Representing the Rainbow

Would you ever hire a white actor to play Martin Luther King? What about an atheist to play the part of a priest? Or a Frenchman to play a Russian? While the answer seems clear cut for the first question, it may not be for the others. Now, to the question I really want to ask: would you hire a straight actor to play the part of a queer character?

Let’s get our hands on some data

I decided to turn to a source that may (or may not) be considered reliable: Wikipedia. What can I say, I trust in the wisdom of crowds (and on my thorough screening of said wisdom). Using Wiki as a starting point, I compiled two lists: one of the actors and actresses who have publicly come out as LGBTQ (being it as gay, lesbian, bisexual, trans,…) and a list of all queer characters in both TV and Radio and Movies.

I then started the (quite time consuming) task of cleaning the data and gathering information on each actor sexual orientation. Refer to the end of the article for details on how I went about it.

What does *insert LGBT term here* even mean?

Here’s a handy glossary with some of the terms I’ve used. Liberally copied from Stonewall. I like to be PC on this 😉 Feel free to skip and refer back to this in case you find any term you are unfamiliar with.

• Bisexual (or Bi): refers to an emotional and/or sexual orientation towards more than one gender
• FtM: an abbreviation for female-to-male, a term used to describe someone who is assigned female at birth but identifies and lives as a man. Also transgender man or simply trans man.
• Gay: refers to a man who has an emotional, romantic and/or sexual orientation towards men. Also a generic term for lesbian and gay sexuality – some women define themselves as gay rather than lesbian
• Gender expression: how a person chooses to outwardly express their gender. A person who does not confirm to societal expectations of gender may not identify as trans.
• Gender identity: a person’s innate sense of their own gender, whether male, female or something else, which may or may not correspond to the sex assigned at birth
• Lesbian: refers to a woman who has an emotional, romantic and/or sexual orientation towards women
• LGBTQ: the acronym for lesbian, gay, bi, trans and queer.
• MtF: an abbreviation for male-to-female, a term used to describe someone who is assigned male at birth but identifies and lives as a woman. Also transgender woman or simple trans woman.
• Non – Binary: an umbrella term for a person who does not identify as only male or only female, or who may identify as both
• Queer: in the past a derogatory term for LGBT individuals. The term has now been reclaimed by LGBT young people in particular who don’t identify with traditional categories around gender identity and sexual orientation
• Sexual orientation: a person’s emotional, romantic and/or sexual attraction to another person
• Straight: refers to a person who has an emotional, romantic and/or sexual orientation towards people of the opposite gender
• Trans or transgender: an umbrella term to describe people whose gender is not the same as, or does not sit comfortably with, the sex they were assigned at birth.

Now that you know what the terms mean, let’s dive in, shall we?

Who’s out, out there?

Not that many people, unsurprisingly. And mostly gay man (unsurprisingly?). There are many more LGBTQ actors than actresses out publicly. It is also interesting to note that there are more bisexual women than bisexual man, which chimes with the idea that women’s sexuality is more fluid compared to men’s. There are also more trans women compared to trans men actors, more than 11x.

A rainbow of characters

I run this analysis on the full list (1500+ characters). If we look at what “flavour” of queer characters is more popular, the gays win: more than 60% of all LGBTQ roles are for gay men and lesbians. Bisexuals come in second. Trans folk only account for 6% of total LGBTQ roles. If we consider than almost half of the pie is taken up by one color of the rainbow, I guess it is fair to talk about lack of representation…

If we consider the split by gender, we see that there are more bi roles for women than for man. Almost 40% of roles for women are in fact bisexual. Unfortunately, I do not have data on the endgame… maybe this should be something I should look into next (¬‿¬)

Playing gay

More than 80% of queer characters are played by straight actors, and results are similar for both genders.

Let’s zoom in a bit: about 1/5^th of all roles are played by LGBTQ actors. However, when we look at actors who are playing a character with their same sexual orientation (for example, a lesbian playing a lesbian character), the proportion falls quite significantly. For example, although there are 81 LGBTQ roles played by non-straight women (17%), only 52 of those roles are played by actresses identifying with the same sexuality as the character they are portraying (11%). This is mostly driven by there being quite a few actresses identifying as bisexual who are playing lesbian characters.

