Any transaction between two parties which is not to end in mutual frustration requires that each makes clear what is required of the other. In the case of principal and agent, both must agree on the method for judging the agent's performance, such as the type of ‘outcome’ which is relevant to assessing performance. This will require that the outcome is observable by the two parties and capable of being measured on an agreed scale. In the case of the shareholder-manager relationship of a joint-stock enterprise it is usually assumed that the shareholder is interested in the value of his or her shares on the market and therefore in the flow of profit generated by the enterprise. Profit is not easy to measure and will depend upon accounting conventions, but the existence of trusted auditors and the ability to construct a single summary measure of ‘success’ means that the conditions of observability and measurability of the outcome are normally satisfied.
A fundamental difficulty with public enterprise is to construct similar observable and measurable indicators of performance. An instruction to an agent to ‘maximise social welfare’ is to choose an objective which satisfies neither criterion. This is not to say that the agent will not attempt to operate an enterprise in the public interest, but simply that even if such an attempt were honestly made there would be little hope that the other party to the contract would recognise it. In the 1960s, as we saw in Section 8.4.1, an attempt was made to clarify objectives by establishing pricing and investment rules. These rules depended, however, on being able to measure marginal costs and to agree on the future social benefits flowing from investment projects. They were therefore unenforceable. As Littlechild (1979) put it ‘an industry could defend virtually any pricing and investment policy it wished to adopt as being a reasonable interpretation of the rules and consistent with its own view of the future’.
To replace the pricing and investment rules of the 1970s the 1978 White Paper The Nationalised Industries (Cmnd. 7131) introduced financial and productivity targets. The financial target was set individually for each concern, while the investment programme as a whole was intended to achieve a required rate of return (RRR) of 8 per cent from 1989 (originally 5 per cent). How the financial target was to be determined was not specified, but the main concern appeared to be that public enterprises should at least break even.
The imposition of a financial target can be supported on a number of grounds:
It represents a clear and easily measurable criterion of performance.
It improves management morale by clarifying objectives, and avoids the psychological disadvantages which are associated with managing loss-making concerns.
It provides a crude ex post check that, in total, consumer benefits are as great as the costs of production.
Critics point to several less desirable features, however.
In monopoly situations, the meeting of a financial target may be achieved by setting prices which are allocatively inefficient.
An overall financial target is only a very indirect check on productive or technical efficiency, particularly in monopolistic conditions.
The financial target ignores the wider social purposes of the public enterprise unless these are explicitly and separately financed (for example by special payments to keep branch railway lines open).
The determination of financial targets is likely to involve bargaining between government and industry in which strategic behaviour will play a large part.
Recognition that financial targets were not a sufficient means of controlling natural monopolies has led to the development of other indicators of efficiency. These usually concentrate on productive efficiency and take the form of cost targets or standards of service. To be useful as providers of incentives these targets must be fairly straightforward and verifiable, but simplicity runs the risk of distorting productive effort. In particular, any measurement of productive efficiency will take the form of ‘output per unit of input’ and will therefore imply what ‘output’ is considered relevant and the nature of the ‘inputs’ required to produce it. Imposing specific objectives of this type may then distort the activities of an enterprise in unexpected ways.
The Efficiency Target for the Post Office
The Post Office is set an objective of an annual percentage reduction in real unit costs. For the Counters Services this is expressed in terms of cost per ‘basic transaction hour’ (BTH). For each type of transaction this measure of output is the physical number of transactions multiplied by a standard transaction time expressed in hours. Expressing output in transaction hours in this way enables aggregation of very different transactions to occur (e.g. motor taxation, pensions, the sale of stamps, etc.). It has some curious properties, however. A fall in achieved time per transaction, for example, will reduce the level of measured output for a given number of transactions. Thus a particular qualitative attribute of output has a perverse influence. Further, as explained by an MMC report (1988a) ‘All counter services require a significant amount of back office support … and this is not taken into account in BTH times; the extent of this support varies significantly between product lines. It follows that if the mix of transactions varies from year to year, unit cost comparisons based on the BTH can be distorted’ (p. 44).
The next example illustrates more specifically the distortions which may be introduced by singling out particular ‘observable’ factors to monitor.
