Adjusting Corruption Perception Index Scores for National Wealth

My post two weeks ago discussed Transparency International’s newly-released 2017 Corruption Perceptions Index (CPI), focusing in particular on an old hobby-horse of mine: the hazards of trying to draw substantive conclusions from year-to-year changes in any individual country’s CPI score. Today I want to continue to discuss the 2017 CPI, with attention to a different issue: the relationship between a country’s wealth and its CPI score. It’s no secret that these variables are highly correlated. Indeed, per capita GDP remains the single strongest predictor of a country’s perceived corruption level, leading some critics to suggest that the CPI doesn’t really measure perceived corruption so much as it measures wealth—penalizing poor countries by portraying them as more corrupt, when in fact their corruption may be due more to their poverty than to deficiencies in their cultures, policies, and institutions.

This criticism isn’t entirely fair. Per capita income is a strong predictor of CPI scores, but they’re far from perfectly correlated. Furthermore, even if it’s true that worse (perceived) corruption is in large measure a product of worse economic conditions, that doesn’t mean there’s a problem with the CPI as such, any more than a measure of infant mortality is flawed because it is highly correlated with per capita income. (And of course because corruption may worsen economic outcomes, the correlation between wealth and CPI scores may be a partial reflection of corruption’s impact, though I doubt there are many who think that this relationship is so strong that the causal arrow runs predominantly from corruption to national wealth rather than from national wealth to perceived corruption.)

Yet the critics do have a point: When we look at the CPI results table, we see a lot of very rich countries clustered at the top, and a lot of very poor countries clustered at the bottom. That’s fine for some purposes, but we might also be interested in seeing which countries have notably higher or lower levels of perceived corruption than we would expect, given their per capita incomes. As a crude first cut at looking into this, I merged the 2017 CPI data table with data from the World Bank on 2016 purchasing-power-adjusted per capita GDP. After dropping the countries that appeared in one dataset but not the other, I had a 167 countries. I then ran a simple regression using CPI as the outcome variable and the natural log of per capita GDP as the sole explanatory variable. (I used the natural log partly to reduce the influence of extreme income outliers, and partly on the logic that the impact of GDP on perceived corruption likely declines at very high levels of income. But I admit it’s something of an arbitrary choice and I encourage others who are interested to play around with the data using alternative functional forms and specifications.)

This single variable, ln per capita GDP, explained about half of the total variance in the data (for stats nerds, the R2 value was about 0.51), meaning that while ln per capita GDP is a very powerful explanatory variable, there’s a lot of variation in the CPI that it doesn’t explain. The more interesting question, to my mind, concerns the countries that notably outperform or underperform the CPI score that one would predict given national wealth. To look into this, I simply ranked the 167 countries in my data by the size of the residuals from the simple regression described above. Here are some of the things that I found:

