The Great Leveraging

Transcription

1 1/61 The Great Leveraging Moritz Schularick University of Bonn SAFE, 17 June 2016 Macrohistory Lab Bonn

2 When you combine ignorance and leverage, you get some pretty interesting results. Warren Buffett 2/61

3 3/61 Explicandum 1 financial crises happen: why? Crisis frequency since 1800, all economies Share of countries in a financial crisis (%) High-income economies (N=23) Middle- & low-income economies (N=44) Year

4 4/61 Explicandum 2 financial crises are painful: why? log real GDP log real GDP GAP = jan jan jan jan jan jan jan jan jan2012 EZ pretrend trend

5 5/61 Explicandum 2 financial crises are painful: why? log real GDP log real GDP GAP = jan jan jan jan jan jan jan jan jan2012 US pretrend trend

6 6/61 Motivation Core questions What makes financial systems crisis-prone? What makes financial crises costly? Have these things changed over time?

7 6/61 Motivation Core questions What makes financial systems crisis-prone? What makes financial crises costly? Have these things changed over time? Our approach Combining international macroeconomic history and empirical macroeconomics. First-order questions: history has much to offer. Long-run perspective: rare events and stylized facts.

8 7/61 Outline A Presentation in Three Parts The financial hockey stick Origins of financial crises Consequences of financial crises References: Schularick and Taylor, Credit booms gone bust (AER 2012) Jordà, Schularick, and Taylor, When credit bites back (JMCB 2013) Jordà, Schularick, and Taylor, Sovereigns versus banks (JEEA 2015) Jordà, Schularick, and Taylor, Betting the house (JIE 2015) Jordà, Schularick, and Taylor, Leveraged bubbles (JME 2015) Jordà, Schularick, and Taylor, The great mortgaging (EP 2016) Funke, Schularick, and Trebesch, Going to extremes (EER 2016) Knoll, Schularick, and Steger, No price like home (AER, cond. acc.)

9 In the brief history of macroeconomics, the subject of money and banking has witnessed wide fluctuations in both its internal consensus and external influence. The crisis of 8/61 Credit Booms Gone Bust: Monetary Policy, Leverage Cycles, and Financial Crises, By MORITZ SCHULARICK AND ALAN M. TAYLOR The crisis of the advanced economies in has focused new attention on money and credit fluctuations, financial crises, and policy responses. We study the behavior of money, credit, and macroeconomic indicators over the long run based on a new historical dataset for 14 countries over the years , using the data to study rare events associated with financial crisis episodes. We present new evidence that leverage in the financial sector has increased strongly in the second half of the twentieth century as shown by a decoupling of money and credit aggregates. We show for the first time how monetary policy responses to financial crises have been more aggressive post-1945, but how despite these policies the output costs of crises have remained large. Importantly, we demonstrate that credit growth is a powerful predictor of financial crises, suggesting that such crises are credit booms gone wrong and that policymakers ignore credit at their peril. It is only with the long-run comparative data assembled for this paper that these patterns can be seen clearly. (JEL: E44, E51, E58, G01, G20, N10, N20.)

10 9/61 Contribution The original ST AER 2012 paper A new long-run annual panel database Key financial variables for advanced countries

11 9/61 Contribution The original ST AER 2012 paper A new long-run annual panel database Key financial variables for advanced countries Building on that paper we have moved on... Developed, refined, and extended the dataset Applied to other macro-finance questions Methodological refinements (local projections, synthetic control groups, etc.)

12 10/61 Where we stand now 17 advanced economies extending Schularick and Taylor (2012) Annual since real and nominal variables MFI credit to private non-financial sector: mortgages vs. non-mortgages; households vs. firms Safe and risky interest rates House prices and rents Equity prices and returns Remarks: Many series available but scattered: now one source 150+ pages online appendix with sources Soon data and appendix at

13 11/61 Table 1: A new macro-financial dataset. Available samples per variable and per country Country R. GDP N. GDP Cons. Inv. C. Acc. Gov. Exp. Gov. Rev. CPI Nw. mon. S. rate L. rate Stocks Ex. rate Pub. debt Bank lend. Mort. lend. Australia Belgium Canada Switzerland Germany Denmark Spain Finland France UK Italy Japan Netherlands Norway Portugal Sweden USA

14 The financial hockey stick 12/61

15 13/61 The financial hockey stick Figure 1. Aggregates Relative to GDP (Year Effects) year Bank Loans/GDP Broad Money/GDP Bank Assets/GDP

16 14/61 Trends Age of Money ( s) Money and credit were tightly linked and maintained a fairly stable relationship relative to GDP Recovery from the collapse from 1940s to 1970s in a period of low leverage/financial repression/regulation (with no financial crises)