Here’s a complete table showing which characters are played by which actors. I find it quite weird that there are 2 straight guys playing lesbians…

Typecasting and shows full of unicorns

Actors have a fear that, if they accept to play a gay character, they will be typecasted in the future. This does not seem to come to pass: only 91 out of 994 actors who have played queer roles have portrayed more than one LGBTQ character. The only two actors who have played 4 LGBT characters have played all those roles in the same show. And they are both straight (Tatiana Maslany of Orphan Black fame and Griffin Mcelroy, podcast and Tv show host, in case you were wondering who they were). However it is true that it is more likely for a member of the LGBTQ community to play a queer character more than once: 33% of actors who have played gay actually identify as such. Whether that’s typecasting or just gays wanting to be themselves on screen, who can tell…

Are shows more likely to have one lone gay character or do they go full out, with a full rainbow cast? Looks like the gays are destined to be lonely. Out of 725 unique shows, there is only one LGBTQ role in roughly 60% of them. And in almost half of these shows, the lone LGBT character is a gay guy. GBF anyone?

Here’s the list of shows which feature more than 10 LGBTQ characters. Excluding Hollyoaks and Eastenders (each of which counts more than 4000 episodes and a huge  cast), I can’t say I’m surprised by who took the top spot!

So what?

It looks like the answer to the question I asked at the beginning is a resounding yes. Whether that is a good or a bad thing, I’ll leave that for you to decide. On the one hand, an actor’s job is to be acting, so playing a gay character is just like playing any other character. On the other hand, I think that it may be quite hard for a straight person to capture the nuances of being part of the queer community (or maybe I’m just being conceited). If a character is not properly understood, they end up just being a stereotype, e.g. the hot lesbian or the flamboyant gay friend, and I would venture to say, this happens quite often.

I do now know whether the characters listed are main, secondary or just appear for a short time as a guest star in one lone episode (unless they are in a series or movie I’ve seen) (and I have seen a lot of these…). I think representation of queer people is still lacking in TV and movies, and let’s not even get into the “bury your gays” trope… LGBTQ fans deserve better. I’m not advocating for every show to have a LGBTQ character (although that would be awesome), I would however want to see shows where queer characters are not just there to be a stereotype, or to fill a quota, but because gay people exist “IRL” and thus deserve to be included.

Making good movies and television is an incredibly hard task. I could argue that a director’s or writer’s grasp of a character is more important than an actor’s. It does not matter how smart your waiter looks or how well they serve you, if the chef does not know how to cook, your meal is going to be terrible. At the same rate, if you start with a poorly designed character, even the best actor cannot turn the movie around. We should not forget the importance of directors, producers and writers to invent, develop and show compelling stories and rounded characters.

As the whole world is hailing the golden age of television, I cannot help but be hopeful that queer representation in pop culture, both in terms of roles and of LGBTQ staff involved in production, will improve in the future. And, who knows, maybe one day we will have more movies written by and starring LGBTQ folks.

In the meantime, if you’ll excuse me, I’ll be on the couch watching Carol for the 138th time.

Note on data

To find each actor’s disclosed sexual orientation I did the following:

• If the actor / actress was featured in the list of those that had publicly come out, I labelled them accordingly
• If they were not, I searched online. If they had publicly disclosed their sexuality, I labelled them according to their disclosure
• If no disclosure had been publicly made, I looked if they were marked as married / partnered on Wikipedia
• If no information on marriage or dating was available on Wikipedia, I searched online. If I found more than one source citing them dating, I assumed their sexual orientation to be consistent
• If no information was available online, I excluded them from the sample.

There were a lot more exclusion with TV and radio programs, mostly because of the higher proportion of secondary characters. Here’s what I ended up with:

A (slightly) Depressing Look at European Diversity

When you wake up in the morning and look at yourself in the mirror, what do you see? A woman? A Black man? A Christian? A queer person? If you’ve answered “a human being”, it means you consider yourself to be a generic person, with no race, no class, and no gender. And you are most likely a white male: privilege is invisible to those who have it.