  • The set of “overachievers” (countries with CPI scores notably higher than one would predict based on their wealth) included, perhaps unsurprisingly, several of the countries that are among the top performers in the CPI anyway, like New Zealand–which is not only the least corrupt country according to the CPI, but one with a CPI score (89) that is 30 points better than one would predict based on New Zealand’s per capita GDP. (The three countries in the sample with per capita incomes closest to New Zealand are Italy, Malta, and Israel, which get CPI scores of 50, 56, and 62, respectively.) Denmark, Finland, Sweden, Norway, Switzerland, and the Netherlands are also all in both the CPI top ten and my list of top ten “overachievers” relative to income, with CPI scores 20 points higher than even their high national incomes would predict.
  • But there are also some notable surprise overachievers. Rwanda’s CPI score of 55 may seem like nothing to write home about, but it’s 30 points higher than one would predict based on Rwanda’s low national income. Bhutan is another country where perceived corruption is substantially lower than one would predict if one looked only at its wealth: Bhutan’s CPI score is 67, but its predicted CPI score, given its modest per capita GDP, is closer to 43.
  • Indeed, there are a few countries with CPI scores below the midpoint of 50 (which TI often treats as the threshold for serious corruption) that are nonetheless notable overachievers relative to what one would predict given their overall poverty. These countries include Burkina Faso, Senegal, Liberia, Niger, and Sao Tome & Principe, all of which have CPI scores below 50 but that outperform their wealth-predicted CPI scores by over 15 points. Many of these countries are not usually identified anticorruption as “success stories,” for the good and obvious reason that corruption in these countries is still perceived as widespread and serious. Yet at the same time, these countries do perhaps offer some hidden good news (and perhaps some lessons to be learned?), in that most other countries at a comparable income level are perceived as far more corrupt. Now, one should be careful not to read too much into this. As I cautioned in my last post, simple random error could and likely would produce outliers in both directions, and it would therefore be a mistake to assume, without further inquiry, that there was something special about these countries (such as their anticorruption policies) that explains why their perceived corruption levels, while bad, are better than one finds in similarly poor countries. But it does suggest that further inquiry may be warranted.
  • It may also be worth observing that there are a few countries that do quite well on the CPI, but whose performance looks somewhat less impressive when their national wealth is taken into account. Singapore and Luxembourg, for example, regularly rank in the CPI top ten, as they did in 2017, but while both countries are also modest overachievers (by about 16 and 12 points, respectively), neither is a super-overachiever the way New Zealand and Rwanda are. There are also some countries in the upper end (though not the very top) of the CPI range that actually underachieve when one takes their incomes into account. Most notably, the oil-rich states Qatar and Brunei Darussalam get decent CPI scores (63 and 62), but those scores are several points lower than one would predict based on their (exceptionally high) per capita incomes.
  • Continuing on the theme of resource-rich CPI underachievers, Saudi Arabia, Kuwait, Equatorial Guinea, and Turkmenistan all get CPI scores that are notably lower than what one would predict from their per capita incomes. Saudi Arabia’s CPI score is a middling 49, which is about 13 points lower than what one would predict based on Saudi Arabia’s wealth. Kuwait is even more striking: it’s the fifth-wealthiest country in my sample (just behind Brunei), but its CPI score of 39 is about 27 points lower than one would expect from such a rich country. And Equatorial Guinea’s CPI score of 17—one of the lowest in the world—is a whopping 37 points below what one would predict based on the country’s mid-range per capita GDP. Turkmenistan is similar: its dreadful CPI score of 19 is 30 points lower than the level its GDP would predict. The concentration of resource-rich states among the notable underachievers might be due to the “political resource curse,” the controversial hypothesis that posits substantial resource wealth, at least in countries that are not robust democracies, fosters corruption. Alternatively, the explanation might be that when a country is blessed with abundant natural resources, especially hydrocarbons, its GDP may be much higher than what one would expect based on its history, geography, and political institutions. So maybe it’s a bit misleading to say that Kuwait, Saudi Arabia, Equatorial Guinea, Turkmenistan, Qatar, and Brunei “underachieve” on the CPI; rather, these countries may “overachieve” (in a manner of speaking) on per capita GDP.
  • The set of CPI underachievers also includes some large, important countries, such as Russia (CPI of 29, 24 points below its wealth-predicted level), Mexico (also a CPI of 29, 20 points lower than expected), Turkey (CPI of 40, approximately 14 points below the CPI score for comparably wealthy countries), Argentina (CPI of 39, about 12 points below expectations), and Brazil (CPI of 37, about 11 points below expectations). Though TI doesn’t typically frame the issue in quite this way, it might be worth focusing more attention on the fact that these countries have notably worse perceived corruption than other countries at a comparable income level.

