Governments and Asset Prices. Pietro Veronesi University of Chicago Booth School of Business

Transcription

1 Governments and Asset Prices Pietro Veronesi University of Chicago Booth School of Business

2 Pietro Veronesi Governments and Asset Prices page: 2 Political Uncertainty and Volatility Source: Bloom The Impact of Uncertainty Shocks, Econometrica, 2009

3 Pietro Veronesi Governments and Asset Prices page: 3 Political Uncertainty and Volatility: Germany Source: Bittlingmayer Output, Stock Volatility, and Political Uncertainty in a Natural Experiment: Germany, , Journal of Finance, 1998

4 Pietro Veronesi Governments and Asset Prices page: 4 Political Cycles and Stock Prices Stock market prefers Republicans. Kerry vs Bush close election. FIGURE I The S&P 500 is Higher under a Bush versus Kerry Presidency Source: Snowberg, Wolfers, Zitzewitz Partisan Impacts on the Economy: Evidence From Prediction Markets and Close Elections QJE, 2007

5 Pietro Veronesi Governments and Asset Prices page: 5 Political Cycles and Stock Prices Stock market prefers Republicans... FIGURE III Equity Markets have Historically Preferred Republican Presidents Source: Snowberg, Wolfers, Zitzewitz Partisan Impacts on the Economy: Evidence From Prediction Markets and Close Elections QJE, 2007

6 Pietro Veronesi Governments and Asset Prices page: 6 Political Cycles and Stock Prices...but excess return are higher during Democratic presidencies Sample: Rep Dem t-diff Average Excess Returns (%/year) Average Real Div Growth (%/year) Average P/D Ratio (logs) Average Volatility (%/year) Median Excess Return (%/year) Median Nominal Dividend Growth (%/year) Median P/D Ratio Median Volatility (%/year) See also: Santa Clara and Valkanov Political Cycles and the Stock Market Journal of Finance, 2003

7 Pietro Veronesi Governments and Asset Prices page: 7 Political Cycles and Stock Prices: Belo, Gala, and Li (2009) BGL sort stocks depending on a their industry exposure to government spending. Use Benchmark Input-Output Accounts released by Bureau of Economic Analysis to compute GOV i = Sales in industry i generated by government spending Total sales in industry i Captures both direct and indirect effects Compute average returns over democratic and republican presidents across firms with large and small exposure to government. Perform standard asset pricing tests.

8 Pietro Veronesi Governments and Asset Prices page: 8 Table 1: Summary Statistics of Sample Variables This table reports the summary statistics of selected variables in our sample across all years and across presidential terms. Firm is the average number of firms in each month, rew and rvw are the equally-weighted and value-weighted average excess return respectively, Size is firm s market capitalization, B/M is the ratio of book equity to market equity, Mom is momentum, Lev is leverage ratio and Yb is profitability. It also reports selected moments of the Gov factor, the measure of the firms level of market interaction with the government. Here 10th and 90th are the corresponding average percentiles of the cross-sectional GOV distribution. The data is from July 1948 to December Gov factor (in %) Years President Firms rew rvw Size B/M Mom Lev YB Mean Std Med Max Min 10th 90th All Republican Democratic Truman(D) Eisenhower(R) Eisenhower(R) Kennedy/Johnson(D) Johnson(D) Nixon(R) Nixon/Ford(R) Carter(D) Reagan(R) Reagan(R) Bush(R) Clinton(D) Clinton(D) Bush(R) Bush(R)

9 Pietro Veronesi Governments and Asset Prices page: 9 Figure 2: High Minus Low Excess Returns of Government Portfolios Across presidential Terms The figure plots the average value-weighted excess returns across presidential terms and for small and large firms of a spread portfolio that goes long on firms that have a high exposure to the government sector short on firms that have a low exposure. The sample data are monthly from July 1948 to Dec 2007.

10 Pietro Veronesi Governments and Asset Prices page: 10 Table 2: Industries with Highest and Lowest Exposure to the Government Sector This table reports the ten Fama-French 48 industries with highest and lowest average exposure to the government sector. The average GOV for each of the Fama-French 48 industries is computed as the totaloutput weighted average of GOV at the 4-digit SIC level. Panel A reports the top ten highest GOV industries, and Panel B reports the top ten lowest GOV industries. The data is from July 1948 to December Panel A: Top 10 Industries with the Highest Exposure to the Government Sector Rank Industry GOV (%) 1 Aircraft Defense Shipbuilding and railroad equipment Electronic equipment Construction Non-metallic and industrial metal minning Precious metals Fabricated products Measuring and control equipment Steel works 15.8 Panel B: Top 10 Industries with the Lowest Exposure to the Government Sector 1 Tobacco products Candy and soda Beer and liquor Apparel Food products Retail Entertainment Utilities Agriculture Textiles 4.7

11 Pietro Veronesi Governments and Asset Prices page: 11 Table 3: Summary Statistics of Selected Macroeconomic Variables This table reports the summary statistics of selected macroeconomic variables across all years and across presidential terms. GDP is the growth rate of gross domestic product, Consumption is the growth rate of consumption (Nondurables and Services), Gov Spend. is the growth rate of government consumption expenditures and gross investment, Gov Spend. is the growth rate of Federal government spending, Defense is the growth rate of national defense spending, Non Defense is the growth rate government of non defense spending, Rf is the annualized nominal one-month Treasury bill rate, and Inflation is the annual inflation rate. All values in Panel B are in percentage and all growth rates are per annum, per capita and in real terms. Mean and Std are the average and the standard deviation of each variable respectively. The data is from July 1948 to December GDP Consumption Gov Spend. Defense Non Defense Rf Inflation Years President Mean Std Mean Std Mean Std Mean Std Mean Std Mean Std Mean Std All Republican Democratic Truman(D) Eisenhower(R) Eisenhower(R) Kennedy/Johnson(D) Johnson(D) Nixon(R) Nixon/Ford(R) Carter(D) Reagan(R) Reagan(R) Bush(R) Clinton(D) Clinton(D) Bush(R) Bush(R)

12 Pietro Veronesi Governments and Asset Prices page: 12 Table 5: Summary Statistics of Government Portfolios This table reports the time series averages of average returns and median characteristics of six government portfolios double sorted on GOV and firm size across all years (Panel A) and across Republican (Panel B) and Democratic (Panel C) presidential terms. The cutoffs used to form portfolios are 33.3 th percentile and 66.7 th percentile of GOV at the industry level, and the median for firm size. rew and rvw are the equally-weighted and value-weighted average excess return respectively, Mkt Share is the average portfolio market share (across all firms in CRSP), Firms is the average number of firms in each month, GOV is the firm s exposure to the government sector, Size is the firm s market capitalization, B/M is the ratio of book equity to market equity, Lev is the leverage ratio, Yb is profitability and IK is the firm investment rate. The sample data are monthly from July 1948 to Dec Panel A: All Years (714 Months) Portfolios Size GOV rew rvw Mkt Share Firms G-Exp Size B/M Lev YB IK Small Low Small Mid Small High Large Low Large Mid Large High Panel B: Republican Years (420 Months) Small Low Small Mid Small High Large Low Large Mid Large High Panel C: Democratic Years (294 Months) Small Low Small Mid Small High Large Low Large Mid Large High

13 Pietro Veronesi Governments and Asset Prices page: 13 Table 8: Fama-MacBeth Cross-Sectional Regressions This table reports the results from Fama-MacBeth cross-sectional regressions of monthly stock returns on lagged variables during Republican presidential terms (Panel A) and across Democratic presidential terms (Panel B). GOV is the firm level exposure to the government sector, Size is the natural log of firm size, B/M is the natural log of firm book-to-market ratio and Mom is firm momentum. The table reports the average loading across each cross sectional regression and the corresponding t-statistic of the loading in parenthesis with Newey-West correction. Adjusted R 2 are reported in the last column. The sample data are monthly from July 1948 to Dec Panel A. Republican Years Intercept GOV Size B/M Mom Adjusted R 2 (%) (3.28) (-0.36) (3.37) (-1.72) (3.50) (4.30) (3.43) (0.29) (-1.75) (3.48) (4.22) Panel B. Democratic Years (4.75) (3.68) (3.98) (-3.02) (1.40) (5.19) (3.77) (3.71) (-3.03) (1.51) (5.02)

14 Pietro Veronesi Governments and Asset Prices page: 14 The Need for a Benchmark Theoretical Model Rampant political uncertainty in the past few years U.S.: Debt ceiling, fiscal cliff, Fed policy, Wall St & health care reforms, etc. Europe: ECB policy, banking union, structural reforms, bailouts, etc. Political news move markets Yet, no role for political news in finance theory How do political news affect stock prices? Cash flow channel: political news affect expected future cash flows Discount rate channel: political news affect the discount rate (cost of capital) Does political uncertainty carry a risk premium? How large?

