Documente Academic
Documente Profesional
Documente Cultură
Executive Statistical models for predicting takeover targets by using publicly available infor-
mation, specifically historical accounting information, has attracted considerable
Summary academic endeavour. These empirical studies draw from the stylized fact that has
unequivocally emerged from literature on performance of mergers and acquisitions:
that target firms gain abnormal returns when a takeover announcement is made.
Hence, it has been hypothesized that early prediction of takeover targets can stimu-
late strategic trading that can consistently ‘beat the market’, and make abnormal
returns. While it has now been generally proven that such a strategy cannot succeed
within semi-strong efficient markets, attempts continue to construct such prediction
models to identify potentially valuable firms that can at least provide higher returns
under a new management with synergistic propositions. Besides, the characteristics
identified by a robust model are also used for preliminary exploration for a poten-
tially good target by acquirers.
Following this strand of literature, this paper builds a prediction model for acquisi-
tion targets in India using logistic regression. For the estimation of the logistic re-
gression, 122 target firms of acquisitions during the three year-period from 2002 to
2005 were considered, and matched with non-acquiring, non-target firms that had
similar promoters’ holdings and belonged to the same industry as the target. Results
from logistic regression indicate that, a typical target is inherently strong with high
growth and large free cash flow, in spite of high debt levels, but encumbered by an
inefficient management, who are probably disciplined by takeover market. Tradi-
tional determinants of US and UK studies, viz., size and growth-resource imbalance
are not significant in the Indian context.
Methodological care was taken at various steps to avoid known biases. Estimation
period was taken for a modest three year period rather than a longer period to en-
KEY WORDS sure minimal changes in the macro-economic landscape that might have a bearing
on the target characteristics. Further, both raw accounting ratios, and industry ad-
Acquisition justed ratios were used to account for non-normality of such data. To build the pre-
Prediction diction model, cut-off values were calculated using two methods, one that minimized
statistical errors and another that maximized returns; again, the latter was found to
Promoter Holding
be superior. Finally, the prediction model was tested on an out-of-sample database
Industry Relative Ratio of acquisitions that took place during 2005-2006 and was found to yield prediction
Cut-off Calculation accuracies up to 91 per cent.
3 Brar, Giamouridis and Liodakis (2009) could predict portfolios that 4 We are grateful to an anonymous referee for suggesting the inclusion
earn significant returns not explained by conventional risk factors. of this variable.
Choice of targets: All targets in the estimation and pre- In addition, the object of this study was to eliminate
diction sets were obtained from three sources: the M&A friendly acquisitions. Since in India there is no specific
module of CMIE database, the website6 of SEBI7 where database that mentions whether an acquisition is friendly
all open offers are listed, and the Indian Business In- or hostile, it was decided that, it would be logical to ig-
sight Database8. In each of these databases, only those nore deals where acquisitions of shares have taken place
deals were chosen where through preferential allotment. Hence, only non-pref-
erential allotment of shares was considered for the pur-
• initial stake of the acquirer was less than 15 per cent
poses of this study. Finally, following Bartley and
• targeted acquisition of shares was greater than 20 per
Boardman (1990), any takeover that had been attempted
cent was included in the sample, irrespective of the final out-
• mode of acquisition is through non-preferential al- come.
lotment of shares.
Choice of non-targets: In this study, non-targets were
The initial stake of 15 per cent is chosen because of the
chosen by matching. Matched samples are justified when
provisions of Takeover Regulations of SEBI. Regulation the matching factor is likely to have a large effect on the
12 deems that, in order to gain control over a target com- likelihood of takeover, but itself is not of primary re-
pany, all acquirers have to make a public announcement search interest. Thus, the matched sample design helps
to acquire shares, where various disclosures are man- to focus on variables of primary interest while applying
dated, including the object and purpose of acquisition control over the ‘nuisance’ variables (Cram, Stuart and
of shares. However, Regulation 10 allows a prospective
Karan, 2007). In earlier literature, this matching was done
acquirer to build up a holding of up to 15 per cent with-
with respect to one of the variables of primary interest,
out making any public announcement to acquire shares.
like size, or market capitalization, thus reducing the in-
What emerges from these Regulations is that, up to ac-
formation content of the data. In this study, the vari-
quisitions of 15 per cent of the voting capital, it is diffi- ables to be matched were taken as percentage share-
cult to differentiate whether the intention behind buying holding of promoters in the company, and the industry.
an initial holding is takeover, or simply investment. But It is worthwhile to mention here that, the first criterion
initial holdings of more than 15 per cent clearly signal that for matching is unprecedented in the literature. There
the firm has already been identified as a target, and the were two reasons behind the choice of using promoters’
intention is to gain substantial control of the firm. Hence
holdings as the matching variable in this study – one
the choice of maximum initial holding of 15 per cent.
