Economics is hard.
When survey researchers ask people to guess what proportion of the wealth the top 1% in their respective countries own, they’re often way out, as seen in Ipsos’ Perils of Perception studies. Britain and France were the worst at guessing. The actual proportion of wealth owned by the top 1% in both countries is the same, at 23%. But Brits thought the top 1% owned 59% of wealth while the French thought it was 56%.
Although people in some surprising countries did underestimate this concentration of wealth. The average guess in Russia was that the top 1% own 53% of the wealth... the reality is far worse, the top 1% in Russia actually own an extraordinary 70% of the country’s wealth, three times the level of the UK and France.
So, why are we so out of kilter with reality? A big part of the explanation lies in our emotional reaction, rather than our lack of knowledge of the facts. Social psychologists have a term for this phenomenon, they label it: ‘emotional innumeracy’. This means that when we ask people to guess at the scale of an issue, cause and effect run in both directions: we’re not just worried about things that we think are big issues, we exaggerate the scale of issues that we’re already worried about.
Emotional innumeracy points to a vital conclusion: because our misperceptions of reality are partly emotional, simple “myth-busting” – telling people they’re wrong on the facts – will not really help that much, because it misdiagnoses the cause. Our misperceptions are often not about ignorance, but more about a world view driven by our underlying attitudes and beliefs.
We also asked people what they thought the top 1% should own, and this reveals some fascinating patterns. First, people are mostly not calling for complete equality of wealth – with the averages running from people saying the top 1% should have 14% of wealth in Israel to 32% in China. Some people in all countries do want complete wealth parity though – and Britain is highest on this: 19% of Brits said the top 1% should only have 1%, followed closely by Russia, where 18 % said they should only have 1%.
But, second, if we go back to the averages of what we say inequality should be and compare that with the actual level, something immediately jumps out: on the surface, people seem pretty comfortable with the current state of wealth inequality in many countries. For example, the French said the top 1% should own 27%, when they actually ‘only’ currently own 23%. A simplistic reading of this is that the French – a country that has ‘égalité’ (equality) as part of its national strapline – are saying give the wealthiest a bit more.
Of course, that’s the completely wrong interpretation. We know from our question on how people see the ‘reality’, that the French on average thought the top 1% currently own 56% of the country’s wealth. So, what they’re really saying is that the very wealthy should have about half what they currently have.
http://www.bbc.com/capital/story/20181107-what-we-just-dont-understand-about-the-1
How about a more sensible explanation: the actual figure is not accurately or widely reported.
ReplyDeleteThere are several problems here, right?
ReplyDelete1. does anyone know what the total household income of their country is?
2. how national vs. global is their news coverage?
3. what country do international elites belong to? Taxes for US citizens are always payable in the US no matter where you live. Other places have other rules.
4. what income is required to count in the top 1% in each country? Presumably the 1% population is different depending on whether you count per country or globally. If people knew what was actually owned, how would their opinions change?
....
Edward Morbius That's a fair point. But it may be even more interesting to compare accuracy of estimates on the absence of information, and the explanation may still hold.
ReplyDeleteWhat would be really interesting would be to also compare which countries have the most frequent reporting of the value.
Rhys Taylor What would be really interesting would be to also compare which countries have the most frequent reporting of the value.
ReplyDeleteThis.
19% of Brits said the top 1% should only have 1%
ReplyDeleteSo is there 40% support for communism in Labor, or are people just bad at math?
what exactly is wealth? I guess the power to buy/own -- and why should some group (any group) possess more of that power than any other?
ReplyDeletelaurie corzett what exactly is wealth?
ReplyDeleteThat turns out t be a nontrivial question. Adam Smith's definition ("the annual poduce and labour of the nation") is in fact a flow rather than a stock, as numeous other, including his biographer Edwin Cannan, have noted.
Generally it refers to financialised asset holdings, a large share of which is real estate.
why should some group (any group) possess more of that power than any other?
Welcome to the rabbit hole.
laurie corzett do you plan on outlawing savings? Taxing any wealth gains from appreciated property? How do you plan on ensuring that everyone's net outflows of wealth equals their inflows?
ReplyDeleteChris Greene Wealth, land, and estate taxes, along with living wages, rights to organise, rights to equity, needs-based social suports (education, healthcare, pensions, welfare) and guaranteed income are a good start.
ReplyDeleteEdward Morbius that will still have a non flat wealth distribution. That is the only point I'm arguing here.
ReplyDeleteI do not intend anything is this regard. I asked questions.
ReplyDeleteChris Greene Flatter and fairer is preferred. Perfectly flat is an unlikely ideal.
ReplyDeleteCouple of relevant links. First (slightly tangential) on metaknowledge :
ReplyDeletehttps://plus.google.com/u/0/+RhysTaylorRhysy/posts/ExUtrBorfWs
Second more directly relavent on meritocracies and wealth distribution :
https://plus.google.com/u/0/+RhysTaylorRhysy/posts/4cXYwfSjfEZ?cfem=1
I quickly tried plotting the guess difference against the desired wealth fraction of the 1%, but I couldn't see any obvious correlation. But there does appear to be one very interesting trend : a clear anti-correlation between the actual wealth fraction and the size of the error in the guess.