15 Pietro Veronesi Governments and Asset Prices page: 15 Outline 1. Summarize a simple GE model of government policy choice (Pastor and Veronesi (2012 JF, 2013 JFE)) and discuss its implications for asset prices 2. Present some preliminary evidence using Baker, Bloom, and Davis Policy Uncertainty Index 3. Present evidence using options, based on Kelly, Pastor and Veronesi (2015, JF)

16 Pietro Veronesi Governments and Asset Prices page: 16 Model: Pástor and Veronesi (2012, 2013) Finite horizon [0, T]; continuum of equity-financed firms i [0, 1] Firm i s profitability (= growth rate of capital): db i t /Bi t = (µ + g t) dt + σdz t + σ 1 dz i,t g t = impact of government policy on average profitability Government can change policy at time τ, 0 < τ < T, choosing from N potential new policies g t = g 0 for t τ g 0 for t > τ if old policy is retained g n for t > τ if new policy n is chosen, n {1,...,N}

17 Pietro Veronesi Governments and Asset Prices page: 17 Impact Uncertainty Both g 0 and g n are unknown Prior beliefs: g 0 N ( 0, σ 2 g ) g n N ( µ n g,σ2 g,n ) for n = 1,..., N Posterior beliefs (Kalman Bucy filter): g t N ( ĝ t, σ 2 t ), where d ĝ t = σ 2 tσ 1 dẑt, σ 2 t = 1 σ 2 g + t σ 2 1 If government changes policy at time τ, beliefs about g t change from posterior of g 0 to prior of g n

18 Pietro Veronesi Governments and Asset Prices page: 18 Political Uncertainty Investors maximize E W 1 γ T 1 γ, where γ > 1, W T = 1 BT i di 0 Quasi-benevolent government maximizes max n {0,1,...,N} Cn E τ W 1 γ T 1 γ policy n C n = political cost of choosing policy n (C n > 1 cost; C n < 1 benefit) C n is unknown for n = 1,...,N before time τ Uncertainty about C n political uncertainty Prior beliefs: c n = log (C n ) N ( 1 2 σ2 c, σ2 c ) E 0 [C n ] = 1

19 Pietro Veronesi Governments and Asset Prices page: 19 Learning about Political Costs For t (t 0, τ), agents observe signals about C n : ds n t = c n dt + h dz n c,t for n = 1,...,N Steady flow of political news Posterior beliefs: c n N ( ĉ n t, σ 2 c,t ) where dĉn t = σ c,th 2 1dẐn c,t, σ2 c,t = ( 1 σc 2 + t t ) 1 0 h 2 dz n c,t = political shocks Orthogonal to economic shocks dz t, dz i,t

20 Pietro Veronesi Governments and Asset Prices page: 20 Learning about {c 1,É, c N } Òpolitical shocksó 0 Learning about g 0 {c 1,É, c N } revealed " T Government chooses policy n " {0,1,É,N} Learning about g n Agents consume

21 Pietro Veronesi Governments and Asset Prices page: 21 Utility Score Result: Given any two policies m, n {0, 1,...,N}, E τ W 1 γ T 1 γ policy m > E τ W 1 γ T 1 γ policy n µ m > µ n where µ n = utility score of policy n: µ n = µ n g σ2 g,n 2 (T τ) (γ 1) n = 1,...,N µ 0 = ĝτ σ 2 τ 2 (T τ) (γ 1) Higher mean and lower variance of g deliver more utility

22 Pietro Veronesi Governments and Asset Prices page: 22 Optimal Government Policy Choice Result: Government chooses the policy n {0, 1,...,N} to where c n = c n / (γ 1) (T τ) max n µ n c n Corollary: Government changes its policy iff ĝ τ < max { µ n c n } + σ τ 2 (T τ) (γ 1) n {1,...,N} 2 i.e., if the current policy is perceived as sufficiently unproductive Government provides put protection to the market Investors don t know c n cannot fully anticipate a policy change

23 Pietro Veronesi Governments and Asset Prices page: 23 Three Types of Shocks Result: Before time τ, SDF follows the process dπ t = γσdẑt }{{} π t Capital shocks + σ π,0 dẑt }{{} Impact shocks + N n=1 σ π,ndẑn c,t }{{} Political shocks 1. Capital shocks: Fluctuations in aggregate capital (db t ) 2. Impact shocks: Learning about policy impact (dĝt) Capital + Impact shocks = Economic shocks (dẑt) 3. Political shocks: Learning about political costs (dĉn t ) Orthogonal to economic shocks σ π,n 0 when ĝt

24 Pietro Veronesi Governments and Asset Prices page: 24 Stock Prices Firm i s stock is a claim on the firm s liquidating dividend B i T Market value of stock i: π T Mt i = E t BT i π t Complete markets State price density: π t = 1 λ E t [ ] W γ T, where WT = 1 BTdi i 0 Risk-free rate = 0 (equivalently, use risk-free bond as numeraire) Because no intermediate consumption

25 Pietro Veronesi Governments and Asset Prices page: 25 A Two-Policy Example Consider policies H and L, with σ g,h > σ g,l Choose µ g,h > µ g,l so that both policies yield same utility Parameters: σ g σ c µ σ σ 1 T τ γ h σ g,l σ g,h 2% 10% 10% 5% 10% % 1% 3%

26 Pietro Veronesi Governments and Asset Prices page: 26 The Probability of a Policy Change Old policy New risky policy New safe policy Percent Economic conditions (ĝ t )

27 Pietro Veronesi Governments and Asset Prices page: 27 The Level of Stock Prices: Economic vs Political Shocks 3 New risky policy more likely New policies equally likely New safe policy more likely M/B Economic conditions (ĝ t )

28 Pietro Veronesi Governments and Asset Prices page: 28 The Equity Risk Premium and Its Components 6 5 Capital shocks Impact shocks Political shocks 4 Percent per year Economic conditions (ĝ t )

29 Pietro Veronesi Governments and Asset Prices page: 29 Policy Changes Allowed vs Precluded Percent per year A. Total Risk Premium High H Med H Low H No change Economic conditions (ĝ t ) M/B B. Market to Book Ratio 2.2 High H Med H 2 Low H No change Economic conditions (ĝ t ) Percent per year C. Return Volatility High H Med H Low H No change Percent D. Correlation High H Med H Low H No change Economic conditions (ĝ t ) Economic conditions (ĝ t )

30 Pietro Veronesi Governments and Asset Prices page: 30 Empirical Analysis Test model s predictions about political uncertainty (PU) PU is higher in weaker economic conditions PU commands a risk premium, larger when economy is weak PU makes stocks more volatile and more correlated, especially when the economy is weak Proxy for PU: Baker, Bloom, and Davis (2011) Weighted average of 3 components: News coverage of policy-related uncertainty Number of expiring federal tax code provisions Disagreement among forecasters of inflation and govt spending

