Very basic politics #4: how do we measure the strength of the economy?

“The economy is the central, critical, irreducible core of this election”, wrote David Cameron in the run-up to the election. “Everything depends on a strong economy.” And although I am not inclined to agree with everything Cameron says, this seems pretty much unarguable. If we as a country want to do the things that civilised countries do — educate our young, heal our sick, feed our poor, care for our elderly — we need money to do it.

872c71b4-364f-49f2-bc45-efcf074441c9-2060x1236

But what exactly do we mean by “a strong economy”? How do we measure whether the economy is strong or not? And whether it’s getting stronger or weaker?

The ubiquitous answer is the GDP (Gross Domestic Product) — the total value of all goods and services produced in a year. As Wikipedia explains, “The pattern of GDP growth is held to indicate the success or failure of economic policy and to determine whether an economy is in recession.”

Obviously, larger countries have more economic activity, and therefore a higher GDP. So we often use per capita GDP, which is simply GDP divided by population size. This gives us a measure of prosperity by which different countries can be compared. For example, the UK’s GDP in 2013 was 2.678 trillion US dollars, across a population of 64.1 million. So per-capita GDP was 2678000000000/64100000 = 41778 US dollars — about £27589.

(£27,589 seems intuitively in the right ballpark for the average income; but remember, per-capita GDP is not about income nor restricted to earners. It’s an average across all the population, including children — so the per-capita GDP of a family of five, if you like, is £137,945.)

how-norway-disrupted-sushi

Is a per-capita GDP of 41778 US dollars good? Yes, pretty good. The United Nations National Accounts Main Aggregates Database is awkward to use: select all countries, then “GDP, Per Capita GDP – US Dollars”, then 2013; download the XLS, load into LibreOffice and re-sort the table. You can see a snapshot of the results in the 3rd column of this Wikipedia page.

The UN uses slightly different numbers from what I calculated above, though it’s close. Its results show the UK as 23rd out of 194 countries (or 32rd out of 241 areas). Monaco is top, with $173,377 per person, followed by Liechtenstein and Luxembourg — all tiny countries with disproportionately wealthy citizens. Perhaps more surprisingly, Norway is fourth, with $103,585 per person. (Somalia is last, with $133.27 per person.) Our $42423.40 per person puts us about the same as France, a little way behind Germany and Western Europe as a whole, and some way behind the USA ($52391.85 per person).

We’re doing well in the UK.

So the key question is this: if the UK economy is strong enough to be generating £137,945 per year for each family of five, why is it that a typical family of five lives on perhaps a quarter to a fifth of this?

unequal-sushi

The answer of course is inequality. Some people have much more than others.

I don’t for the moment want to get sidetracked into the question of how much inequality is acceptable. I am not a communist, and I agree with Paul Graham that a society needs to allow at least some inequality as an incentive. But I can’t accept a society where some people don’t have enough to live on, and that is pretty clearly what we currently have in the UK. Child poverty figures are so appalling under the current government that the government’s response is simply to scrap the child poverty targets. And this is happening despite consistent year-on-year growth in GDP which tells us that the economy is performing.

So here is the problem: GDP tells us that we’re doing well, but the reality is 3.7 million children in poverty (for some definition of the word) and just over a million families forced to rely on food banks in the last year.

And my conclusion: GDP measures the wrong thing.

When we want to determine whether a country is succeeding, it doesn’t suffice to ask whether its total economic activity is increasing. On that basis, if a thousand low-paid families each make a thousand pounds less this year, while a billionaire makes an additional two million, the GDP increases by a million pounds — which looks good — but by any rational measure the actual wellbeing of the country has reduced.

main_img_1409879014_RiceBowl_b_1

So I think we need to measure something else.

But what?

Those of you who actually know any economics might be able to tell me whether there are already better measures out there, which are not yet widely used but ought to be. But in the absence of an existing Right Answer, here are two approaches that occur to me.