was derived from economic theory that showed that it
At any point of initial shareholding, the acquirer has two was a critical factor in takeovers and the other was to
options: one, to make a substantial acquisition along with focus on variables representing important hypotheses
a public announcement, or two, to make a creeping ac- explaining takeovers without losing information.
quisition. Regulation 11 allows creeping acquisition of
Economic theory suggests that, irrespective of the finan-
5 per cent every year up to 55 per cent holding without
cial characteristics of firms, it is the shareholding pat-
invoking the requirement of public announcement to tern that defines takeover dynamics. In an influential
acquire shares. On the other hand, if the acquirer wants article, Grossman & Hart (1980) argue that, no raider
to make a substantial acquisition, he has to make public will ever find it profitable to take over a company with
offer to the shareholders of the target company for a completely dispersed shareholding. Shleifer & Vishny
minimum twenty per cent of the voting capital of the (1986) show that the presence of a large shareholder can
5
actually facilitate takeovers. Subsequent theoretical de-
Centre for Monitoring Indian Economy
6 velopment (Baron, 1983; Giammarino & Heinkel, 1986;
http://www.sebi.gov.in/Index.jsp?contentDisp=DataTakeOver
7 Securities and Exchange Board of India Khanna, (1997) justify resistance and delaying tactics,
8 ibid.informindia.co.in to increase the gains to the target shareholders, at the
dustry values. As pointed out by Cudd and Duggal turing those industry-specific dispersions, they also ob-
(2000), this approach incorrectly assumed that, the stand- tained significant additional information from industry
ard deviations of the industry values were same. By cap- relative ratios.
Maximization of Returns Method: Following Powell As has been suggested in the literature, these cut-off
(2001), the logit models are used to compute predicted probabilities are indeed greater than those obtained from
probabilities of all firms in estimation sample. These Palepu’s (1986) method.
firms are then sorted in ascending order of probabili-
ties. Ten portfolios are constructed by dividing this Prediction Results
ranked list equally. The portfolios are then examined The above cut-offs are now used in prediction tests.
for target concentration, which is computed as number Acquisition probabilities of all firms in the prediction
of targets in each portfolio divided by the number of set are calculated using estimates of Models 2 and 4.
firms in that portfolio. The portfolio for which the tar- Based on the relevant cut-offs, the firms are classified as
get concentration is highest is then the critical portfolio. target and non-target. Table 6 presents the summary of
The first acquisition probability of this portfolio becomes the prediction tests for both models, including the pre-
the cut-off probability. diction accuracy.
Model 2, which is based on raw financial data, produces still too high to provide any scope of abnormal returns.
a prediction accuracy of 62.59 per cent when the cut-off
from Minimization of Errors method is applied. On the Another point worth observing here is that, when accu-
other hand, when the cut-off from Maximization of Re- racies within models are compared, irrespective of the
turns method is used, Model 2 yields a significantly method for cut-off being employed, Model 4, with in-
higher prediction accuracy of 77.91 per cent11. The rea- dustry weighted financial ratios performed significantly8
son is that, the criteria for cut-off is higher in the latter better than Model 2, which employed raw financial data.
method and hence, the number of targets correctly pre-
CONCLUSION AND IMPLICATIONS
dicted as targets reduced marginally whereas, the
number of non-targets correctly predicted as non-tar- The objective of the study is to build a takeover likeli-
gets increases substantially. The portfolio size also re- hood model in the Indian context employing logistic
duces by a large amount. regression on nine variables that represent nine theo-
ries commonly cited as takeover strategies. The method
Similar observations may be made for Model 4. With used in this study to select non-targets is by matching
Maximization of Returns method, the cut-off was 0.69 targets, rather than randomly selecting non-targets.
vis-à-vis a cut-off of 0.482 with Minimization of Errors However, unlike previous literature, where matching is
method. Again, the number of targets correctly predicted done with respect to one of the hypothesized variables,
was lesser in the former method by only 10 firms, while leading to loss of information, in this study, matching is
the number of correctly predicted non-targets increased done with respect to promoters’ holdings in the firm.
by 483 firms. Hence the net increase in correct predic- Irrespective of the attractiveness of a firm, promoters’
tion increased by 483-10=473 firms! This resulted in a holdings have been shown to affect actual takeover dy-
total prediction accuracy of 90.79 per cent. namics in the literature. By selecting non-target firms
Thus, it can be concluded that, as claimed by Powell with similar ownership structures as the targets, the ef-
(2001), the prediction accuracy resulting from fect of such an important factor is isolated, without re-
Maximization of Returns method is driven by its ability ducing the information content of the model.
to peg the cut-off at higher levels so that, the number of Following previous literature, two types of data are used
non-targets in the target portfolio reduces drastically, – raw financial data of each firm, and the same data
thereby reducing misclassification error. However, the scaled by industry averages. It is found that, the explana-
per cent of misclassified non-targets in the portfolio is tory ability of industry scaled data is much higher than
that of raw data.
