I made a thing. Mostly, I made this thing to prove a point to a friend that was complaining about proposals for higher income tax. But I’m quite proud of this, given that it was about 25 minutes work using publicly available data, so I thought I’d share it.
Effect of Income Tax Dashboard
You can interact with this better at the actual site.
TLDR; The higher up, the more the country pays in tax. The further to the right, the better the country is. The strongest correlations are that if your country pays more in tax, you’re more likely to have broadband, you’ll live longer and you’ll work fewer hours (I will let you speculate as to whether that is the correct order of importance…) Keep in mind that this is a very random selection of countries that contains some outliers.
The dashboard was intended to show correlations between higher rates of income tax and a bunch of indicators. These indicators and countries were fairly randomly chosen based upon availability of data and ease of access. The countries are all countries which the OECD holds published figures for income tax as a % of GDP for. I then went through the OECD data, picked a bunch of datasets that I thought would be interesting to test for correlation, and merged the data sets. All the data is publicly available on their website. If you’re a data nerd that cares about politics or economics, there’s a ton of stuff there.
In order to control for the massive variance in scale across indicator, and to present the data consistently, I have used rankings rather than absolute values. In other words, I sorted the list from worst to best (or least to most), and gave the countries with data a number according to their rank. So a high rank means you pay a lot of taxes, or you have whatever is considered “good” for the indicator. Life expectancy, a high ranking means your citizens live longer. Poverty rate, it means you have fewer people living in poverty.
This created a dataset that made for easy, consistent visualisation despite the massive differences in the type and scale of the data. Total time to create the dataset: about 15 minutes. Most of this was finding the data sets that I wanted to use.
I didn’t want to spend much time on this, and I needed to be able to share this quickly to prove a point, so to visualise it I turned to Tableau Public. This is a fantastic little tool to quickly visualise data – and is free as long as you don’t care about sharing the data with the visualisation (I don’t), or being limited in size (not an issue here).
On the Y-Axis, we have a ranking of countries by how much they pay in income tax, as a % of GDP. On the X-Axis, we have whatever indicator is tested for. By standardising the dataset using rankings using worse to better (rather than less to more), the data visualisations are nicely consistent and comparable.
I took about 5 minutes to get the template that I wanted working correctly (I forgot how to make the points coloured by country). Each copy of that then took seconds to make as you can just copy the old format and change the variable. Overall, the dashboard took something less than 10 minutes to make.
Total time to complete: 25 minutes on a bored Friday afternoon.
What does this mean?
Some of the indicators don’t really have very strong correlations. If you exclude some of the countries that are likely to be quite different then you get stronger correlations, but I didn’t want to cherry pick results (if you play with the tool you can remove any you aren’t interested in). So I’ve left in both indicators that didn’t show anything I wanted and countries that might have skewed my results.
However, Life Expectancy at Birth, availability (and use) of Broadband internet and Hours Worked all have strong correlations. The short answer is: in the countries that pay more tax, they are a lot better about each of these.
Is that directly because of tax? Who knows. I think you’d need to do more than ~25 minutes work on an analysis to prove that point.
I have no particular plans to advance this any further. However, if anyone wants me to add any indicators to this list, just provide me with a data source (preferably but not necessarily from the OECD) and I’ll happily chuck them into the mix.