Proposal 1. Sum the square root of incomes

GDP is, roughly, equal to the sum of everyone’s income (since economic activity generally results in someone getting paid). But the true value of increasing a low income is much higher than that of increasing a high income by the same amount. (A thousand pounds is a glorious windfall to a low-paid family, but just pocket-money to a billionaire.) So instead of summing incomes, maybe we should sum something like the square roots of incomes? Let’s call the result sqGDP for now.

Worked example: consider that £1000 increase in income, added to a family whose income is £10k and to a billionaire whose income is £1,000,000. In the first case, the contribution to sqGDP grows from sqrt(10,000) = 100 to sqrt (11,000) = 104.88 — a growth of 4.88. In the second case, the contribution to sqGDP grows from sqrt(1,000,000) = 1,000 to sqrt(1,001,000) = 1000.50 — a growth of 0.50. On that measure, the increase for the low-paid family is about ten times as valuable to the country, which seems to be in the right ballpark.

Obviously this measure could be tuned: we don’t need to take the square root of incomes — we could take the 2/3 power, or the 1/3 power, or whatever we felt gave us a result that best reflected what we want to achieve.

Proposal 2. Incorporate an existing measure of inequality

There are plenty of existing measures of inequality. The most common is probably the Gini coefficient, which varies from 0.0 (when everyone has the same) to 1.0 (when a single person has all the wealth). So we could use a measure which combines GDP with Gini in some way.

How, exactly? Well, we could just divide GDP by Gini, but that would give us a measure where exact equality is seen as desirable, which we probably don’t want. We’d need to be a bit cleverer — perhaps divide GDP by one plus five times the Gini, or something. Again, this can be tuned to give the results that best correspond to what we actually want to see.

In a sense the details are unimportant. Because …

live-sushi-1

Conclusion

What matters, I think, is getting away from the mindset that GDP is a good measure of how an economy is performing. If we move to a better measure, or even seriously discuss moving to a better measure, then doing so will force us to think about what we’re actually trying to achieve with our economy rather than giving carte blanche to those who are happy just to keep increasing the wealth of the top 0.001%.

We are a rich country by any measure. A rich country shouldn’t have poor citizens. It certainly shouldn’t have millions of children in poverty, by any definition of the word. Measuring GDP (and therefore optimising for it) is one of the things that leads to this. So let’s stop.

27 responses to “Very basic politics #4: how do we measure the strength of the economy?

  1. Thought-provoking stuff; thanks Mike. One nitpick: you’ve occasionally written “GPD” instead of “GDP”.

    [Mike says: thanks for spotting that, Andrew; now fixed.]

  2. Pingback: Interesting Links for 02-10-2015 | Made from Truth and Lies

  3. We are a rich country by any measure. A rich country shouldn’t have poor citizens. It certainly shouldn’t have millions of children in poverty, by any definition of the word

    Well, except if you define poverty in relative terms relating to average incomes then you will always have millions of children in poverty, because in a country of 70 million there will always be millions of people earning below the average (unless you pay everybody the exact same, whatever they do).

  4. (In fact I’m pretty sure that if you measure poverty in that way then it would be theoretically possible to decrease inequality while increasing poverty: if you give, say, the absolute worst-off families an extra £100 a year, you will reduce inequality; but you will also raise the average income, and therefore a load of people who were previously above the poverty line will now be below it. Result: less inequality, more poverty. Is that good or bad?)

  5. Most actual relative poverty measures are satisfiable without paying everyone the exact same.

    If your relative poverty measure is ‘less than 50% of average income’, and you have one person on £100k and ten people on £10k, then you’ve done it – the average there is just over £18k, and half of that is just over £9k, which is less than the lowest salary – even though the highest salary is still 10x the lowest.

  6. Most actual relative poverty measures are satisfiable without paying everyone the exact same

    Well, okay, replace ‘paying everyone the exact same’ with ‘forcing incomes to be within an unrealistically strict band’.

    It doesn’t have to be the exact same, but you can’t do it with reasonable real-world levels of variation that take into account relative supplies/demand ratios for different skills (eg, someone with a rare, highly-in-demand skill is clearly going to command a price not ten times but possibly one hundred or more times as much as someone whose only skills are such that there are more people who can do the job than positions needing filled).