I was feeling uncharacteristically happy, so I decided to dampen by mood a bit by looking at these wonderful statistics from the European Union on discrimination. Admittedly, the data is almost 2 years old, however I imagine attitudes would not have shifted massively over the period.

The demography of minorities

How diverse is Europe, really? The following charts show the percentage of survey respondents that consider themselves part of a minority in their respective countries. I have graphed all to the same scale, to make the statistics more easily comparable across different diversity groups. We can immediately note that race and religion diversity are much higher than the prevalence of different sexual orientations or disabilities (on average by a factor of 3x).

I was interested in comparing the prevalence of minority groups to the perception of discrimination in every country. I grouped the perception of discrimination in quartiles, and charted it on the top: the red dots indicate countries in the bottom quartile, that is, countries residents believe to be quite discriminatory. The green dots on the other hand mark countries in the top quartile, that is, countries that are more accepting. Are perception of discrimination and diversity by type correlated? I checked with a quick regression: the R^2 are very low, sub 5%, however there appear to be a couple of patterns:

• People could be reluctant to disclose their belonging to a minority in countries where discrimination is high. If this were true, we should observe higher prevalence of “concealable” minorities (we can quite easily hide homosexuality, you cannot hide the colour of your skin) in more accepting countries. We can see a higher prevalence of “green dots” to the right in the sexual orientation chart: this means more queers in more tolerant countries, substantiating the point. (It could also be that sexual minorities move to more tolerant countries, thus inflating the numbers…).
• Higher prevalence of diversity could lead to higher acceptance of such diversity. This seems to be the case for sexual orientation and disability (more green dots to the right), however, worryingly, not in the case of race: a lot of red dots in countries where there is a sizeable ethnic minority. I cannot assess whether this is an indication of a backlash against recent immigration or not without trend data however.

A survey on hypocrisy

Have a look at the following tables. These show the average score people reported when answering the following two questions:

• Regardless of whether you are actually working or not, using a scale from 1 to 10, how comfortable you would feel if one of your colleagues at work belonged to each of the following groups?

• Regardless of whether you have children or not, using a scale from 1 to 10, how comfortable you would feel if one of your children was in a love relationship with a person from each of the following groups?

I highlighted scores below 5 in red (ie. not comfortable) and scores below 6.5 in yellow (ie. slightly comfortable). People may be broadly accepting to work with folks who are different from them, however they are clearly not ok with their kids dating a member of a minority. Especially if they happen to be Trans, Muslim or Gay. ‘nuff said. I think the percentage of people who are not ok with one of their children dating a same sex partner is even more problematic, as parents are implicitly saying they could not accept that their kid was gay if they found out.

Kissing in public is gross*

*If you are Gay

How comfortable would you feel, on a scale of 1 to 10, if a couple were showing affection in public (e.g. kissing or holding hands)? On average, Europeans say 7.5. But if the couple happens to be two girls, we go down to 5.8, and if it’s two guys, 5.5. I wanted to compare the variance in percentage terms, as the simple difference would not take into account the fact that in some countries any type of PDA may be considered inappropriate.

The map above shows the percentage difference for gay vs straight in different Europeans counties: the higher the percentage (the darker the country), the less acceptable it is to kiss your boyfriend in public if you happen to be a guy. You might want to check it out before deciding on your next holiday destination, if you are queer and plan on travelling with your significant other…

You got a friend in me

Does having a “minority friend” (man, that sounds awful) make you more accepting? Evidence is mixed. The table below shows the R^2 output for the regressions where the independent variable is the share of survey respondents who say they have a friend who belongs to a minority group and the dependent variable is one of the following:

• Is discrimination against [appropriate minority group] rare?
• Would you feel comfortable if the highest elected political position in your country was a member of [appropriate minority group]?
• Should school lessons and material include information about diversity in terms of [appropriate minority group]?

