Real and Financial Industry Booms and Busts

Real and Financial Industry Booms and Busts Gerard Hoberg University of Maryland Gordon Phillips University of Maryland & NBER Paper presentation to Insead, May 2007 Motivation • Industry booms and busts are recurring phenomenon according to WSJ, NY Times • • • • • • Railroads 1860s, 1870s Petroleum exploration, paper mills (70s) Agriculture commodities (80s, 90s): “Hog Cycles” Winchester Disk Drives, 1982-83 Non-store retailing (catalogs), 1991-1992 Internet, Telecommunication, 1998-00 Insead, May 2007   August 9, 2006 NEW YORK (CNNMoney.com) -- Homebuilder Toll Brothers said the current slump in residential construction is unlike any it has seen in 40 years as it became the latest to warn of a glut in new homes for sale and a slowdown in the closely watched real estate market. "… it seems to be the result of an oversupply of inventory and a decline in confidence," he added. "Speculative buyers who spurred demand in 2004 and 2005 are now sellers; builders that built speculative homes must now move their specs; and nervous buyers are canceling contracts for homes already under construction." Insead, May 2007 Our Questions • • • How important are industry booms? Outcomes: How do firms’ cash flows and the stock market respond to industry booms? Test how competition impacts subsequent outcomes following booms and busts. • • “Excess/high” competition in concentrated industries? Schumpeter (1941), Perry (1984) Mankiw and Whinston (1986) Excess competition in competitive industries? Scharfstein (1988), Reinganum (1989), Hou and Robinson (JF, 2005) Insead, May 2007 Related Literature  Large literature on reversals, momentum and misvaluation, rational herding and informational cascades. • Some examples: De Bondt and Thaler (1985), Scharfstein and Stein (1990), Welch (1992), Jegadeesh and Titman (1993), Shleifer and Vishny (2003) Rhodes-Kropf and Viswanathan (2004) (rational misvaluation given asymmetric information and mergers) Both rational and behavioral explanations.  No industrial organization, microeconomics unspecified. Insead, May 2007 Related Literature 2 Rational Booms and Busts:    Irrational Booms and Busts:   Pastor and Veronesi (2005): Switch of uncertainty from idiosyncratic to systematic causes what appear to be bubbles. DeMarzo, Kaniel, Kremer (2005), “Overinvestment in new technology”, “may invest to the point that its expected return is negative” given consumption insurance. Also Gala (2005). Neg. Stock returns following high equity issuance: Baker and Wurgler (2000). Neg. Stock returns following high investment: Titman, Wei, and Xie (2004) and Polk and Sapienza (2004) for crosssectional results and Lamont (2000). Insead, May 2007 Related Literature 3 Real Options    Exercise of real options causes changes in risk and expected returns – Carlson, Fisher, Giammarino (2005) Beta declines faster with demand increases in competitive industries as firms in these industries exercise growth options faster as in Aguerrevere (2006). When demand decreases during bust, Beta increases given increased operating leverage and failure (rationally) to internalize positive effects of exiting. How product market competition affects incentives to gather information, volatility and returns (Gaspar and Massa (2005), Hou and Robinson (2005) and Peress (2006)). Insead, May 2007 Effect of competition on stock markets:  Conclusions     Booms and busts are not just in high tech industries Market competition & financing are crucial in understanding outcomes of booms. Transitions out of booms are more likely if relative investment and new financing are high. In competitive industries: • • • Subsequent operating performance and abnormal stock returns are negatively related to valuation metrics & industry new finance. Particularly true in high systematic risk industries. Risk changes only partially mitigate (but do not eliminate) these effects. Effects persist in high valuation industries. Results consistent with high competition in competitive industries not being internalized by the stock market. Insead, May 2007 Contrast in Concentrated Industries  In concentrated industries:  Little evidence of ex post predictability in operating cash flows or stock returns. Evidence consistent with firms in concentrated industries internalizing the effects of their production on industry outcomes.  