Information management with specific investments and cost-based transfer prices

Abstract

We analyse information flows in a profit-centre organisation with internal trade between two risk-neutral divisions. Prior to production, the divisions make unverifiable investments in intrafirm synergies. After investments are made, the selling division announces a cost-based transfer price which includes a mark-up on variable costs. The buying division then decides what quantity to purchase at that unit cost. From the head office's perspective, the key issues are to influence both, divisional investments and the seller's manipulation of the mark-up. To do so, the head office can fund a pre-decision information system before divisional investments are made. The system produces forward-looking information that can be used to improve the divisions' investment decisions, but which cannot be used in evaluating their performance. Our analytical framework allows us to identify cost and revenue structures for which pre-decision information either supports or destroys intrafirm synergies by motivating or discouraging divisional investments, thereby resulting in an increase in, decrease in or in no impact whatsoever on, firm profit. Among our most interesting findings is the counterintuitive result that pre-decision information can undermine the incentives of risk-neutral agents to invest specifically. Our results add to earlier agency models that found different, albeit equally dysfunctional effects of pre-decision information. Contrary to these studies, our findings are not driven by either risk aversion or rent extraction.

Notes

¹ In what follows, we exclude two-part transfer pricing from our analysis. As will become evident below, such a scheme would be optimal only in the case of one-sided investment by the producer, whereas it would be ambiguous if investments were to be made by both divisions or by the marketer only. Two-tariff transfer pricing is tantamount to a negotiated transfer pricing scheme in which the producer holds all the bargaining power (see Baldenius et al., 1999, p. 81). Since in this case, the two-tariff transfer pricing system is just a special application of Pfeiffer's (2004) study on information and negotiated transfer prices, we omit this scheme in our analysis.

² See, for instance, Edlin and Reichelstein (1995, p. 279) and Baldenius et al. (1999, p. 70), for the same assumption about managerial behaviour.

³ There is ample reason as to why this might be plausible in practice: first, information systems typically provide detailed information to lower levels of the organisation, while only a summary of this information is furnished to top management (Baiman and Sivaramakrishnan, 1991). Secondly, tasks are delegated because managers are more specialised than the head office. Central to specialisation is superior knowledge and expertise, which endows a person with a higher absorptive capacity when it comes to interpret and act on new information. Therefore, the head office may lack the specialisation and the necessary expertise to understand and adequately interpret the signal (Hart and Moore, 1988; Hart, 1995). Thirdly, forward-looking information is typically soft information, because variables and cause-and-effect chains are inherently difficult to specify in an unambiguous manner (Stein, 2002). Suppose, for instance, that the information system of a biotech firm reveals that a new drug component successfully passed all required tests. What does this mean for the division's costs and revenues? Clearly, the division manager possesses superior knowledge regarding these factors and thus is more capable of planning on the division's behalf than is the head office. Since soft information is not verifiable, it cannot be used for performance evaluation. A contract that conditions variables on such soft information would not be enforceable in court (Hart, 1995). Another reason might be that the head office may simply be too busy to re-run the whole planning process once the signal is made available. Besides its inferior absorptive capacity due to the lack of specialisation, a large control span prevents the head office from devoting as much time and effort to re-running a division's planning process as its associate division could. This argument is supported by empirical evidence showing that the head offices of large corporations may supervise up to several hundred organisational units.

⁴ Remember that both states are equally probable, therefore, . Likewise, .

⁵ This follows immediately from equations (5) and (6), i.e. v(·)=(a _θ − c _θ + I)/2b _θ and q(·)=(d _θ + I)/2b _θ.