31 Pietro Veronesi Governments and Asset Prices page: 31

32 Pietro Veronesi Governments and Asset Prices page: 32 Correlation (percent) Recession Political uncertainty Stock correlation Panel A. Political Uncertainty vs Stock Correlation Month Standard deviation (percent per year) Recession Political uncertainty Stock volatility Panel B. Political Uncertainty vs Stock Volatility Month

33 Pietro Veronesi Governments and Asset Prices page: 33 Is PU Higher in a Weaker Economy? Table reports estimates of b and their t-statistics for Specification 1: Specification 2: PU t = a + be t + e t PU t = a + be t + cpu t 1 + e t Measure of Economic Conditions CFI -REC IPG P/E -DEF Specification (-7.24) (-5.12) (-4.10) (-3.38) (-8.61) Specification (-3.90) (-2.75) (-1.85) (-1.58) (-3.06)

34 Pietro Veronesi Governments and Asset Prices page: 34 Are Volatility and Correlation Higher When PU Is Higher? Table reports estimates of b and their t-statistics for Specification 1: Specification 2: V C t = a + bpu t + e t V C t = a + bpu t + cv C t 1 + e t Correlation Volatility EW VW Realized Implied Specification (9.81) (7.25) (4.81) (5.27) Specification (6.43) (5.14) (3.45) (2.53)

35 Pietro Veronesi Governments and Asset Prices page: 35 Are VOL and COR More Linked to PU in a Weaker Economy? Table reports estimates of b and their t-statistics for Specification 1: V C t = a + bpu t E t + cpu t + de t + e t Measure of Economic Conditions CFI -REC IPG P/E -DEF Correlation: EW (-2.41) (-0.96) (-2.36) (-0.00) (-0.08) Correlation: VW (-1.92) (-0.60) (-2.03) (-0.26) (1.28) Volatility: Realized (-5.46) (-4.39) (-4.52) (-3.74) (-3.17) Volatility: Implied (-4.50) (-3.69) (-3.18) (-5.48) (-1.91)

36 Pietro Veronesi Governments and Asset Prices page: 36 Are VOL and COR More Linked to PU in a Weaker Economy? Table reports estimates of b and their t-statistics for Specification 2: V C t = a + bpu t E t + cpu t + de t + ev C t 1 + e t Measure of Economic Conditions CFI -REC IPG P/E -DEF Correlation: EW (-2.04) (-1.07) (-1.97) (0.05) (-0.05) Correlation: VW (-1.48) (-0.79) (-1.54) (-0.10) (1.13) Volatility: Realized (-4.11) (-3.86) (-3.11) (-2.77) (-2.58) Volatility: Implied (-2.81) (-3.71) (-0.36) (-2.76) (-2.70)

37 Pietro Veronesi Governments and Asset Prices page: 37 Is Political Risk Premium Higher in a Weaker Economy? Table reports estimates of b and their t-statistics for R t+1,t+h = a + bpu t E t + cpu t + de t + e t. Measure of Economic Conditions Horizon CFI -REC IPG P/E -DEF 3 months (-1.30) (-1.24) (-0.71) (-2.17) (-1.19) 6 months (-2.09) (-1.53) (-1.17) (-3.18) (-1.97) 12 months (-2.41) (-1.78) (-1.76) (-2.85) (-1.69)

38 Pietro Veronesi Governments and Asset Prices page: 38 Conclusions We develop a theory in which political news moves stock prices Political uncertainty commands a risk premium that is larger in a weaker economy reduces the value of the government s implicit put protection increases stock volatilities and correlations, especially when the economy is weak and policy heterogeneity is large

39 Pietro Veronesi Governments and Asset Prices page: 39 Brogaard and Detzel (2015): Further evidence Motivated by Baker, Blume, and Davis (2013) and Pastor and Veronesi (2012, 2013), Brogaard and Detzel (2015) further study the impact of political uncertainty on stock returns. They begin with investigating the determinants of EPU index

40 Pietro Veronesi Governments and Asset Prices page: 40 Table 1 Economic Determinants of EPU Index Panel A: Standardized regressions of EPU on uncertainty and business cycle variables (1) (2) (3) (4) (5) (6) (7) (8) VXO VAR TERM DEFAULT RREL CFI Log D/P N Adj.-R (Source: Brogaard and Detzel: The Asset-Pricing Implications of Government Economic Policy Uncertainty, Management Science, 2015)

41 Pietro Veronesi Governments and Asset Prices page: 41 Changes to EPU and Market Returns Regress log excess return xr t on changes in EPU and controls: xr t = α + β EPU t + γcontrols t + ε t Panel B: Regression of log excess market return on first differences of EPU and controls (1) (2) EPU VAR VXO TERM DEFAULT RREL Log D/P N Adj.-R (Source: Brogaard and Detzel: The Asset-Pricing Implications of Government Economic Policy Uncertainty, Management Science, 2015)

42 Pietro Veronesi Governments and Asset Prices page: 42 Shocks to EPU and Market Returns Because EPU t is persistent, BD extract shocks to EPU as the real pricing factor for stock returns, through the estimation EPU t = α + β 1 EPU t 1 + β 2 EPU t 2 + β 3 EPU t 3 + β 4 EPU t 4 + ε EPU t where theys use the Bayesian Information Criterion for the number of lags. ε EPU t is the shock to EPU. In addition, BD use factor mimicking portfolios: For instance, for EPU one run ε EPU t = a EPU + β EPU xr t + η t where xr t is a set of excess returns on some basis assets. Then defines F EPU,t = β EPU xr t Finally, inference uses a standard GMM procedure: From no-arbitrage where the pricing kernel is and factors are portfolios of stocks. E[R it m t ] = 0 m t = a + b(f t E(f t ))

43 Pietro Veronesi Governments and Asset Prices page: 43 The Economic Policy Risk Premium Table 5 GMM Estimations of Linear Asset-Pricing Models with Economic Policy Uncertainty (1) (2) (3) (4) (5) (6) (7) (8) Panel A: Pricing kernel coefficients MKT SMB HML UMD LIQ F EPU F VXO F VAR (Source: Brogaard and Detzel: The Asset-Pricing Implications of Government Economic Policy Uncertainty, Management Science, 2015)

44 Pietro Veronesi Governments and Asset Prices page: 44 The Economic Policy Risk Premium Panel B: Estimated risk premiums (in % per month) MKT SMB HML UMD LIQ F EPU F VXO F VAR HJ distance Std. err (Source: Brogaard and Detzel: The Asset-Pricing Implications of Government Economic Policy Uncertainty, Management Science, 2015)

45 Pietro Veronesi Governments and Asset Prices page: 45 Kelly, Pastor, and Veronesi (2015): Political Uncertainty Empirically analyze whether / how uncertainty associated with political events (national elections and global summits) is priced in the option market Why options? Short maturities Different strikes Why elections and summits? Can result in major policy shifts Exogenous variation in political uncertainty Guided by an existing theoretical model of government policy choice Derive in closed form the price of a European put option whose life spans τ Analyze the price risk, variance risk, and tail risk associated with political events, and their dependence on economic conditions and uncertainty

46 Pietro Veronesi Governments and Asset Prices page: 46 Option Pricing Implications of Pastor and Veronesi (2013) Option pricing formula under the assumptions of Pastor and Veronesi (2013). Formula is one page long. See appendix in paper. Compute three critical quantities: 1. Black - Scholes Implies Volatility: IV (AT M) 2. Variance Risk Premium: V RP = IV (ATM) 2 E[V ariance(t)] 3. Implied Volatility Slope: Slope = IV (5%OTM) IV (5%ITM) Two Policy example Each of these quantities depend on the state variables (ĝt, ĉl t, ĉh t ) Fix ĉh t = prior = 1 2 σ2 c, and plot the variables on other dimensions