11 Significance is tested by applying the one tailed test for differences
between proportions. The model that results from the logit analysis using in-
So, here, p1 = 0.6259, p2 = 0.7791, n1 = n2 = n = 6429
H0 : p1 = p2; H1: p1 < p2 dustry weighted data give credence to the hypothesis
Then, overall proportion of success, p = ½ (p1 + p2) = 0.7025 that, in the Indian context, the typical target is one which
Also, estimated standard error of the difference between two propor-
tions = σ p1-p2 = √(2 p q / n) = 0.0081 has high growth potential along with high levels of free
z1-2 = (p1 - p2) / σ p1-p2 = -19.00 ⇒ evidence to accept the alternate cash flow in spite of high leverage, that is, an inherently
hypothesis
APPENDIX
47
REFERENCES
Akhigbe, A., and Madura, J. (1999). The industry effects re- rate operating performance. downloaded from http://
garding the probability of takeovers. Financial Review, www2.bc.edu/~tehranih/TehranianInstOwner10-06-
34(2), 1-18. 05.pdf
Ambrose, B. W., and Megginson, W. L. (1992). The role of as- Cram, D. P., Stuart, I., and Karan, V. (2007). Review of choice
set structure, ownership structure, and takeover defenses based and matched sample studies in auditing research.
in determining acquisition likelihood. Journal of Financial available at http://www.nhh.no/for/seminars/accounting-
& Quantitative Analysis, 27(4), 575-590. management-science/2007-spring/140307.pdf
Barber, B., Lyon, J., and Tsai, C. (1999). Improved methods for Cudd, M., and Duggal, R. (2000). Industry distributional char-
tests of long run abnormal stock returns. Journal of Finance, acteristics of financial ratios: An acquisition theory ap-
54(2), 165-201. plication. Financial Review, 35(1), 105-119.
Barnes, P. (1990). The prediction of takeover targets in the U.K. Dencic-Mihajlov, K., and Radovik, O. (2006). Problems in pre-
by means of multiple discriminant analysis. Journal of dicting target firms at the undeveloped capital markets.
Business Finance & Accounting, 17(1), 73-84. Economics and Organisation, 3(1), 59-68.
Barnes, P. (1998). Can takeover targets be identified by statis- Dietrich, J. R. (1984). Discussion of methodological issues re-
tical techniques? Some UK evidence. Journal of the Royal lated to the estimation of financial distress prediction
Statistical Society: Series D (The Statistician), 47(4), 573-591. models. Journal of Accounting Research, 22(1), 83-86.
Barnes, P. (2000). The identification of U.K. takeover targets Elyasiani, E., and Jia, J. (2010). Distribution of institutional
using published historical cost accounting data. Some ownership and corporate firm performance. Journal of
empirical evidence comparing logit with linear discrimi- Banking & Finance, 34(3), 606-620.
nant analysis and raw financial ratios with industry rela- Giammarino, R.M., and Heinkel, R.L. (1986). A model of dy-
tive ratios. International Review of Financial Analysis, 9(2), namic takeover behavior. Journal of Finance, 41(2), 465-
147-162. 480.
Baron, D. (1983). Tender offers and management resistance. Grossman, S., and Hart, O. (1980). Takeover bids, the free-rider
Journal of Finance, 38(2), 331-343. problem, and the theory of the corporation. Bell Journal of
Bartley, J. W., and Boardman, C. M. (1990). The relevance of Economics 11(1), 42-64.
inflation adjusted accounting data to the prediction of Jandik, T., and Makija, A. K. (2005). Leverage and the com-
corporate takeovers. Journal of Business Finance & Account- plexity of takeovers. The Financial Review, 40(1), 95-112.
ing, 17(1), 53-72.
Jennings, R.H., and Mazzeo, M. A. 1(993). Competing bids,
Bebchuk, L. A. (1999). A rent-protection theory of corporate target management resistance, and the structure of take-
ownership and control. Harvard Law and Economics over bids. Review of Financial Studies, 6(4), 883-909.
Discussion Paper No. 260, available from http://papers.
ssrn.com/sol3/papers.cfm?abstract_id=168990 Jensen, M.C., (1987). The free cash flow theory of takeovers: A
financial perspective on mergers and acquisitions and the
Belkaoui, A. (1978). Financial ratios as predictors of canadian economy. available from http://papers.ssrn.com/AB-
takeovers. Journal of Business Finance & Accounting, 5(1), STRACT=350422
93-108.
