House of Commons - Regulatory Reform

Select Committee on Regulatory Reform Written Evidence

Memorandum submitted by Professor Rob Baldwin of LSE

The reform agenda contains a number of unresolved tensions between "Better", "Less is More" and "Risk—Based" Approaches—as discussed below.

TENSIONS WITHIN THE BETTER REGULATION MESSAGE

A first message of the better regulation movement is that regulation can be improved by the application of a series of regulatory improvement tools[1]—tools that are conceived of with some consistency by UK, EU and OECD policymakers. Within the vocabulary of the OECD there are seven main regulatory improvement tools:

— Regulatory Impact Assessments (RIAs)

— Consultations and Transparency.

— Reducing Burdens and Red Tape: Administrative and Regulatory Simplification.

— Enforcement Guidelines.

— Alternatives to Traditional Regulation.

— Sunset Provisions.

— Regulatory Policies and Reviews.

It is, however, the RIA that is seen as the centrally important tool of regulatory improvement in the UK, the EU and the OECD.[2]

Since the RIA system was introduced, around a thousand RIAs have been conducted at a current rate of about 160-200 per year.

When the BRE published its consultation on The Tools to Deliver Better Regulation in July 2006 the only such tool to be discussed was the RIA—which was described by the Minister for the Cabinet as "the cornerstone of our approach to better regulation."[3] Little could have been done to make made the primacy of this tool more plain. The problem with the RIA, however, is that the processes that it involves may be at tension with the requirements of "better" regulation—if better regulation is seen as lighter touch control as per the language of "smarter" regulation.[4]

At this point it is necessary to pause in order to recap on the requirements of "smart" regulation.[5] Proponents of "smart" regulation have argued convincingly that designing good regulatory systems demands a central focus on how best to combine different institutions and techniques.[6] Smart regulation thus moves beyond state controls and looks to mixes of control methods as applied not merely by public bodies but by other institutions and actors including trade associations, pressure groups, corporations and even individuals. It advocates deploying those combination of instruments that will be most appropriate in a given setting and designing strategies that mix instruments and institutional actors to optimal effect.[7]

The messages of "better" and "smart" regulation appear at first glance to be consistent. If, however, we investigate the capacity of the current better regulation movement to deliver smart regulation on the ground, we see that the route to delivery is not unproblematic. Here it is useful to measure the better regulation movement against the five core principles for smart regulatory design.[8] These principles can be summarised as follows:

1. Prefer policy mixes incorporating a broader range of instruments and institutions.

2. Prefer less interventionist measures.

4. Empower participants which are in the best position to act as surrogate regulators.

Regarding the first principle, the better regulation movement might be expected to perform well since it repeatedly emphasises the need to consider alternative, more imaginative, ways of regulating.[11] In practice, however, a number of factors may militate against its delivery of imaginative regulation. First, the RIA occupies a central place in "better regulation" but the evidence is that RIA processes are not highly effective in leading policymakers to consider alternative ways of regulating. RIA processes, moreover, are more attuned to measuring the effects of traditional "command" systems of control than "alternative" methods and this may positively discourage the canvassing of more imaginative regulatory strategies—especially those `softer' strategies involving voluntary and incentive- driven controls where predicting effects (and hence calculating costs and benefits) is extremely difficult. Second, smart regulation is about cumulative regulatory effects and the coordination of regulatory systems with widely varying natures. RIA processes, however, are best suited to looking at the costs and benefits associated with a single, given, regulatory proposal rather than combinations of approach. They are most attuned to the "single strategy."[12] Those officials who are charged to carry out RIAs would find it very difficult to calculate the costs and benefits of a simultaneously acting combination of very different regulatory strategies and institutions. It would, for instance, be extremely hard for proponents of a combination of, say, state, corporate and trade association laws, codes and guidelines to predict how all the relevant actors will draft, design and apply their different control strategies. This would make calculations of costs and benefits a matter of heroic guesswork and the ensuing uncertainties would undermine the essential value of the RIA. Smart regulation involves too many variables, estimates and judgements to lend itself to the RIA process.