I highlighted the cells with different colours depending on whether the slope coefficient was negative (inverse relationship) or positive (direct relationship), and checked if the slope was significant.

• When considering perception of discrimination, it looks like having a “minority friend” will not impact positively. On the contrary. For the two cases where the slope is significant, the reverse is true: having a minority friend will lead you to believe discrimination in your country is more widespread. Maybe it’s because you realize how much worse the rest of society is treating your friend compared to how they are treating you?
• Things look rosier for the other two statistics: in countries where more people have a friend who has a different ethnic background, sexual orientation, gender identity or religion believing that schools should teach diversity and that a gay or Buddhist president would not be such a bad thing is more common. (Or it could just be that more open people just tend to have more diverse friends…)

Bonus point: Do people who live in countries where it is common to have a more diverse group of friends feel more satisfied with their life? Looks like they do! Especially if their friend happens to be gay (see chart below). Guess the whole GBF thing actually has some merits (¬‿¬)

So what?

I’m no statistician, these are simple observations gathered from the data and I have not controlled for any other variable while coming up with my conclusions. There are a myriad of different issues that may impact the outcome of the analysis (demographic and social composition in the different countries, openness and willingness to reply truthfully to a survey, age, …) however I do believe the conclusions are relevant and important. The data shows we are still living in a society that is not as accepting as we might want it to be, and sometimes, living in a large city like London, it is easy to forget that there are still places in Europe where whoever is different from the majority still faces discrimination. There is some hope though: more visibility, especially through friendship and representation in the media, can help change things. Let’s hope this will continue to be the case in the future and let’s look forward to when it will not be necessary to have surveys such as this.

God and Gender

It is not that uncommon to hear people referencing the link between religion and the treatment of women, and usually the two things are not positively correlated. I wanted to check whether or not this stereotype stands up to scrutiny. And to data. This one’s going to be short (but not so sweet). Let’s have a look, shall we?

The Sources

I decided to use three different indices to see if the findings were consistent across a number of measures of inequality. Here is a handy dandy little table with the data sources I used, along with a brief description of each index:

Index	Publisher	Description
GGGR	World Economic Forum	This index focuses on – gaps rather than levels (for example, the Index measures the size of the gap between male and female enrolment rates, but not for overall levels of education). – outcome rather than input variables (for example indicators related to country-specific policies, rights or culture, factors considered “inputs”, are not included). – measuring gender equality rather than women’s empowerment (rewards when outcomes for women equal those for men, but neither reward nor penalize cases in which women are outperforming men)
GII	UNDP	The GII is an inequality index. It shows the loss in potential human development due to disparity between female and male achievements in empowerment and economic status. Overall, the GII reflects how women are disadvantaged in these dimensions
GDI	UNDP	The GDI measures differences between male and female achievements health, measured by female and male life expectancy at birth; education, measured by female and male expected years of schooling for children and female and male mean years of schooling for adults ages 25 and older; and equitable command over economic resources, measured by female and male estimated earned income
Religiosity index	WIN – Gallup	A survey by WIN/Gallup International which explores religious beliefs of over 66.000 people in 68 countries across the world.

 

The Data

The world is God’s playground. At least, that is what the Win/Gallup pool data shows. On average, 60% of the population in any one country is religious; however there are sizeable differences between nations. The least religious countries are China, where only 9% of the population say they believe in God, and Japan (13%). Compare this to Thailand, where 98% of the population is religious and where your religion is written on your ID card, and Nigeria, where 97% of the population is made up of believers.

The prevalence of religious beliefs does not appear to be clustered geographically. However developed regions do tend to have lower levels of religiosity overall, although this is not the case for all countries. For example, Italy or Greece have quite a high proportion of believers, at 70% and 73% of the population respectively, while some developing countries have a lower level: take Estonia or China for example, where the figures are 28% and 9%.

Since all indices are measured on a scale, I decided to rebase them to 1. For example, I assume the country with the lowest score in religiosity to be a 0, and the one with the highest to be 100. I then scale the values for all countries.

