Real and Financial Industry Booms and Busts

1
Real and Financial Industry Booms and Busts

• Gerard Hoberg
• University of Maryland
• Gordon Phillips
• University of Maryland NBER

Paper presentation to Insead, May 2007
2
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

3

• 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."

4
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)

5
Related Literature

• Large literature on reversals, momentum and
  misvaluation, rational herding and informational
  cascades.
• Both rational and behavioral explanations.
• 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)
• ? No industrial organization, microeconomics
  unspecified.

6
Related Literature 2

• Rational 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).
• Irrational Booms and Busts
• 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 cross-sectional results and Lamont
  (2000).

7
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.
• Effect of competition on stock markets
• How product market competition affects incentives
  to gather information, volatility and returns
  (Gaspar and Massa (2005), Hou and Robinson (2005)
  and Peress (2006)).

8
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.

9
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.

10
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.

11
Outline

• Hypotheses
• How we construct our measure of relative industry
  valuation or Booms and Busts
• Characteristics of Booms and Busts
• Relation between ex post outcomes and Booms and
  Busts. We examine ex post
• Operating Cash Flows
• Adjusted or Excess Stock Returns

12
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.

13
Risk-Based HypothesesCan 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)

14
Rational Theories of Booms and BustsCan
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).

15
Toll Brothers Stock
16
Booms and Outcomes

• Size of the
• Pre-Boom Boomt Ex
  post
• Estimation period time t char. using
  Outcomest1
• t-10 to t-1 t-10 to t-1 coefficients
  f(Boomt, NFt)_
• (1) Valuation Model ex post
• (Kothari, 2001, RRV, 2005) Dcash flow,
  returns
• (2) M/B model
• (3) Simple P/E Model

17
Valuation Model

• Market Valuation
• Estimation period
• t-10 to t-1
• 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

18
Investment Model

• Predict Investment
• Estimation period
• t-10 to t-1
• Estimated by industry, save coefficients each
  industry.
• Predict investment in period t using period t RHS
  variables.

19
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.

20
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.

21
Competitive Industries Booms
22
Competitive Industries Booms - 2
23
Outcomes

• Ex post
• Outcomest1
• f(Boomj,t, NFj,t, RIj,t, Boomi,j,t, NFi,j,t,
  RIi,j,t,)
• ex post
• Dcash flow, returns

24
T4 Regressions Predicting Firm-level cash
flows CFi,t1-CFi,t (industry-year
adj.)(scaled by each period assets)
25
T4 Regressions Predicting Firm-level cash
flows CFi,t1-CFi,t (industry-year adj.)
26
T5 Firm-level cash flow regressions (high
market risk tercile)
27
Characteristic AdjustedMonthly 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 t1
• - 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

28
Abnormal Returns FF 4 factor model
Mitchell/Stafford Alphas

• Step 1 Group firm/years as July year t to June
  year t1
• Step 2 Get intercepts FF 93 MOM (12 monthly
  obs./regression)
• Step 3 Control for non-linearity
• Style matching based on same 125 style
  portfolios, same lag structure.

29
T6 Regressions Predicting Firm-level Abnormal
Stock Returns ai,t1
30
T7 Return RegressionsHigh Relative Valuation
Tercile
31
T8 Return RegressionsHigh Market Risk Tercile
32
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?

33
Univariate risk changes
34
Tables 910 Regressions Predicting Chg. In
Market Risk Dependent Variable D?
35
Tables 910 Regressions Predicting Chg. Idio.
Risk Dependent Variable D? idio.
36
?Risk Adjusted Abnormal ReturnsGoal Can ?risk
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 Take residuals from the
following in-sample regressions
Define above residual as AR?Ri,t and run our
previous regression AR?Ri,t ? ?1 Valuation
?2 Investment ?3 Finance ?
37
Table 11 Regressions Predicting ?Risk Adjusted
Abnormal Stock Returns? Do results go away?
38
Table 12 Firm-level Quintile Returns
39
Annual Profitability Excluding Internet Yrs
Relative Industry-level Valuation Quintile
Returns
40
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.

41
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.