This image is dedicated to all those Tory small business owners who appear on election programmes to claim that paying a higher minimum wage will mean they can't afford the new VAT charge on their children's school fee (yes, it really happened):
This is the source.
And I stress, whilst correlation does not prove causation, what that downward sloping line suggests is that as minimum wages increase unemployment rates go down. Undertaking such analyses is always difficult, but let's be clear that purchasing power parity has been taken into account in this case. And that makes the link look significant to me.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
I think you mean afford….and yes I saw that clip!
I did, and it’s been changed
No, it doesn’t suggest that – R^2 of 0.054 means no correlation.
I have long felt that such evidence is usually bogus. It places too much evidence on pure data and too little on something far more important, which is our ability to interpret data.
I am well aware that is controversial. I am quite unconcerned.
You think regressions are “usually bogus”? That’s what I expect to hear from the anti-global warming crowd. Not controversial – just silly and anti-intellectual.
No it not anti-intellectual and its crass to say so in social science
Correlation is likely to be a significant and reliable tool in science where data is more controlled, noise less likely and experimentation less likely to be manipulable with replication also being easier. In this case reliance on data makes a lot of sense
Social science is not like that. I made clear that measurement in this area is hard because the noise in the data is considerable and any measure, form assessing PPP onwards is open to considerable doubt. When that is the case judgement on data findings is required. They can at best only mildly reinforce hunches. Correlation coefficients can be as spurious as the quality of the data that underpins them. Thinking they, by themselves, determine an issue is absurd in that case. Indeed, I’d suggest there are few cases in social science where that is true bit think otherwise in physical science
Blind believe in correlation coefficients in social science is to reduce the subject to an absurd belief that people are rational. They’re not. That’s true in their actions. It’s true in their data selection. It’s true in the way they ask questions. And so it’s true in their findings. In that case saying correlation coefficients are evidence is just wrong. They’re a guide, maybe. But so too is a simple best fit line.
No correlation is good! It means that you can increase the minimum wage without increasing unemployment.
On the Pearson correlation coefficient – reducing a data set to a single number is always dangerous! In Physics we do not use it!
Thanks Charles
Whilst increasing wages will increase costs, it will also equally boost incomes and therefore spending power. Some businesses will win through increased sales, some will loose. What never ceases to amaze me in such discussions is that the double entry is frequently ignored. This particularly applies to the propaganda surrounding the supposed necessity for austerity. Never mind the need to teach basic economics we need to teach double entry bookkeeping.
That is on my list of subjects
This was a comment I wrote on Left Foot Forward, replying to someone who had a similar opinion to the small business owner Richard mentioned in his blog-
“Jack,
If the increased taxes are used to invest in the UK economy, the jobs and increased wages produced will likely improve the spending power of the workforce. As such, more people will be able to afford to buy your products/services, meaning overall you increase your profits.
Mainstream economics appears to have “forgotten” a key maxim — employees wages are a companies income. If less people can spend on what you are trying to sell, you’ll sell less of it and make less profit. It’s time Economics remembered these facts”
Very true Daniel
Right on Daniel!
Or as Paul Krugman says:
“Everyone’s’ wages is someone else’s wages”.
This is how the REAL economy works as opposed to the asset and merger economy.
“Blind believe in correlation coefficients in social science is to reduce the subject to an absurd belief that people are rational.”
What on earth is the connection between people being rational and correlation coefficients?
If you don’t get that I despair
Not only do I not get that, but no economist I’ve ever read has got that. Perhaps you need to explain to Piketty, Stiglitz, Krugman et al why their use of correlation coefficients involves a mistaken assumption of rationality?
Not at all
If you think data can answer a social science question in isolation you don’t understand the question
Or the people it impacts
Which is why most of the real economic greats never went near a correlation
They did not need to
It’s safe to say – “There is no evidence higher minimum wages cause greater total unemployment [possibly the reverse].” TOTALLY 🙂
Higher minimum wage seems a neutral factor, so why not just do it to boost the economy! Of course benefits must be carefully considered as part of policy not to leave the worker in an unchanged financial position.
You might want to google “Gibson’s Paradox”.
So? Have you read the Heneral Theory?
How any correlations are in it
(And yes, I have radar Leynes on maths)
Have you read what Charles Adam wrote?
Ciaran
Correlations and all that aside………..
You seem to be in denial about something I am au fait with there in the public sector.