Insead, May 2007 Importance of competition  Why does industry competition have such a large effect?   Cash flows: Economics 101 – produce more until prices re-equilibrate. Mean reversion quick given elastic supply response. Stock market: Why do prices especially in the stock market go so high initially? Overoptimism/Consumption hedge /Failure or lack of incentives to anticipate elastic supply response. Insead, May 2007 Outline 1. Hypotheses 2. How we construct our measure of relative industry valuation or “Booms” and “Busts” 3. Characteristics of Booms and Busts 4. Relation between ex post outcomes and Booms and Busts. We examine ex post:   Operating Cash Flows Adjusted or Excess Stock Returns Insead, May 2007 Main Hypotheses  H1: In concentrated industries with high valuations, high investment and high financing decrease industry cash flows and stock returns. H2: In competitive industries with high valuations, high investment and high financing decrease industry cash flows and stock returns.  Insead, May 2007 Risk-Based Hypotheses “Can we make predictability go away?”  H2b: Changes in stock returns can be explained given changes in sensitivity to priced risk factors / we also add a competition risk factor – Hou and Robinson, Peress.  H2c: During (following) industry booms, systematic risk decreases (increases) more in competitive than in concentrated industries. (Real Options Models) Insead, May 2007 Rational Theories of Booms and Busts “Can Rational Models explain our findings?  H3a: Booms have high idiosyncratic risk given technological uncertainty. Subsequent busts have increased systematic risk and decreased idiosyncratic risk (Pastor and Veronesi).  H3b: Consumption Hedging: In industries with high systematic risk, subsequent stock returns will be negatively related to industry investment, valuation and financing (DeMarzo, Kaniel, Kremer). Insead, May 2007 Toll Brothers Stock Insead, May 2007 Booms and Outcomes Pre-Boom Estimation period t-10 to t-1 Size of the Boomt time t char. using t-10 to t-1 coefficients Ex post Outcomest+1 =f(Boomt, NFt)_ (1) Valuation Model (Kothari, 2001, RRV, 2005) (2) M/B model (3) Simple P/E Model Insead, May 2007 ex post Dcash flow, returns Valuation Model Market Valuation Estimation period t-10 to t-1 (1) MVequityijt   j,0   j,1*BVequityijt   2,j*(abs(NI))  3,j *dummy NegNIijt   4,j *Levijt (2) MVequityijt /BVequityijt   j,0  1,j *(AT)   2,j*abs(NI)  3,j *dummy NegNIijt Estimated (1) and (2) by industry in logs, save coefficients each industry. Predict valuation in period t using period t RHS variables. Average and median R2 approx = .75 Insead, May 2007 Investment Model Predict Investment Estimation period t-10 to t-1 ln(Capx)i, j,t   j,0   j,1 * qi, j,t -1   2, j * ln(Assetsi, j,t -1 )   3, j * ln(abs(OIBDP j,t )) i,   4, j * dummy NegOIBDPi, j,t Estimated by industry, save coefficients each industry. Predict investment in period t using period t RHS variables. Insead, May 2007 Relative Valuation /Investment Relative MVi,j,t (RMV)= ln(MVi,j,t)Predicted Valuation (time t char. using coefficients from t-10 to t-1 data) Gives firm-level relative valuation RMVi,j,t = ln(MVi,j,t) - PMVi,j,t  Average over firms to get relative industry-level Booms.  