1

Causality and the Diversification Discount

1 Introduction
Does corporate diversification, i.e. the expansion of a firm’s business operations into unrelated areas, destroy shareholder value? The wealth effects associated with conglomerates have been controversially discussed in scholarly journals ever since the seminal papers of Lang and Stulz (1994) and Berger and Ofek (1995) suggested that diversification reduces shareholder value. Both find that conglomerates are attributed with a lower market value than a portfolio of comparable focussed firms operating in the same businesses as the conglomerate. This finding seemed to suggest the hypothesis of a “diversification discount”. In line with this Scharfstein and Stein (2000) postulate
"it has become almost axiomatic among researchers in finance and strategy that a policy of corporate diversification is typically value reducing.”
Yet, subsequently financial scholars have challenged this dogma of a diversification discount. They did so with respect to the method used (Mansi and Reeb (2002); Glaser and Müller (2010)) and the causal interference (Graham et al. (2002); Campa and Kedia
(2002); Villalonga (2004)). Taking these latest developments into account, the empirical evidence on the value effects of corporate diversification is mixed.
The controversy that has evolved around these wealth effects provides a suitable setting to investigate the pitfalls associated with causal analysis and interference in empirical financial economics. The aim of this paper is to structure the discussion of the diversification discount with respect to causal interference and to lay out why the traditional methods upon which the hypothesis of a diversification discount is based, do not allow for a robust conclusion on the wealth effects of corporate diversification.

2

2 Casual analysis of diversification and firm value
2.1

Development of the “Diversification Discount”
The analysis of the causal interference from diversification to firm value is closely

linked to the empirical method used. Lang and Stulz (1994) and Berger and Ofek (1995), whose empirical findings shaped the idea of a diversification discount, applied the socalled “surrogate portfolio” approach. Under this approach, an “excess value” ( EV ) is calculated as the difference between the actual market value of the diversified conglomerate (Vc | DIV  1 ) and its hypothetical value given the conglomerate’s I
I

individual segments would be operating as standalone entities (  (Vc i | DIV  0) . i 1

(1)

EV  (Vc | DIV  1)  (Vc | DIV  0)
I

 (Vc | DIV  1)  (  (Vc ,i | DIV  0)

(2)

i 1

EV:

Excess value

Vc :

Market-value of the conglomerate

Vc i :

Market value of the ith segment of the conglomerate

Since (Vc i | DIV  0) is unobservable, for each of the conglomerate’s segments an imputed value of the segment as stand-alone firms is assigned. This value is derived by the use of a multiple valuation using accounting items (AI) such as asset, sales or EBIT multiples (Berger and Ofek (1995)).1
I

 (Vc | DIV  1)  (  ( AI i  (Ind i (

(3)

i 1

AI i :

Ind i (

1

V
)))
AI

Accounting item of the conglomerate’s segment i

V
) : Reference ratio of the market value to the accounting
AI
item for the single segment firms in the industry i

Equation (3) is modelled following the outlines in the appendix of Berger and Ofek (1995).

3

If the excess value is negative, the conglomerate is worth less than the sum of its parts. Using this approach Berger and Ofek (1995) quantify this diversification discount to be approximately 15%. The finding of a lower valuation of multi-segment firms as compared to a portfolio of single-segment firms has been shown to be robust for different time-spans and most geographic regions (Servaes (1996); Lins and Servaes
(1999)). In figure (1) we depict the causal relationship in a directed acyclic graph as suggested by Pearl (2009).
Figure 1: Causal relationship between diversification and firm value as by “diversification discount”

Decision to diversify Act of diversification Multi-segment firm Negative
Excess Value

The graph consists of three nodes that are connected by directed links. The decision to diversify and the subsequent act of diversification lead to firm with more than one segment which in turn results in a negative excess value of the conglomerates.

2.2

Casual interference
The surrogate portfolio method itself and the interpretation of the results have been

criticized in recent research for methodical reasons. In particular causality issues were raised. Here, we will focus on the causal inference made. The early research on corporate diversification in the late 1990s posited the strong proposition that diversification destroys firm value. The empirical evidence, however, only suggests that diversified firms trade at a lower value relative to firms that are focussed. But can we interfere from this evidence that diversification caused the lowering of firm value? This

4

issue has been addressed along two basic lines of research: on the one hand, linear causality is challenged; on the other hand the direction of causality is questioned.