Some of our clients have tried working in low paid jobs but give up (and thus contribute to the unemployment figures again) because after taking into account travel expenses (or trying to run a car) they are hardly making enough money from work to live on.
When we see people get better paid jobs, they tend to stay in the job longer. We know about this because we have to help such people pay their rent to stay in their home. They also spend more in the community and that helps employment there too.
Another observation is that good wages incentivise everyone to work – not just the overpaid CEO’s and fat cats in the banking sector.
Not disagreeing with you in the slightest, Richard, but until we rationalise the cost of that basic necessity of keeping a roof over our heads, a meaningful living wage becomes impossible to set. My finances were transformed when my unemployed husband died and the mortgage was redeemed through insurance.
This is absolutely true Carol. Even if labour build a lot of housing, the prices might not come down that much -it has been suggested by some that house building ALONGSIDE credit controls will be the more effective way of bringing prices down slowly.
see: http://www.rt.com/shows/renegade-inc/387555-gentrification-alternative-socially-cleansed/
from 2.25-4.00
Hi Richard
I’m new here and don’t know much about Tax Research.
How do you set the pay rates for the staff at Tax Research, particularly for unskilled workers, trainees etc.? Is there a method that those of us on the left can ‘copy’?
I have no employees
When I had I always sought to pay above going rates and be proactive with training
It always paid
Let me see if I’ve got this right. You say:
“Higher minimum wages seem to be good for unemployment”
When the chart itself says:
“Minimum wages do not correlate with higher total unemployment”
So the chart is showing there is no correlation *either way* with higher or lower unemployment and minimum wages. So you simply cannot then state there is one – as there is NO evidence to show as much from the data.
You then go on to say that correlation coefficients can be spurious, and don’t work in the social science setting? Because frankly thats the answer you want, rather than what is presented in the data.
I’m really not sure what to take from this. Are you seriously am economist? Any mathematical training at all?
I am a political economist
And I am entirely happy with the conclusions drawn
I have already stated my reasons why
I am no economist but I worked for a healthcare charity for the last 5 years. As a charity we did increase the hourly rate of the lowest paid to the living wage because we are well supported by our local community and we had the resources. Further increases will have a significant impact on the budget of the charity as pay differentials will also have to be addressed. On top of this the removal of the public sector pay-cap may also require funds be diverted from patient care to salaries. I am not saying that the healthcare staff don’t deserve a pay increase but for charities it may mean service rationalisation in the future.
If we had a decent government there wouldn’t be a need for healthcare charities.
Even South Korea who are recognised as providing the best healthcare in the world have medical/healthcare charities.
Once we abandon evidence and reason we have nothing left at all.
I haven’t done either
Don’t be stupid
I just don’t think that’s the only evidence
Of course since the graph leans my way I am tempted to accept it. But I see obvious problems with it. Populations of different sizes are treated equally and that alone, to my mind, invalidates it. And of course the R^2 says that the regression line only “explains” about 5% of the variance anyway.
So while I happen to think that increased minimum wages will lead to better outcomes all round and will pay for themselves, I’m afraid this graph is not evidence one way or the other.
So unemployment is unlikely to be created by higher minimum wages
That’s good, usn’t it?
And as I said, the technique is not perfect. But if it adds some weight to other evidence (increased minimum wages have not caused unemployment based on UK observation) I accept it as part of a whole range of evidence to support what is, anyway, a conviction
I am not suggesting I would ever use it as the sole basis for a case
I thought it interesting
That’s it
I agree with you Richard that the correlation here is not useful.
This, however, is because it just looks at the minimum wage in absolute terms and does not take account of the average wages in the particular country concerned. For example, looking at the extremes Luxembourg has a much higher minimum wage than Mexico per the graph. However, the *average* annual wage in Luxembourg is about $60,000 in Luxembourg but only $15,000 in Mexico. If Mexico adopted a *minimum* wage equivalent to Luxembourg’s the minimum wage in Mexico would rise to approximately $22,500 (12.5 hourly rate for a 40 hour week and 45 *working* weeks per year). Given that is 50% higher than Mexico’s current average wage I would suggest that would have a profound effect on the Mexican economy and jobs.
Therefore, as you say, the data in isolation or used incorrectly sheds very little light on minimum wages and unemployment … and so you can’t draw any conclusions from it. Here the data is being used incorrectly and missing a key variable (median wage in each country).
It is no coincidence that, as a general rule, richer countries are on the right of the graph and poorer countries are on the left.
As I said – these things are hard to do and there is a lot of noise
I said so for a reason….