Similar procedure for relative industry investment. Insead, May 2007 Booms and Outcomes by Industries Tests: • Are booms and outcomes different in concentrated and competitive industries? • Herfindahl measured (public firms), predicted using public and private firms. • We examine changes in operating cash flows and risk and style-adjusted ex post stock returns. Insead, May 2007 Competitive Industries: Booms Insead, May 2007 Competitive Industries: Booms - 2 Insead, May 2007 Outcomes Ex post Outcomest+1 =f(Boomj,t, NFj,t, RIj,t, Boomi,j,t, NFi,j,t, RIi,j,t,) ex post Dcash flow, returns Insead, May 2007 T4: Regressions Predicting Firm-level cash flows: CFi,t+1-CFi,t (industry-year adj.) (scaled by each period assets) Insead, May 2007 T4: Regressions Predicting Firm-level cash flows: CFi,t+1-CFi,t (industry-year adj.) Insead, May 2007 T5: Firm-level cash flow regressions (high market risk tercile) Insead, May 2007 Characteristic Adjusted Monthly Abnormal Returns ARi,t = RAWi,t – STYLEi,t for firm i in month t - Per Daniel/Grinblatt/Titman/Wermers (1997) and Wermers (2004) - Construct 125 style portfolios, rebalanced July 1 (size NYSE). - Conditional sorts on Size, Industry Adj. B/M, past 12 month return Lags: Davis/Frama/French (2000): - For monthly observations b/t July year t to June year t+1 - Size = CRSP market cap from June of year t - Accounting data from fiscal years ending in year t-1 - Past 12 month return from June of year t-1 to May of year t Insead, May 2007 Abnormal Returns FF 4 factor model + Mitchell/Stafford Alphas Step 1: Group firm/years as July year t to June year t+1 Step 2: Get intercepts FF ’93 + MOM (12 monthly obs./regression) RawAlphai,t   i   i MKTt   i  L   i SMB   iUMD   Step 3: Control for non-linearity MSAlphai ,t  RAWAlpha i ,t  STYAlphai ,t - Style matching based on same 125 style portfolios, same lag structure. Insead, May 2007 T6: Regressions Predicting Firm-level Abnormal Stock Returns: i,t+1 Insead, May 2007 T7: Return Regressions High Relative Valuation Tercile Insead, May 2007 T8: Return Regressions High Market Risk Tercile Insead, May 2007 Risk Tests  Examine if changes in risk occur following industry booms.   Industry technological adoption: Pastor and Veronesi (2005) We examine market beta and idiosyncratic risk. Real Options: Grenadier (2002), Aguerrrevere (2006)  We examine change in total risk.  Do any changes explain abnormal returns? Insead, May 2007 Univariate risk changes Percent of Risk that is Systematic (Standard Deviation) 0.3 0.295 0.29 0.285 0.28 0.275 0.27 0.265 0.26 0.255 0.25 0.245 -5 -4 -3 -2 -1 0 1 2 3 4 5 Insead, May 2007 Tables 9+10: Regressions Predicting Chg. In Market Risk: Dependent Variable “D” Insead, May 2007 Tables 9+10: Regressions Predicting Chg. Idio. Risk: Dependent Variable “D idio.” Insead, May 2007 ΔRisk – Adjusted Abnormal Returns Goal: Can Δrisk can explain return patterns? - Per Daniel/Grinblatt/Titman/Wermers (1997) and Wermers (2004) ARi,t = RAWi,t – STYLEi,t for firm i in month t ARi ,t     [DMKT  ]i ,t   [D L ]i ,t  Take residuals from the following in-sample regressions: [DSMB ]i ,t  [DUMD  ]i ,t   i ,t Define above residual as ARΔRi,t and run our previous regression: ARΔRi,t =  + 1 [Valuation] + 2 [Investment] + 3 [Finance] + Insead, May 2007 Table 11: Regressions Predicting ΔRisk Adjusted Abnormal Stock Returns? Do results go away? Insead, May 2007 Table 12: Firm-level Quintile Returns Insead, May 2007 Annual Profitability: Excluding Internet Yrs Relative Industry-level Valuation Quintile Returns Insead, May 2007 Conclusions    Booms and busts are not just in high tech industries Market competition & financing are crucial in understanding outcomes of booms. In competitive industries: • • • Subsequent operating performance and abnormal stock returns are negatively related to high relative industry valuation & industry new finance. Risk changes only partially mitigate these effects. Effects still present in highest valuation industries. Results consistent with stock and product market failure to internalize effects of high competition and investment / consumption hedge as in DeMarzo et. al. Insead, May 2007 Conclusions   In concentrated industries: Little evidence of ex post predictability in operating cash flows or stock returns. Evidence consistent with firms in concentrated industries internalizing the effects of their production on industry outcomes.  Insead, May 2007