Jensen, M.C., and Ruback, R.S. (1983). The market for corpo-
Betton, S., and Eckbo, B.E. (2000). Toeholds, bid jumps, and rate control: The scientific evidence. Journal of Financial
expected payoffs in takeovers. Review of Financial Studies, Economics, 11(1), 5-50.
13(4), 841-882.
Kamath, Radhika (2006). Changes in promoter holdings:
Brar, G., Giamourdis, D., and Liodakis, M. (2009). Predicting What’s at stake for the shareholder? Business Line, Octo-
European takeover targets. European Financial Manage- ber 15.
ment, 15(2), 430-450.
Khanna, N. (1997). Optimal bidding for tender offers. Journal
Bushee, B. J., Carter, M. E., and Gerakos, J. (2009). Institutional of Financial Research, 20(3), 323-342.
investor preferences for corporate governance mecha-
nisms. available from http://papers.ssrn.com/sol3/ Kumar, R., and Rajib, Prabina (2007). Characteristics of merg-
papers.cfm?abstract_id=1070168&download=yes ing firms in India: An Empirical Examination. Vikalpa,
32(1), 27-44.
Cheh, J., Weinberg, R., and Yook, K. (1999). An application of
an artificial neural network investment system to predict Lyons, P. J., and Persek, S. C. (1991). Integrating neural net-
takeover targets. The Journal of Applied Business Research, works and expert systems for merger and acquisition
15(4), 33-45. analysis. IEEE Proceedings of the First International Confer-
ence on Artificial Intelligence on Wall Street, 200-205, avail-
Chen, C. and. Su, R. (1997). Do cross-border acquisitions of able from http://ieeexplore.ieee.org/xpl/tocresult.jsp?
U.S. targets differ from U.S. domestic takeover targets? isnumber=6074&isYear=1991&count=52&page=1&Result
Global Finance Journal, 8(1), 71-82. Start=25
Cornett, M. M., Marcus, A. J., Saunders, A., and Tehranian, H. McConnell, J. and Servaes, H. (1990). Additional evidence on
(2005). The impact of institutional ownership on corpo- equity ownership and corporate value. Journal of Finan-
Panigrahi, P. K. (2004). An alternative predicting model for Walter, R. M. (1994). The usefulness of current cost informa-
corporate mergers and acquisitions. Vilakshan – XIMB tion for identifying takeover targets and earning above-
Journal of Management, 1(1), 12-21. average stock returns. Journal of Accounting, Auditing &
Finance, 9(2), 349-377.
Powell, R. G. (1997). Modelling takeover likelihood. Journal of
Business Finance & Accounting 24(7/8), 1009-1030. Wansley, J. W.; and Fayez, E. (1986). Determinants of Gains to
Security Holders from Mergers. Financial Review, 21(3),
Powell, R. G. (2001). Takeover prediction and portfolio per- 91-92.
formance: A note. Journal of Business Finance & Account-
ing 28(7/8), 993-1011. Wansley, J., W. R. Lane, H. Yang, (1983), Abnormal Returns to
Acquired Firms by Type of Acquisition and Method of
Powell, R. G. (2004). Takeover prediction models and portfo- Payment. Financial Management, 12(3), 201 – 218.
lio strategies: A multinomial approach. Multinational Fi-
nance Journal, 8(1/2), 35-72. Wansley, J,. W. Roenfeldt, R. L., and Cooley, P. L. (1983). Ab-
normal returns from merger profiles. Journal of Financial
Platt, H. D., and Platt, M. D. (1990). Development of a class of & Quantitative Analysis, 18(2), 149-162.
Parama Barai is an Assistant Professor at the Vinod Gupta Pitabas Mohanty is a Professor of Finance at XLRI School of
School of Management, Indian Institute of Technology, Business and Human Resources. He is a Fellow of IIM Banga-
Kharagpur. She is a Fellow from XLRI School of Business and lore. He is also a cost accountant and CFA. He was a Fulbright
Human Resources, Jamshedpur, and has more than six years Visiting Scholar at Stern School of Business, New York Uni-
of industry experience. Her research interests include Merg- versity in 2009-10. He has more than 15 years of teaching ex-
ers and Acquisitions, Corporate Governance and Valuation. perience in Corporate Finance, Investments, Company
She has published her work in national and international jour- Valuation and Financial Modeling. His research interests in-
nals, and also participated in a number of conferences. clude M&A, Market Efficiency and Asset Pricing.
e-mail: parama@vgsom.iitkgp.ernet.in e-mail: pitabasm@xlri.ac.in