All this is pretty crude and preliminary. For those who are interested, and want to do more (and hopefully more sophisticated) analyses, I encourage downloading the data. My objective here hasn’t really been to try to explain over- and underachievement, even though I did offer a few conjectures in passing. Here I’m more interested in following the “league table” approach that TI and many journalists emphasize in their coverage of the CPI, but looking at performance relative to income level rather than overall performance. So, in case folks are interested, I’ll provide below my own “league table,” ranking 167 countries in the order of their performance on the 2017 CPI relative to their predicted performance from a simple regression of the CPI score on natural log of purchasing-power-parity-adjusted GDP:

Country CPI Score 2017 ln(2016 PPP-adjusted per capita GDP) Predicted CPI score (based on ln per capita GDP) Gap (actual CPI minus predicted CPI)
New Zealand 89 10.56 58 31
Rwanda 55 7.56 25 30
Denmark 88 10.80 61 27
Finland 85 10.68 60 25
Bhutan 67 9.09 42 25
Sweden 84 10.80 61 23
United Kingdom 82 10.66 60 22
Canada 82 10.71 60 22
Norway 85 10.98 63 22
Switzerland 85 11.06 64 21
Netherlands 82 10.83 61 21
Germany 81 10.80 61 20
Uruguay 70 9.98 52 18
Barbados 68 9.80 50 18
Burkina Faso 42 7.48 24 18
Senegal 45 7.85 28 17
Australia 77 10.74 60 17
Singapore 84 11.38 68 16
Cabo Verde 55 8.79 39 16
Iceland 77 10.82 61 16
Estonia 71 10.30 56 15
Liberia 31 6.70 16 15
Niger 33 6.89 18 15
Sao Tome and Principe 46 8.08 31 15
Belgium 75 10.75 60 15
Chile 67 10.05 53 14
Hong Kong 77 10.98 63 14
Japan 73 10.65 59 14
Austria 75 10.83 61 14
Saint Vincent and the Grenadines 58 9.35 45 13
Luxembourg 82 11.54 69 13
Vanuatu 43 8.03 30 13
Dominica 57 9.30 44 13
Georgia 56 9.21 43 13
Bahamas 65 10.02 52 13
Benin 39 7.68 26 13
Solomon Islands 39 7.71 27 12
United States of America 75 10.96 63 12
Lesotho 42 7.99 30 12
Botswana 61 9.74 49 12
Timor-Leste 38 7.67 26 12
Malawi 31 7.06 20 11
Ethiopia 35 7.46 24 11
France 70 10.63 59 11
Costa Rica 59 9.72 49 10
Togo 32 7.31 22 10
Central African Republic 23 6.55 14 9
Ireland 74 11.18 65 9
Saint Lucia 55 9.47 46 9
Sierra Leone 30 7.30 22 8
Portugal 63 10.33 56 7
Burundi 22 6.66 15 7
Namibia 51 9.27 44 7
Tanzania 36 7.93 29 7
Gambia 30 7.42 24 6
Ghana 40 8.36 34 6
Jordan 48 9.11 42 6
United Arab Emirates 71 11.19 65 6
Democratic Republic of the Congo 21 6.69 15 6
Poland 60 10.22 55 5
Seychelles 60 10.25 55 5
Mozambique 25 7.10 20 5
Mali 31 7.66 26 5
Grenada 52 9.56 47 5
Comoros 27 7.33 23 4
Slovenia 61 10.40 57 4
Latvia 58 10.15 54 4
Israel 62 10.53 58 4
Zambia 37 8.28 33 4
Côte D’Ivoire 36 8.21 32 4
Lithuania 59 10.30 56 3
Nepal 31 7.82 28 3
Jamaica 44 9.08 42 2
Guinea 27 7.58 25 2
Madagascar 24 7.32 22 2
Uganda 26 7.51 25 1
India 40 8.79 39 1
Cyprus 57 10.40 57 0
Czech Republic 57 10.46 57 0
Spain 57 10.50 58 -1
Morocco 40 8.97 41 -1
Mauritius 50 9.96 52 -2
Malta 56 10.54 58 -2
Haiti 22 7.49 24 -2
Swaziland 39 9.03 41 -2
Kenya 28 8.06 31 -3
Guyana 38 8.97 41 -3
Kyrgyzstan 29 8.18 32 -3
Tunisia 42 9.36 45 -3
Vietnam 35 8.75 38 -3
South Africa 43 9.49 47 -4