17 14/61 Trends Age of Money ( s) Money and credit were tightly linked and maintained a fairly stable relationship relative to GDP Recovery from the collapse from 1940s to 1970s in a period of low leverage/financial repression/regulation (with no financial crises) Age of Credit (1970s 2008) Continued and unprecedented rise of leverage and growth of non-monetary liabilities of banks Decoupling of credit from money

18 15/61 Households leveraging up more than businesses Ratio of debt to GDP Corporate debt Household debt

19 16/61 The Great Mortgaging Ratio of bank lending to GDP Nonmortgage lending Mortgage lending

20 17/61 Change in bank lending to GDP ratios, (1) (2) (3) (4) (5) Country Total lending Mortgage Non-mortgage Households Business Netherlands Denmark Australia Spain Portugal USA* USA Sweden Great Britain Canada Finland Switzerland France Italy Belgium Germany Norway Japan Average Fraction of Average

21 Business history Deutsche Bank 18/61

22 Business history Deutsche Bank 19/61

23 20/61 Banks no longer do what you think they do Reality check: the textbook model vs. modern banking. What advantages do banks have in providing mortgage credit?

24 Liability structure 21/61

25 Capital 22/61

26 23/61 Great Leveraging = Great Mortgaging Sharp increase of aggregate debt driven by real estate borrowing of households Three margins (extensive / intensive) More people borrow to buy houses (ownership) People buy more expensive houses (price effect) People borrow more against house value (leverage)

27 24/61 Owner occupied share of housing in percent CAN GER FRA ITA CHE U.K. U.S. Avg

28 Long-run house prices 25/61

29 26/61 House prices Compiled a new dataset that allows the systematic study of real house prices in 14 advanced economies between 1870 and Since the 1870s: Australia, Belgium, Germany, Denmark, France, the Netherlands, Norway, Sweden. Since the 1890s/1900s: UK, US, Finland, Japan, Switzerland. Since the early 1920s: Canada. Decompose house prices into its two main components: structure and land. Long-run construction costs proxy the replacement value. Compare imputed land prices with observed prices.

30 27/61 The housing hockey stick Real house prices Real house prices

31 28/61 Decomposing house prices: land and structure Housing sector (competitive) production function: F(Z, X) = (Z t ) α (X t ) 1 α, α (0, 1) combining land Z and residential structures X Profit maximization yields: p H t = B(p Z t )α (p X t )1 α with B := (α) α (1 α) (1 α) house price depends on the price of land p Z t and the price of (quality-adjusted) residential structures p X t Using house price data and the price of structures, we back out the growth rate of land prices: ( ) 1 ) α 1 α = p Z t+1 p Z t p H t+1 p H t α ( p X t+1 p X t

32 Decomposing house prices: Construction costs 29/61

33 Imputed land prices 30/61

34 House prices vs. rents 31/61

35 32/61 Leverage Table : Leverage U.S., U.K., France, and Germany. (a) All wealth, all lending (b) Housing wealth, mortgages (r)2-5 (r) Loans/GDP U.S U.K France Germany Average Loans/Wealth U.S U.K France Germany Average Wealth/GDP U.S U.K France Germany Average

36 The origins of crises 33/61

37 Credit and crises: UK 34/61

38 Credit and crises: Switzerland 35/61

39 Credit and crises: Sweden 36/61

40 37/61 The origins of crises Economic conditions at t 1, t 2,... crisis at time t logit(p it ) = b 0i + b 1 (L) logcredit it + b 2 (L)X it + e it where ( ) logit(p) = ln p 1 p is the log odds ratio of a crisis b i (L) is a polynominal in the lag operator L We have also tried a linear probability specification (and a variety of fixed effects), but the results are robust.

41 38/61 Baseline Model TABLE 4 BASELINE MODEL AND ALTERNATIVE MEASURES OF MONEY AND CREDIT (6) (7) (8) Specification Baseline Replace Replace (Logit country effects) loans with loans with broad narrow money money L.Dlog(loans/P) (2.05) (2.94) (1.37) L2.Dlog(loans/P) 7.215*** 5.329** (1.99) (2.52) (1.68) L3.Dlog(loans/P) (1.83) (2.63) (1.77) L4.Dlog(loans/P) (1.49) (2.51) (1.65) L5.Dlog(loans/P) * (1.78) (2.30) (1.82) Observations Groups Avg. obs. per group Sum of lag coefficients 10.02*** 12.23*** se Test for all lags = 0, χ *** 18.35*** p value Test for country effects = 0, χ p value Pseudo R Notes: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors in parentheses.