Bureaucratic incentives should also be borne in mind. An official who is contemplating regulating and knows that the RIA process has to be undertaken will experience little impetus to propose complex combinations of regulatory institutions and strategies with all the attendant predictive and calculative difficulties. Rather than aim for a "smart" form of regulation, he or she will incline towards a simpler regime that can be predicted to pass a RIA. Such bureaucratic incentives may, moreover, militate against the application of a high level of "regulatory craft" as advocated by Malcolm Sparrow.[13] Such craft calls for the placing of problem-solving at the centre of regulatory design.[14] Problems, when identified are, on this view, to be responded to with a variety of strategies in a manner consistent with smart regulation.[15] For an official who faces a potential RIA, however, the incentive to adopt a problem—centred approach maybe weak, not merely because this would require the evaluation of costs and benefits regarding a variety of institutions and strategies, but also because it may demand an unpacking of the way that a host of existing regulatory regimes impinges on a problem and an examination, within the RIA process, of potential ways to reshape and re-deploy those regimes in combination with any new regulations.[16] The political and bureaucratic implications would be daunting—the proponent of the RIA would often be questioning the way that numbers of established regulators go about their jobs in order to evaluate his or her proposed regulation. A far more attractive proposition will be to take any existing controls as givens and consider whether the addition of a new regulation will pass a cost benefit test.[17] The consideration of alternatives is liable, accordingly, to straitjacketed by existing regulatory frameworks. Overall, then, if RIA processes are retained at the centre of better regulation, they may not conduce to smarter regulation as encapsulated by its first principle.

The second principle of smart regulation holds that less interventionist measures should be preferred. Again, it might be anticipated that the `better regulation' approaches will encourage less prescriptive, less coercive modes of influence. The record of the RIA procedures, however, does not demonstrate that they are well attuned to either the measurement, or the consideration, of alternative, less interventionist controls. Smart regulation, moreover, demands that attention is paid to enforcement strategy, not merely the formal design of regulatory laws.[18] It is, after all, how a regulatory power is used on the ground that tends to determine its essential character. Here again, however, the RIA process draws attention away from how regulations are applied in practice. First, it is the case (as was noted above) that very many RIAs do not attend to implementation and enforcement issues at all well[19]. Second, there are structural reasons why RIAs cannot be expected to come to grips with enforcement strategy in a routinely well-informed manner—RIAs tend to focus ex ante on the general design of regulation and it may be impossible to predict how any regulator or set of regulatory bodies will go about deploying the powers that they are to be given in a proposed regulation.

Smart regulation's third principle urges that "responsive" strategies of regulation should be adopted and should be employed across mixes of regulatory strategies—as applied by numbers of different institutions and instruments. Smart regulation thus holds out the possibility of escalating degrees of coercion through the interaction of different but complementary instruments and parties.[20] It is difficult, however, to argue that "better regulation" approaches sit easily alongside this principle. The RIA process, as noted, encounters difficulties in dealing with either questions of enforcement or with "combined" strategies of regulation. These difficulties are likely to be compounded by attempts to evaluate incremental, and coordinated escalations up the three sides of a strategic pyramid that reflect the combined use of quasi-regulatory and corporate self-regulatory as well as state controls.[21]

Smart regulation, fourthly, advocates the empowerment of those participants who are in the best position to act as surrogate regulators. This, accordingly, favours using the influence of quasi-regulators, such as industry associations and pressure groups, where appropriate. Here again the better regulation toolkit might be expected to prompt consideration of alternative regulatory methods but it may fail, for reasons discussed above, to come to grips with "combined" regulation where quasi-regulatory functions have to be evaluated alongside other regimes. This suggests that it is one thing to develop a set of regulatory improvement tools but another to use those tools harmoniously. In practice the application of the toolkit may be hindered by tensions between different tools—as between the RIA tool and the consideration of alternatives. The better regulation toolkit, moreover, may fail to empower quasi-regulators optimally for cultural and political reasons—regulators and politicians may produce "single solution" proposals because, for instance, they see a problem in traditional ways due to their prior commitment to a certain course of control or because important interests distrust non-traditional regulatory methods. It may also be the case that the existence of a toolkit for evaluating regulatory strategies will not necessarily drive the use of surrogate regulators and produce "smarter" proposals because the messages to be gleaned from, say RIA processes may be poorly heeded within the political and administrative systems that drive policies—a matter to be returned to below.

Yet another reason why "better" regulation may be slow to make best use of surrogate regulators and to come to grips with "combined" sets of controls is because empowering quasi-regulators or corporate self-regulatory controls within combined regimes of control may require an incremental approach to regulatory design in which key actors negotiate and adjust the roles of different controlling institutions and influences over behaviour. This kind of regulation—as is envisaged by smart regulatory theory—involves a reflexive, dynamic approach in which regulatory strategies are constantly revised and "tuned" to changes in circumstances, preferences and so on. Such ongoing processes are not amenable to evaluations in a "one-shot," policymaking process. Issues will have to be revisited as different control systems adjust to each other. Those control systems, moreover, will have to deal collectively with changes. This can be portrayed as a process of continuous regulatory coordination and change rather than the operation of a single fixed design. The better regulation toolkit does envisage the use of a variety of regulatory controls but it is difficult to see how ongoing regulatory coordination with all its flexibilities, can be tested in advance by a RIA process as if it is a static single-shot system.

