• Selected
• Entire List

Abstract:
We examine the determinants of the optimal "cost of capital" in residual income for measuring performance in a moral hazard setting with both firm-specific and systematic risks. The recommendation in the literature is that the "cost of capital" should be the riskless interest rate adjusted for systematic risk. We show that, if the manager is risk averse and can personally trade in market securities, and if there is firm-specific risk, the optimal "cost of capital" for performance evaluation is the riskless interest rate adjusted for firm-specific risk.

The manager provides unobservable and personally costly effort. Its inducement creates a need for performance evaluation based on the firm's (residual) income which is affected by both firm-specific and systematic risk. The manager also selects a capital investment that is costly to shareholders and increases the firm's risk. If the manager can personally trade in market securities, he can offset the systematic component of the firm's risk. He bears his efficient share of the market risk and his risk-return preferences for that type of risk are perfectly aligned with those of well-diversified shareholders. However, he bears the firm-specific component of the firm's risk, which is of no direct concern to the well-diversified shareholders.

Therefore, the "cost of capital" levied in residual income by shareholders must mitigate the manager's incentive to under-invest in capital due to its impact on the firm-specific risk in the residual income. If the manager receives no private firm-specific information, the optimal "cost of capital" is lower than the riskless interest rate. On the other hand, if the manager receives perfect firm-specific information after signing the incentive contract but before choosing the capital investment, he will chose first-best investments if he is charged the riskless interest rate. However, in this setting, the optimal "cost of capital" is set higher than the riskless interest rate to induce lower variations in capital investments in order to reduce the risk premium paid to the manager for ex-ante firm-specific information risk. We conclude the paper by examining a setting in which the manager can allocate the capital provided by shareholders among investments in the firm's production technology and in market securities. If the allocation is not contractible, the optimal "cost of capital" is the riskless interest rate so as to avoid arbitrage opportunities.

Cost of capital; Residual income; Performance evaluation

Abstract:
Recently, much of the research into the relation between market values and accounting numbers has used, or at least made reference to, the residual income model (RIM). Two basic types of empirical research have developed. The "historical" type explores the relation between market values and reported accounting numbers, often using the linear dynamics in Ohlson (1995) and Feltham and Ohlson (1995, 1996). The "forecast" type explores the relation between market value and the present value of the book value of equity, a truncated sequence of residual income forecasts, and an estimate of the terminal value at the truncation date. The analysis in this paper integrates these two approaches. We expand the Feltham and Ohlson (1996) model by including one- and two-period-ahead residual income forecasts. This can be viewed as the use of forecasts in the "historical" model to infer "other" information about future revenues from past investments and about future growth opportunities. Alternatively, it can be viewed as the use of historical data (and the assumed dynamics) in the "forecast" to develop forecasts beyond the truncation period. In either case, the result is in a model in which the difference between market value and book value of equity is a function of current residual income, one- and two-period ahead residual income forecasts, current capital investment and start-of-period operating assets. The existence of both persistence in revenues from current and prior investments and growth in future positive NPV investment opportunities leads us to hypothesize a negative coefficient on the one-period ahead residual income forecast and a positive coefficient on the two-period-ahead residual income forecast. Our empirical results strongly support our hypotheses with respect to the forecast coefficients.

Abstract:
We examine a firm's choice of a measurement system designed to serve two distinct objectives; provide forward-looking information about future firm productivity and ex post information about past managerial performance. A firm can have two separate measurements, one for each purpose, or a single measure that simultaneously serves both objectives. In a two-period principal-agent model, we illustrate how implicit incentives can lead firms to prefer a single dual-purpose measure.

accounting measures, incentives, multi-period

Abstract:
The relationship between the informational environment and the cost of equity capital has received considerable interest in finance and accounting research as well as in financial reporting regulation. Recent papers have demonstrated that increased public disclosure may decrease firms' cost of capital, at least if the additional information pertains to systematic risk. The discussion has focused on the impact of information on the cost of capital subsequent to the release of the information (the ex-post cost of capital). We show that the reduction in the ex-post cost of capital is offset by an equal increase in the cost of capital for the period leading up to the release of the information (the preposterior cost of capital). Thus, within the class of models framing the recent discussion, there is no impact on the ex-ante cost of capital covering the full time span of the firm. The extent to which information is made publicly or privately available affects the timing of the resolution of uncertainty and when the information is reflected in equilibrium prices, but there is no impact on initial equilibrium prices.

In efficient economies with only public information, there is no impact of the information system choice on the investors' ex-ante expected utilities either. In the partially revealing rational expectations equilibrium of an economy with private investor information, however, the rational investors may actually benefit from a higher ex-post cost of capital (at the expense of the liquidity traders).