42 39/61 A distraction: public debt Classifier logit model (1) (2) (3) (4) (5) Change in private credit/gdp 21.79*** 21.34*** 26.63** (5-year moving average) (5.39) (5.44) (13.00) Change in public debt/gdp (5-year moving average) (1.88) (3.68) (3.29) Lagged level of private credit/gdp (0.63) Lagged level of public debt/gdp (0.29) (Lagged level of private credit/gdp) (Lagged level of public debt/gdp) (9.34) (3.02) Observations Area under the curve (AUC) (0.03) (0.03) (0.03) (0.03) (0.03) Public debt does not predict financial crises, private credit does. Public debt (typically) rises after crises, not before.

43 40/61 Asset price bubbles and financial crises Logit model to predict a financial crisis event Full sample Post-WW2 sample (1) (2) (3) (4) (5) (6) Benchmark Credit Full Benchmark Credit Full only model only model Credit (0.11) (0.17) No bubble credit (0.18) (0.35) Equity bubble credit (0.18) (0.29) Housing bubble credit (0.20) (0.30) Both bubbles credit (0.30) (0.50) Pseudo-R AUC (.05) (.05) (.05) (.07) (.06) (.06) Observations

44 Leverage, debt overhang and and consequences of crises 41/61

45 42/61 What about asset price amplification? [O]ver-investment and over-speculation are often important; but they would have far less serious results were they not conducted with borrowed money. Irving Fisher, 1933 All of us knew there was a bubble. But a bubble in and of itself doesn t give you a crisis... It s turning out to be bubbles with leverage. Former Federal Reserve Chair Alan Greenspan, CNBC Squawk Box, 2013

46 Bubble trouble Journal of Monetary Economics 76 (2015) S1 S20 Contents lists available at ScienceDirect Journal of Monetary Economics journal homepage: Leveraged bubbles $ Òscar Jordà a,b, Moritz Schularick c,d, Alan M. Taylor b,d,e,f,n a Federal Reserve Bank of San Francisco, USA b Department of Economics, University of California, Davis, USA c Department of Economics, University of Bonn, Germany d CEPR, UK e Graduate School of Management, University of California, Davis, USA f NBER, USA article info Article history: Received 30 November 2014 Received in revised form 24 August 2015 Accepted 26 August 2015 Available online 15 September 2015 Keywords: Boom Bust Bank lending Debt overhang Crises Local projections abstract What risks do asset price bubbles pose for the economy? This paper studies bubbles in housing and equity markets in 17 countries over the past 140 years. History shows that not all bubbles are alike. Some have enormous costs for the economy, while others blow over. We demonstrate that what makes some bubbles more dangerous than others is credit. When fueled by credit booms, asset price bubbles increase financial crisis risks; upon collapse they tend to be followed by deeper recessions and slower recoveries. Credit-financed housing price bubbles have emerged as a particularly dangerous phenomenon. & 2015 Published by Elsevier B.V. [O]ver-investment and over-speculation are often important; but they would have far less serious results were they not conducted with borrowed money. 43/61

47 44/61 What we ask Are all bubbles alike? Idea that credit boom + bubble can be dangerous. What s the evidence? What we find Housing bubble + credit buildup deeper recession + slower recovery Equity bubbles less damaging

48 45/61 What is an asset price bubble? Deviation from fundamentals? Many definitions available... Lack of consensus + lack of data a pragmatic solution. Let z be a log asset price, z HP denotes its HP cycle. Price Elevation Signal it = I(z HP it > standard deviation of z HP in country i ) Price Correction Signal it = I(z i,t+3 z it < 0.15 for some year t within the episode) Bubble Signal it = d it = (Price Elevation Signal it Price Correction Signal it )

49 46/61 Examples of famous bubble episodes Using our bubble signal definition Housing bubble: AUS, Housing bubble: SWE, Housing bubble: USA, Equity bubble: JPN, Equity bubble: USA, Equity bubble: USA,

50 47/61 Empirical strategy Treatment-response framework: how is the path of the economy through expansion and recession affected by debt overhang? Regress change in log real GDP per capita from peak to year h on treatments: normal/financial recession (N, F), credit and bubble indicator Use semiparametric approach for added flexibility and to examine non-linearities easily To do all this use methods of local projections (Jordà 2005)

51 48/61 Benchmark LP path Full sample: Equity bubbles Housing bubbles Percent (100x log) Recession Bubble, low credit Bubble, high credit Year Percent (100x log) Year

52 49/61 Conclusions Leveraged bubbles spell trouble: ignore at your peril The combination of asset price bubbles and credit booms substantially increases financial fragility House price bubbles are less frequent than equity bubbles and more often end up in financial crises Credit fueled housing bubbles are particular costly: they lead to deeper recessions and slower recoveries

53 Political consequences of financial crises 50/61

54 Goodbye financial crisis, hello political crash 51/61

55 52/61 The question There are clear signs of political radicalization in recent years Our question: How does the political landscape change after financial crises? Does governing become more difficult, crisis resolution harder, and policy uncertainty more pronounced?

