Policy Design

Define the Cause of the Problem

It is surprisingly easy to leap from policy problem to policy solution without testing your assumptions along the way.  This example might help make this point.

Louise Richardson, in What Terrorists Want, points out that "If one believes terrorism to be caused by poverty, dispatching the military to defeat it will not solve the problem.  Conversely, if one believes terrorism to be the work of maniacal religious fanatics, all the best social problems in the world won't reduce the threat."

Chris Yiu says that a good problem definition is specific, measurable, action-oriented, relevant and time-bound (SMART). Think "How do we eliminate carbon from surface transport in London by 2030?", not "I wonder what we should do about cars" or "Let's fix cities".

And - as I also stress elsewhere - you absolutely have to assemble data, other evidence, insights and contrary views before getting anywhere near a conclusion.

Avoid Magical Thinking

Chris Yiu again:-  Faced with a difficult problem, it's easy to hope that if you just read enough and think hard enough the answer will eventually reveal itself. Don't do this. Form a testable hypothesis, break it down, prioritise and work smart to validate or refine it.

And remember that too many policy-makers - in business and in politics - believe in what David Allen Green calls 'the magical thought' that by 'banning' something you actually get rid of it, as if you have pointed a wand and said aloud some spell of prohibition. ... and 'puff' the 'banned' thing disappears.  Spoiler - it doesn't! 

Be aware, too, that large organisations may commit to change without necessarily being able to take their workforce or their customers with them.  It is well worth being familiar with the basics of principal agent theory.

Beware of Complex Systems

After a human tragedy - the Grenfell fire, the Windrush story - there is a rush to judgement.  People are looking for simple causes and for people to hold accountable (or to blame).  Much later, after inquiries, investigations or judicial process, it usually turns out the problem is complex, with multiple causes of failure and no easy allocation of blame.  The Better Government Institute has published a very good discussion paper (Complex Systems and Why They Go Wrong) which looks at the factors which make the design and implementation of government policy complicated and why complex systems need to be well understood and carefully managed.

Don't Try to Control Too Much

If you are managing only one or two inexperienced people, it can make sense to define exactly how they should do a job, and the results you expect.  But you cannot replicate this on larger scales, let alone across a whole country. 

In general, therefore, it is far better for a central authority or head office to define the outcomes of new policies, and provide appropriate resources, but then leave it to local managers, local politicians and front line staff to decide how best to achieve those outcomes given their knowledge of local circumstances. 

In education, for instance,. it has proved much better to set standards to be achieved at particular ages, without prescribing every lesson plan in great detail.

Ditto overseas development aid.

The Need for Good Guidance

It is all too easy to forget that our policies have to be understood and implemented by somewhat reluctant or uninterested colleagues and members of the public.  Here are some thoughts on guidance material and ensuring compliance with new protocols and other rules.

Whether we are designing a public policy, or a policy to be followed by others in our company or organisation, it is vital to remember that very few people read guidance material, however well written. New procedures, laws and regulations therefore need to be easy to understand, and clearly communicated, including via the media, if they are to be generally accepted and easily enforced. The main purpose of detailed guidance material should be to answer questions about the application of the new rules to non-standard situations.

We must therefore design policies that are easily communicated. Furthermore, presentation needs to be considered right at the beginning of the policy process, not near the end, for - as noted above - policies that are hard to understand, and hard to defend, are generally flawed. The time to think about this is before you choose your policy option, not right at the end of the process. Plenty of company bosses have found that their staff can be amazingly resistant to change. So explanatory material, White Papers, consultation documents etc. should make it crystal clear how each proposal will make life better for customers, voters, business or other sectors.

See also my advice on effective communication. Much of it applies equally to the private and other sectors.


Without enforcement, policy instruments are merely wind instruments' - Marie Jacuzzi

It is vital that compliance and enforcement issues are considered before any decisions are made about the scope and nature of any new policies and regulations. The initiative can be well-meant, but it will quickly fall into disrepute if it meets resistance, or is undermined, and then has to be policed in an obtrusive way, or if the compliance costs, or enforcement costs, are out of proportion to its benefit.

Felix Martin summarised one aspect of the problem rather well: 'When trying to direct social behaviour, policy interventions that do not properly account for the innate ability of human beings to understand, adapt to, and ultimately also to game those interventions are doomed to failure.'

Formal regulations should always be transparent, targeted, consistent, and in proportion to the risk, and the regulator must be publicly accountable.

In the case of risk regulation, compliance is often best assured by providing incentives to encourage those causing the risk to change their behaviour. Where possible, therefore, the cost or impact of the policy should fall upon the person or entity causing the risk, not the person suffering it. If that is not possible then any numerical targets (e.g. for local enforcement bodies) should be concerned with reductions in the occurrence in the risk (e.g. fewer outbreaks of food poisoning) rather than increases in enforcement action (e.g. numbers of prosecutions). And don’t make it obligatory for small firms to keep papers for 40 years – and, yes, such a regulation did recently exist.

Watch out, by the way, for the implications for middle class journalists. For instance, when designing policies affecting employees, think carefully about their impact on au pairs. Or when changing education policy, how will it affect Montessori schools? You attack the freedom of the press at your peril!

Finally, make sure that your solution can be implemented by those who have not been immersed in the issue in the way that you have been. Don’t design systems which are subtle, clever or difficult to understand, and don’t plan staff numbers and implementation timetables on the basis that all involved will be geniuses. Also, make sure that your solution is understood by those at whom it is aimed. Research can help, of course. The Business Department, investigating ineffective product warnings, found that many young people could not define the word ‘fatal’. A new warning ‘solvent abuse can kill suddenly’ was substituted.

Making Laws That Work

I strongly recommend David Goddard's book with the above title . Here is an abbreviated extract from one chapter:

Some legislation simply fails to produce the desired result.  The legislation is enacted but the world doesn't change - or doesn't change much.

Resources, Timescales

Follow this link for advice on how to ensure that policies can be delivered on time and within resource constraints by those on the front line.


Martin Stanley

Spotted something wrong?
Please do drop me an email if you spot anything that is out-of-date, or any other errors, typos or faulty links.