EDIT : I originally wrote that the richer the 1%, the worse everyone does at estimating their wealth. But then I realised this isn't true, because everyone also does badly when the 1% don't own so much of the wealth.
https://docs.google.com/spreadsheets/d/1gZdKZ3nc0THLLO7Vqu87N38O63R0KH2AIKnvCzC8_xE/edit?usp=sharing
https://lh3.googleusercontent.com/ALe7Ig4A5uJeKul0Duzsp6Tra8q4JcPILIoQM1B4sPSDgQ4Ck7bkyV8oq206ZKblm_ARlZDunj_w8J5P6n6ahxyLgRSrWwNV1iNy=s0
Rhys Taylor The guesses regress to the mean -- it's essentially random noise. Where actual delta is high, the guesses are consistently low, where actual delta is low, the guesses are consistently high.
ReplyDeleteThe guesses:
n: 29, sum: 1377, min: 32, max: 59, mean: 47.482759, median: 49, sd: 8.069434
%-ile: 5: 32, 10: 36, 15: 39, 20: 40, 25: 40, 30: 41, 35: 44, 40: 46, 45: 46, 55: 49.5, 60: 50, 65: 53, 70: 54, 75: 55.5, 80: 56, 85: 56, 90: 57, 95: 59
The actual:
n: 29, sum: 990, min: 18, max: 70, mean: 34.137931, median: 34, sd: 12.529178
%-ile: 5: 18, 10: 19, 15: 22, 20: 23, 25: 23.5, 30: 25, 35: 26, 40: 27, 45: 31, 55: 34, 60: 36, 65: 38, 70: 39, 75: 43, 80: 44, 85: 47.5, 90: 53, 95: 54
The deltas:
n: 29, sum: 387, min: -17, max: 36, mean: 13.344828, median: 15, sd: 15.899856
%-ile: 5: -16, 10: -13, 15: -7.5, 20: -1, 25: 1, 30: 4, 35: 7.5, 40: 11, 45: 13.5, 55: 16.5, 60: 20, 65: 21, 70: 23, 75: 29, 80: 30, 85: 32, 90: 33, 95: 35
The data, sorted by actual inequality, low to high:
50 18 32 Belgium
50 18 32 New Zealand
41 19 22 Japan
56 21 35 Australia
46 23 23 Italy
56 23 33 France
59 23 36 Great Britain
40 24 16 Netherlands
45 25 20 Norway
55 25 30 Canada
40 27 13 Ireland
56 27 29 Spain
59 30 29 Germany
46 32 14 Sweden
38 34 4 Poland
43 34 9 Columbia
49 34 15 South Korea
36 36 0 Mexico
57 37 20 USA
32 39 -7 Israel
56 39 17 China
49 43 6 South Africa
54 43 11 Chile
46 44 2 Argentina
32 47 -15 Peru
40 48 -8 Brazil
40 53 -13 India
53 54 -1 Turkey
53 70 -17 Russia
And finally, a 3-way plot, guess vs. actual, guess vs. delta, and actual vs. delta. The key is that guess vs. actual is very nearly random.
i.imgur.com
Edward Morbius That's interesting. I have to think about that one. I've seen such apparent trends emerge from noise before, but for some reason I'm having a hard time wrapping my head around this one.
ReplyDeletePeople are always horrible at guessing. Nothing new there.
ReplyDeleteRobert Miller The point is that the BBC's story is shite.
ReplyDeleteRobert Miller It's maybe not so much about guessing so much as it is measuring what people are mistaken about. Knowing that people are sometimes mistaken is one thing. Knowing exactly what they're mistaken about and just how mistaken they are is another. For one, it tells you which things to concentrate on when correcting their mistakes. For another, it might give you clues as to why they support policies that defy the evidence (or perhaps if their knowledge is correct but their decision seemingly is not, then something else may be going on).
ReplyDeleteEdward Morbius I'm still struggling to see a statistical explanation as to why there should be a trend in error in the guess versus the actual value.
ReplyDeleteIt's clear that there's no tend in guess versus actual. There is a rough, positive correlation between the guess and the error in the guess. I do not understand why there is a clear, negative correlation between the actual and the error in the guess.
Where actual delta is high, the guesses are consistently low, where actual delta is low, the guesses are consistently high.
It looks to me like the opposite is true, unless I'm misunderstanding "actual delta". It is entirely possible that I'm being very dense or missing something obvious.
Edward Morbius I think I finally see it and feel slightly silly.
ReplyDeleteLet the distribution of actual values be truly random and delta be defined as (guess - value). Guesses and actual values can only be positive and between 0-100.
If the actual value is 0, the maximum delta is 100 and the minimum 0.
If the actual value is 50, the delta range is -50 -> + 50.
If the actual value is 100, the delta range has to be -100 -> 0.
Hence the lower and upper bounds on delta must vary as a function of the actual value and an apparent correlation results.
Rhys Taylor Bingo.
ReplyDeleteAnd the BBC chose the least meaningful measure (delta) of the three to highlight, in the least useful way possible.
Sorting by actual and highlighting the guess should show that the guesses are little better than random.
Edward Morbius I'm hesitant to ask, but... doesn't the data at least indicate an interesting systematic difference between countries ? Though it doesn't show any measure of dispersion of the guesses, if everyone is guessing randomly within each country, I would guess the mean values ought to be extremely similar.
ReplyDelete... on the other hand, the mean guess is 47 with a standard deviation of 8, and that's hardly a massive variation (I haven't checked how many people they asked).
I think I just answered my own question.
Rhys Taylor The most interesting thing is that the lowest guesses tend to be pretty good ones. Mexico is spot on, Israel and Poland aren't far off. Peru, the lowest guess, is fairly inaccurate.
ReplyDeleteActually, there is one specific delta that stands out markedly, the maximum: 36.
It's for Great Britain.
With which the BBC might have some familiarity.