Nort
rancis
National Universiq of Singapore
Unique data availability and institutional arrangements for new issues in Singapore allow a direct
test of the empirical implications of Rocks model of pricing unseasoned new issues. Our empirical
results are consistent with the model. Specifically we find that the unseasoned new issues anomaly
disappears when the rationing associated with new issues is incorporated into the analysis. The
winners curse is evident in allocation patterns used in Singapore.
lications of Rocks
etails of all new issues since the inc ration of the Singapore
Exchange (SSE) in 1973 listed on the SSE
monthly SSE journals. Seventy new issues resu
70 companies at the SSE library to extract the details of
adopted and find data for 48 issues. Before 1980 few firms routinely filed these
details with the exchange. A comprehensive search is made of the daily
business newspapers (the Business Times and the Straits Time) for the period
between he application-closing date stated in the prospectus and the lis
date. This produces data on the allocation basis for a further ten issues.
call the remaining companies and their underwriters for the data (wit
subsequent written confirmation of the request), and establish a further eight
data points. Thus, for 66 of the 70 firms we are able to establish information
on the allocation system adopted. This allows us to calculate the probability of
an allocation, conditional on the size of an application, for each is
Other data are relevant to our experiment. First-day sales prices
and last sale) are extracted from the daily trading summaries of the SSE. The
issue price per share, the number of shares offered, the underwriter and the
applications-close date are collected from rke prospectus. TIPe risk-frez rate of
interest and the prime rate prevailing during this listing lag
close date to listi re extracted from publications of the
ity of Singapore
thod
F. Koh and T. Walter, The pricing of unseasoned issues
,,=ln -
I
Excess
Return 0.271
i
0.27 -j
0.269
1
0.267
Number of hots (of one thousand shares each) in the Application Strategy
Fig. 1. Average initial-listing-day excess return for 66 new issues made between January 1973 and
June 1987 on the Singapore Stock Exchange for investment strategies in the range of one to 1,000
lots (of 1,000 shares each) on the assumption that no rationing occurs. The estimated t-statistics
that these excess returns are greater than zero are in the range of 4.01 to 4.24.
hese returns are then averaged for the sample of 66 issues for which we have
data.
The assumption in fig. I is that a full allocation of the requested shares is
granted, though we incorporate the cost of an application (i.e., interest costs
and fixed costs) as d~scrih.i
. in the previous section.
Irrespective of the investment strategy ado
anomaly is present in fig. 1.
rate (with associated t-statis
The discontinuities in these results at 11, 155, and 163 lots are caused by three issues in which
the maximum allowable application size was 10, 154, and 162 lots, respectively. These limits were
to comply with regulations that limit the maximum individual shareholding in newspaper
es (two cases) and in a chemical company.
for the issues in this study is in the range of S$l to S$5, with a mean of
oximately U$l, so an application for one lot involves an amount of
Probability
of an 0.5 Overpricedksues(9)
Allomtion
0.
-l___
Fig. 3 Probability of receiving an allocation of shares in 66 new issues made between January
1973 and June 1987 on the Singapore Stock Exchange for investment strategies in the raulge of one
to 1,ooO lots (of X.000 shares each) and subdivisions according to whether the issue was
underpriced (57 issues) or overpriced (9 issues).
The allocation bias also depends on the extent of unde ricing, although
these results are not reported in fig. 2. The probability of receiving an
allocation in new issues where the excess return is positive but less than 30
percent is in the range of 0. plication for one lot) to 0.23 (fo:r 1,000
lots). The corresponding probabilities for issues vwith initial returns greater
than 30 percent are 0.16 and 0.06, respectively, less than one third the Gze.
Fig. 3 results plot average returns (after incorporating ratio
tion costs) against the ntumber of lots in the applicatio
(Xi, S) are calculated as follows,s and these are then summ
Number of Lots (of one thousand shares each) in the Application Strategy
Fig. 3. Average initial-listing-day excess return for 66 new issues made between January 1973 and
June 1987 on the Singapore Stock Exchange for investment strategies in the range of one to 1,000
lots (of 1,000 shares each) after incorporating the actual rationing process adopted in each
new issue. The estimated t-statistics that these excess returns are greater than zero are all less
than 1.0 with the exception of that for an investment strategy of one lot, where the estimated
t-statistic is 1.51
,+Si(N,-ai . ,)-I, 1.
