Economists used to tell governments to fix their policies. Now they tell them to fix their institutions. Their new reform agenda covers a long list of objectives, including reducing corruption, improving the rule of law, increasing the accountability and effectiveness of public institutions, and enhancing the access and voice of citizens. Real and sustainable change is supposedly possible only by transforming the "rules of the game" - the manner in which governments operate and relate to the private sector.

Good governance is, of course, essential insofar as it provides households with greater clarity and investors with greater assurance that they can secure a return on their efforts. Placing emphasis on governance also has the apparent virtue of helping to shift the focus of reform towards inherently desirable objectives.

Traditional recommendations like free trade, competitive exchange rates, and sound fiscal policy are worthwhile only to the extent that they achieve other desirable objectives, such as faster economic growth, lower poverty, and improved equity.

By contrast, the intrinsic importance of the rule of law, transparency, voice, accountability, or effective government is obvious. We might even say that good governance is development itself.

Unfortunately, much of the discussion surrounding governance reforms fails to make a distinction between governance-as-an-end and governance-as-a means. The result is muddled thinking and inappropriate strategies for reform.

Economists and aid agencies would be more useful if they turned their attention to what one might call "governance writ small". This requires moving away from the broad governance agenda and focusing on reforms of specific institutions in order to target binding constraints on growth.

Poor countries suffer from a multitude of growth constraints, and effective reforms address the most binding among them. Poor governance may, in general, be the binding constraint in Zimbabwe and a few other countries, but it was not in China, Vietnam, or Cambodia - countries that are growing rapidly despite poor governance - and it most surely is not in Ethiopia, South Africa, El Salvador, Mexico, or Brazil.

As a rule, broad governance reform is neither necessary nor sufficient for growth. It is not necessary, because what really works in practice is removing successive binding constraints, whether they are supply incentives in agriculture, infrastructure bottlenecks, or high credit costs.

It is not sufficient, because sustaining the fruits of governance reform without accompanying growth is difficult. As desirable as the rule of law and similar reforms may be in the long run and for development in general, they rarely deserve priority as part of a growth strategy.

Governance writ small focuses instead on those institutional arrangements that can best relax the constraints on growth. Suppose, for example, that we identify macroeconomic instability as the binding constraint in a particular economy.

In a previous era, an economic adviser might have recommended specific fiscal and monetary policies - a reduction in fiscal expenditures or a ceiling on credit - geared at restoring macroeconomic balances.

Today, that adviser would supplement these recommendations with others that are much more institutional in nature and fundamentally about governance. So he or she might advocate making the central bank independent in order to reduce political meddling, and changing the framework for managing fiscal policy - setting up fiscal rules, for example, or allowing only an up-or-down legislative vote on budget proposals.

Macroeconomic policy is an area in which economists have done a lot of thinking about institutional prerequisites. The same is true in a few other areas, such as education policy and telecom regulation.

But in other areas, such as trade, employment, or industrial policies, prevailing thinking is either na�ve or non-existent. That is a pity, because economists' understanding of the substantive issues, professional obsession with incentives, and attention to unanticipated consequences give them a natural advantage in designing institutional arrangements to further the objectives in question while minimising behavioural distortions.

Designing appropriate institutional arrangements also requires both local knowledge and creativity. What works in one setting is unlikely to work in another.

While import liberalisation works fine for integrating with the world economy when import-competing interests are not powerful and the currency is unlikely to become overvalued, export subsidies or special economic zones will work far better in other circumstances.

Similarly, central bank independence may be a great idea when monetary instability is the binding constraint, but it will backfire where the real challenge is poor competitiveness.

Unfortunately, the type of institutional reform promoted by, among others, the World Bank, IMF, and the World Trade Organisation is biased towards a best-practice model, which presumes that a set of universally appropriate institutional arrangements can be determined and views convergence towards them as being inherently desirable.

But best-practice institutions are, by definition, non-contextual and cannot take local complications into account. Insofar as they narrow rather than expand the menu of available institutional choices, they serve the cause of good governance badly.

Good governance is good in and of itself. It can also be good for growth when it is targeted at binding constraints.

Too much focus on broad issues, such as rule of law and accountability, runs the risk that policymakers will end up tilting at windmills while overlooking the particular governance challenges most closely linked to economic growth.