Having a Law and Doing Anything

A state can have every legal power on paper and be unable to collect the tax it has legislated, register a business in under a year, keep teachers in classrooms, or enforce a contract in a court.

State capacity is the ability to actually do things, and it is distinct from both the size of government and its formal authority. A small state can be highly capable; a large one can be almost inert. Timothy Besley and Torsten Persson model fiscal capacity and legal capacity as investments that accumulate over decades, which is why capacity cannot be acquired by passing a law; the law is the easy part.

The starkest illustration of what low capacity means in practice is informality.

In many poor countries, more than half the workforce operates outside the tax and regulatory system entirely, and in some it is above 80%. These are not criminals; they are street traders, smallholders, workshop owners, drivers and domestic workers, operating in an economy the state cannot see.

The consequences run through everything. Labour law does not apply to them. Minimum wages do not reach them. Payroll-based social insurance and pensions exclude them by construction. The tax base rests on a narrow formal sector, so tax rates on that sector are high, which increases the incentive to stay informal. And every policy instrument that a rich-country adviser is trained to use assumes a formal economy, so the advice addresses a minority of the labour market.

Raising capacity is therefore partly about bringing activity into view, and the technologies that have done this (mobile money creating a digital transaction record, digital identity systems, electronic invoicing) have arguably done more for tax collection in several countries than any reform of the revenue authority.

Which brings up the most useful concept in this literature, and the most uncomfortable.

Lant Pritchett, Michael Woolcock and Matt Andrews describe what they call isomorphic mimicry: a state adopting the form of a modern institution without its function. An anti-corruption commission is established, procurement law is rewritten to international standard, a performance management framework is introduced, and nothing about what actually happens changes.

The reason the pattern is so persistent is that mimicry pays. Donors and international bodies reward the adoption of recognised forms with funding and legitimacy, and they can observe forms far more easily than functions. A government that adopts the template gets the money. A government that solves the underlying problem in an unfamiliar way may not.

So the incentives run toward the appearance. Their proposed alternative (problem-driven iterative adaptation) starts from a problem that local people have actually identified, tries something, measures whether it worked, and adapts, rather than importing a design that worked elsewhere.

The same distinction explains a case that once seemed settled.

Hernando de Soto argued in The Mystery of Capital that the poor hold enormous wealth in untitled property (houses and land they occupy without formal deeds) which cannot be used as collateral and is therefore dead capital. Give people title, and a wave of borrowing and investment follows.

It is an elegant argument and it drove large titling programmes across Latin America, Africa and Asia. The evaluations have been disappointing. Titling produces modest gains in tenure security and, in several studies, in labour supply, because people no longer need to stay home to defend their property. Effects on credit access are generally weak.

The reason is that banks lend to people who can repay, not to people who hold a deed. A title on a house in an informal settlement with no local land market, where foreclosure would be politically impossible and resale difficult, is not collateral a bank wants. And in some cases titling has enabled poorer households to be bought out, which is the opposite of the intended effect.

One last finding worth carrying, because it reframes a subject everybody thinks they understand.

Corruption is damaging to growth, which is not surprising. The more interesting result is that predictability matters more than magnitude. A known, stable bribe is a tax: unpleasant, and something a firm can calculate and plan around. An unpredictable bribe, which may or may not deliver what was paid for, cannot be planned around at all, and firms respond to that uncertainty by not investing.

Comparisons of countries with similar measured corruption levels but different predictability find the predictable cases growing faster. Which means an index measuring how much corruption exists is missing part of the mechanism, and that a reform which reduces the number of bribes while making the remaining ones more arbitrary can leave things worse.