The final principle of smart regulation suggests that opportunities for win-win outcomes should be maximised—so that, for instance, corporations can behave more responsibly and maximise profits at the same time.[22] Win-win outcomes, however, may not always be possible and, in certain circumstances, there are tensions between corporate profit-seeking and some regulatory objectives.[23] In practice, therefore, regulators will need to identify areas and issues that will lend themselves to win-win outcomes if certain stimuli are applied. The targeting of regulatory approaches will, accordingly, be central to success. Regulators of a variety of kinds will have to deploy a wide range of strategies and aim these at different kinds and sizes of enterprise, as well as activities, in order to maximise win-win outcomes. This, again, is exactly the sort of flexible and adaptive regulatory strategy that is extremely difficult to set out and evaluate in advance according to the RIA-centred better regulation toolkit as now encountered.

For the above reasons, it can be contended that better regulation is at tension with the prescriptions of smarter regulation. A further tension within the notion of "better" regulation should, however, also be noted. This issue is prompted when asking the question: "Better for whom?" It is clear from field research that "better" regulation is conceived of differently by different interests—such as by small and by large businesses.[24] The former, for instance, put an especially high premium on regulatory systems that pay attention to their special sensitivities to certain regulatory effects.[25] Potential dangers of RIA processes are that, as with all cost- benefit analyses, they skew calculations in favour of those parties who benefit from the existing distribution of wealth[26] and that they place emphasis on overall costs an benefits rather than distributional issues—and, accordingly, underplay the question of who bears the costs and who enjoys the benefits. What may be better regulation for large enterprises or urban consumers of services may not be better regulation for small enterprises or rural consumers.

TENSIONS BETWEEN "BETTER", "LESS IS MORE" AND "RISK—BASED" APPROACHES

A second set of tensions exists between the messages of the "better", "less is more" and "risk-based" approaches to regulation. The publication of the Less is More[27] report in 2005 marked, as noted above, a change in emphasis within UK government—away from the pursuit of better regulation and towards less regulation through the imposition of lower costs on business. Regulators were charged to use the Standard Cost Model (SCM) to assess the costs that its regulations place on businesses and to devise ways to meet reduction targets—expressed as percentage decreases in those costs.[28] The SCM estimates the costs of completing activities that are driven by regulation on the basis of a few basic cost parameters- so that the cost per activity (or data requirement) is the product of price (wage costs plus overheads/costs for external services) and time (the time taken to complete the activity) and quantity (the population of businesses affected and the frequency of the requirement per year).

The emphasis, it is to be noted, rests on the reduction of administrative costs (ie information providing costs) rather than policy costs (ie the costs of carrying out the required action, such as fitting the guards or filters). Such an approach to burdens reduction, argues the SCM Network, brings the advantages of: showing how changes in information requirements will affect costs; identifying parts of legislation that are particularly burdensome for business; and assisting in the calculation of cumulative cost burdens.[29]

An initial difficulty with the SCM approach is that it presupposes that it is possible to separate out policy costs from administrative costs. A second problem is identifying those costs that are imposed by a regulatory requirement and go beyond those costs that will be incurred by competent management in the ordinary course of business. These are issues that involve contentious assumptions and combine to produce the danger that, if industry is asked to cost administrative burdens (as is the case) they will tend to conflate policy and administrative, as well as regulatory and managerial costs. The effect will be grossly to exaggerate both the costs of informational burdens and the potential gains to be made by removing a regulation.

Not only is the SCM fraught with difficulty but it produces problems for regulators whose activities and proposals stand to be evaluated by means of RIAs. The SCM offers a particular approach to the calculation of costs but there are dangers. The criteria and assumptions employed in the SCM may not always correspond to those that are used when the actual cost burdens are calculated through the RIA process. A dilemma arises—do regulators use "notional" costings as demanded by the SCM or do they operate with the costings approaches that they have developed on their particular ground. If they adopt the former approach, the RIA may prove the poorer for it—especially when the SCM criteria are perceived as in need of adjustment. If they take the latter course, there is likely to emerge a divergence between SCM and "real" costings methods and results.