47 Pietro Veronesi Governments and Asset Prices page: 47 Option Pricing Implications of Pastor and Veronesi (2013) A. Implied Volatility x 10 3 B. Variance Risk Premium ĉ L t 2 2 ĝ t ĉ L t 2 2 ĝ t 0 2 C. Slope D. Jump Risk Premium at τ ĉ L t 2 2 ĝ t ĉ L τ 2 2 ĝ τ 0 2

48 Pietro Veronesi Governments and Asset Prices page: Implied Volatility IV Econom ic Conditions (ĝ)

49 Pietro Veronesi Governments and Asset Prices Slope Skewness of Returns Skew ness 0.02 Slo pe page: Econom ic Conditions (g ) Jump Risk Premium 0.03 JRP I V 2 E[V ar] Variance Risk Premium Econom ic Conditions (g ) Econom ic Conditions (g ) Econom ic Conditions (g ) 0.02

50 Pietro Veronesi Governments and Asset Prices page: 50 Data Options: 20 countries (from OptionMetrics) Put options on each country s premier stock market index (for 15 countries) or, if unavailable, ETF on the country s MSCI index (for 5 countries) Political events National elections: parliamentary, presidential Global summits: G8, G20, European Economic conditions GDP : Realized real GDP growth (from OECD) FST : Forecast of real GDP growth (from IMF) CLI: Composite leading indicator (from OECD) MKT : Stock market index return (from Datastream) Political uncertainty: for elections only UNC: minus the poll spread (most recent opinion poll spread before the election; if unavailable, then ex-post election margin)

51 Pietro Veronesi Governments and Asset Prices page: 51 Table 1: Option sample Start End Country Index Date Date Australia ASX Belgium BEL Brazil MSCI Brazil Canada MSCI Canada Finland OMXH France CAC Germany DAX Italy FTSE MIB Japan NIKKEI Korea Kospi Mexico MSCI Mexico Netherlands AEX Singapore MSCI Singapore South Africa MSCI South Africa Spain IBEX Sweden OMXS Switzerland SMI Taiwan TAIEX UK FTSE USA S&P

52 Pietro Veronesi Governments and Asset Prices page: 52 Table 2: Number of political events Elections Summits Total Total Parl. Pres. Total Euro G8/G20 All Australia Belgium Brazil Canada Finland France Germany Italy Japan Korea Mexico Netherlands Singapore South Africa Spain Sweden Switzerland Taiwan UK USA

53 Pietro Veronesi Governments and Asset Prices page: 53 Option-Market Variables Issue: Option variables may be high/low for reasons unrelated to political events = control using neighbouring options that do not span political event control treatment control a s τ: political event date a, b, c: option expiration dates a b s b c s c τ Implied volatility difference: IV D τ = IV b 1 2 (IV a + IV c ), where IV b = average implied volatility before τ. Same computations for other quantities: Implied Volatility Slope difference: SlopeD. Variance Risk Premium difference: VRPD

54 Pietro Veronesi Governments and Asset Prices page: 54 Put Implied Volatility IV no event RV no event VRP Moneyness ( )

55 Pietro Veronesi Governments and Asset Prices page: 55 Put Implied Volatility IV no event RV no event IV event RV event VRP Moneyness ( )

56 Pietro Veronesi Governments and Asset Prices page: 56 Put Implied Volatility IV no event RV no event IV event RV event VRP Moneyness ( )

57 Pietro Veronesi Governments and Asset Prices page: 57 Put Implied Volatility IV no event RV no event IV event RV event VRP Moneyness ( )

58 Pietro Veronesi Governments and Asset Prices page: 58 Table 3: Mean implied volatility differences Weak minus strong economy All MKT GDP F ST CLI Panel A: All political events Mean (4.43) (3.79) (3.34) (3.78) (4.61) Obs Panel B: Elections only Mean (3.13) (2.73) (1.78) (2.34) (2.39) Obs Panel C: Summits only Mean (3.76) (3.27) (3.17) (3.56) (4.30) Obs

59 Pietro Veronesi Governments and Asset Prices page: 59 Table 3: Mean implied volatility differences Weak minus strong economy All MKT GDP F ST CLI Panel A: All political events Mean (4.43) (3.79) (3.34) (3.78) (4.61) Obs Economic significance: ATM treatment-group options are more expensive by 5.1% compared to control-group options, on average

60 Pietro Veronesi Governments and Asset Prices page: 60 Table 3: Mean implied volatility differences Weak minus strong economy All MKT GDP F ST CLI Panel A: All political events Mean (4.43) (3.79) (3.34) (3.78) (4.61) Obs Panel B: Elections only Mean (3.13) (2.73) (1.78) (2.34) (2.39) Obs Panel C: Summits only Mean (3.76) (3.27) (3.17) (3.56) (4.30) Obs

61 Pietro Veronesi Governments and Asset Prices page: 61 Table 3: Mean implied volatility differences Weak minus strong economy All MKT GDP F ST CLI Panel A: All political events Mean (4.43) (3.79) (3.34) (3.78) (4.61) Obs Economic significance: ATM treatment-group options are 7.1% to 9.0% more expensive compared to control-group options when the economy is weak ATM treatment-group options are 0.1% to 1.2% more expensive compared to control-group options when the economy is strong

62 Pietro Veronesi Governments and Asset Prices page: 62 Table 3 (cntd.): Variance risk premium and implied volatility slope: Mean differencess Variance risk premium (V RPD) Implied volatility slope (SlopeD) Weak minus strong economy Weak minus strong economy All MKT GDP F ST CLI All MKT GDP F ST CLI Panel A: All political events Mean (2.61) (3.51) (2.80) (3.54) (3.63) (3.59) (3.11) (2.52) (3.08) (2.97) Obs Panel B: Elections only Mean (2.59) (2.62) (1.11) (2.20) (1.25) (2.08) (3.69) (1.08) (1.71) (1.28) Obs Panel C: Summits only Mean (2.15) (3.15) (2.92) (3.49) (3.74) (3.16) (2.58) (2.54) (2.83) (2.80) Obs

63 Pietro Veronesi Governments and Asset Prices page: 63 Table 3 (cntd): Variance risk premium and implied volatility slope: Mean differencess Variance risk premium (V RPD) Implied volatility slope (SlopeD) Weak minus strong economy Weak minus strong economy All MKT GDP F ST CLI All MKT GDP F ST CLI Panel A: All political events Mean (2.61) (3.51) (2.80) (3.54) (3.63) (3.59) (3.11) (2.52) (3.08) (2.97) Obs Economic significance (variance risk): ATM treatment-group options are 48.1% more expensive relative to the Black- Scholes model, on average; control-group options are 36.5% more expensive

64 Pietro Veronesi Governments and Asset Prices page: 64 Table 3 (cntd): Variance risk premium and implied volatility slope: Mean differencess Variance risk premium (V RPD) Implied volatility slope (SlopeD) Weak minus strong economy Weak minus strong economy All MKT GDP F ST CLI All MKT GDP F ST CLI Panel A: All political events Mean (2.61) (3.51) (2.80) (3.54) (3.63) (3.59) (3.11) (2.52) (3.08) (2.97) Obs Economic significance (tail risk): 5% (10%) OTM treatment-group options are more expensive by 9.6% (16.0%)