If we analyse the data distributions (see charts above), religiosity seems to be pretty evenly distributed across the spectrum, while GII and GDI, the two UNDP indicators, present a skew towards positive values. Of the tree gender inequality indicators, the GGR is the one whose distribution most resembles a normal one. The values for the three indices have been rebased to 1, with higher values indicating lower gender disparity. I used simple (that is, not rebased) religiosity data for this chart. Thus we can say that, for example, there are 11 countries where between 20 and 30% of the population believes in God.

The analysis

I have about 65 data points for each of the indices, the data looks to me fairly distributed, there are no obvious extreme outliers skewing the results. I simply run a regression with religiosity as the independent variable and one of the inequality measures as the independent variable.

The findings are consistent: the higher the religiosity, the lower the country scores in terms of gender equality. The power of the regression as measured by the R^2 is quite high.

I was brought up catholic, however I now consider myself a flaming atheist. I think everybody should be entitled to believe in whatever they want, however belief systems should be kept out of social and political life, and religion should not play a part in defining laws nor have any bearings on how people who do not believe live their lives. It can be hard to disentangle the effects of religion from culture and traditions, however I think that such efforts are warranted, especially if we want our society to be pluralistic, open and accepting.

With my opinion of religion out of the way, let’s have a look at the results of this analysis. Correlation does not imply causation, however it is interesting to note the existence and the strength of negative relationship between the variables. It is also worth noting that the correlation is present when using all three of the indices, and albeit the power of the relationship varies, all the correlation coefficients are statistically significant. From looking at the charts, we can say that higher religious prevalence in a country is correlated with higher gender inequality. I will let you draw your own conclusions from this. Over and out.

Of Cougars and Boy Toys: age differences in couples

Melania and Donald Trump and Emanuel and Brigitte Macron: are these situations common? No, I’m not talking about being president or being a billionaire (or both), I’m talking about marrying someone much older (or younger). In other words, are there many cougars or manthers (which is apparently how we should refer to men going for much younger women) in the jungle? Let’s have a look at the data.

Man marrying younger women?

The fact that man tend to marry younger women is a stereotype but it is grounded in truth. If we look at the age difference distribution between husbands and wives, and compare it to a random normal distribution, we immediately see that wives tend to be younger than their husbands. However the difference in ages is usually 0-10 years, and most people actually marry within + or – 5 years from their own age. In only 0.3% of cases are wives 20 or more years older than their husbands, and only 2% of married man are over two decades older than their spouse.

Does that change with age? Obviously, when you are 20, it is very hard (not to mention illegal) to marry someone 10 years younger than you, and when you are 90 it is very unlikely you will find a dashing 100 year old to take to the altar. If we look at the data, we can however notice two distinct patterns.

If we consider marriages by the wife’s age, we can see younger women prefer older guys. But as they get older they tend to go for younger partners: 31% of under-20 year old women, for example, marry guys that are over 10 years older than them, and this drops to 14% by the time they are 60.

Guys on the other hand, maintain a preference for younger women. If we segment the data according to husbands’ ages, only about 2-3% of any age cohort marry women significantly older than they are.

Has this changed over time? That is, did people use to marry with wider (or narrower) age gaps in the past compared to today? It does not seem to be the case. The mean age difference has remained stable, although the dispersion has slightly increased. It appears that there is a higher number of marriages where either the husband or wife are significantly older than their partner. This could however be simply due to increasing life expectancy.

Do these marriages last?

Do people who marry a partner of their same age stay together for longer compared to people marrying someone much older or younger?

If we look at the overall proportions, we can see that the percentage of marriages and divorces by age difference are quite close. For example, 41% of people marry someone who’s up to 5 years younger or older than they are, and out of 100 divorces, 42 are between people within 5 years of age of each other.

According to a more rigorous 2008 study from the ONS (Wilson and Smallwood: “Age Differences at Marriage and Divorce”.) there does not seem to be a difference in the proportion of marriages that have ended in divorce when looking at age difference, even when accounting for mortality. In couples where one of the partners is significantly older, the marriage may be more likely to end because of their death and not because of divorce, and this could skew the numbers, showing a lower proportion of divorces than would be expected. However this does not seem to be the case.

