks
for Bp. nationwide
for
banking
banking currency
crisis
crisis
crisis
4%
64%
-128%
-128%
-64%
-128%
64%
128%
-64%
-128%
32%
NA
-128%
-128%
NA
-128%
32%
NA
-128%
64%
NA
-32%
8%
NA
-64%
-64%
-64%
-128%
-128%
-128%
-128%
1%
-64%
-64%
32%
-64%
-256%
-128%
-128%
-64%
-64%
-128%
8%
16%
1%
32%
32%
-16%
-64%
-64%
-64%
-16%
-16%
-16%
Sensitivity to the change in
tranquil period average monthly
change
Indicato
Indicator
Indicato
r ticks
ticks for
r ticks
for Bp. nationwide
for
banking
banking currency
crisis
crisis
crisis
-4%
-64%
128%
256%
256%
256%
-32%
-64%
32%
256%
-64%
NA
2048%
2048%
NA
256%
-64%
NA
128%
-64%
NA
-32%
8%
NA
256%
256%
256%
1024%
1024%
1024%
1024%
-1%
512%
256%
-128%
256%
-1024%
-1024%
-1024%
-64%
-64%
128%
-8%
-16%
-1%
-64%
-32%
32%
128%
128%
256%
16%
16%
16%
Sensitivity to the change of
standard deviation threshold
Bp.
bankin
g crisis
4%
-128%
64%
-128%
-128%
-128%
-128%
-256%
-64%
-128%
-128%
-64%
-256%
-128%
8%
32%
-64%
-128%
Nationwide
banking
crisis
64%
-64%
128%
32%
-128%
32%
64%
-128%
-64%
-128%
1%
32%
-128%
-64%
16%
32%
-64%
-128%
Currenc
y crisis
-128%
-128%
-64%
NA
NA
NA
NA
NA
-64%
-128%
-64%
-64%
-128%
-128%
1%
-16%
-64%
-128%
53