Veterans Administration and Proprietary Hospitals
Lindsay (1976) considers monitoring by Congress of Veterans Administration Hospitals in the United States. He argues that monitoring will result in the selected observable characteristics of output being overproduced and unobservable characteristics taking a lower priority. This model therefore takes account of the important facts that output can have several dimensions and that quantity is difficult to summarise in a single figure. An interesting implication of Lindsay's analysis is that measured costs per unit will be lower in public enterprises than private ones. His investigation of proprietary and Veterans Administration Hospitals is consistent with the view that managers of public hospitals will have less incentive to incur costs to produce non-observable benefits such as reassurance to patients, closer attention to personal comfort and so forth.
The observability of performance depends not only on the nature of the performance criteria adopted but also on the costs of monitoring and collecting information. The control methods discussed below are concerned with improving the flow of information to the principal concerning the agent's performance.
These were introduced in the UK in the Competition Act 1980 and were to be carried out by the Monopolies and Mergers Commission at the request of ministers. The idea was to concentrate in depth on a particular part of the business of a public enterprise. The MMC report on the Counters Service of the Post Office quoted in Example 8.10 is one of an increasing number of such reports.
Information about the performance of an enterprise can be generated if there are many other examples of similar enterprises which allow comparisons to be drawn. This is sometimes mentioned as a reason for breaking down a single organisation into several smaller ones, even if it is technically a natural monopoly and the constituent parts do not compete directly with one another (for example regional utilities). Where a natural monopoly is considered to be national in scope, international comparisons of operations can convey information about levels of productive performance. Perelman and Pestieau (1989), for example, use data on rail and postal services in 19 countries to estimate technical efficiency and the change in productivity in the years 1981–3. The managers of an enterprise whose output appears low by international standards relative to the inputs it is using will find themselves under pressure to explain the findings, presumably by reference to overlooked ‘quality’ variables or other special considerations.
If monitoring of particular decisions or of a manager's effort is extremely costly, the natural solution to this contractual hazard is to link the manager's remuneration to the ‘ outcome’. We saw in Section 8.5 that in the private sector, profit bonuses and stock options are available.
In the public sector there is no reason in principle why managerial remuneration should not be linked to results. The most straightforward bonus scheme would be related to profit targets. Where a firm faces competition in the product market and is constrained to act broadly as a price-taker (as for example, British Steel), remuneration linked to profits would give a powerful incentive towards productive efficiency. At the same time, it is not clear why such an enterprise should be publicly owned in the first place, since regulated or indeed unregulated private operations would be as likely to produce economic efficiency.
Product market competition has been introduced as a deliberate policy choice in recent years in some areas. In the UK, the deregulation of long-distance coach services and then local bus services before the privatisation of the subsidiaries of the National Bus Company are cases in point. However, in the UK this form of management discipline has been associated with the process of transition to the private sector rather than with the establishment of long-term arrangements in the public enterprise sector.
Written performance contracts between governments and state-owned enterprises have been introduced widely with the encouragement of the World Bank since the mid-1970s. A survey of developing countries in 1995 found 565 contracts in 32 countries. In China, performance contracts (PCs) became a national policy for the State-Owned Enterprise (SOE) sector between 1987 and 1994 with over 100 000 such contracts in operation by the end of the period (see Shirley and Xu, 2000). Given the problems of running SOEs on public interest lines without specific contracts with the managers there is clearly a case for experiment. However, studies of the effects of performance contracts have produced disappointing results. Shirley (2000, p.4) reports on a set of case studies of monopoly public utility SOEs and an econometric study of over 400 SOEs in China. There was no evidence that performance contracts had systematically improved efficiency.
The reasons for this lack of success derive ftom fundamental contracting problems. Clear and measurable objectives must be specified (Section 8.7.1); rewards and penalties must be quantitatively significant and linked to these objectives; and managers must believe that the terms of the contract will be enforced. These conditions were rarely satisfied. Case studies showed contracts with multiple objectives that changed from year to year. They were often unrelated to productive performance, only very weakly linked to managerial bonuses and virtually unenforceable. Where contracts are properly constructed, however, there is evidence that they can work. Shirley and Xu, (2000, p.20) conclude that although PCs did not improve productivity of state enterprises on average in China, the provisions of the contract mattered. ‘PCs can improve productivity when they provide high powered incentives, use targets less vulnerable to information problems (profit orientation), and signal commitment through longer terms – and when they are implemented in a more competitive environment’.
