Reason for misallocation of corporate billions may be as simple as trick-or-treating
September 1, 2008
For more information, contact: Benjamin Haimowitz, HHaimowitz@aol.com
A continuing puzzle about corporate management is the well-documented lack of logic in the way multi-unit companies allocate capital among divisions, so that the companies commonly overinvest in relatively weak enterprises at the expense of more promising ones.
Among explanations that have been offered for this anomaly, one is that managers of weak divisions have more incentive than managers of stronger units to lobby top management for support. Another is that top management lacks full and accurate data about investment opportunities of the divisions, and managers of underperformers have a strong incentive to overstate these opportunities.
But new research presented at the annual meeting of the Academy of Management suggests that the answer may be simpler -- almost as simple as children's trick-or-treating on Halloween.
"Executives and teams of executives who make allocation decisions are susceptible to an unconscious bias toward even allocation," write the study's authors, David Bardolet of Bocconi University, Craig R. Fox of the UCLA Anderson School of Management, and Daniel Lovallo of the University of Sydney and UC Berkeley.
In other words, resources in multi-unit firms are, to a significant extent, divided evenly among units, leading the study's authors to observe that such "naive diversification'" has been observed "in numerous studies of judgment in the behavioral decision-making literature."
One such study, for example, found naive diversification in the way employees allocate their money among types of investments in defined-contribution retirement savings plans. Another study found evidence of the same phenomenon among Halloween trick-or-treaters. Some children approached two adjacent houses sequentially on Halloween and were offered a choice in each one between Three Musketeers and Milky Way candy bars. Others approached a single house where they were asked to choose whichever two candy bars they liked, large piles of both brands being in evidence so that the children would not think it rude to take two of the same. Yet, every child in this group selected one of each candy, while only half the children in the two-house group did so.
Is a similar mechanism at work in the allocation of many billions of dollars among units of large corporations? To test whether it might be, the three professors employed two research approaches -- 1) an analysis of actual corporate data and 2) a series of behavioral experiments involving executives.
In the first, the authors investigated the allotment of capital in a large sample of single-business firms and in comparable units of multi-business companies. For the sake of analysis, they assembled the single-unit firms into virtual multi-unit companies similar to the actual multi-unit companies. As the professors put it, "The virtual sample matched the major characteristics of the real sample except that it lacked a layer of corporate management allocating capital over multiple business units."
The researchers found that "capital expenditures on the focal business unit decrease as the number of business units in the rest of the firm increases, holding firm characteristics such as size constant...Not surprisingly, this result disappears for virtual firms, consistent with the notion that diversification bias requires the hand of management."
They add that "these patterns are consistent with a tendency of multi-business firms to naively diversify their assets over all business units (i.e., a bias to allocated 1/n of the capital to each of n units)."
In one behavioral experiment, 64 participants in an executive MBA program were asked to act as the top manager in charge of capital allocation in a hypothetical international consumer-products company. Half the executives were asked to allocate funds directly to three main divisions (home care, beauty care and health care); the other half were asked to allocate funds directly among six geographical subdivisions -- three home-care divisions, in the U.S., Europe, and Latin America; two beauty-care divisions, in the U.S. and Europe; and a single health-care subdivision, in the U.S. A set of instructions for all participants provided basic financial details about the various units, including estimates for future returns of each division and subdivision.
As predicted, allocations in the group that was asked to allot funds to the three main divisions were biased toward one third for each division, while those in the second group were biased toward one sixth for each subdivision. For example, the first group allocated an average 33% to the health-care unit, reflecting the fact that it was one of three company divisions, while the second group allotted just 19% for this very same unit, reflecting the fact that it was just one of six company subdivisions.
In another experiment, involving other participants, 18 executives were presented with a chart consisting of geographic divisions and product subdivisions, while 19 execs were presented with an organizational chart of the same firm with product divisions and geographic subdivisions.
Having viewed one organizational chart or the other, participants were instructed on one page to indicate the amount they would allocate to each division and on the next page were asked what percentage of that amount they would allot to each subdivision.
Again, allocations between the groups showed differences that were strongly suggestive of naive diversification: for example, the group with geographic divisions allocated an average of 40% to home care Latin America, reflecting the fact that Latin America was one of three geographic divisions and home care was the only unit there (1/3 X 1); in contrast, the group with product divisions allotted home care Latin America a mean of only 17%, reflecting the fact that home care was one of three company divisions and that, in turn, it was subdivided into three geographic units (1/3 x 1/3).
Wondering if their experimental results would be different if participants were held accountable for their decisions, the researchers replicated the first experiment with 144 executives in an MBA program, half of whom were told that their responses would remain confidential and half of whom were advised that they might be selected to explain and justify their choices in front of the class. As the professors put it, "Participants in the 'high accountability' condition made allocations that were virtually indistinguishable from participants in the 'low accountability' condition."
A reason for this may be that participants in the experiments were scarcely aware that naive diversification was at work in their decisions. After both experiments, participants were asked to provide brief explanations of their allotments. About half cited perceived growth potential, many referring to the estimates of future returns they were given as part of their instructions. In both experiments, however, fewer than 10% of the participants cited the 1/n rule as a factor in their decisions.
Herein, the study's authors believe, rests an opportunity. To the extent that firms become aware of this bias, they "might develop decision analytic tools to help them expunge [it]."
The study, entitled "Naive Diversification and Partition Dependence in Capital Allocations Decisions: Field and Experimental Evidence," was among several thousand research reports at the Academy of Management meeting. Founded in 1936, the Academy is the largest organization in the world devoted to management research and teaching. It has more than 18,000 members in 92 countries, including more than 10,000 in the United States. This year's annual meeting drew more than 9,000 scholars and practitioners to Anaheim, California from August 10th to 13th for sessions on a host of subjects relating to business strategy, corporate organization and investment, the workplace, technology development, and other management-related topics.