Motivation Model Comments and Extensions
Why Do Firms Announce Open-Market Repurchase Programs?
Jacob Oded, (2005)
Presented by R. Djeliov
Boston College
PhD Seminar in Corporate Finance, Spring 2006
Outline
1
Motivation The Problem Previous Work
2
Model Setup
Empirical Results
3
Comments and Extensions
Motivation Model Comments and Extensions
The Problem Previous Work
Outline
1
Motivation The Problem Previous Work
2
Model Setup
Empirical Results
3
Comments and Extensions
What is the main problem in this paper?
Presentation focus on most interesting part of paper:
separating equilibrium in which good firms announce repurchase programs and bad firms do not
Price increase accompanies announcement of
open-market stock repurchase program (OMRP), even
though no commitment and often no actual repurchases
Oded constructs two-type signaling model that delivers
announcement returns, because in separating equilibrium,
good firms don’t incur cost when announcing, but mimicry
costly for bad ones
Motivation Model Comments and Extensions
The Problem Previous Work
Outline
1
Motivation The Problem Previous Work
2
Model Setup
Empirical Results
3
Comments and Extensions
Literature Review
Repurchase programs increasingly popular relative to other common forms of payout: appr. 90 percent of all announced repurchases (only 10–15 percent in 1980s)
"Most buybacks are stated, not completed" – WSJ, 1995, speculated actual repurchase rates of 30-40 percent Stephens and Weisbach (2000), and with Jagannathan (1998), suggest actual repurchase rate of 70-80 percent Grullon and Michaely (2002) show growth in repurchase programs in addition to dividend payouts
Netter and Mitchell (1989) report after market crash in
1987, many firms announced programs, but most did not
Motivation Model Comments and Extensions
The Problem Previous Work
Literature Review
Miller and Modigliani (1961) demonstrated that in perfect markets, firm’s payout policy irrelevant
Theories tried to explain why firms distribute cash in general and why they choose specific payout forms
Information-signaling theory suggests firms better informed than investors, so good firms can distinguish themselves by sending costly signal about their type
Daniel and Titman (1995) review signaling models in
corporate finance
Literature Review
Bhattacharya (1979), John and Williams (1985), and Miller and Rock (1985), present signaling with dividends
Ofer and Thakor (1987), and Persons (1994) study models of signaling with tender offer repurchases
Ikenberry and Vermaelen (1996) view repurchase program as option that gives firm ability to exchange market value for "true" value if prices fall
Announcing program grants firm option whose value
reflected in announcement return
Motivation Model Comments and Extensions
The Problem Previous Work
Literature Review
Brennan and Thakor (1990) suggested stock repurchases transfer wealth from small uninformed investors to large informed investors
Oded suggests that Ikenberry-Vermaelen option generates announcement returns because trading gains allow (good) firms to signal their value
regards OMRP as non-dissipative signaling tool in spirit of
Ross (1977), Heinkel (1982), and Brennan and Kraus
(1987), and partly Bhattacharya (1980)
Outline
1
Motivation The Problem Previous Work
2
Model Setup
Empirical Results
3
Comments and Extensions
Motivation Model Comments and Extensions
Setup Empirical Results
Model Intro
Timeline consists of three dates: t
0, t
1, t
2Agents risk neutral, zero interest rate, no taxes or transaction costs
Firm entirely equity financed: value of assets, V
t, follows exogenous random process
At t
0firm owns assets in place with fixed value θ and risky
project with random payoff value ˜c
Model Intro
N outstanding shares owned by "original" shareholders Firm type characterized by project value C
jwhere j{B, G}, C
j(0, 1) and C
B< C
G(notice type "increasing" in volatility) At t
1, with probability q >
12project value realized to C
j, and with (1 − q) to −C
j, where 0 < C
j< θ
Fraction p of firms Bad type, and (1 − p) Good type
Good firms have more investment opportunities than bad
firms, but investment opportunities associated with higher
risk than assets in place
Motivation Model Comments and Extensions
Setup Empirical Results
Model Intro
After project value realized, some original shareholders become constrained and sell K shares at market price through auction (price bidding for entire block of shares) At t
2whole firm sold (dismantled), so N − K shares of original shareholders sold along with those of new shareholders, and proceeds represent terminal wealth At all dates firm has all available information
At t
0knows its type and can announce OMRP, i.e. option to
repurchase fixed number γ < K of shares at t
1Model Intro
Firm can repurchase only if announcement made, gets priority allocation, and remaining K − γ shares allocated to market according to price bids
Market cannot tell firm type at t
0(first asymmetry), but knows distribution of firms and value processes
At t
1firm knows realization of ˜c but market gets no new information (second asymmetry)
At t
2all information public
Motivation Model Comments and Extensions
Setup Empirical Results
Figure 1. - Timeline
Case (1) – No program-signaling ability
Firm knows realization of ˜c but knowledge worthless because firm cannot participate in t
1market, consequently P
0= E
p[E [V
2]]/N = (θ + (2q − 1)C)/N, where E
p[.]