T-FC
i .s -Ri
3 N,
l
1
Table 1
Correlations between initial-listing-day excess returns and the application and allocation patterns
in 38 new issues made between January 1973 and June 1987 on the Singapore Stock Exchange for
which sufficient data exist to split total applications and allocations into subcategories based on
application size.
issue is 1.51.9 Adopting this strategy would have produced 54 positive returns
and only 12 negative returns. Recall however from fig. 2 that the probability of
receiving an allocation of overpriced securities is approximately three times
that of receiving an allocation in underpriced securities, and thus statistical
significance would be overstated by using a binomial test on the sign of the
initiai return.
Table 1 provides additional evidence on the bias in the allocation system
faced by the uninformed investor. Thirty-eight of the 66 new issues provide
sufficient data for this test. VJe are concerned with the question of whether
large- or small-investor applications and allotment patterns differ for issues
that are over- and underpriced. If the size of an investors application is a
reasonable proxy for his information, ocks analysis su;2ests there shoul
a negative correlation between underpricing and the proportion of an issue
subscribed for and allocated to small (uninformed) investors. This association
is caused where informed investors withdraw from overpriced issues (negative
initial-listing-day excess returns) and thus cause a high proportion of the issue
to go to the uninformed, but swamp the underpriced
(positive initial-listing-day excess returns) and thus ca
the issue to go to the u
should dominate the und
between underpricing a
receive. inally, general
by a strong correlation
Skewness coefficients for g. 3 are in the range of - 0.11 to 0.72 a suggest return.\ arc
generally positively skewe
260 F. Koh urui T. Walter, The pricing 01 unseasoned issues
The average underpricing for the 38 issues for which sufficient data exist to
determine both application and allotment patterns is 34.7 percent. These
issues are on average oversubscribed 40 times, with a range of 2.2 to 248.1.
The Spearman rank correjlation between oversubscription levels and initial-day
returns is 0.952, which is significant at 1 percent.
In general the correlations in table 1 are significant and in the expected
direction. There is a negative correlation of - 0.088 between the proportion of
an issue applied for by small investors and initial-day returns. This negative
correlation is much more pronounced and is statistically significant (the
correlation of -0.363 is significant at 5 percent) for medium/small investors.
Small investors application strategies tend to be inversely related to the
success of the issue. Recall however that there is a systematic preference in the
allocation system for small investors and that on average new issues are
underpriced. These factors produce a lower negative correlation (- 0.064) for
small investors between allocated proportions and returns than between appli-
cation proportions and returns. Small (naive) investors application strategies
are cushioned by the allocation process. lo The applications of large investors
are positively and significantly correlated (0.275) with the issues outcome, as
expected. Large investors follow wealth-increasing application strategies at the
small investors expense?
These correlations do not perfectly capture the responsiveness of investors
in the various size categories to expected underpricing. The highly significant
correlation between oversubscription levels and initial-listing-day excess re-
turns is not driven solely by demand from a particular class or category of
investor; rather, it is a phenomenon that pervades the applications of aft,
investor classes. It is less pronounced, however, for small investors and more
pervasive for large investors.
able 2 explores the issue of investor responsiveness in more detail. Using
the same size-of-application definitions for small and large investors employed
in table 1, we calculate the proportion that total shares applied for by
investors in each category bears to the total shares offered. This proportion is
used as the dependent variable in four regressions (i.e., four size-of-application
investor categories) in which initial-day returns are the independent variable.
The dependent variable, which is a subscription rate by category, is a measure
of the extent to which investors in a particular size-of-application class are
There is always the possibility that informed investors might attempt to take advantage of the
preferential allocations afforded small applicants by breaking up their purchases into smaller lot
sizes. Indeed this practice occurred in the early years of the SSE. It is now illegal. howvcr. to
submit multiple applications. Further, the prospectus states that all applications bv an investor arc
declared void when an investor submits multiple applications. The cvidcncc in table 1 is consistent
with relativeIy little gaming of the allocation mechanism by informed investors.
The results in fig. 1 are insensitive to different definitions of boundaries for application
strategy size.
