There is a NEETs crisis – but it is being deliberately created by the government
Sure, higher min wage, no firing of dullards, more taxes on employing someone, too many degrees and not enough training, hte nationalisation of apprenticeships. Sure, all being deliberately created.
The analysis following that title somewhat fails of course:
The fact is that the entire foundation of UK macroeconomic policy rests on the belief that some unemployment is necessary for inflation to be controlled. Neoliberal economists call this the NAIRU, the non-accelerating inflation rate of unemployment. Behind the technical language is a very simple proposition. The economy must always contain a pool of people who cannot find work.
Cretin. The point of Nairu is that changes to the microeconomic structure of the economy – wages, taxes on emplyment, ease of hiring and firing, all that sort of stuff – change what the Nairu is. If you’ve a high Nairu then you’ve a stultifying micro-set of rules. The answer is to change those rules.
Entirely possible to have a Nairu at the 2% or so that is really zero unemployment – gotta give people time to change jobs etc – but that does mean having a red in tooth and claw free market employment market.
The first victims are usually young people. They have the least experience. They are the newest to the labour market. They are the easiest to exclude. They have little chance of complaining because they are, economically, just about the weakest group in society, and, when many have few dependents, they are also considered the easiest group to sideline. When economic opportunities contract, they are pushed aside first. The NEETS issue is then deliberately constructed. It exists as a policy choice.
So, vicious free markets to aid the young!
The Churchill/Eden/MacMillan government in the 1950s reduced unemployment to *less than* 2% showing that frictional unemployment can be less than the 2% assumed by economists.
I suspect that may be an artifact of how they were counting.
Rather like the current rate, there’s a lot of provisions that discount people not working but not officially unemployed.
Unemployment stats are always heavily affected by the way the statistics are compiled. Because it is based on the denominator of “people in the labour market”, and so people not in work and not seeking to work don’t count. That can still be a useful stat BTW and is economically relevant. But there’s also some subjectivity there. How hard does someone have to be looking? What about people who would be looking if wages were higher and chances of success finding a job were greater, but are simply discouraged from trying to find work by the current state of the labour market?
So it’s best not looked at in isolation – it should be viewed together with employment rate (where the denominator is all working age people, regardless of whether they see themselves as part of the labour market or not) combined with average hours worked (to deal with part-time workers and especially those working just a few hours per week).
The situation differed from today in that there was no national minimum wage and the dole was much less generous, so some jobs that would be uneconomic today were still worth doing. There was also the expectation that mothers with small children would stay at home or work part-time so that they could look after them,
Nevertheless it showed that frictional unemployment could be less than 2%.
Unfortunately they didn’t have use surveying then that allowed the employment rate to be calculated, those modern stats only go back to 1971. https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/timeseries/lf24/lms
Reconstructed rates for the 1950s seem quite similar to the present, in the low to mid seventy-something % range. As you say it varied between men and women though. https://www.ons.gov.uk/economy/nationalaccounts/uksectoraccounts/compendium/economicreview/april2019/longtermtrendsinukemployment1861to2018
Your point about frictional employment being low, about 1%, is correct, I’m just looking at the overall employment rate for more context.
They had data on the unemployment rate, which is that referred to be NAIRU.
The overall employment rate tells us quite a lot of useful information but is almost entirely irrelevant concerning NAIRU.
In the 1970s (and, to a lesser extent in the preceding post-was decades) the primary driver in inflation was the leapfrogging demands by union leaders to transfer the majority of the benefits of technological progress by ICI and Hawker Siddeley and … into the pockets of *their* members rather than the pockets of members of other unions (let alone those who made the technological improvements). NAIRU was the level at which enough employers were emboldened to refuse the most outrageous demands because enough workers were scared of losing their jobs to vote against a strike. The number of housewives doing a part-time job had a negligible impact.