Zimbabwe 22 7.61 26 -4
Montenegro 46 9.78 50 -4
Korea, South 54 10.51 58 -4
Croatia 49 10.06 53 -4
Bangladesh 28 8.18 32 -4
Brunei Darussalam 62 11.26 66 -4
Pakistan 32 8.56 36 -4
Romania 48 10.04 53 -5
Papua New Guinea 29 8.34 34 -5
Mauritania 28 8.26 33 -5
Moldova 31 8.58 36 -5
Chad 20 7.60 26 -6
Slovakia 50 10.32 56 -6
Belarus 44 9.80 50 -6
Honduras 29 8.46 35 -6
Guinea Bissau 17 7.38 23 -6
Greece 48 10.20 54 -6
Serbia 41 9.58 48 -7
Philippines 34 8.96 41 -7
Bolivia 33 8.89 40 -7
Suriname 41 9.61 48 -7
Albania 38 9.35 45 -7
Armenia 35 9.09 42 -7
Cameroon 25 8.19 32 -7
Myanmar 30 8.65 37 -7
China 41 9.65 48 -7
Bosnia and Herzegovina 38 9.41 46 -8
Bulgaria 43 9.86 51 -8
Malaysia 47 10.23 55 -8
Sri Lanka 38 9.44 46 -8
Indonesia 37 9.36 45 -8
Italy 50 10.56 58 -8
Qatar 63 11.76 72 -9
El Salvador 33 9.06 42 -9
Tajikistan 21 8.00 30 -9
Hungary 45 10.19 54 -9
Peru 37 9.47 46 -9
Mongolia 36 9.41 46 -10
Laos 29 8.79 39 -10
Afghanistan 15 7.57 25 -10
Colombia 37 9.56 47 -10
Nigeria 27 8.68 37 -10
Nicaragua 26 8.62 37 -11
Brazil 37 9.62 48 -11
Ukraine 30 9.02 41 -11
Cambodia 21 8.23 32 -11
Yemen 16 7.83 28 -12
Argentina 39 9.90 51 -12
Thailand 37 9.74 49 -12
Egypt 32 9.32 45 -13
Ecuador 32 9.33 45 -13
Guatemala 28 8.98 41 -13
Macedonia 35 9.61 48 -13
Saudi Arabia 49 10.90 62 -13
Turkey 40 10.14 54 -14
Paraguay 29 9.17 43 -14
Algeria 33 9.62 48 -15
Maldives 33 9.64 48 -15
Trinidad and Tobago 41 10.40 57 -16
Panama 37 10.04 53 -16
Congo 21 8.65 37 -16
Uzbekistan 22 8.78 39 -17
Gabon 32 9.80 50 -18
Azerbaijan 31 9.76 49 -18
Dominican Republic 29 9.63 48 -19
Sudan 16 8.46 35 -19
Lebanon 28 9.57 47 -19
Angola 19 8.77 39 -20
Mexico 29 9.76 49 -20
Iran 30 9.90 51 -21
Kazakhstan 31 10.14 54 -23
Russia 29 10.12 53 -24
Kuwait 39 11.22 66 -27
Turkmenistan 19 9.73 49 -30
Iraq 18 9.76 50 -32
Equatorial Guinea 17 10.17 54 -37

 

4 thoughts on “Adjusting Corruption Perception Index Scores for National Wealth

  1. As usual, interesting food for thought. Nicholas Ambraseys and Roger Bilham used a similar exercise for their 2011 article in Nature, “Corruption Kills.” They found that countries with higher-than-predicted corruption accounted for 83% of deaths from collapsed buildings during earthquakes.

  2. This is great stuff, Matthew. Another reason why this has genuine validity is as an indicator of the relative resources available to a country to fight corruption, irrespective of what the causal relationships (if any) may be between wealth and corruption themselves. So, the fact Australia comes in at 17th on a scale where perceived corruption levels is adjusted for relative wealth (as against 13th on the basic CPI ranking) has face validity, to me… Our decline since 2012 is combined with evidence that comparatively, we could be doing more. Or maybe I’m reading a little too much into it. Anyway, thanks again.

  3. Pingback: Adjusting Corruption Perception Index Scores for National Wealth | Matthews' Blog

  4. An empowering finding that anti-corruption drives don’t necessarily depend on high levels of development to succeed.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.