"One area where the evidence is less clear-cut than the findings presented in World Development Report 1997 is the issue of public sector pay and its effects on corruption. There is plenty of anecdotal evidence that the low wages available to civil servants in many developing countries drive them to take bribes in order to supplement their incomes. While at least one study has found systematic cross-country evidence of higher corruption being associated with lower wages in a sample of 28 countries, other studies covering more countries fail to do so. Many of these studies also do not distinguish between countries where petty corruption (which is more likely to be influenced by salaries) and grand corruption (which is less likely to be influenced by salaries) are important.

Careful country-specific analysis is beginning to provide more nuanced evidence on the relative importance of wages and other factors for corruption. For example, a study of procurement contracts in public hospitals in Buenos Aires, Argentina, found that a 10 percent increase in the salary of procurement officers was associated with a 1.2 percent reduction in prices paid for hospital supplies. However, this relationship between pay and performance was apparent only after a crackdown on corruption had been in effect for a period of six months. The crackdown itself also had significant effects on procurement prices, initially lowering them by an average of 18 percent—although this effect weakened over time. Interestingly, this particular crackdown achieved significant results without threats of penalties for wrongdoing. Instead, the staff of the health secretary simply collected data on the procurement prices of basic hospital supplies from each hospital and then circulated this information among all hospitals on a regular basis.

Political institutions that restrain politicians from arbitrary actions, and institutions that hold politicians accountable for their actions, can help reduce the opportunities and incentives for corruption...

The effectiveness of elections as a disciplining device depends on two factors. The first is the extent to which elections are free and fair. Without this minimum condition, elections cannot serve to discipline politicians and sanction them for corrupt practices. Second, provided that elections are in fact free and fair, there is evidence that the design of electoral rules themselves influence the accountability of individual politicians to their constituents.

Recent research has focused on two dimensions of electoral rules that matter for accountability. The first is the extent to which electoral systems reward or punish individual candidates relative to political parties. When legislatures are selected by proportional representation, with candidates chosen from party lists, voters can vote only against particular parties and not against individuals whom they perceive as corrupt. As a result individual politicians have less reason to fear that they will be punished at the ballot box for engaging in corrupt practices. The second is the extent to which electoral rules create barriers to entry for new political parties. When new parties find it difficult to gain representation in the legislature, it is more difficult for them to challenge corrupt incumbents. One factor determining the ease of entry for new political parties is the number of representatives per electoral district, since it is easier for smaller parties to win seats in districts with multiple representatives.

Press freedom and civil society
Lack of information breeds corruption. When the actions of public officials are not subject to scrutiny by the general public, opportunities for official misconduct become more attractive... Without widespread information on the extent of public wrongdoing, the public disgust with corruption that is essential to implementing reforms is slow to form...

The media can help provide information by vigorous investigation and reporting of allegations of public malfeasance. For the media to be effective in this role, they need to be free from political pressures that prevent investigation and reporting of scandals that would embarrass those in power. Across countries, there is a clear association between indicators of press freedom and absence of corruption. An important factor in this regard is media ownership. When the media are controlled by the state, they are more likely to be subject to political pressures (chapter 10).39 The quality of media coverage is also likely to be important in determining the extent to which decentralization will lower corruption."

--- World Bank, World Development Report, 2002