Abstract:
The analysis in this paper extends the single-agent/multi-task LEN model in Feltham and Xie (1994) to a multi-agent/multi-task context. A key feature of the paper is that we consider centralized contracting with both full and limited commitment. The former refers to settings in which the principal specifies the contracts for each agent and is assured that those contracts will be implemented. The latter refers to settings in which the principal cannot preclude inter-agent negotiation to change the terms of their initial contracts. The results from centralized contracting with limited commitment are also obtained from decentralized contracting, in which the principal sets the terms of an aggregate compensation pool and delegates the allocation of the pool to one of the agents.

There are two basic frictions that result in a reduced payoff with limited commitment. One is inefficient risk sharing, which occurs because the agents will choose to share their incentive risk and this results in reduced induced effort. The other is inefficient allocation of effort, which occurs because when the agents allocate the aggregate incentives they ignore the principal's payoff and focus on inducing actions that maximize their compensation. One of the interesting results is that the impact of an agent's risk tolerance and the precision of a performance report on managerial incentives may be in the opposite direction for full and limited commitment.

Incentives, Inter-agent negotiation, Multi-agent contracting, Performance evaluation

Abstract:
The analysis in this paper extends the single-agent/multi-task LEN model in Feltham/Xie (1994) to a multi-agent/multi-task context. A key feature of the paper is that we consider both full- and limited-commitment contracts. The former apply to settings in which the principal can specify the contracts for each agent and is assured that those contracts will be implemented. The latter apply to settings in which the agents can collude to effectively change the terms of their contracts. Equivalently, limited contracting occurs if the principal can set the terms of the aggregate compensation pool, but must let the agents choose how the pool is allocated among the agents. From the principal's perspective, full-commitment contracts dominate limited commitment contracts. There are two basic frictions that result in a reduced payoff with limited commitment. One is inefficient risk sharing which occurs because the agents will choose to share their incentive risk and this results in reduced induced effort. The other is inefficient allocation of effort, which occurs because when the agents allocate the aggregate incentives they ignore the principal's payoff and focus on inducing actions that maximize their compensation.

multiple agents, limited commitment, incentive contracts

Abstract:
We consider a single-principal/multi-agent model to investigate the principal's preferences over delegated contracting. The analysis extends the single-agent/multi-task LEN model in Feltham and Xie (1994) to a multi-agent/multi-task context. We consider full-commitment contracts, i.e., the principal is assured that his contract for each agent is implemented, and analyze the principal's delegation choice to authorize a senior agent to contract with a junior agent based on reports that are only available to the agents. Despite the full-commitment assumption, the agents' incentives to write a decentralized contract are driven by their incentives to share risk, and by their incentives to re-allocate their effort choices, i.e., the two frictions that arise in a limited-commitment setting (Feltham and Hofmann, 2007a). We find that there are benefits and costs to delegated contracting, i.e., establishing a decentralized performance evaluation in addition to a centralized performance evaluation may be detrimental to the principal. However, a decentralized performance evaluation is likely to be beneficial if the senior agent is evaluated based on an aggregate report.

Delegation, Incentives, Multi-agent contracting, Performance evaluation

Abstract:
The analysis in this paper extends the single-agent/multi-task LEN model in Feltham/Xie (1994) to a multi-agent/multi-task context. A key feature of the paper is that we consider the impact of alternative reporting systems on full- and limited-commitment contracts. With a centralized reporting system all performance measures are reported to the principal, while some measures are not reported to the principal with a decentralized reporting system. Under limited commitment, the principal may prefer to restrict the agents' access to certain measures. While additional reports are weakly preferred if there is full commitment, these reports may have a negative effect if there is limited commitment.

Abstract:
This paper provides a general version of the accounting-based valuation model that equates the market value of a firm's equity to book value plus the present value of expected abnormal earnings. Prior theoretical work (e.g., Ohlson 1995; Feltham and Ohlson 1995, 1996) assumes investors are risk neutral and interest rates are nonstochastic and flat. Our more general analysis rests on only two assumptions: no arbitrage in financial markets and clean surplus accounting. These assumptions imply a risk-adjusted formula for the present value of expected abnormal earnings. The risk adjustments consist of certainty-equivalent reductions of expected abnormal earnings. A key issue deals with the capital charge component of abnormal earnings. It is measured by applying the (uncertain) riskless spot interest rate to start-of-period book value. Risks do not affect the rate used in the capital charge, and accounting policies do not affect the formula?s constructs. An application of the general formula shows how the classic risk-adjusted expected cash flows model derives as a special case.

Abstract:
This paper examines a principal-agent model in which the agent receives a sequence of two signals about the future outcome from his actions. Conditions are identified under which sequential communication (signals reported when received) is strictly preferred to simultaneous communication (signals only reported after all are received). If the second signal does not provide additional information about the outcome, then it can only be valuable if its report is verified. If the first signal is informative about the second and the second provides additional information about the outcome, then there exist settings in which sequential unverified reporting is strictly valuable.