E Koh and T. Walter, The ; ricing of uns xS
L ortea isstres 261
Table 2
Regression results for four regressions that employ the initial-listing-day return as the independent
variable and subscription level achieved within various subcategories of the total application pool
as the dependent variable for 38 new issues made between January 1973 and June 1987 on the
Singapore Stock Exchange for which sufkient data exist to split total applications into various
subcategories based on application size.
--
Demand by investors applying for
2,OC! to 100,000 to
99,000 shares 249,000 250,000 shares
1,000 shares (ix., medium/ shares (i.e., or more (i.e.,
Regression results and (i.e., small small medium large
related statistics investors) investors) investors) investors)
Rock (1986, p. 185) argues that it is essential to establish that uninformed demand expands as
issue price is reduced because unless an issuer can increase the chance of a full subscription in an>
state of the world by attracting uninformed investors to the offering, there is no point whatsoever
in pricing the shares at a discount. This evidence is consistent wit cman
uninformed investors.
262 F. Koh and T. Walter, The pricirlg of unseasoned issues
a subscription price w
d alone Will fully su
el a rational issuer would not rice an issue below
scribe the issue, so
seen as an additional test of the descriptive validity
I 10.102.
rice i5 ca e f rice
e3
Number of issues where the uninformed d alone fully subsc the issue f ative
levels of uninformed demand (in lots of hares) for 38 issues e between 1973
and June 1987 OR t mgapore Stock Exchange.
1 2oa
10 29
11 30
12 31
115 32
21 34
50 38
Using the strictest definition of uninformed demand allowed by our data produces 20 (out of
38) cases which are consistent with rational issuer behavior.
bAll 38 issues are characterized by rational issuer behavior if uninformed demand includes
applications up to 50,000 shares.
potential way to
ation data in tab
shares are (comparatively)
264 F. Koh and T. Wulter, The pricing oj wnsemoned issues
Awerags
Initial O-02
Litting
Day
Exe= 0.01
Return
-0.01
-0.02
200 300 400 580 600 700 800 900 1
Number of Lots bf one thousandshareseach) in the Application Strategy
Fig. 4. /iwrqge initia!-!istmg-day excess return for 66 new issues made between January 1973 and
June 1987 on the Singapore Stock Exchange split by underwriter identity into 25 issues underwrit-
ten by the Development Bank of Singapore, Ltd. (DBS), 22 issues underwritten by Wardley
Limited, United Chase Merchant Bankers Limited, and Singapore International Merchant Bank
Limited (frequent issuers), and 19 issues underwritten by others (otL:r issuers) for investment
strategies in the range of one to 1,OQOlots (of 1,000 shares each) after incorporating the actual
rationing process adopted in each new issue. The estimated t-statistics that these excess returns are
greater than zero are (i) all less than 1.2 tor the DBS issues, (ii) in the range of 1.01 to 1.54 for the
frequent issuers, and (iii) all less than 0.13 and generally negative for the other issuers.
Av
Ini
Fig. 5. Average initial-listing-day excess return for 66 new issues made between Januaq 1973 and
June 1987 on the Singapore Stock Exchange split by size of the issue into the 22 largest issues, the
medium issues, and the smallest issues for in- Jstment strategies in the range of one to 1,000 lots
(of 1,000 shares each) after incorporating the actual rationing process adopted in each new issue.
The estimated t-statistics that these excess returns are greater than zero are (i) ah less than 0.95 for
the huge issues, (ii) less than 0.61 and generally negative for the medium issues, and (iii) in the
range of 0.63 to 1.33 for the small issues.
mo
etwee~
investors
16
re iss
med -siz ivel
266 F. Koh ad T. Walter. The pricirtg of unseasorted issues
are reserve
268 F. Koh and T. Walter, The pricing of unseasoned issues
and 20 percent of
lying for quotatio
expected to meet the following criteria:
(3) minimum percentage of the issued and paid-up capital is in the han
areholders each holding nsi less than 500 shares and not more
10,000 shares:
Sam
ANNOUNCEMENT
The Ikzctors of Avimo Singapore Limited are pleased to announce that at the close of the
Application List at 12 noen, 21 April, 1987, there were 108,787 applicants applying for
1,754,99O,Ooc) shares, total!Ing $3,071,232,5OO. The offer for sale is approximately 104 times
subscribed compared with the 16.875,OOOshares available to the public for subscription at $1.75
per share.