  7. … and in fact no, actually, that’s not the problem with your example, your problem is that you’ve forgotten the middle class and only included those at the top and the bottom. If you have a more normal income structure, with a few people at the top, a few people at the bottom, and most in the middle — which we could represent in your example by adding in, say, 30 people on £25K — then you have poverty again.

    (Hm, forgetting about the middle class and thinking everybody is either poor or rich, that seems to be exactly what Labour’s doing, isn’t it? And given that the vast majority of people are neither poor nor rich but somewhere in the middle, it’s hard to see it being a winning strategy electorally)

  8. I think most measures of poverty are at least a little more sophisticated than that. For example, the Child Poverty Action Group that I linked to uses this definition: “This survey sets the poverty line in the UK at 60 per cent of the median UK household income. In other words, if a household’s income is less than 60 per cent of this average, HBAI considers them to be living in poverty”. Note the use of median rather than mean — a harder measure to game.

  9. I think something like your Proposal 1 is the Right Thing. I suggest summing log(income + smallish_constant) instead; I think it’s generally been found that the relationship between money and other measures of wellbeing is roughly logarithmic. (Probably even slower than logarithmic, at the high end, but logarithmic should be close enough up to, say, £1M/year.)

    I suggest a smallish constant because a person’s welfare isn’t entirely down to the income they earn; if you lose your job you don’t generally actually starve to death. There are state benefits and family and friends and charities. But they don’t provide a huge amount and no one wants to be depending on them if they can help it; the right constant might be something like £3k/year.

    A couple of caveats. Firstly, wealth matters as well as income. Someone with £1M in savings and an income of £10k/year is much, much, much better off than someone with no savings and an income of £12k/year, even if we assume that economic growth has stalled and those savings don’t grow or produce dividends or interest or anything. Secondly, lots of valuable things go on that don’t appear in measures of income. Imagine (not super-plausibly, but never mind) two families in which both parents work; in fact, one parent in each family works as a childminder or nanny or something for the other family’s children. One day they come to the staggering realization that they can just look after their own children. Suddenly, much less money is changing hands, but actually everyone is a little better off in practice. Any measure of aggregate income will have gone down, but the decrease is purely an artefact of measuring only formally paid work.

  10. If simplicity were no object, I think we would approach the problem of defining poverty along the following lines. There are two aspects to poverty. First, being unable to afford basic necessities like food and shelter. Second, being much poorer than others around you, which (1) tends to be demoralizing and (2) may mean disproportionately little access to scarce things like political influence.

    Accordingly, if there’s to be a poverty threshold then it needs to take into account both typical other-people’s wealth and absolute wealth. Perhaps something like threshold = max(cost of living a minimally-decent life, 1/3 of median income).

    It isn’t true that such a notion of poverty (even without the component that deals with absolute wealth) necessarily makes lots of people poor unless income is staggeringly evenly distributed. For instance, if you define it in terms of median income, you can make the upper half of the population as rich as you please and it won’t increase the nominal amount of poverty.

  11. T, it’s quite correct that “relative” measures of poverty will identify poverty whenever you have smallish numbers of people who are much poorer than a “middle-class” majority. But why shouldn’t they? I can think of a few possible reasons, but I find them all unconvincing.

    “If you measure poverty that way, then no matter what you do there will be millions of ‘poor’ people, which just indicates that your metric is too pessimistic.” This isn’t true. Let’s suppose we define “poor” to mean “below 1/3 of median income”, and suppose (not very realistically) that incomes are uniformly distributed in the range from 0 to some maximum which I’ll call 1 unit.Then median = 1/2, poverty threshold = 1/6, average income of the poor = 1/12, and getting all of those people just barely out of poverty costs 1/6 of median income * 1/6 of the whole population. The richest 1/6 of the population in this model have an average income of 11/6 * median, so an extra 1/11 ~= 9% income tax on those people would do it.

    (Note 1. In the real world, the income distribution is somewhat different, but the extra tax burden to get everyone out of poverty with that sort of definition would still be of that order of magnitude. Note 2. I am not claiming that that’s actually the best way to get people out of poverty; the point is just that this sort of definition doesn’t necessarily lead to so much poverty that fixing it is unthinkable.)