What about civil partnerships?

Since civil partnerships have been recognised in the UK in 2005, the ONS has collected data on both partnerships and dissolutions. Are there significant differences between ages of partners in same sex couples compared to heterosexual couples?

Lesbian couples tend to be closer in age than male gay couples, with up to 16% of gay male couples having one of the partners be more than 10 years older than the other, compared to only 7% for gay females couples.

As they age, lesbians continue to form civil partnerships with someone close to their own age, while guys show an increasing preference for younger partners: 48% of gay men marry someone of their own age when they are 35 to 45 year old, however this drops to 24% by the time they are 55.

Finally, is there a correlation between age difference and divorce rates in same sex couples? It appears that there is a higher rate of divorce for couples closer in age (when compared to the “marriage rate”), although conclusions may only be relevant for this particular dataset and cannot be generalized.

So what?

Other than having some trivia to bore guests at dinner tables with, what are the implications from this analysis?

• Guys tend to marry younger women, regardless of how old they are
• Girls prefer more mature man, unless they are cougars—ehm, older, in which case they go for a boy toy
• Couples with larger age difference do not divorce more frequently (maybe the younger party just waits for the other to drop dead and run away with the inheritance)
• There are larger age differences between gay man couples than between lesbian couples
• Same sex couples with lower age difference appear to “divorce” more frequently

There is one remaining question: why? To address that, we would need data on a number of parameters for couples, chiefly (I think) on income and education. Until I can get my hands on that dataset, the question will remain unanswered (or just a google search away if you are curious).

Caveats and sources

• I tried diversifying my sources a bit, and was looking at Eurostat and American census data. However I could not find easily accessible datasets until I went back to my favourite source: the UK Office for National Statistics. And, lo and behold, the data was there.
• Some of the statistics are more robust and generalizable than others, simply because the sample size is much larger for marriages happening between the ages of 25 and 35 than for marriages happening when the bride and groom are over 60.
• The same comment holds true for civil partnership, where the sample size is much smaller and conclusion may not be as broadly applicable as with heterosexual marriages.
• The data on civil partnerships dissolutions is not as robust as for divorces in heterosexual couples, with a total of only 6000 civil partnership dissolutions over 10 years

Regions, Revenues and Valuation

Does the market attribute a higher valuation to companies that have more diversified revenues? I approximated the degree of diversification using the percentage of revenues that are earned outside the domestic market. Companies with higher foreign revenues should be perceived as more diversified and thus receive a higher valuation in terms of CAPE. Let’s see if this is true.

The Data

• Company revenue split by currency was obtained from financial reports. I sed companies from the SP 500, EUROSTOXX50, Topix and FTSE 100 indices, for a total of 200 companies across a number of sectors.
• Data necessary to estimate CAPE (cyclically adjusted PE) was downloaded from Bloomberg.
• I used EPS before extraordinary to compute the CAPE
• To account for outliers, CAPEs which are above 2x StDev from the mean are substituted with the median

Analysis

Quartiles

I run the analysis looking at both mean and average across the entire sample. As I am using adjusted before extraordinary EPS to calculate the CAPE, the difference between averages and medians across the sample is not very significant, thus the conclusions below are applicable in both scenarios.

There does not appear to be a significant difference between averages in different quartiles. Running a more formal analysis, we can observe that the only significantly different average is the third quartile one. The difference is significant at the 90% level.

This could be due to both the presence of outliers and the concentration around the third quartile of sector which have a higher valuation (consumer discretionary, Financials and IT)

Conclusion: companies that derive between roughly 50 and 75% of their revenues in their domestic market appear to achieve higher valuations, but the relationship is not very robust.

Sectors

There does not appear to be a direct correlation between revenue source, sector and CAPE. In the rare cases when there is a positive or negative correlation between share of domestic revenues and sector, the relationship is driven by one or two outliers. To conduct more formal tests, it would be necessary to gather more data, as for some sectors there are only a few data points available.

