Fighting Corruption: What Actually Works

What has actually reduced corruption, what has not, and what an individual officer can do inside a system he did not design.

Start with what has worked

Reform proposals are more credible when they begin from measures with a demonstrated effect rather than from a wish list, and several are available.

Removing the intermediary. Direct transfer of benefits into accounts eliminated the point at which a share could be extracted during handover. Chapter 24 noted the qualification, that the difficulty relocated to enrolment and authentication, but the reduction at the delivery point is real and should be acknowledged.

Removing discretion over sequence. Online applications with recorded timestamps and published queues remove the power to let a file sit, which Part 2 identified as the product being sold in most petty corruption.

Competitive allocation. Auctions and open tendering for scarce public resources replace discretionary distribution, which Chapter 17 identified as the structural condition that made large allocation scandals possible.

Statutory service guarantees. Chapter 19 examined these: a named officer, a fixed period, an appeal and a penalty convert a delay that could be sold into a default that costs the office.

The common feature is worth stating plainly, because it is the organising idea of the whole Part: each measure removes the occasion rather than appealing to virtue. That is why they work at scale and why exhortation does not.

What has worked less well

An answer gains credibility by naming the measures that have disappointed and explaining why.

New institutions have been created repeatedly, and Chapter 23 explained the pattern: creating a body is visible and announceable, while making an existing one effective requires years of unglamorous attention. The result is overlapping authorities sharing the same scarce investigative capacity.

Harsher penalties address severity when the binding constraint is probability. Where detection is unlikely and adjudication slow, raising the maximum sentence changes the expected cost very little, which is the arithmetic from Part 2.

Codes and training alone address knowledge, which Chapter 16 established is almost never the constraint. The officer who diverted funds knew he should not.

The general lesson is that measures failing to change either the occasion or the probability of detection will underperform however well intentioned they are.

The reform agenda, organised

Set out by the term of the arithmetic each addresses, which is a structure examiners reward.

Reduce the occasion: simplify procedures and reduce the number of approvals; publish criteria for every discretionary decision; require recorded reasons; separate the functions of specifying, selecting, receiving and paying; rotate officers in sensitive postings; and prefer direct transfer over intermediated delivery.

Raise the probability of detection: proactive disclosure so patterns are visible without a complaint; social audit for delivery, since beneficiaries detect what reconciliation cannot; data analysis across transactions to identify anomalies; protected and credible channels for those who report; and leniency provisions for participants in collusion, which is the only reliable way to break arrangements with no victim to complain.

Make consequences timely: specialist investigative capacity, dedicated adjudication with time limits, and departmental action that does not wait for the conclusion of criminal proceedings. Chapter 16 stated the principle: certainty and speed deter, severity alone does not.

Address the upstream causes: transparency in political funding, since Part 2 identified the recovery pressure it generates; predictable postings and tenures, which reduce the value of cultivating individuals; and adequate remuneration paired with, not instead of, the measures above.

What an individual officer can do

Most of the agenda above lies beyond a single officer, and the honest answer to what he can do is nonetheless substantial.

Within his own charge he can publish criteria, require recorded reasons, separate functions, verify physically before payment, refuse advance against undelivered work, rotate staff in sensitive seats and answer grievances with reasons. Chapter 23 made the point that these supply the standard, information and answerability elements locally even where national consequence is weak.

He can also decline personally, using the method Chapter 18 set out: depersonalise the refusal by citing the criterion, put the position in writing, brief superiors so the matter is institutional rather than a private stand, and report an explicit inducement to the appropriate authority. And he can protect subordinates who refuse, since a junior who declines a demand and is left exposed teaches the whole office that refusal is unaffordable.

What he cannot do is clean a system by personal example alone, and an answer claiming otherwise is not credible. What he can do is ensure that within his charge the occasion is reduced and refusal is survivable, and that when he leaves the arrangements are in process and record rather than resting on his own supervision.

The closing proposition

This chapter completes Part V, and the argument that has run through all of it can be stated in a sentence worth carrying into any answer on governance.

Corruption is not primarily a problem of bad people in a good system. It is what a system produces when discretion is wide, criteria are unpublished, detection is improbable, consequences are slow and refusal is costly. Each of those is a design choice, and each can be changed. That is simultaneously the most demanding and the most hopeful conclusion available, because it means the problem is neither a feature of national character nor beyond the reach of the officers who administer the system. Chapter 26 turns to applying all of this to unseen cases.

Citizens, media and the demand side

Every measure so far addresses the supply of corrupt opportunity inside the administration. The demand side deserves treatment because answers routinely omit it and because it is where an officer has least control and most need of allies.

Citizen awareness matters in a specific rather than a general way. A person who knows the fee, the timeline and the officer responsible is far harder to extract from than one who knows none of these, which is why the publication measures in Chapters 19 and 22 are anti-corruption instruments and not merely service ones. Awareness of the entitlement is worth more than awareness that corruption is wrong.

Collective action matters more than individual refusal. A single applicant who declines to pay is simply not served; a group that collectively refuses, or a community conducting a social audit, changes the calculation entirely because the official cannot exclude everyone. This is the structural reason social audit works where individual complaint does not.

Media scrutiny raises the cost of exposure, with the qualification Chapter 16 recorded, that coverage driven by sensation can damage honest officers and push administrations toward visible rather than correct responses.

Why this is a generational project

A final realism worth including, because it distinguishes a considered answer from an optimistic one. Countries that substantially reduced corruption did so over decades, through sustained enforcement, steady removal of discretionary opportunity and slow change in what was regarded as normal.

Two implications follow. Expecting rapid transformation from any single measure leads to the conclusion that nothing works, which is itself corrosive and empirically wrong, since the specific measures listed above did produce measurable change. And durability matters more than intensity, because a campaign that raises enforcement sharply and then relaxes teaches that waiting is a viable strategy.

The officer’s realistic ambition is therefore not to eliminate corruption but to make his own charge one where the occasion is reduced, refusal is survivable and the arrangements outlast him. Enough officers doing that, for long enough, is what national change has actually consisted of everywhere it has occurred.