2.2.1

Linear causality and INUS condition

The linear causal relationship in figure (1) proposes that the lower value of conglomerates is a direct result of being diversified. The results of the study by Graham et al. (2002) show that this conclusion is premature. For their subsample of conglomerates that diversified by means of acquisitions, they showed that the firms acquired unrelated targets which were trading at a discount prior to the transaction – possible the reason why these targets were up for sale on the market for corporate control. Graham et al. (2002) further find evidence that the lower value of the conglomerates post-acquisition can be traced back to the addition of these already discounted firm segments. For firms that diversified through internal growth, Graham et al. (2002) do not find any diversification discount, which fosters the hypothesis that the lower excess value is linked to other reasons than only diversification. Figure 2 shows the proposed causal relationship as proposed by Graham et al. (2002). The relationship between diversification and the negative excess value is mediated by other circumstances. 5
Figure 2: Causal relationship between diversification and firm value following the reasoning of Graham et al. (2002)

Decision to diversify Acquisition of discounted target

Act of diversification
(acquisition)

Negative
Excess Value

Multi-segment firm The results of Graham et al. (2002) suggest that diversification fulfils the requirements of an INUS condition (Insufficient, but necessary part of an unnecessary but sufficient condition) first introduced by Mackie (1965). The concept of an INUS condition is associated with the theory of necessity and sufficiency of conditions. Following
Brennan (2011), P is a necessary condition for Q, if the presence of Q necessarily implies the presence of P (Q  P). Likewise, P is a sufficient condition for Q, if the presence of P necessarily implies the presence of Q (P  Q). Mackie (1965) derives the INUS condition by describing a situation in which a short-circuit leads to a house burning down. The short-circuit itself is neither necessary for the house catching fire (something else could have initiated the fire) nor sufficient (the short circuit itself does not imply the occurrence of a fire – e.g. the presence of flammable material is required). Instead, the cause of the fire is a more complex set of individual conditions (short-circuit, presence of flammable material, absence of sprinklers, etc.). Each of these conditions is itself neither necessary nor sufficient, but “an indispensable part of a complex sufficient
(but not necessary) condition of the fire” (Mackie (1965)). Taken together, however, these conditions are sufficient for the fire to start.
An analogous argumentation can be applied to the case of corporate diversification and the valuation discount. The fact of being a diversified firm is neither necessary nor

6

sufficient for the lower valuation of conglomerates. The lower valuation does neither imply the diversification, nor does diversification imply a valuation discount.
Nevertheless, we can argue that a more complex set of conditions exists that has a causal relationship to the valuation discount. Following Graham et al. (2002) for example, this set of conditions may consist of the decision to diversify, diversification by acquisition and the acquisition of an unrelated and already discounted target.
Hence, being a diversified firm is a non-redundant part of this complex set of conditions, which is itself sufficient, but not necessary. A lower valuation can still stem from other reasons.

2.2.2

Reverse causality

A second argument centres around the question whether the benchmarking of a conglomerate’s value with the value of a portfolio of focussed firms in the same business is valid. Villalonga (2004) argues that “simple average differences in outcomes between treatment [diversification] and control

groups [no diversification] is only an

unbiased estimate of the treatment effect when units are randomly assigned to the treatment.” Hence, similar to the conditions of randomized controlled trials in clinical studies, we may only interfere from the excess value, if the investigated firms have been assigned to the group of multi-segment firms and the control group of singlesegment firms on a random basis. In case of a non-random assignment of the treatment
“diversification”, the estimation of the excess-value is biased due to self-selection.
Villalonga (2004) and Campa and Kedia (2002) both model a firm’s propensity to diversify and find that firms diversify as a response to exogenous changes in the firm’s environment. Among others, the propensity to diversify increases if the firm has a lower than average profitability, low growth opportunities and high dividend payouts.
Likewise, agency conflicts seem to be more prominent in firms that choose to diversify.
Hence, the decision to diversify is endogenous and not random. Once controlling for

7

this self selection in the cross-sectional analysis of the excess value, the diversification discount disappears and sometimes turns into a premium.
Figure (3) graphs this relationship. Specific firm characteristics such as agency conflicts and a weak historical performance reduce the value of the firm. Resulting is an increase in the likelihood that the firm will diversify. Hence diversification as a strategy is the descendant of a lower value and not the ancestor. A reverse causality may explain the diversification discount.
Figure 3: Reverse causal relationship between diversification and firm value

Decision to diversify Diversification

Negative
Excess Value

Firm-specific characteristics We can conclude that the wealth effects of corporate diversification are much more multilayered than proposed by initial research. On the one hand, we do not have evidence that diversification alone causes the lower valuation of conglomerates. On the other hand it remains unclear whether diversification is the cause or a result of the negative excess value. In fact both may be true sometimes and sometimes neither.
While some studies have found that the diversification discount turns into a premium if properly accounting for endogeneity and firm-specific characteristics
(Campa and Kedia (2002); Villalonga (2004)), recent working papers on corporate diversification (Ammann et al. (2011); Hoechle et al. (2011)) advocate that a significant discount exists even after accounting for firm-specific characteristics and self selection.

