Did Peter Gorringe get it right? reflections on the Williamson diagnosis¹

by

Robin Johnson

(formerly Deputy-Director, Economics Division, Ministry of Agriculture)

The Institute of Policy Studies has performed a public service by publishing some of the papers on public administration that the late Peter Gorringe wrote while an official in the New Zealand Treasury². The papers show a breadth and scope that is unusual for a public servant. They also provide a useful commentary on the public policy issues that engaged policy makers in the late 1980s and early 1990s. The book adds to the earlier book on the economic reforms by Boston, Martin, Pallot and Walsh³.

As is well known the author made considerable use of the new institutional economics in his diagnosis of New Zealand's economic problems of the period. The papers are full of references to Coase, Buchanan, North and Williamson. This literature makes common use of the particular insights of Ronald Coase with regard to transaction costs and property rights and the constitutional and institutional implications that follow when transaction costs are non-zero.

It appears to me that Gorringe applies the transaction cost model (to use this phrase as an overall descriptor) to the policy issues of the day rather than the policy process. This involves taking the constitutional structure as a given and recommending policy change within this paradigm. For example, if one takes his discussion of the State Sector Act 1988 and the Public Finance Act 1989 (pp 161-5), the discussion is limited to the devolution of the management of government departments to chief executives, and away from central agencies; and the need for a system of accountability for the goods and services core government agencies produce. While there is considerable discussion of the agency problem in this and other contexts, there is no reference (in the published papers) to the transaction costs of the government policy process itself.

In previous times this would have been discussed under the rubric of dead-weight losses in a welfare framework. More recent work by Dixit and Horn has brought policy implementation into an institutional framework⁴. Dixit asks what organisational forms will overcome the monopoly of information held by bureaucrats, the opportunistic behaviour of civil servants, the problem of divided principals, and government-owned asset specificity? Horn, independently of Dixit, bases his analysis on bounded rationality, opportunism, agency theory and transaction costs in a parliamentary system. He analyses the transaction costs involved in the political and administrative process including the cost of achieving durability or lasting impact (the latter turns out to be a coordination problem).

I believe these frameworks complement the work done by Gorringe (they appear to have been developed after most of his work was complete). In this article I set out what the transaction cost model can offer in this area and suggest some useful applications in the New Zealand case. I first discuss the interpretation of institutional economics made by Dixit and Horn and then follow with a discussion of some of the implications for public policy analysis.

The Dixit-Horn hypothesis

The role of institutions

The transaction cost model of the political decision process sees politicians and interest groups resolving their differences by adopting policy measures that suit both parties. Politicians favour policies which extend the life of the party[ies] in power while interest groups favour policies which deliver benefits to them now and in the future. Interest groups are powerful precisely because they can influence voting patterns and party finances. Major transaction costs arise from the difficulties and uncertainties of maintaining control of the process over time.

Transaction cost economics derives from the work of Ronald Coase who first posited that when it is costless to transact, the efficient competitive solution of neoclassical economics obtains⁵.

It does so because the competitive structure of efficient markets leads the parties to arrive costlessly at the solution which maximises aggregate income regardless of the institutional arrangements. But with information deficiencies and imperfections in input and product markets further transaction costs are incurred by the firm in the conduct of its business. It is no longer costless to transact and the institutional arrangements have a bearing on the outcome as the base line is no longer the neoclassical solution.

Proponents of the institutional approach therefore focus on alternative governance structures in firms and ask what are the net gains after costs of information gathering and maintaining contracts have been taken into account⁶. The focus is on organisational alternatives rather than the marginal conditions. The marginal conditions are seen as some ideal benchmark which can never be achieved in the real world.

This approach is the basis of Oliver Williamson's analysis of hierarchical organisations and the economics of vertical and horizontal integration⁷. It stresses the role of a number of transaction costs that prevent the pure model of competition working in an imperfect world. In particular, costs of maintaining and policing contractual obligations and asset specificity are the determinants of the final outcome.

Can this reasoning be applied to government policy-making and organisation? North⁸, Horn and Dixit are among those who have applied transaction cost economics to political decision making in the realm of economic policy. There is a new recognition of the role of governance structures in policy making institutions. In turn, these constructs enable policy formation and implementation to be identified separately from the political process of which they are part.

Dixit's approach

Dixit develops an organisational approach following the writings of Buchanan⁹, North¹⁰, and Williamson¹¹. In terms of the political process, the standard normative approach to policy analysis views this whole process as a social-welfare-maximising black box, exactly as the neoclassical theory of production and supply viewed the firm as a profit-maximising black box.....Economists studying business and industrial organisation have long recognised the inadequacy of the neoclassical view of the firm and developed richer paradigms and models based on the concepts of various transaction costs (op cit p.9). Dixit therefore suggests that a better model of the policy forming process should be based on an analysis of the relevant transaction costs.

