Some 30 participants took part in a joint CEPR workshop with HM Treasury on `Discount Rates and Rates of Return in the Public Sector' in London on 14 May. The workshop was organized by Colin Mayer, Co-Director of CEPR's Applied Microeconomics programme, and Michael Spackman, Under-Secretary and Head of the Public Expenditure Economics Group at the Treasury. The workshop was prompted by Spackman's recent Treasury Working Paper, `Discount Rates and Rates of Return in the Public Sector: Economic Issues'¹, which set out the technical background to the present UK regime as announced in April 1989.

Opening the first session, on `Time Preference and Opportunity Cost', David Newbery (Department of Applied Economics, Cambridge, and CEPR) questioned the paper's major conclusion that the rates of social time preference (STP) and social opportunity cost (SOC) of public expenditure were for practical purposes equal. Although he agreed with Spackman's derivation of STP by adding a utility time preference rate to a factor for marginal income's declining utility as consumption rises over time, he considered 6% too high. For annual consumption growth of 1.5-2.5%, an elasticity of marginal utility of 1.5 and a utility time preference rate of 1%, he calculated a discount rate of about 4% for marginal consumption. He suggested that the government use overseas borrowing costs as a cost of capital alternative.
Newbery maintained that the World Bank's practice of applying a 10% real rate of return to projects in developing countries led to serious resource misallocation, particularly in Africa. The UK Overseas Development Administration's appraisal rates based on local costs of capital were better, and any valuation of a project's costs and benefits should also consider their distribution.

Alan Williams (University of York) noted that an STP rate cannot be derived from market rates, since many time preference decisions are not transacted in the market. If STP is lower than the marginal social return from private investment, this should be remedied by measures to increase private investment rather than to restrict public investment.

Maurice Scott (Nuffield College, Oxford) opened the second session on `Required Rates of Return for Public Enterprises' by noting that the present UK regime assumes indexation of capital values to general inflation. He argued that the target of earning an 8% return on `new' but not on `old' investment was an unhelpful complication, and that the government should instead set an average return of 8% on the whole of an agreed base-period capital stock. Annualized real rates of return (RRRs) cannot adequately measure the performance of monopolies, which also require price controls, pressures to improve efficiency and measures of output quality; but RRRs should in principle be sufficient for enterprises in competitive markets. In practice, the public sector applies external financing limits and reviews corporate plans for competing as well as monopoly activities, which also reflects practice in the private sector. Scott agreed with Spackman's conclusion that average returns on private sector investment exceed marginal returns, but he questioned the conclusion that this justified price discrimination by monopoly nationalized industries.

Spackman replied that the Treasury sought to recover more than the cost of capital from the nationalized industries on average, in part because controlling them with an average return would in practice lead them to price different activities to achieve different returns. With an average target equal to the cost of capital, some activities would therefore recover less than this rate, which would be inefficient.

Geoffrey Whittington (Fitzwilliam College, Cambridge, and CEPR) said that goodwill should maintain equity values above physical asset values, but he questioned the robustness of comparing capital costs derived from financial market data with returns on assets derived from data and depreciation assumptions published by the Central Statistical Office (CSO). Graham Houston (National Economic Research Associates, London) commended the paper's innovation in proposing a persistent gap between average returns and the cost of capital that was not competed away by new entrants. Alan Williams noted that individual nationalized industries' financial targets were in practice set on a pragmatic basis bearing little relationship to private sector returns or capital costs. They reflected government pricing policies and targets for productivity growth and quality.
Paul Grout (University of Bristol and CEPR) opened the third session on `Risk' by explaining that specific risks in a large equity portfolio were diversified away in applications of the capital asset pricing model. The return on a particular stock was the risk-free rate plus a factor reflecting its `systematic risk', i.e. the covariance of its returns with the market average. Grout noted that the relevant set of assets for the public sector was not the stock market, but the nation's total wealth; although he questioned Spackman's conclusion that the covariance of public sector costs and benefits with national wealth was too small to justify different systematic risk premiums for different public sector activities. A firm using a single cost of capital would tend to engage in projects with high systematic risk; and many public sector projects have high fixed costs, which may exacerbate any cyclical variability of returns. Grout agreed that using higher discount rates was a bad way to counter `appraisal optimism', especially since many estimating errors involve capital costs. He suggested sensitivity analysis as an alternative.

Spackman explained that he had derived an average equity risk premium of only 2-4% lower than the figures cited in many textbooks by basing it on the geometric mean of long-term past returns (rather than the arithmetic mean of annual returns) and on judgements about investors' expected long-term risk-free rate (rather than Treasury Bill rates, or ex post nominal bond yields). Charles Goodhart (LSE) maintained, however, that this premium was still far too high to be explained plausibly by public aversion to variability of income or wealth. The conclusion that most equity risk premium arose from specific characteristics of the equity market seemed plausible, although the issue remained unproven.

Paul Grout suggested that mean reversion in the stock market might suggest using a hybrid of arithmetic and geometric means of past returns to project future returns, but the chosen rate would remain close to the arithmetic mean. Spackman replied that the arithmetic mean would be suitable for a truly random series, but even weak mean reversion with very wide fluctuations around a smooth underlying trend made the geometric mean the more suitable.

David Newbery stressed that much public project risk associated with optimistic bias derived from adverse selection. Charles Goodhart noted the importance of bankruptcy and insolvency costs in the private sector and suggested greater monitoring of the equivalent risks in the public sector, noting the importance of political risks and risks of technical innovation. Spackman replied that these issues were increasingly being recognized and reflected in public sector appraisal and monitoring, and that technical innovation was especially prone to optimism. On the other hand, the recently introduced monitoring of capital costs for relatively mundane construction projects with conventional technology had so far shown reassuringly little appraisal optimism.

In some final comments on points from the earlier discussion, Spackman agreed with Newbery that an STP rate of 6% seemed high, although within the plausible range, while 6% was plausible for the SOC, which was estimated in the range of 4-6% (for a risk- free rate of 3-4%). Such a figure could be justified by the desire to avoid downward bias in costing government activities. He welcomed Newbery's comment that the distribution of costs or benefits affects their value, and he agreed with Scott that monopoly public bodies should not be given the effective power to tax. Setting average returns above the cost of capital was justified, however, for bodies whose individual charges were not controlled, and this should not be seen as a tax. In closing the workshop, Colin Mayer (City University Business School, London, and CEPR) agreed with Newbery that the use of international interest rates for public sector appraisal had many attractions, and he noted that public and private financing also differ in their distribution of risk, transactions costs and the appropriate asset base for systematic risk.


*Government Economic Service Working Paper No. 113 (Treasury Working Paper No. 58), January 1991, available from Treasury Publishing Unit, Tel: (44 71) 270 4558.