The following table sets out tile details on the applications received and the bask of allotment:
No. of
Denomi- shares
nations Total no. allot ted Percent
of shares ot shares Percent of per succ- of the
applied No. of applied total Balloting essfJ1 No.of offer
for applicants for applications ratio applicant shares for sale
cOOO1 (000) (%1; (000) (000) (Tr)
______- .--- ~ . ______ --____ ---
l-4 93,207 --103.719 5.9 1:27 1 3,391 20.1
5-9 2,022 10,784 06 1:12 1 16X 1.0
10-49 7,464 88,404 5.0 2: 15 1 935 5.9
W-99 : pz1z3
i,\,// ?QC
56,Jt?_l f2 519 1 610 3.6
100-199 3,629 367,300 2b:9 - 1 3,629 21.5
2CW299 134 29,418 1.7 - 2 268 1.6
300-399 19 5,737 0.3 - 3 57 0.4
400-499 10 4 W 0.2 - 4 40 0.2
5OO-749 633 317:61; 18.1 - 5 3,165 18.8
750-999 9 7,850 0.5 - 6 54 0.3
1.OOO-1,999 482 $86,133 27.7 - 7 3,374 20.0
2.OOO--2,999 44 91,530 5 - 1I 484 2.9
?,WO--3,999 5 15,100 0:; - 14 70 0.4
3 12,000 0.7 - 16 48 0.3
21 105,013 0.0 _- 1x 37n 2.2
6,Wl-- 7,999 I 6,000 0.3 - 20 20 0.1
KOOO-9.999 I 8,000 0.5 -- 24 24 0.1
B0,04io- 10.0~~I --A-- 0 002 2.3 .- 25 100 -- 0.6
14~8,787 1.754.990 100.0 16.875 100.0
--
The Directors of Avi 0 Singapore Limited wish plicaats for their confidence in the
Company. Unsucces lkations and refun (partially) successful app
will be sent to applicants as soon as possible.
e Announcement,
PRESS RELEASE
The Directors of Hotel Tai-Pan Limited (The Company) are pleased to announce that at the
time of the closing of the Application List at 2:00 p.m. on 23rd July 1981, a total of 74,148
applications in respect of 288,463,OOOshares in the company had been received. T
monies amounted in aggregate to $375,001,900. The issue was therefore subscribed approximately
13 tumes.
No. of shares
allot ted per
successful
No. of shares applied for Balloting ratio application
woo 1:lO 1,600 -
2,000 to 5,000 I:5 1,000
6,000 to 10,000 1:2 1.OOO
?l,m! to 2$000 no balloting 1.000
21,rDOOto 50,000 9, 2.000
Sl.,OOOto 100,m 9. 4,000
101,000 to 150,m 9. 5,000
,r
315i,wOto 200,000 9,084)
201,000 to 300,000 ,* 14,000
3u1,OOO to 400,000 9) 16,000
401,000 to 500,000 9, 24,000
501,000 to 5CjL900 .1 30,000
I,OOO,OOand abtive .) 55,000
Every effort will be made to ensure the cxpe ious processing of succc
return of the application onies to unsucce~ i and partially successfu
Reatty, R.P. and J. Ritter, 1986, Investment banking, reputation, and the underpricing of initial
gs, Journal of Financial Economics 15, 213-232.
Higham, 1986, The performance of unseasoned new equity issues-cum-stock
lags in Australia, Working paper (University of Queensland, Brisbane).
Ibbotson, R.G., 1975, Price perform.ance of common stock new issues. Journal of FinanciaI
Economics 2,235-272.
Ibbotson, R.G. and J. Jaffe, 1975, Not issue markets, Journal of Finance 30, 1027-1042.
Koh, F. and T. Walter, 1987, Ahocation patterns for new issues in Singapore, Working paper
(National University of Singapore).
tter, J., 1984, The hot issues market of 1980, Journal. of Business 57, 215-240.
Rock, IL, 1956, Why new issues are underpriced, Journal of Financial Economics 15, 187-212.
Smith, C.W., 1986, Investment banking and the capital acquisition process, Journal of Financial
Economics 15, 3-29.