Dixit argues for the inclusion of the political parameters in the analysis of the policy process. As a crude but effective caricature, one can say that normative policy analysis began by supposing that the policy was made by an omnipotent [having infinite power], omniscient [infinite knowledge], and benevolent dictator. The work on second best removed the omnipotence. That on information removed the omniscience. However, the assumptions of benevolence and dictatorship have remained unaffected by all these improvements in our understanding of the limits on instruments and information....In reality, a policy proposal is merely the beginning of a process that is political at every stage - not merely the process of legislation, but also the implementation, including the choice or formation of an administrative agency and the subsequent operation of this agency (op cit pp 8-9) (italics in original).

Dixit asks what organisational forms will overcome the monopoly of information held by bureaucrats, the opportunistic behaviour of civil servants, the problem of divided principals, and government-owned asset specificity? He identifies transaction costs involved in overcoming the asymmetric distribution of information between parties (signaling and screening costs, costs of monitoring and incentives, auditing costs and costs of misrepresentation), costs involved in managing agents (monitoring, incentives, and contractual obligations), costs of agents responding to multiple principals (coordination of policies, playing off one principal against another), and costs related to asset specificity (irreversible investments and lack of durability).

Dixit takes the view that the election process creates a contract between the politicians (individuals or parties) or administrators (regulatory agencies etc), and the citizens (individuals or interest group organisations). The contract is a promise of a policy (or programme) in return for votes (or contributions). Unlike private contracts, such policy contracts are difficult to enforce¹².

First, political contracts are rarely between two clearly identifiable contractors; they have multiple parties (voters or lobbyists) on at least one side of the relationship. Second, their terms are generally much more vague (unspoken) than those of economic contracts. They leave much room for interpretation, and many loopholes for escape and opportunities to blame third parties or force majeure for failure to deliver. Thus a promise to cut taxes rarely specifies by how much, and can be rescinded when the budget situation turns out to be much worse than expected. In the electoral contract, the citizens have granted the elected representatives certain powers to collect taxes and make decisions on their behalf, the only safeguards in this process remains the free vote and, if this goes, abuse of power is likely to follow.

Horn's approach

Horn puts more emphasis on the excessive cost of coordination in the political process. He spells out some of the details of the exchange politicians and citizens enter into¹³. His interpretation is also driven by bounded rationality, opportunism, agency theory and transaction costs. He introduces some of the specifics of the political process and the accompanying transaction costs, including the cost of durability or lasting impacts. This is a wider definition of transaction costs than described by Dixit above.

Horn's transactions approach examines the relationship among three sets of actors with different roles and motivations. Legislators are the elected representatives who perform both legislative and executive functions. Legislation can only be enacted by an enacting coalition of individual legislators - that is, a group of legislators that is large enough to guarantee the passage of a bill into law. Constitutional differences between countries affect such groupings and the solutions they might seek. Administrators are appointed rather than elected. They answer to the legislature. Horn assumes that they do not bring policy preferences to their work! Constituents conveys the idea of particular groups in society that have a [vested] interest in the policy making process. Constituents enjoy the benefits - or suffer the costs - of legislation, offer support or opposition to legislators, and, ultimately, elect legislators to office. While these stereotypes tend to simplify the complexity of the political/administrative process in any one country, Horn maintains that they are sufficiently general to explain decisions legislators make about the organisation of public administration on a broad front.

The approach focuses on the difficulties political decision makers have in securing continued electoral support on a longer term basis. Legislators are regarded as self-seeking in their use of legislation to increase their net political support and lasting power. Their opportunities are limited by the transaction costs of achieving agreement on their proposals. These costs are the time and effort it takes to reach agreement on legislative refinements and any time and effort that affected private interests subsequently have to devote to participating in implementation and administration; political uncertainty that the legislation will last; uncertainty that the legislation will be administered as intended; and uncertainty about the distribution of private benefits and costs. If such relationships are regarded as the hazards of the exchange between the respective actors, then they are generically similar to Williamson's far-sighted but incomplete contracting.

The elected/political appointees who are most likely to remain in power are those who are most successful in overcoming these transaction problems in the political process, such as those who are best able to reassure their supporters that the benefits of legislation will not be lost to administrators in implementation, or undone by subsequent legislatures. Successful implementation will depend on administrative agents who do not necessarily share the objectives of the enacting coalition and its constituents; these divergences create transaction costs to do with monitoring the performance of such agents and devising a system of rewards and sanctions that improve performance. There is no reference to asset specificity in this version of the model though government administrations are likely to depend on labour specificity in the administrative staff. Williamson¹⁴ notes that asset specificity is negligible in most public sector departments but human assets in many public bureaucracies involve considerable specificity (for example, nontransferable training and social conditioning).