56 53/61 What we do We turn to quantitative economic history to study rare crisis events and their political aftermath We code a new large dataset covering 800 general election results in 20 countries since 1870 We then quantify the political after-effects of more than 100 financial crises

57 54/61 What we find 1 Hard right turns: The political far right is the biggest beneficiary from financial crises. 2 Governing becomes more difficult: Crises are associated with shrinking government majorities, greater fractionalization, and increasing street protest. 3 The effects appear specific to the aftermath of financial crises: Not observable in either normal recessions or severe economic downturn without a financial crisis.

58 55/61 Why is the reaction to financial crisis different? The public may not see crises as excusable events: Rather the result of policy failures, moral hazard and political favoritism: desire for accountability. At the same time, policy responses to financial crises often have the opposite effects by design: Dissatisfaction with bail-outs. Order amidst fear and chaos: The appeal of authoritarianism.

59 Political variables: voting and parliament Variable Description Sources Government vote share Vote share of governing party or coalition in the most recent general elections to the national parliament (lower chamber) Mackie and Rose (1974), Nohlen and Stöver (2010), Döring and Manow (2012) and countryspecific sources Opposition vote share Combined vote share of all opposition parties, excluding independents, in the most recent general elections to the national parliament (lower chamber) Mackie and Rose (1974), Nohlen and Stöver (2010), Döring and Manow (2012) and countryspecific sources Far-right/ far-left vote share Combined vote share of all farright (far-left) political parties with more than 0.1 % of total votes in the most recent general elections to the national parliament (lower chamber) Bertelsmann Foundation (2009), Betz (1994), Capoccia (2012), de Bromhead, Eichengreen, and O Rourke (2012), Minkenberg, (2001, 2008), Mudde (2000, 2005, 2007) and country-specific sources Fractionalization The probability that two representatives picked at random from among the parties in the legislature will be of different parties; range: [0;1] Mackie and Rose (1974), Nohlen and Stöver (2010), Döring and Manow (2012) and countryspecific sources Number of parties The number of parties elected into the legislative branch in the most recent general election to the national parliament (lower chamber) Mackie and Rose (1974), Nohlen and Stöver (2010), Döring and Manow (2012) and countryspecific sources 56/61

60 57/61 The tragedy of the left Percentage of total votes Pre crisis Post crisis Far left (Avg.) Far right (Avg.) Total (Avg.) Notes: The figure shows average vote shares of far-left (white columns) and far-right (black columns) political parties. The grey columns represent the sum of the two. The left panel refers to average vote shares in the five years before the start of a financial crisis and the right panel shows average shares in the five years after. The differences are statistically significant at the 5% level, except for the far-left vote share, which does not significantly increase in the post-crisis period. All crises from 1919 to Years that are simultaneously pre-crisis and post-crisis years, years in which financial crisis erupts and World War II years are excluded.

61 58/61 Time path: right wing votes Full sample Pre WWII sample Post-WWII sample Year Year Year Notes: Each path shows local projections of the cumulative change in 100 times the logged variable relative to peak for years 1-5 of the recession/recovery period. The red line refers to the average path in financial crisis recessions and the shaded region is a 90% confidence interval. The controls are contemporaneous and 1-year lagged values of the growth rate of real GDP per capita and the CPI inflation rate at peak. The left panel cover the years , with World War II being excluded, the middle panel , and the right panel The dependent variable is the combined vote share of all electorally successful far-right political parties in the most recent general election.

62 59/61 Crises and other (severe) recessions How does the political fall-out from crises compare to normal recessions and severe (non-financial) macroeconomic slumps? Benchmark 1: all normal recessions without financial crisis Benchmark 2: sub-sample of particularly severe non-financial recessions, macro-disasters Definition: annualized peak-to-trough percentage decrease of real p.c. GDP more severe than the average financial recession

63 60/61 Implications 1 Slump politics are different. The political economy of financial crises poses particular challenges and puts a strain on effective governance when it is needed most. 2 The danger that politics go off track is real. Fragility of democracies not only a 1930s story. 3 Preventing financial crises means reducing the probability of a political disaster. Regulators carry a large responsibility for political stability.

64 61/61 Summary 20th century macrofinancial history is essentially a story of increasing debt secured against higher asset prices with higher leverage. The holy grail is to understand the interaction of debt, home ownership and house price dynamics and its consequences. Many if not most of the big debates seem to converge here: Wealth distribution Financial (in-)stability Growth of the financial sector Transmission of monetary policy