This post is a very slightly scrubbed up comment on Paul Frijters’ comment on my recent column on regulation review and DIY super.
I have no silver bullets, but I think the whole area is dominated by a kind of category mistake. It has been assumed – by the reg review crowd and by the BCA and by regulation consultants and many commentators – that the problem we have is ‘over regulation’ and thus that the solution is the imposition of constraints on regulation.
The thinking is that if we erect barriers to poor regulation, the result will be better regulation. In a sense this is tautologically true – if you think that reg analysis will leave only cost effective regulation, what’s there not to like? Still regulation is one of the few major activities of government. Why is our scrutiny of regulation only a scrutiny of over-regulation. Why are we not focused not just on lowering its costs but on optimising its benefits?
I call this the Michelangelo theory of regulation after Michelangelo’s suggestion that sculpture involved take a block of marble and removing all the marble that wasn’t supposed to be in the sculpture you’re carving. Voila.
Beyond this problem, and even in the terms RIA is cast, the real questions are
1) if there are incentive issues producing the problems in the first place, does RIA solve them? I’ve argued no.
2) are RIAs a cognitively efficient way of working out how to regulate better?
On this second question there’s a presumption in all this that one can design systems which surveil regulation making from the top of some command structure. This is flattering to those at the top and those who like to pontificate about policy as if nutting it out from the top enables one to impose the right policy and then things jolly well sort themselves out. When the BCA produces work on regulation it hires some consultants and they produce very high level reports on, for instance which states have the ‘best’ reg review systems. Indeed, they produce a ‘scorecard?‘ (pdf). Everyone loves a scorecard. Pity that to produce a good one you’d have to know more than any consultant or any organisation could ever know.
What if it’s not really like that? If you honestly do an RIA and think about it, you rapidly come to realise that they’re completely unsuited to the job they’re supposed to be doing. For instance, Lateral Economics did an RIS of pharmacy regulations. It was a pretty Mickey Mouse affair with about three real questions. How long should the ‘preceptorship’ – internship after graduation – be? Should there be a special private area in pharmacies and what would be the best way to provide for it? What equipment should be mandated? How would you handle those in an RIS? We had no idea. No-one had any idea. The literature had virtually nothing to say on the questions. The ‘answer’ is, I think that the regulatory system should itself operate so as to continually generate data and reflect on those issues so they can be optimised. That’s what we proposed – to the howling indifference of everyone who had been through a year of reg design and consultation and wanted their RIS ticked off. But some mug economist sitting in an office running his ruler over the activities of the regulators like a philosopher king is absurd (or was in this case).
Further, the analysis we actually did on those three questions ran to 60 odd pages. Imagine if you really were doing an analysis of all aspects of a substantial piece of regulation. In addition to being relatively clueless ‘in principal’ advice, the RIS would run to thousands of pages examining each regulatory option chosen. Go read some RIS’s if you think this is drawing a long bow. For instance here’s the Stronger Super RIS. It’s 83 pages long (rtf). It’s got umpteen sections each supposedly focusing on a major aspect of the relevant legislation/regulation. But the way all the complexity is handled – is to boil all the decisions down to a choice between (almost invariably) three different options – as called for in the RIS handbook. Very rarely is that a sensible or reasoned way to come to a decision. Often one or even two of the three options are silly or not really contenders. And guess which option that leaves? Usually it’s a silly way to reason one’s way towards decisions.
I think the metaphor of regulation should be a different one and have written about some of this in this Lateral Economics paper on Regulation and Innovation (pdf). I think we should be thinking of regulation as a complex system and one that requires high quality maintenance and governance right down to the level of the micro-detail. We have some examples of how this is done in business. It’s not done by getting consultants in or by command and control systems in which the managers and the engineers work it all out and then pass it down to the proles to implement.
It’s done by optimising the micro-detail. And that requires that those in the chain of command buy into the mission and have the authority and the tools and the incentives (usually these are social, reputational and promotional rather than financial incentives) to optimise their own effectiveness. The Toyota Production System is emblematic of this kind of thing. Now it would be tautologically true to say that every change in the production line in a Toyota factory should be cost effective. But of course they don’t run cost benefit analyses on every change, or even any of them. They try to measure what they do and create the scope for those on the line to understand what they’re doing and together to optimise it.
Is this kind of thing a pipe-dream in regulation? Well perhaps doing it really well is, but it’s at least an appropriate model of what would be a system that worked. If regulation is important and if the volume of regulation is considerable – which it is in most areas – then minimising unnecessary regulation might be worthwhile, but optimising what’s there (for instance quickly enabling new approaches to emerge where they are worth trying and actually regulating to try to generate results – like safer, higher return superannuation funds, better informed markets etc) should be at least equally important though my guess would be that it would be a whole lot more important. And anyway, even if one were trying to achieve ‘minimum effective regulation’ we know that the current dysfunctional – or perhaps it would be fairer to just say ‘non-functional’ – system doesn’t come close to delivering it. So there’s not a lot to lose.