The only exception is industrials, which appear to have higher valuations when they are focused on international markets (see graph below)

Conclusions: there does not appear to be a significant relationship tying sector, revenue diversification and valuations.

Regions

Finally I looked at the split by region, to see whether when subdividing the sample into the index components, a relationship could be found.

The most significant conclusion is that there is a negative relationship between the proportion of domestic revenues and the valuation as proxied by CAPE. This relationship is not so markedly evident when we consider the full sample. This is due to the effect of the mild negative relationship in the Topix and SP500 constituents.

The graph has been zoomed in to exclude outliers and to better show the negative relationship

If we increase the adjustment for outliers, decreasing the number of standard deviations above which we substitute the median value for the original one, the strength of the negative relationship decreases but is still present.

The explanatory power of the relationship is however low, with the exception of the EUROSTOXX constituents, the only to have an R^2 larger than 20%.

Conclusions: companies in Europe appear to achieve a premium valuation if they are more revenue diversified. The relationship however is not robust.

So what?

Overall, companies that have a more geographically diversified source of revenues do not appear to receive a markedly better valuation on the market.

 

 

These taxing Taxes

Who is really supporting the burden of taxes? Are the rich really paying too little (or too much)? How is income distributed by tax bracket?

I have always been interested in finding an answer to these questions, and have conducted a short analysis to see if I could get to the bottom of it.

The Data

I downloaded data from the Italian , the UK , and the US tax agencies on income tax distributions. The data is in thousands of LCU, therefore comparisons across salary levels are not completely accurate. However we can assume, for simplicity sake, that earning 15k EUR per year is basically equivalent to earning 15k GBP or 15k USD. Additionally, I have assumed equally distributed taxes and earnings within income tax groups to compute quartiles. I have tried to be as accurate as the data granularity would allow, but consider all figures to be approximations.

The Analysis

From looking at the overall distribution, we can immediately see that there are considerably more people earning salaries in excess of 100k in the US than in the other two countries. Additionally, when comparing redistributive effects of taxes, again the US has a much more redistributive system, meaning that a larger proportion of taxes is paid by those earning the highest salaries compared to the other two countries.

This becomes more evident when we analyse the data on the dividing the population in quartiles. The richest 25% earn more than 50% and pay more than 70% of all taxes in all three countries. This is somewhat to be expected, however I was surprised by the extent of the concentration of both earnings power and taxation in the top quartile.

I decided to investigate further and focused on the top 1 and 5% richest taxpayers, or at least, of those that declare the largest income. A couple of interesting conclusions from this graphs are that the UK seems to be taxing the richest more, when we look at the difference between percentage of income and taxes earned by the richest 1 and 5%. This has to be caveated with the fact that this analysis is only looking at income tax, and thus not providing a complete picture of the tax burden or actual income distribution.

Finally, I wanted to have a look at visualizing equality as a “GINI curve”. The larger the area between the 45* line in the graph and the underlying income and taxes distribution, the more unequal the system is. Overall, we can appreciate that the income inequality is stronger in the US, however the tax system appears to be accounting for that difference and increasing the burden of paying taxes to the richest.

Overall, it appears to me that the assumption that the rich are not paying their share is not reflected in the numbers. On the contrary, I might suggest that 25% of the richest population supporting over 70% of the tax burden may even be excessive. This is to be coupled with the fact that, at higher earnings, and with higher tax rates, the incentive is to find ways to avoid paying those taxes, especially if they are perceived to be unjustly high and / or are not reflected in the level of services the state provides. In this context, I would like to point out the fact that only 1% of the population in Italy declares to be earning more than 100k EUR per year, compared to 2.5% in the UK and a whopping 25% in the US…. An additional point is the number of people paying income tax as compared to the total population. In Italy, about 40m people are paying income tax, on a population of 60m. In the UK the figure is even lower, with 30m income taxpayers on a population of 64 m. Finally, in the US, there are 320 m people, but only 150m are paying income tax. And considering that in all governments , income tax accounts for anywhere from 30 to 50% of the national budget, you can draw your own conclusions…