    “If you measure poverty that way, then we can make everyone better off but increase the amount of poverty.” Yes, that’s true. For instance, if 75% of the population are earning X and 25% are earning just a teeny-tiny fraction over X/3, and then you give the top 75% a 5% pay rise and the bottom 25% a 4% pay rise, you suddenly have 25% of your population in poverty. But all this means is that you need to be careful about definitions with hard thresholds in them. Which, indeed, you do, but outside such contrived examples they don’t usually work so badly.

    “If you measure poverty that way, then in any realistic society there will be some poverty.” That may well be true. But why’s that a problem with the definition? In any realistic society there will be some cancer; is that a problem with the definition of cancer? In any realistic society there will be some criminals; is that a problem with the definition of crime? Not so far as I can see.

    “If you measure poverty that way, then poverty is not something I see much reason to care about.” The idea presumably being either that relative wealth shouldn’t matter, and that being below (say) 1/3 of median income isn’t a problem if the median income is nice and high; or else that relative wealth should matter, but the point at which we should start worrying is much lower. To these I say (a) that relative wealth does matter, because prices adjust themselves according to what people can pay and if you’re much poorer than most you’re going to find yourself unable to afford things that everyone else can, which is demoralizing and possibly worse than that, and (b) that until such time as our societies get far richer than they are now, a figure like 1/3 of median income is actually really bad and one actually can’t live decently on it. I’ll be very happy to consider revising my opinion some day in the future when the median person is much better off than at present, but that day is not now and doesn’t seem likely to come soon.

  12. Lots of good points there from the enigmatically named g. I agree with pretty much everything he or she says here, and especially regarding the importance of wealth as well as income. Taxing wealth seems to be harder to do: I’d be interested to know what wealth taxes actually do already exist in the UK (or indeed elsewhere) and how they’re working out.

    On what should count as “poverty”: it’s true that any numeric definition will have unsatisfactory edge-cases, and we could probably have a fruitful discussion on which definitions yield the least-worst representation of reality. That said, there are plenty of poverty specialists who have gone through the same sequence of thoughts in much more detail than us, so it may be that the Child Poverty Action Group’s definition is actually better than we realise.

    What is indisputable is this: in a country where the average economic value generated by every man, woman and child is £27,000, we have people literally dying because they can’t afford actual necessities like housing, food and heating. Any measure of a country’s health that doesn’t register that fact is inadequate.

  13. For example, the Child Poverty Action Group that I linked to uses a defintion of “if a household’s income is less than 60 per cent of this average, HBAI considers them to be living in poverty”. Note the use of median rather than mean — a harder measure to game.

    I don’t know what you mean by ‘game’, but I don’t think switching from mean to median makes much difference. As most people are middle-class, then the median must be one of the middle-class incomes (by definition, if a particular contiguous subgroup is more than half the population, the median must lie in that subgroup). So the median will be a middle-class income, and anyone lower-class who earns less than 60% of that will be ‘in poverty’.

    In the Very Simple Example of one person on £100K, thirty on £25K, and ten on £10K, the median income is £25K and all those on £10K are ‘in poverty’ (regardless of their actual standard of living).

    in a country where the average economic value generated by every man, woman and child is £27,000, we have people literally dying because they can’t afford actual necessities like housing, food and heating.

    Do we? Seriously? Honestly? There are, in Britain, otherwise-healthy people (ie not, for example, people who are homeless due to having mental health issues or being addicted to drugs) literally starving to death?

    I am sceptical. I would like to hear actual concrete examples of this happening.

  14. A common measure during the Cold War was in terms of what a median hour of labor could buy. Most people make their living by selling hours of their time, so the hour is basic fungible unit. Ever since the start of the Reagan / Thatcher era an hour of labor at median pay has bought a smaller and smaller share of the GDP, that is, the sum total of all goods and services produced by that work.

