Short Answer: Maybe economic near-equivalence is a lousy explanatory variable for corruption.
Long Answer: This is as much a question about the CPI as it is about politics. Scales like the CPI or the Failed States Index attempt to measure states by a rigorous set of metrics over a scale on the order of 100, with varying granularity. For the cases like Canada and New Zeland, they may happen to fall into different colors on the map, but their scores aren't all that far apart (6 points). It's possible that the CPI has a less granular set of clusters that better describes the corruption in one country relative to another. In other words, maybe the scale is bad because it's too precise. In the Canada/New Zealand case, this would mean that both of these countries fall into the same cluster, even though they didn't fall on the same point in the number line or bin in the choropleth.
The question becomes more interesting when we look at pairs like Bhutan/Nepal or Chile/Argentina. These countries have a much wider spread in their scores. I don't have a substantive background in either region (and I don't feel like picking through the statistics at this moment) but I would question the implicit assumption that their economic situations are similar because their sizes are similar or they have close geographic proximity to one another. If this happens to be the case, then I would proceed to the short answer above: maybe the economic character of a state simply has little relationship with the people's perception of corruption in reality.