A further tension between the RIA and the burdens reducing processes may arise out of the Government's desires (a) to reduce quite significantly the burdens of supplying information that regulators impose on them and (b) to ensure that regulators target their enforcement activities more precisely in order to take up less business time.

The problems are, first, that targeting enforcement demands that inspections and other actions are based on intelligence, and second, that, if the obligations of businesses to supply information to regulators are reduced, it is increasingly difficult for regulators to engage in targeting without generating intelligence independently. Such independent generation of data may, of course, prove hugely expensive for regulators—indeed far more expensive for them than for the businesses that they are controlling (who may have the information quite readily to hand).

This is especially likely to prove an issue when, in the wake of the Hampton Review, regulators are expected to target their enforcement actions at those businesses or service providers who pose greatest risks. Risk-based systems are built on risk analyses, which are founded, in turn, on the collection of quantities of good data.

TENSIONS BETWEEN BETTER REGULATION AND THE POLICY PROCESS

A third serious tension lies between the prescriptions of the RIA and the realities of normal governmental policy processes. As a result, a key challenge presented to the proponents of RIAs is that of locating the RIA within the policy process. On this front, a first issue is whether the RIA can be carried out to a high technical standard. The UK holds itself out to be a world leader in carrying out RIAs but it appears that the current fit between the RIA and policy processes does not conduce to the production of high quality RIAs. A succession of reports from the National Audit Office (NAO) and the British Chambers of Commerce(BCC) has revealed a number of weaknesses. The National Audit Office (NAO) looked at 23 "test case" RIAs in 2001[30] and reported on 10 further sample RIAs in 2004.[31] The NAO revealed in 2004 that only half the RIAs examined included "a reasonably clear statement of objectives" and seven out of 10n did not consider any option for regulation other than the one preferred by the department. None of the 10 RIAs considered what would happen in the absence of the regulation. All acknowledged a level of uncertainty about the data that were used for estimates but such uncertainties were not always reflected in the cost and benefit figures used, which presented single point estimates rather than ranges. Only one out of 10 gave the results of sensitivity tests and only three out of the 10 contained quantified estimates of benefits (often no market for benefits existed making quantification difficult). Most RIAs, accordingly, did not offer a quantified comparison of expected costs and benefits. As for considering the likely effects of regulations on the ground, only half of the sample RIAs considered enforcement and sanctioning effects. Most RIAs, moreover, described how the regulation would be monitored but "often in a very brief and vague way" and only four stated that there would be a formal review to evaluate the success of the regulation. The NAO's 2005-06 evaluation of RIAs[32] stated that only two out of twelve RIAs analysed levels of compliance well and the same report produced the finding that: "The purpose of RIAs is not always understood; there is a lack of clarity in the presentation of the analysis ; and persistent weaknesses in the assessments."[33]

The NAO's findings were broadly in line with, though perhaps less critical than, the British Chambers of Commerce (BCC) studies of 2003 and 2004 which looked respectively at 499 and 167 RIAs produced by government in the two periods studied (1998-2002 and 2002-03).[34] The BCC studies noted a series of problems with RIAs and concluded that ministerial statements that benefits justified costs were not in general supported by the evidence in the RIAs. Some departments, indeed, were under-resourced or badly managed for conducting RIAs. On choice of regulatory strategy the BCC found that the option of not regulating was considered in only a minority of cases (11% in 1998-2002 and 23% in 2002-03) and less than half of RIAs (44%) quantified all the options considered. RIAs are supposed to pay acute attention to business (and especially SME) compliance costs but the BCC reported that costs for business were only quantified in 23% of RIAs and a quarter of RIAs did not consider effects on SMEs at all. A substantial minority of RIAs contained little factual data about consequential costs and benefits and "scant attention" was given to "sunset" clauses or to subsequent monitoring or evaluation. Nor was the BCC impressed by new efforts to improve RIAs—it found that the RIA process showed little recent evidence of improvement.[35]

Such criticisms suggest prima facie that the way that RIA processes are accommodated within policy-making procedures is not conducing to technically impressive assessments. Clearly there is room to improve the technical quality of RIAs and such improvements may be necessary if RIAs are to conduce to better regulation. It would be a mistake, however, to assume that technical improvements in RIAs will be sufficient to improve regulation. Those RIAs would still have to be located within in legislative and regulatory policymaking processes in a manner that allows them to influence emergent laws and policies. There are, however, a number of reasons why RIAs tend to prove less influential than might at first be supposed.