Furthermore, the amount of net electoral support legislators receive from promoting a piece of legislation depends on the net flow of costs and benefits that private interests expect it to generate over time. Implementation bears on this calculus because private interests are sufficiently forward looking to anticipate how decisions on implementation will affect the flow of benefits and costs. Political support will cease unless they are well pleased (over time).

Horn's model supplements Dixit's model and brings back the broader concepts of public choice into main focus. The uncertainties identified by Dixit (information asymmetry, monitoring of agents and opportunism) create costs in the political process that have to be overcome if lasting control of the Treasury benches is to be achieved. The model posits that the endeavours of the legislators/politicians will continuously be directed to their own survival. This will be achieved by paying greater attention to the wishes of their supporters in a longer term framework. One term of office is never sufficient.

Now in a certain sense, these endeavours can be regarded as an optimal arrangement of the nation's affairs. The model posits that effective public administration requires that the transaction costs be minimised in determining and pursuing society's goals. There will be pressure from constituents to find least cost solutions to the problems of the day. There will be limits on the revenue and challenges from alternative expenditure items including government programmes. Debate and bargaining will tend toward least cost solutions.

Discussion

The insight from the transaction cost model of the political market is that it mimicks how policy decisions are reached in government polities rather better than previous constructions. The black box is opened out. Outcomes are based on the balancing of interests and the respective power bases of the participants. The balancing of interests extends to policy implementation and performance. Influential participants have a vested interest in who is to implement a policy and on what terms and will influence the political decision accordingly.

The model recognises that power is unequally distributed and that silent majorities tend to be forgotten. The balancing process assumes the interested participants can anticipate future benefits and costs that affect their welfare and that they will act in this light. The model assumes that self-interest is dominant to the national interest in most circumstances.

The model does not require any deep analysis of political motives but rather sets out a useful framework that explains the process in which economic policy is conceived and executed. It clearly posits that decision making has to be a political process subject to electoral demands and interest group pressure on the one hand, and national interest concerns contained in rational advice from civil servants (who may or may not be disinterested) on the other.

It can be utilised to examine the structure of government bureaucratic organisations. Williamson offers a detailed analysis of the US Department of State¹⁵. He asks: could the organisation of foreign affairs by the Department of State be replaced by a private bureau? Not surprisingly, perhaps, he comes to the conclusion that it cannot and that there are specific tasks for which a government bureau is more suited. These tasks include situations where hazards of probity (lack of honesty) are to be avoided and organisational matters such as dependence on labour asset specificity (social conditioning and reliance).

The model indicates that blackboard exercises involving producer and consumer surpluses are of limited use in policy making. The presence of transaction costs will suggest different solutions to a policy problem depending where they fall. The case for government intervention in the market failure case is not clear-cut if the costs of intervention outweigh the possible benefits. There may be little or no relationship between the costs and benefits estimated by the outside observer and the evaluations that individuals place on alternatives in actual choice situations¹⁶.

Another application is to ask whether the policy advice function in a government bureau could be contracted out? The Canadian department of agriculture, Agriculture Canada, explored this option and sponsored a study of a similar proposal in New Zealand¹⁷.

While the study recognised that alternative sources of advice should be sought, for informational reasons, it came to the conclusion that confidentiality and probity concerns made the function more suitable for a public bureau.

More widely, the international flavour of the analysis overcomes the dominance of US writing on the subject dependent as it is on interpretations of the US constitution^18>. Horn, in particular, while developing these ideas at Harvard, has incorporated the Westminster tradition into his version of the model and broadened its application to include state-owned profit-making organisations as well as government-financed bureaus.

While there is an emphasis on democratic governmental decision making, particularly with regard to maintaining electoral support, the focus on institutional economics broadens the application to non-democratic regimes and offers a fruitful avenue of further research. It also provides an entry point to the analysis of malfeasance and corruption in government administrations¹⁹.

Another application is in the distribution and administration of foreign aid. Many IMF and World Bank programmes are focused on particular national policy programmes or projects which depend on local initiatives and execution. The strengths and weaknesses of national administrations very soon come to the fore when visiting economists seek answers to poor response rates to massive aid packages. Institutional economics offers useful guidance to those seeking reform in E. Europe²⁰.

I have argued elsewhere for a more realistic policy-making framework as a guide to better policy making²¹. National and international advisors in ministries would benefit from a better understanding of the institutional background. University training could benefit from the teaching of more institutional economics and its practical applications. More university people could with advantage work in the civil service. General economists could re-assess their welfare models and voluminous writings in the professional literature and pay greater regard to the conditions under which their policy recommendations will have to operate.

3572 words

References