The Ranking

Table 1, below, shows the results of the BPI 2006, with additional statistical information to indicate the degree of agreement among respondents in one particular country: the smaller the standard deviation, the broader the consensus. The scores are close together, despite the countries’ differing rankings. Higher scores reveal a lower propensity of companies from a country to offer bribes or undocumented extra payments when doing business abroad.

Table 1: The full results of the BPI 2006

 

The margin of error at 95 percent confidence is provided to demonstrate the precision of the results. The confidence level indicates that there is a 95 percent probability that the true value of the results lies within the range given by the margin of error above and below each score. The BPI results and confidence intervals, indicated by vertical bars, are also shown on the graph in figure 1.

Cluster analysis on the BPI 2006

Cluster analysis ¹ of the BPI results groups countries that exhibit similar behaviour in terms of their companies’ propensity to bribe abroad. As the differences in scores between adjacent countries on the ranking are small, this analysis provides further material with which to interpret and understand the results.

This analysis produces four clusters (or groups) of countries. Cluster 1 comprises the countries from which companies are least likely to bribe when doing business abroad, and cluster 4 comprises those that are most likely to bribe, according to the BPI 2006.

Cluster 1: Switzerland, Sweden, Australia, Austria, Canada, UK, Germany, Netherlands, Belgium, US, Japan

Cluster 2: Singapore, Spain, United Arab Emirates, France, Portugal, Mexico

Cluster 3: Hong Kong, Israel, Italy, South Korea, Saudi Arabia, Brazil, South Africa, Malaysia

Cluster 4: Taiwan, Turkey, Russia, China, India.

It is important to note that although cluster 1 represents the best performers of the 30 countries, the results of the BPI highlight that companies from all countries in the survey show a considerable propensity to pay bribes. Countries in the top clusters should therefore not view this as an endorsement of their companies’ behaviour.

Once there are several comparable iterations of the BPI, these clusters will allow changes in the countries’ ratings to be tracked across time.

 

 

---

1 This analysis uses an agglomerative hierarchical cluster procedure to form four clusters. The decision to use four clusters was made using a graphical approach, ‘the elbow criterion’, to examine the reduction of variance.