Governments, for instance, may be committed to certain regulatory steps and strategies for ideological reasons, or because of manifesto commitments or because a political settlement has been made with various interests. They will, accordingly, not be minded to pay too much attention to RIAs that send contradictory signals. Ministers, for example, tend to be predisposed towards legislative solutions and, if they have promised to legislate in order to address a problem, they will not respond enthusiastically to RIAs that propose non-legislative solutions. The costs and benefits of regulation, moreover, tend to be difficult to quantify[36] and the perceived "softness" of RIAs may reduce their impact on the policy or legislative process. This is liable to be the case especially where costs and benefits can only be calculated on the basis of guesses about the use that various regulatory actors will make of their powers or about the strategies that will be deployed to apply regulatory rules. In 2006 the NAO stated that weaknesses in assessments meant that: "RIAs are only occasionally used to challenge the need for regulation and influence policy decisions"[37]

It is also frequently the case that the full nature of the regulatory proposal is unclear from any given item of legislation (eg a framework Act) because the real substance will follow in secondary legislation. The effect will be that regulation escapes a good deal of parliamentary and RIA scrutiny. It might be responded that the secondary legislation will, in all likelihood, be RIA-tested in its own right at a later date, but this may be no complete answer to the point. The framework regulatory strategy within which that secondary legislation is to operate will in most instances be established by the primary legislation that has "escaped" RIA influence. Many key regulatory issues will already have been decided by the time the "secondary" RIA is carried out.

Another concern is that even when the secondary legislation is RIA-tested, it may be extremely difficult to assess the substance of a regulatory proposal because its nature will still depend on the use that will be made of the delegated powers involved. Additionally, of course, secondary legislation will not be debated in Parliament in the way that primary legislation is[38] and any RIA-based messages are, accordingly, the less likely to influence decisions in the legislature.

A further problem that arises in the legislative process is that amendments of laws and rules may be introduced at a late stage in the progression of legislations and the proposals involved may, for that reason alone, escape RIA attention.

The culture of governmental policymaking may itself prove resistant to the influential use of RIAs. This has been a special concern to the NAO, which found in 2006 that RIAs were often seen by officials as a bureaucratic task rather than being integral to the process of policy-making. The NAO, accordingly, recommended that a number of steps would have to be taken to change that culture—notably:

— It should be made make clear to policymakers that the RIA is necessary and that the level of effort put in to preparing the RIA reflects its importance.

— Policymakers should start impact assessment early and use the RIA to project manage the decision-making process.

— Policymakers should make greater and earlier use of departmental expertise and, as far as possible, embed expertise into policy teams.

Cultural changes, however, are easier pleaded for than achieved. The recommendation that RIAs should be used earlier in the policy process than presently may, for instance, prove more difficult to implement than might be assumed. A real problem in some areas may arise from tension between the politics of a process and the RIA principles. Within the RIA process policymakers are supposed to consider and compare the array of regulatory routes to a policy objective but in the real world, a proposal may be the product of a process of political negotiation. It arises when compromises and concessions have been made between different interests and, as such, it may be the only feasible option politically.

To compare this proposal with an array of alternatives via the RIA procedure may be to compare a live horse with a number of dead non-runners. (Such a comparison is also likely to be seen by relevant policy-makers as an exercise too far.) This is not to say that RIAs have no value—is, however, to point out that there may be strict limits to the extent to which RIA processes can be fully "embedded" within policy processes so that they can influence political decision-making.

RE-THINKING THE PHILOSOPHY—TOWARDS REVIEW

The above discussion raises some worrying points, notably that better regulation, as pursued via RIAs, may not produce "smarter" regulation or conduce to lighter touch regimes of control; that it may be at tension with desires for lower information burdens on business and with the Standard Cost Model approach; and that it may impact on policy and legislative processes far less than many of its proponents might imagine.

The way forward is, as indicated, not to focus simply on carrying out RIAs in a technically superior manner. Nor is it to settle for embedding the RIA within the policy-making process more fully. Such steps would not ensure that RIAs would impact on legislative processes in a satisfactory manner or that they would lead to smarter regulation. This is why the Better Regulation Executive's 2006 report, The Tools to Deliver Better Regulation is excessively narrow in its approach to improving RIAs and their impact. The report's "key" recommendations are staggeringly modest in highlighting the needs to: take steps to make RIAs more transparent; shorten the RIA guidance; and have RIAs signed off by Chief Economists as well as Ministers.