expectation with respect to distribution of firm population
P
1= P
0and P
2ji= V
2ji/N = (θ ± C
j)/N, where j{B, G},
type, and i{L, H}, realization at t
1: e.g. P
2BL= (θ − C
B)/N
and P
2GH= (θ + C
G)/N
Motivation Model Comments and Extensions
Setup Empirical Results
Figure 3(a) – Price paths without program signaling
Case (1) – No program-signaling ability
Types unobservable at t
1, so short-term shareholders of both types get KP
1= K [θ + (2q − 1)(pC
G+ (1 − p)C
B)]N regardless of type and value realization
At t
1short-term shareholders of bad firms gain at expense of those of good firms
At t
2long-term shareholders get prices according to type and value realization (all information known then)
Competitive market, expected gain of new shareholders 0 They pay KP
1, and get expected terminal wealth
KE
p[E (P
2)] = KP
1= K [θ + (2q − 1)(pC
G+ (1 − p)C
B)]N
(no quantity risk without program)
Motivation Model Comments and Extensions
Setup Empirical Results
Case (2) – Program signaling possible
At t
0firm can announce repurchase program and condition t
1bid on value realization (but market cannot do so) Seek separating equilibrium in which only good firms announce program at t
0and repurchase or not at t
1, depending on realization
Bad firms don’t announce at t
0and can’t repurchase at t
1Case (2) – Program signaling possible
Oded assumes existence of such separating equilibrium, so prices must equal (when types known to market):
At t
0, P
0j= V
0j/N = (θ + (2q − 1)C
j)/N where j{B, G}
At t
1, P
1G=
qK −γ
N−γ(θ+CG)+(1−q)KN(θ−CG)
qγ(K −γ)
N−γ +K −qγ
and P
1B= P
0BAt t
2, P
2GL= (θ − C
G)/N,
P
2GH=
(θ+CN−γG)−γP1G(firm repurchased γ shares at t
1)
P
2BH= (θ + C
B)/N, and P
2BL= (θ − C
B)/N.
Motivation Model Comments and Extensions
Setup Empirical Results
Figure 3(b) – Price paths with program signaling
Case (2) – Program signaling possible
At t
0good firms announce program and enjoy positive announcement returns
Bad firms do not announce, and their stock price falls
Post-announcement, t
0stock price higher for good firm,
and lower for bad one (can separate types already)
Given separation, P
0j= expected value of stock type
At t
1, good firms repurchase if high value realization and
do not repurchase if low
Motivation Model Comments and Extensions
Setup Empirical Results
Case (2) – Program signaling possible
For good firm, t
1price lower than expected value because market takes informed trading into account (firm takes γ shares if high value, less available for market)
For bad firm, t
1price equal to expected value (no quantity risk), lower than in no-program economy
At t
2, for good firm with high realization price reflects gains
from informed trading, higher than in no-program case (?)