65 Pietro Veronesi Governments and Asset Prices page: 65 Table 3 (cntd): Variance risk premium and implied volatility slope: Mean differencess Variance risk premium (V RPD) Implied volatility slope (SlopeD) Weak minus strong economy Weak minus strong economy All MKT GDP F ST CLI All MKT GDP F ST CLI Panel A: All political events Mean (2.61) (3.51) (2.80) (3.54) (3.63) (3.59) (3.11) (2.52) (3.08) (2.97) Obs Panel B: Elections only Mean (2.59) (2.62) (1.11) (2.20) (1.25) (2.08) (3.69) (1.08) (1.71) (1.28) Obs Panel C: Summits only Mean (2.15) (3.15) (2.92) (3.49) (3.74) (3.16) (2.58) (2.54) (2.83) (2.80) Obs

66 Pietro Veronesi Governments and Asset Prices page: 66 Implied Volatility Difference IVD Economic Conditions

67 Pietro Veronesi Governments and Asset Prices page: 67 Variance Risk Premium Difference VRPD Economic Conditions

68 Pietro Veronesi Governments and Asset Prices page: 68 Implied Slope Difference SlopeD Economic Conditions

69 Pietro Veronesi Governments and Asset Prices page: 69 The Impact of Political Uncertainty IVD VRPD SlopeD UNC UNC UNC

70 Pietro Veronesi Governments and Asset Prices page: 70 Extensions Multi-country setting with spillovers and FX Cross-border policy uncertainty spillovers in equity-linked markets Implications for foreign exchange options Theory and empirics Asymmetric information Government observes private signal of policy effects on macroeconomy Potentially relevant empirical distinction: elections vs. summits Reduced-form model comparison Our model nested in general jump-diffusion option pricing model Restrictions from pol. uncertainty absent from no-arbitrage pricing models These are strongly supported by data Event study Correlation among IV D, V RPD, SlopeD, ECON, UNC and Ret line up closely with model-implied correlations

71 Pietro Veronesi Governments and Asset Prices page: 71 Conclusions from Kelly, Pastor and Veronesi (2013) Political uncertainty is priced in the option market in ways predicted by theory Options whose lives span political events offer valuable protection against price risk variance risk tail risk associated with major political events This protection is more valuable when the economy is weaker when political uncertainty is higher

72 Pietro Veronesi Governments and Asset Prices page: 72 Veronesi and Zingales (2010): Paulson s Gift The 2008 financial crisis witnessed the largest intervention of the U.S. government in the financial sector. The stated goal was to restore confidence to our financial system, through a massive transfer of resources from the taxpayers to the banking sector. From an economic point of view, such an intervention is justified only in the presence of a market failure. If this market failure is present, then the government intervention should create, not just redistribute, value. Did this intervention create value or was it simply a massive transfer of resources from taxpayers to financial institutions? If it did create value, why? What can we learn about the possible cost of financial distress in financial institutions?

73 Pietro Veronesi Governments and Asset Prices page: 73 What we do We estimate the costs and benefits of the U.S. government plan announced on Monday, October 13, The plan included a $125bn preferred equity infusion in the ten largest U.S. commercial banks A three year Government guarantee on new unsecured bank debt issues. Compute effect on the enterprise value of banks Estimate changes in the value of existing debt using liquid CDS spreads Controlling for the market and GE capital variation in CDS spreads with find: 1. The intervention increased the value of banks financial claims by $131 billion 2. Taxpayers cost of $25 -$47 billions = net benefit between $84bn and $107bn. 3. Net benefit arises from a reduction in the probability of bankruptcy, which we estimate would destroy 22% of the enterprise value. 4. The big winners were the Citigroup s and investment banks bondholders. The losers were JP Morgan shareholders and the U.S. taxpayers.

74 Pietro Veronesi Governments and Asset Prices page: 74!! " # $ % & ' ( )!! ' * + * * (, - ( ',.!.! / 0 1 ( * ( + '! * * + *! / 0 ( (!! % ( (! * 1! - / 0 ( ) ( ( 2 : ;, ',, 1 ( * * / 0,! * ( < ) *, 1 ( * ) ( ( <! = 4 % < ) ( ( ) ( (, ( ( / 0 - (,, 2 6 : ; ( (, 1!! - (, ( - ( > A D E F H I J K J A C L C M N O B N P I J K Q B R A C N S A M F M N C H H O B N K A L N T U V T W V X U U Y L Z M F H L [ M F F M N C L K M G C J F O H \ C ] ^ _ ^ ` a b c d e f g h i j e f f l i e l l i m j g e n i o p q r s t u v w x y z { q } ~ } } ~ ƒ ƒ ƒ ~ ˆ Š Œ Ž š œ ž Ÿ œ ª «± ² ³ µ µ ³ µ µ ³ ¹ ¹ º» ¼ ½ ¾ À Á Â Ã Ä Å Å ½ Æ Æ Ç È É Ê Ë È Ë Ì Í È Í Ë Í È Î Î Æ È Ê Ï Ð Ñ Ð Ò Ï Ð Ó Ò Ò Ð Ô Õ Ó Ö Ø Ù Ú Ù Û Ü Ý Ü Ú Ü Þ Ü Ú Û Û Ø ß à á â ã ä å æ ä ç è é ê ë ê ì ì í î ë ð ë í ë ê ë í î ñ ò ë í î ó ô õ ö ø ù ú ø û ü ý þ ÿ þ þ þ þ ÿ ÿ ÿ þ BC BG BJ BJ ik ÚÝ ÚÝ íï! " # $ % "

75 Pietro Veronesi Governments and Asset Prices page: 75 Why Government Intervention May Create/Destroy Value? Stop a bank run Bank runs may be inefficient (Diamond and Dybvig (1983)) = an intervention to stop a bank run may create value Who may be running? Not the depositors, but short-term creditors, who refused to roll over bank debt. (Gorton and Metrick (2009)). Is there any evidence of a bank run? Use CDS spreads to compute P(n) = Prob(Default at time n No default up to n 1) For each bank, compute the Run Index R i t = P i t(t + 1) P i t(t + 2) Normal times: R i t < 0. Run times R i t > 0.

76 K K Pietro Veronesi Governments and Asset Prices page: 76 + ) & & + ( & & + ' & & + & & & * & & ) & & ( & & ' & & & a _ \ \ a ^ \ \ a ] \ \ a \ \ \ ` \ \ _ \ \ ^ \ \ ] \ \ \ / : ; < 0 = 3 7 < A &, + &, ' & & ) +, + *, ' & & - (, ' *, ' & & - *, ), ' & & - + +, + (, ' & & - ', ' ', ' & & * ), +, ' & & *.,., ' & & * + ', + *, ' & & * G E B B G D B B G C B B G B B B F B B E B B D B B L M N O P Q L R S S T U L V W O X R Y R V L S Z S R M [ V Q M C B B B G B H G B H C B B E G H G F H C B B I D H C F H C B B I F H E H C B B I G G H G D H C B B I C H C C H C B B F E H G H C B B F J H J H C B B F G C H G F H C B B F e f g h i j k l j m n o p f q r j k l s j k g t u p t q q v g g w u k m n a \ b a \ b ] \ \ _ a b a ` b ] \ \ c ^ b ] ` b ] \ \ c ` b _ b ] \ \ c a a b a ^ b ] \ \ c ] b ] ] b ] \ \ ` _ b a b ] \ \ ` d b d b ] \ \ ` a ] b a ` b ] \ \ `

77 Pietro Veronesi Governments and Asset Prices page: 77 The Run Index Run Index Citigroup Bank of America JP Morgan Chase Wells Fargo /30/2008 6/26/2008 7/23/2008 8/19/2008 9/15/ /10/ /6/ /3/ Goldman Sachs Morgan Stanley Merrill Lynch Run Index /30/2008 6/26/2008 7/23/2008 8/19/2008 9/15/ /10/ /6/ /3/2008

78 Pietro Veronesi Governments and Asset Prices page: 78 Why Government Intervention May Create/Destroy Value? Resolve inefficiency due to debt overhang Excessive leverage makes it hard to obtain new funds to exploit investment oportunities Capital injection that resolves the debt overhang and reduces its deadweight cost may create value = Increase in bank value higher than tax-payer costs. However, government intervention may be costly: Increase inefficiency by restricting banks decisions that may reduce profits. Add political criteria in lending decisions, reducing profitability. Delay or limit the transfer of assets to most efficient managers / firms.