8

Taken together, we have as much evidence that diversification reduces firm value as we have evidence that it does not if we rely on the surrogate portfolio method.

3 Implications for diversification research
The review shows that problems of causal interference are closely linked to the empirical method used. We argue that the use of the surrogate portfolio approach is inappropriate as it does not allow for a robust conclusion on the cause and effect relationship between diversification and firm value.
Instead, we propagate the use of the event study methodology in lieu of the surrogate portfolio method. Following the theory of the semi-strong market efficiency hypothesis, we expect that any negative shareholder wealth effects caused by diversification will be reflected in the abnormal announcement returns of the acquirer upon the announcement of a diversifying acquisition. The abnormal return is measured as the difference between the actual return of the stock and an expected return, which is estimated based on the acquirers’ risk and return characteristics.
The advantage of this approach over the surrogate portfolio method is that the effect of diversification can be evaluated at the time the decision is made. Thus it is a dynamic evaluation than the rather static one as for the case of a portfolio approach.
Further, in a univariate setting we can assess whether the acquirer gains or looses value in the acquisition by comparing the pre-acquisition market value to the postacquisitions value of the same firm. Hence, the evaluation does not require a reference group of focussing acquisition to make a statement about the wealth effects, which helps to attenuate the selection problems. Additionally, in a multiple regression setting, controlling for other deal and firm characteristics known to influence the announcement returns (size, payment method, organizational form, etc.), we are able to isolate the wealth effects that is triggered by the act of diversification. Last, the

9

application of the event study requires fewer assumptions with respect to methodical questions, as we do not have to rely on “sticky” accounting data.

10

References
Ammann, Manuel, Daniel Hoechle, and Markus Schmid (2011), "Is there Really no
Conglomerate Discount?”, (SSRN).
Berger, Philip G., and Eli Ofek (1995), "Diversification's effect on firm value", Journal of
Financial Economics (37), 39–65.
Brennan, Andrew (2011), "Necessary and Sufficient Conditions", in Edward N. Zalta, ed.: The Stanford Encyclopedia of Philosophy .
Campa, Jose M., and Simi Kedia (2002), "Explaining the diversification discount", The
Journal of Finance (57), 1731–1762.
Glaser, Markus, and Sebastian Müller (2010), "Is the diversification discount caused by the book value bias of debt?", Journal of Banking & Finance (34), 2307–2317.
Graham, John R., Michael L. Lemmon, and Jack G. Wolf (2002), "Does corporate diversification destroy value?", The Journal of Finance (57), 695–720.
Hoechle, Daniel, Markus Schmid, and Ingo Y. D. Walter (2011), "How Much of the
Diversification Discount Can be Explained by Poor Corporate Governance?”, (SSRN).
Lang, Larry H., and Rene M. Stulz (1994), "Tobin's "q", Corporate Diversification, and
Firm Performance", The Journal of Political Economy (102), 1248–1280.
Lins, Karl, and Henri Servaes (1999), "International evidence on the value of corporate diversification", The Journal of Finance (54), 2215–2239.
Mackie, John L. (1965), "Causes and Conditions", American Philosophical Quarterly (2),
245–264.
Mansi, Sattar A., and David M. Reeb (2002), "Corporate diversification", The Journal of
Finance (57), 2167–2184.
Pearl, Judea (2009), "Causality. Models, reasoning, and inference”, (Cambridge Univ.
Press, Cambridge).
Scharfstein, David S., and Jeremy C. Stein (2000), "The dark side of internal capital markets", The Journal of Finance (55), 2537–2564.
Servaes, Henri (1996), "The value of diversification during the conglomerate merger wave", The Journal of Finance (51), 1201–1225.
Villalonga, Belén (2004), "Does Diversification Cause the "Diversification Discount"?",
Financial Management (33), 5–27.