With low realization same as in no-program economy
For bad firm, t
2price same as in no-program economy
Case (2) – Program signaling possible
Repurchasing option separates good firms and increases expected terminal wealth of their original shareholders Terminal wealth of original shareholders of bad firms not affected by quantity risk because they don’t announce Compared to no-program economy, original shareholders of good firm better off by (2q − 1)(1 − p)(C
G− C
B)(K /N), and original shareholders of bad firm worse off by
(2q − 1)p(C
G− C
B)(K /N)
Distribution of wealth between short-term and long-term shareholders also different
Good firm transfers wealth from short-term shareholders to
Motivation Model Comments and Extensions
Setup Empirical Results
Figure 3(ab) – Price paths compared
Robustness of equilibrium
Recall without announcement P
1= P
0, but with program under separation P
1G< P
0Gand P
1G< θ/N
All parameters fixed (including C
G> C
B> 0), separating equilibrium exists iff γ(γ
c, K ), where γ
c=
K1+2q−11−q N−KN
If program big enough, t
1equilibrium price decreasing in C
Proposition 3: "in this range, given a marginal increase in
C, the marginal increase in the value of the option to
repurchase is larger than the marginal increase in
expected stock value"
Motivation Model Comments and Extensions
Setup Empirical Results
Robustness of equilibrium
Implies that separating equilibrium exists for all γ(γ
c, K ), because gap between types in option value larger than gap in expected value, and hence enough to deter bad type from mimicking
Good firms announce program and their t
1price reflects informed trading (quantity risk)
Announcement costless for good firms because loss of short-term shareholders exactly offsets gain of long-term shareholders from informed trading
When bad firm announces, short-term shareholders suffer
losses of good firm, but long-term shareholders only enjoy
Robustness of equilibrium
With large enough program, good firms can push their t
1price so low that mimicking bad firm’s long-term
shareholders never recover what short-term ones lose
Firms maximize expected aggregate wealth of original
shareholders, so bad firm better off not announcing, and
separating equilibrium holds
Motivation Model Comments and Extensions
Setup Empirical Results
Figure 4. - Separating Equilibrium
Robustness of equilibrium
Shows "single-crossing property" of indifference curves Announcing program costless for good firm regardless of γ , but for bad firm announcing cost strictly increasing in γ For γ < γ
c, bad firm would mimic good one
At γ = γ
c, benefit from mimicry exactly offsets cost (notice that pooling equilibrium also possible here – need extra assumption to resolve issue)
For γ > γ
c, mimicry results in net loss and separating
equilibrium exists (set γ(γ
c, K ) as close as possible to γ
c)
Motivation Model Comments and Extensions
Setup Empirical Results
Outline
1
Motivation The Problem Previous Work
2
Model Setup
Empirical Results
3
Comments and Extensions
Model Implications
Model predicts announcing firms condition actual
repurchases on future realization of value – consistent with Stephens and Weisbach (1998) findings
Model assumes uncertainty in production technology positively correlated with firm quality – consistent with Jagannathan, Stephens, and Weisbach (2000) findings that announcing firms have a more volatile cash flow than firms that do not announce, and with
Ikenberry and Vermaelen (1996) findings that firms with
higher β are more likely to announce".
Motivation Model Comments and Extensions
Setup Empirical Results
Model Implications
Model predicts post-announcement expected returns relatively low in short run and relatively high in long run – Ikenberry, Lakonishok, and Vermaelen (1995) find
one-year abnormal return in year 3 (4.6 percent), significantly higher than in years 1 and 2 (2 and 2.3 percent, respectively)
Model also predicts long-run returns correlated with actual
level of repurchase – consistent with Ikenberry, Lakonishok
and Vermaelen (2000)
Model Implications
Model suggests that during repurchase period announcing firms "time" market – empirical evidence inconclusive:
Cook, Krigman, and Leach (2004) "find that many firms repurchase following price drops, and
Stephens and Weisbach (1998) found that actual repurchases are timed with temporary undervaluation"
Model shows no relation between program size and announcement return – consistent with findings of Stephens and Weisbach (1998), but
Ikenberry, Lakonishok, and Vermaelen (1995), Ikenberry
and Vermaelen (1996) find positive relation
Motivation Model Comments and Extensions
Setup Empirical Results
Conclusions
OMRPs puzzle explained in part
by model with separating equilibrium where good firms send costless signal about their type, too costly for bad firms to mimic when program size large enough
Equilibrium robust, implications supported in research
Modest contribution to literature
Critique
Some comments already noted during presentation Paper overall very interesting but somewhat lengthy
Results depend crucially on assumptions of severe liquidity constraints, and variance positively correlated with
expected return (in production technology) across types Abstracts from cost of paying out cash, which may create value [Jensen (1986), and Chowdhry and Nanda (1994)]
Some predictions contradict empirical evidence
Motivation Model Comments and Extensions