79 Pietro Veronesi Governments and Asset Prices page: 79 Event Study around the Announcment of Paulson s Plan Typically, event studies concentrate on equity (abnormal) returns. For highly levered entities, such as banks, equity is not sufficient to proxy for enterprise value The impact of the announcement on the value of debt is important Problem: Bonds are quite illiquid, and so it is hard to measure the announcment effect We use variation in liquid CDS spreads to compute the change in value of debt: Corporate Debt + P V (Insurance) = Risk Free Debt = Corporate Debt = (PV 1 (Insurance) PV 0 (Insurance)) Given D(t) = existing debt that will still be outstanding debt at t, then PV i (Insurance) = T CDS i (t) t= D(t)Q i(0, t)z(0, t); i = 0, 1 where Z(0, t) = zero coupon bond with maturity t; Q i (0, t) = risk neutral surviving probabilty up to t extracted from CDS i (t).

80 Pietro Veronesi Governments and Asset Prices page: 80 Event Study around the Announcment of Paulson s Plan We need to control for other events taking place at the time. Use GE Capital CDS variation as control. Compute Adj. PV (CDS) = PV (CDS) PV 0 (CDS) PV GE (CDS) PV GE 0 (CDS) Overall, the bonds gained $120bn in value. The bonds of the three old investments banks gained the most = $87bn. Among the old commercial banks Citigroup stood to gain the most, both in level, $21bn, and in percentage of outstanding debt, 5.3%.

81 Pietro Veronesi Governments and Asset Prices page: 81 x y z { } ~ y ƒ y { ˆ x Š Œ z y Š Ž ˆ ƒ y { { y š š œ š œ ž Ÿ š š š ž š š š š š ž ž ž š š ª š «š ž š ª š ž ž œ š œ ž ««š ± š ² ³ ³ ² ± µ š «Ÿ š š ª š š š š ª š š «Ÿ š š š ³ ¹ ª š º š ž ž ª š š š Ÿ š œ ª š š «ž ž ž «š «ž ª š š ž ž ª ª š ««š œ ž š œ ž ª š š º š œ ª š š «š š š «ž ž» ¼ «½ ž «š š š ¾ ¹ ª š š ª š º š ž š ž ž ž œ À Á Â Ã Ä Å Æ Â Ã Ä Å Æ Ü Ý Þ ß à á â ã ä á å æ ç á æ ä á â è é ê æ è é á ë ê ì ê å í á î Ç È É Ê Ë Æ Ä Å Ì Ç È É Ê Ë Æ Ä Å Í Î Ï Ð Ñ Ò Ó Ô Õ Ö Ò Ø Ù Ú Û Ñ Ò Ó Ü Ý Þ ß Ñ Ò Ó Ü Ñ à á Ô â Ö Ú Û Ñ Ò Ó Ü Ý Þ ß Ñ Ò Ó Ü ã ä å ã ä å ä æ ã ä å ã ç å ä æ Þ è é ê â Þ è é ê â Þ ë Ö Ú Ï ê â Ú Ï ê â ì í î ï í ð ñò ó ô õ ö ô õ ö ø öù öú û ü ý þ ÿ ÿ ÿ ÿ ÿ ÿ! " # $ % & ' ( ) * +,, * -. / * + - * - ) + + * / 0 * ( / * ), + *, - *, - * : ; < = < 9 > = A < B B C < > > A B < A ; < = A < C > B < C? < C? < C D E F F G H I J K L M N O P Q N R P N R O P R S P S M S T P R M P O M P N M S P N S P M S P M U V W X Y Z [ \ ] ^ Y W ` a b a c d e d f c d g f f d h i j k l m n o p q r s t u t v t w u x y y z u y v w s u w w x { u s w x u { w u t w t y u { u x x u w } ~ ƒ ˆ Š Š Œ Ž ˆ Œ Ž Š Ž ˆ ˆ ˆ ˆ Ž š œ ž Ÿ Ÿ Ÿ ž Ÿ ª «ª ± «² «³ µ µ ³ ¹ º»» ¼ ½ ¾ À Á Â Ã Ä Å Æ Ç È Å É Ê Ç Ë Æ Ê Ç Ä Ì Í Î Ï Ð Ñ Ò Ó Ô Ó Ñ Ô Õ Ö Ñ Ø Ñ Ö Ø Ù Ú Û Ö Ó Ô Ú Ö Ø Ô Ó Ö Õ

82 Pietro Veronesi Governments and Asset Prices page: 82 Event Study on Equity ï ð ñ ò ó ô õ ö ø ø ñ ù ú û ü ý þ þ ÿ! " # $ % # & ' ( # & ' ( ) * + $, -. / , , : ; < = 6 A B C D E A F D B G B D F E B D A H B D B C A H D I C D C J K L M N O P Q R S P T U V W W X V Y V X Z Z [ \ X \ ] [ \ X V ^ [ \ X ] ] [ ] \ X W [ ^ ^ X _ ` a b c d e f a g g h i j k h l m i h l n i h g l o i h l k g h n o l h j p q r r s t u v w x y z { y y { } z ~ { ~ { ~ } ~ { ~ } { ~ ~ { ƒ ˆ Š Œ Œ Ž Ž Ž Ž Ž š œ š ž Ÿ ž ž ž Ÿ ª «± ² ³ µ µ µ ² ³ µ ¹ µ µ º ¹» ¼ ½ ¾ À Á  à À Á Ä Å Æ Ç È É Ê Ë Ì É Ç Í Ç É Ì Ê Ç É Ç Î Ç É È Ï Ç Ì É Ë Ç Ç É È Ð Ñ Ò Ò Ó Ô Ô Õ Ö Ø Ù Ú Û Ü Ú Ý Ú Ü Þ ß Ý Ü à á Ý Ü Ý Û â Ý Ü à à à Ü Ú â Ú Ü ã ä å æ ç è é ê é ë ì é í ë î ï ð ë î ñ ò ó ô õ ö ó ø ù ú û ø ü ý û ø ü û ø ù û ÿ ÿ æ ç è é ê ë ì í ê î ï ð ê è ñ ì í ð î ò è ó ê î ì ô î!! " # $ % & % ' $ ( ) * + $ (, -. / : ; < = > A ;? B A ; C D A ; C C C ; < C ; > E F G H I J K L M N O P F Q Q R S T U R Q V U R S S U R Q Q U R S U Q R S S R S W X Y Z [ \ ] ^ _ ` ] a b c d e f g d h c g d i f g d g i g d g j g d g g d h k l m n o p q l r s t u v s w x u s w u u s u t u s u y u s r u s z { } } ~ ƒ ˆ Š Š Š Š Œ Ž Ž š œ ž Ž Ÿ ª «± ² ³ µ µ µ µ ¹ µ µ µ º» ¼ ¼ ½ ¾ ¾ À Á Â Ã Ä Å Æ Ç È Å Ç É Ç Å É Ê Ç Å Ë Ì Ç Å Ë Ì È Å É È Å Ë Í Î Ï Ð Ñ Ò Ó Ô Õ Ö Ô Õ Ô Ó Ø Ù Ú Û Ü Ý Ú Þ ß à á Þ ß â ã Þ ß ä å Þ ß ä à