    In other words, work has been increasingly devalued. Even as nations have gotten more wealthy, workers have seen none of that wealth. This doesn’t mean that they are starving, but that they are getting paid less than their worth. The Soviet Union had this problem. It was easy to get a job, but the pay wasn’t worth very much. No one was starving, but living standards were flat, even as GDP increased. The old saying was “They pretend to pay us, and we pretend to work.” The effect was pernicious. It is a pity that the West chose to copy the Soviet model rather than sticking with the more successful model of the 1950s and 1960s.

  15. They are not difficult to find

    I have read all of those. The only example of someone starving to death that I could find is Mark Wood, who had an eating disorder and ‘food phobia’, and whose GP apparently described him as ‘an extremely fragile individual’.

    I still would like to hear of a single concrete case of an individual in Britain recently who has died, where the cause of death was starvation because they could not afford to eat.

    That is, people without eating disorder, or food phobias, who would eat if they could, but starve, literally starve, to death, because they just do not have the money to buy food.

    I honestly don’t believe such a thing could happen in Britain today. Surely before someone actually starved to death, they would end up in hospital having fainted or something; they would be diagnosed as suffering from malnutrition and the hospital would feed them up before discharging them, no?

    (There’s also ‘David Clapson, a diabetes sufferer who was found dead from acute lack of insulin after his benefits had been stopped’ but I find that odd as I know from personal experience that people suffering from Diabetes do not pay prescription charges, so he should have been getting his insulin free — so I am not inclined to give that report credence without more details about how he got into that state).

  16. ” Surely before someone actually starved to death, they would end up in hospital having fainted or something; they would be diagnosed as suffering from malnutrition”

    So in short, were I to lose my job and spend all my meagre savings before I found another one, I might not actually die because the chances are I’d be admitted to hospital for severe malnutrition? And to think I was worried about that…

    NB Should anyone want to know what actually happened with David Clapson, ie not T or Iain Duncan-Smith (assuming them to be different people), if you google his name the top find says in its first paragraph “without…his jobseeker’s allowance he couldn’t afford to eat or put credit on his electricity card to keep the fridge where he kept his insulin working.”

  17. So in short, were I to lose my job and spend all my meagre savings before I found another one, I might not actually die because the chances are I’d be admitted to hospital for severe malnutrition? And to think I was worried about that…

    The claim was:

    What is indisputable is this: in a country where the average economic value generated by every man, woman and child is £27,000, we have people literally dying because they can’t afford actual necessities like housing, food and heating

    That is, specifically the claim was made that there are people literally dying because they can’t afford food. And not in a way that makes it look like hyperbole to make a point, but in a way that makes it look like this is something that actually happens, if not routinely, at least more than once.

    And yet apparently it doesn’t happen. I have not yet seen one single report of someone otherwise healthy who has ‘literally d[ied] because they can’t afford actual necessities like housing, food and heating’.

    Now, clearly there can be disputes about whether there are people in Britain whose standard of living is unacceptable, and what constitutes an acceptable standard of living, and so on. But there are not people ‘literally dying’ because they can’t afford food, and that claim should be withdrawn.

  18. This would not be a clearer case of sea-lioning if you were bouncing a rubber ball upon your nose.

  19. This would not be a clearer case of sea-lioning if you were bouncing a rubber ball upon your nose

    Well, yes, but that makes no difference whatsoever to the validity of the point, does it?

    Are there, or are there not, people in Britain today dying because they can’t afford food? Yes or no?

    If yes, where are they? If no, will the claim that there are be retracted?

  20. Did Michael O’Sullivan starve to death because he couldn’t afford to buy food? That article doesn’t say he did.

  21. Pingback: How to measure the strength of the economy, redux | The Reinvigorated Programmer

  22. Pingback: Why I’m jealous of my eldest son | The Reinvigorated Programmer

  23. Instead of optimizing for some complex metric that depends on everybody’s income, why not instead just optimize for the wealth of the poorest person in the country? I’ve written a blog post that explores this idea at http://www.reenigne.org/blog/the-mission-of-government/ .

  24. An intriguing idea, Andrew.

    At any rate, it certainly fits my overriding point, which is that we need to stop blindly talking as though optimising GDP is obviously the right thing.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s