What needs to be done is to rethink the better regulation philosophy as a whole. At present, the centrality of the RIA means that better regulation is sought to be achieved by making predictions about the future effects of regulatory regimes, attaching costs and benefit to these, and attempting to convince policymakers and legislators that they should give weight to what is often perceived of as so much guesswork. (Biases against smarter regulatory methods tend to be omitted from consideration.)

The key change has to involve a shift away from the predictive philosophy and towards one that gives centrality of place to review—a process that involves both evaluation and modification. Assumptions of comprehensive rationality have to give way to incrementalist strategies[39] so that it is accepted that regulatory systems cannot be designed ex ante (on the basis of ever more sophisticated analyses) and left alone—that steps must be taken, first, to measure whether regulatory systems are working as well as they can (and better than alternatives) and, second, to bring about changes in regulatory strategy in order to effect improvements.

It might be argued from within the BRE and OECD that a movement "from design to review" has already been instituted. It is true that the UK Government, like the OECD, has expressed a commitment to regulatory reviews.[40] The ongoing challenge, however, is to give new emphasis to that review and adjustment process so that, within government, it is seen as more important than design by RIA. Embedding a review philosophy also means that regulatory policymakers will have to come to grips with the difficulties of post-implementation evaluation and adjustment.

These problems should not be understated. A first difficulty is that reviews and changes may create regulatory uncertainties and trigger adaptation costs. Some regulatory systems may be adjustable at low cost, but others may be difficult to change because high uncertainty and high adjustment costs are involved. All regulators will, accordingly, have to make context-specific judgements about the trade—off between improvements in regulation and the uncertainty costs involved in regime adjustments.

A second problem is that evaluating regulatory performance is extremely difficult technically, not least because formulations of objectives will be contentious, appropriate benchmarks will be difficult to identify and regulatory objectives are liable to change over time.[41] Regulatory statutes tend to allocate large discretions and mandates that are unclear—this means that little legislative guidance on yardsticks can be assumed.

Nor, thirdly, will a move from design to review allow us completely to escape the difficulties of assessing the "regulatory mixes" that were discussed above in looking at "smart" regulation. In multi-actor, multi-strategy regimes, attributions of responsibility will be far more difficult to make than in simpler regimes. Other evaluative problems will be severe when mixes are involved. The application of performance measures always produces a danger of counter-productive effects[42] but in mixed regulatory regimes such effects may be particularly difficult to assess because they will be experienced by a variety of actors in quite different ways. Similarly, the complexity of smart regimes makes it difficult both to evaluate the case for alternative mixes of controls and to bring about adjustments in numbers of institutions and rule systems. What can be said, however, is that evaluations of and adjustments to regulatory mixes, though never simple, will be far more easy to carry out in an ex post incremental manner rather than through an ex ante design approach.

CONCLUSIONS

With these caveats in mind, it can be reasserted that a movement from a philosophy of design to one of regulatory review can be expected to hold out a far more realistic prospect of better regulation than current approaches that give centrality of place to the predictive RIA. Review processes make advancements towards smarter methods of control more realistic. They give greater emphasis to the measurement of results—and improve accountability and transparency. They offer superior ways to evaluate the quality and reliability of regulatory data and they allow policymaking cultures to be reshaped over time rather than presuppose that they can be overridden in a single operation.

At the level of UK regulation, the case for the suggested change in better regulation philosophy seems strong. It should involve not an abandoning of the RIA process or an ending of attempts to improve RIAs and embed them within policymaking practices. A review philosophy would retain such efforts but would not put all eggs in the RIA, or design, basket. The RIA would be seen as the start of the regulatory assessment process rather than the end and the real locus for shaping regulation would become the review stage.

Do such arguments apply to better regulation at the EU level as much as to domestic government? Arguably they have all the more force since EU regulation usually involves a process of delegation to Member States through the use of Directives. This process makes the difficulties of quantifying the anticipated costs and benefits of a regulatory proposal a degree more difficult than is the case in domestic legislation. Even the simplest EU instruction to institute a command and control regime (ie with no "mix" of regulatory methods) might demand that those charged to undertake a RIA should make heroic assumptions about the sorts of powers and sub powers that might be deployed according to some assumed enforcement regime at Member State level.

The better regulation movement involves commendable efforts to reduce the costs of regulation down to absolutely essential levels. It carries within it, however, the potential to place undue emphasis on evaluation mechanisms that are technically difficult, hard to fit into policy and legislative processes, and liable to lead to counter-productive results. A revised philosophy of better regulation is needed if tools such as the RIA can be used positively.

March 2008