83 Pietro Veronesi Governments and Asset Prices page: 83 Aggregate Results õ ö ø ù ú û ø õ ö ø ù ú û ø õ ö ø ù ú û ø õ ö ø ù ú û ø ü ú ý þ ÿ û ø ú ÿ ú ÿ õ ö ú þ ú ö ú þ ú ö ú þ ú þ ú û ø ö ú ÿ þ ø ú ÿ þ ú ý ú ú ø ý ú ø ý! " # $ % # & % & # ' ( # " & # & ) ) # ) ' # * ) # ( ( # ' * # % % + # & & # %, -. / : 9 ; : ; 9 : ; < : = < : > = = : > = : ; < : ; < : < = : 9? < : A B C D E F G H I J K L I M N O P P Q R S Q T P Q R S Q U S Q U O U V Q T W Q X S Q R S Q P U Q Y O P U Q W O W Q T Z [ \ ] ^ _ ` a \ b c d e f d g e d h f d i f d e h d j f d e f d c f d f f d j h d f f d e k l m n p q r s t u v w x v w y y w z v w v v w { y w y w x v w v v w v y w x u v w y v w v } ~ ƒ ˆ ˆ Š Š Š Š Š Œ Š Š Œ Š Œ Š Ž š š š š š š š š š š š œ ž Ÿ ª «««««««± ² ³ µ ¹ º» ¼ ½ ½ ¾ ¾ À Á ½ ¾  à ¾ Ä ¾ Å Â Â ¾ ½ ¾ Ä Ã ¾ ½ ¾ À ¾ Á  à ¾ Á  ¾ Ä Æ Ç È É É Ê Ë Ë Ì Í Î Ï Ð Ñ Ò Ó Ô Õ Ó Ò Õ Ö Ó Ó Ø Ó Ù Õ Ò Ó Ö Õ Ó Ú Õ Ó Ö Ó Ò Ó Û Û Ó Ù Õ Ó Õ Ü no Ý Þ ß à á â ã ä å å æ ç æå ß æ è è à å ä é ê ë ì í î ì ï ð ë ñ ì ò ò ì ó ó ì ï ð ó ó ì ó ó ò ì í ð ñ ì í ñ ì ï ô ï ì ó í î ì ñ ñ ì í õ ö ø ù ú û ü ý þÿ ø ù ý ú! " # $ " % &! ' " ( ( " ) ) " % & ) ) " ) & ) "! & ' " # ' " %! * " # & ' # " $ & " ' +, -. / : ; < = A B C C D E DC = D F F > C B G H I J K L M K N M L K O P K O P K O M Q K J J P K Q L K Q R K Q N P K P P S K P R K P T U V W X Y Z V [ \ ]^ _ ` a W b X c \ Y d e f g h i j h k j i h l m h l m h l j n h g m k h m i h n o h n g g h j g g h n o h g p q r s t u v r w s x y xz s x { { t z } ~ ƒ ƒ ƒ ˆ ƒ ˆ

84 Pietro Veronesi Governments and Asset Prices page: 84 Taxpayer Costs and Aggregate Effect Government receives securities in exchange of $125 bilion equity infusion. Preferred equity with 5% coupon, increasing to 9% after 5 years. 10-year warrants Valuation of these securities is sensitive to assumptions. We consider many scenarios to obtain upper and lower bounds.

85 Pietro Veronesi Governments and Asset Prices page: 85 Taxpayer Costs and Aggregate Effect Š Œ Ž x y z { } { y } ~ š œ ž Ÿ œ š š ª «š ª «š ª «š ª «ª «š ± ª «š ± š ª «² ± ± š ª «³ ± š ± µ ª ª ª ««ª ª ««ª ² ± ± œ š ¹ º» ¼ º ¼ ½ ½ ¼ ¾ ¼» º ¹ ¼ ¹ ¼ À Á Â Ã Ä Å Æ Ç È É Ê Ë Ì Â Í Î Í Ï Ð Ñ Í Ð Í Í Ò Ð Ó Í Ð Ï Í Ï Ð Î Ï Ð Ô Õ Ö Ø Ù Ú Û Ü Ý Þ Û ß à á â á ã ä á á ä å á æ ä á ã ä ç á å ä è é ä ê ë ì í î ï ð ñ ì ò ó ô õ ö ô ò ô ó ö ô ø ô ù ö ô ú û ü ý ý þ ÿ! "! #! $ # #! % & $ ' ( ) * +, * ( ( *,, *, ) * +, * -. / : ; < = 9 < : = < = : < 9 > < A B C D E F G H I F G J K L M N O P O M P N Q P R M P S T P Q S P U V W X X Y Z Z [ \ ] ^ _ ` a b c d a c e e c f ` c b b c g g c h i j k l m n o p q r n r n q s n r q p n n n q t n s q o u v q v o t q w Š Œ Ž Š [ \ ] ^ _ ` ^ \ ` _ a b c d e ƒ ˆ Š Œ Š Ž Š Ž Š Ž Š š Š Œ Š œ Š ž Ÿ Ÿ Š Ž Ž Š Ÿ Ÿ Ž Š Œ ª ««± ² ³ µ ¹ º» ¼ ½ ¾ ½ ¼ º ½» À ½» º ½» À ½» Á Â Ã Ä Å Æ Ç È É Ê Ç Ë Ì Í Î Ï Ð Ñ Ò Ï Ñ Ò Ï Ó Ñ Ô Ô Ñ Õ Ï Ó Ñ Ô Ô Ñ Õ Ö Ø Ù Ú Û Ü Ý Þ ß à á ß â ã ß ã ã ß ä ã ß ã ã ß ä å æ ç ç è é ê ë ì í î ï ð ñ ò ï ð ò ï ð ó ò ï ó ò ï ð ó ò ï ó ò ï ô õ ö ø ù ú û ü ý þ þ ø ö ÿ ÿ ÿ! " #! $! $ % & ' ( ) * +, ) * -. / : 3 4 ; < = = A B C D E F G H I F H J J H K L H J J H K L H J M N O P Q R S T U V W R U X Y U Z W X U S Z T U W W T U Z Z X U Y

86 Pietro Veronesi Governments and Asset Prices page: 86 Cost of Debt Guarantee and Deposit Insurance f g h i j k l m n o p n q p r j s g t u v j h p w x g y g t p j j z y n { } j } h ~ p r j v m ƒ ˆ ƒ Š Œ Ž ˆ Š Œ Š Œ Š ˆ ƒ ˆ ƒ ƒ ƒ ˆ ˆ ˆ ˆ ƒ Ž ˆ Œ ƒ ˆ ˆ Š ƒ ˆ ƒ Ž ˆ ˆ ƒ Š ƒ ƒ Œ ˆ Œ Œ Š ˆ Š š œ ˆ ˆ ž ˆ ˆ Š ˆ Ÿ ƒ Œ Š ƒ ˆ ƒ Š ˆ ƒ ˆ Š œ ƒ ˆ ƒ Ž ˆ ˆ ƒ ƒ Œ Š ƒ Š ƒ Š ƒ ƒ ƒ Ž ƒ ˆ Š ƒ ž ƒ ƒ ˆ ˆ ƒ ƒ Œ Š Œ Š ƒ ˆ ƒ ƒ Œ ƒ ƒ ƒ Ž ƒ ƒ ƒ ƒ ˆ ƒ Œ Š ˆ ˆ ƒ ƒ ˆ ƒ Œ ƒ Š ƒ ˆ ƒ ƒ ƒ Ž ˆ ˆ ƒ ƒ ª Œ Š ƒ Š ƒ Ž ˆ «Š Ž ˆ Š Œ Š Œ Š Œ ƒ Š ƒ ˆ ƒ ƒ ƒ Ž ˆ ˆ ƒ ƒ ª ƒ ƒ ƒ Œ ƒ ˆ ˆ ƒ ƒ ƒ ƒ ƒ ƒ ƒ ˆ Š ª ˆ ƒ Š Œ Ž Œ Œ Š ƒ ƒ Œ ˆ ƒ ƒ ˆ ƒ Š ± ² ³ µ ¹ µ º ± ±» ¼ ½ ³ µ ¹ µ º ± ¾ µ ± º ± À Á  ± ¹ Ã Ä Å Æ Ç È É Å Æ Ç È É Ê Ë Ì Ë Í Î Ï Ð Ñ Ò Ó Ô Õ Ö Õ Ø Ù Ú Û Ü Ý Þ ß à á á â à ã ä å à â æ ç è é ê ë ê ì í î î ï ð ñ ò ó ñ ô õ ñ ö ø ö ó ù ú û ü ý þ ö ÿ! " # $ % & ' & ( ) * +, -. - / / / 4 4 /. 1 5 / 6. / / : ; < = A B 8 C D E F G H I F C C G J F K L F J H E F I L F L K L F K M N O P Q R S T U V S W X Y Y Y Z [ \ Y Z [ Y ] ] Z ^ _ Z ` a Y ] Z a _ Z Y \ _ Z Y b c d e f g h c i j k l i i l k m i n k o p k j i o k i p k p p p k l q r s s t u v w x y z { } ~ ~ z { { { { { { { ƒ ˆ Š Œ Ž š œ ž œ Ÿ ª «± ² ³ µ ¹ º» ¹ ¼ ½ ¾ À Á  À Ã Ä Á Å Æ Á Ç Â Á Ä Á Ä Ä Á Ä Ä Ä Á Ç È É Ê Ê Ë Ì Ì Í Î Ï Ð Ñ Ò Ó Ô Õ Ò Ö Ô Ó Ò Ô Ö Ö Ô Ø Ô Ø Ø Ô Ø Ø Ø Ô Ù Ú Û Ü Ý Þ ß à à á â ã ä å á ä ä æ á ã â ä æ á ã æ á ç â á ç

