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
© Copyright 2024