87 Pietro Veronesi Governments and Asset Prices page: 87 Where Does the Value Come From? One possibility is that the capital infusion and the renewed access to funds enables banks to take advantage of the positive net present value lending opportunities. Yet, we know from Ivashina and Scharfstein (2010) that the discretionary lending of the major banks went down, not up during this period. Another possibility is to stop an inefficient bank run.

88 Pietro Veronesi Governments and Asset Prices page: 88 Enterprise Change and Run Index è é ê ë ì í î ï ð ñ ò í ó ô í õ ð ö ð î í ò ò õ ô í ø ó ù ú í û ð ü ò û ý ó ý õ ù õ î þ û ÿ ú ð "! # $ % & ' ( ) * +, -. / ) 4 )5 6. / * 8 9 : ; < 9 = > : < 9 : 9 ; : > A > <@ : B? C D : E <; ; : F B? C D

89 ƒ k n o Pietro Veronesi Governments and Asset Prices page: 89 Enterprise Change and Past Performance G H I J K L M N O P Q L R S L T O U O M L Q V Q T S L W R X Y L Z O [ Q L \ T Z Y S ] M Z P ^ _ R Q ^ L M [ L Q ` Z Q a R O P L ˆ w x u v l m y z b c d c e f g p q r s t t ~ } { Š Œ Ž š œ ž Ÿ œ Ÿ š Ÿ ž ª «ª ª «± ² «h i j

90 Pietro Veronesi Governments and Asset Prices page: 90 Computing Bankruptcy Costs The value of a firm can be written as the value of an unlevered firm minus the present value of bankruptcy costs. where P = Therefore, we have V = t=1 CF(t) (1 + r) t BC P 5 i<j (1 p i )p j j=1 (1+r) j + 5 i=1 (1 p i )p 5 (r+p 5 )(1+r) 5 V 1 V 0 V 0 =. CF 1 (t) t=1 (1+r) t CF 0 (t) t=1 (1+r) t V 0 BC V 0 P Exploiting the (limited) cross-sectional information, we obtain V i V i 0 = P i = The cost of government intervention is about 2.5% of firm value. = The bankruptcy cost is 22% of firm value.

91 Pietro Veronesi Governments and Asset Prices page: 91 Calibrating a Merton s Model We calibrate a simple Merton (1974) model modified to take into account the possibility of a liquidiyt shock (bank run) at the time of rolling over banks debt. We use this model for three exercises 1. Check the transfer of wealth from equity holders to bond holders due to a pure capital infusion without any other deadweight cost. Model implies transfer of $30bn from equity holders (government) to bondholders. But it cannot explain the changes in value of equity and debt as measured in the data. = Additional benefits from eliminating bankrun 2. Check the model can explain the change in equity and debt around the announcment, and use it to compute the change in the probability of a run and the recovery rate. 3. Use the estimated model to do conterfactuals: Could the government do any better by using other policies?

92 Pietro Veronesi Governments and Asset Prices page: 92 ³ µ ¹ º» ¼ ½ ¾ ¾ À Á À  ¼  à À Ä Å Â Æ ¾ Ç È È É Ê È Ë Ì Í Î Ï Ð Ñ É Ð Ñ É Ò Ê Ó Ï É Ô Ç Ê Õ Ö Ê É Ò É Ó È Ê É Ò Ð Ø Ø Ð Ñ É Ò Ð Ù È Ð Ò Ê Ê É Ò Ï Ú É Û Ê Ü Ý Ú Ú É Þ Ð È Ó Ê È Ô ß Ð à É Ñ É Ò á Ü Ê Ê Ó È Ê Ó Ï É á Ê É Ò É Ó È Ü Ø È Ð Ü Þ Ò Ð Û Ü Û Ó Ø Ó Ê â ã Ð Ù Ü Ø Ó ä å Ó Ú Ó Ê â È Ð æ ç á à Ó æ Ò É Ú å æ É È Ê É Ñ Ü Ø å É Ð Ù Ü È È É Ê È Û â è á Ê Ü Ê Ó È á Ó Ù Ê É Ø Ó ä å Ó Ú Ó Ê â È Ð æ ç Ó Ê È Ê É Ý Ë Ì Õ Ö Î é è Ë Ì Õ Ö ê Î ë ì Ù Ü Ê Õ Ö Ë Ì Õ Ö Î í î Ö á Ê É Ò É Ó È Ú É Ù Ü å Ø Ê Ü Ê Õ Ö Ô ì Ý Ê Ó È æ Ü È É á É ä å Ó Ê â Ü Ý Ú Ø Ð Ý ï Ê É Ò Ï Ú É Û Ê Ð Ø Ú É Ò È Ü Ò É à Ó Þ É Ú Ð å Ê á à Ó Ø É È Ð Ò Ê Ê É Ò Ï Û Ð Ý Ú Ð Ø Ú É Ò È Ò É æ É Ó Ñ É Ë Ì Õ Ö Î Ô ì Ù Ë Ì Õ Ö Î ð î Ö á Ü È È É Ê È Ë Ì Í Î É Ñ Ð Ø Ñ É Ü æ æ Ð Ò Ú Ó Ý ï Ê Ð Ü Ø Ð ï Ý Ð Ò Ï Ü Ø Ï Ð Ú É Ø å Ý Ê Ó Ø Õ ñ Ô Ç Ê Õ ñ á Ú É Ù Ü å Ø Ê Ð æ æ å Ò È Ó Ù Ë Ì Õ ñ Î í î ñ ò ó ô õ ö ø ù ú û ü õ ý õ þ ù ô ÿ ü õ ö þ þ õ ö ÿ ö ù õ õ ú õ þ ô û ù ÿ þ õ ÿ ý ô ÿ ö ù õ õ ú õ ö þ ù ô õ ú õ þ ô ø ü ý ÿ ù ù õ ö þ ý þ þ õ þ ÿ " # $ % & ' ( ) * Ç 0! +, - +. Ç Ê! 1 /, /.