But this all depends a lot on what you mean by “unemployment”. There’s the definition economists want, and there’s the stats that actually exist. The UK Labour Force Survey only came into being in 1973 (and only became annual in 1984, then quarterly in 1992). The unemployment stats created from the LFS are based on the ILO standard so should be internationally comparable, in theory. The unemployment data from the 1950s was not based on these standards and caution needs to be taken comparing those results to today’s figures. This is M’s point and it’s a valid one. The relevance of the employment rate statistics is they show help how the structure of the labour market has changed over time and are immune to definitional changes and discouragement effects in the denominator of “unemployment” – but even there, the 1950s data is reconstructed so care is needed.
https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/methodologies/labourforcesurveylfsqmi#methods-used-to-produce-the-data
https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/methodologies/comparisonoflabourmarketdatasources#unemployment
I don’t know if the ONS has any longer term figures for the male-female employment gap but it’s come down a long way even since the 1970s.
The graph should for men aged 16-64 and women aged 16-59 in the earlier years. Employment for women aged 60-64 when retirement age was 60 was relatively infrequent.
On the contrary, capturing the fact that the employment rate among women was suppressed by them tending to stop work earlier is part of the whole point of the employment rate statistic. You want its definition *not* to be affected by policies or social trends at the time, so changes in employment rate due to shifts in those policies or trends actually get reflected in the figures, rather than erased by the redefinition.
In the range of this data, the education leaving age used to be 14, then 15, then 16, and now 18. But for the same reasoning, the lower age limit shouldn’t be fiddled with over the timespan either. Setting the lower and upper limits has a bit of subjectivity because it’s basically “when is it reasonable that these people were of an age they would likely be physically capable of work”, and these days we are less likely to count eg a 14 year old in that camp, but are more accepting (regardless of the current policy) that 16 year olds could work if needed. Perhaps in 20 years time these stats will be recompiled with 18 as the lower age instead, because the idea of 16 year olds working seems antiquated. But then that would, and should, be applied to the whole data set. Just like we don’t use the age 14+ figures even for those periods kids often left school that age.
It may be the whole point to you and Harriet Harman but to normal people who want comparisons on a like-for-like basis 16-59 gives like-for-like gender comparisons and 16-64 does not. It is comparable to comparing driving speeds on two stretches of road that are otherwise similar but one of which includes 5 miles of a 40 mph zone.
If I had a travel app that told me I could reach my destination at the same time by either of two routes, I wouldn’t be best pleased to arrive at my destination several minutes late only for my app to inform me that what it really meant was “the same time, relative to the fact you travelled on the route with the lower speed limit. After compensating for that, your speed and hence arrival time was the same.”
Don’t lump me in with bloody Harperson, fgs. She’d be banned for speeding anyway.https://www.dailymail.com/news/article-160050/MP-Harman-gets-driving-ban-speeding.html
Suppose there was a year where the 16-59 employment rate was 75% for both males and females, and for 60-65 it was 60% for males and 20% for females. Overall who has the higher employment rate and what’s driving it? The answer is obviously “males, because although at 16-65 their rate is the same, there is a much bigger drop-off in female employment after 60.” And what policy levers might shift that? Well whatever is causing women to stop work at 60 (eg if government raises the female retirement age) or taking the French approach, whatever is causing so many men to keep working after 60 (so we could cut the male retirement age). But your supposedly “like for like” analysis would say males 18-65 average less than the 75% for females 18-59. Even worse, by your logic that is being driven by so *few* males over 60 working, and inequality would fall if more males over 60 worked. Which is completely the wrong way round.
Yes it was a social norm for women to retire early. Or even not return to work after having kids. Just like it was once a social norm for most kids to leave school at 16 and get a job. Most 16 to 65 year old over that time frame would be physically capable of work. It’s a matter of social norms, government policy and individual choices whether they worked or didn’t. So, what proportion of those who could have worked actually did so, and how did that proportion change as those conditions changed? That’s a completely natural question for a “normal” person to ask. Massaging the data to try to “adjust for norms at the time” is hiding the very trends we’re interested in, introduces discontinuities into the data where you change the adjustment applied, and risks the kind of manipulation that M points out the unemployment figures are historically notorious for (not as bad these days we have the Labour Force Survey and the ONS sticks to ILO definitions). I’m sure the idea has crossed many a wonkish mind that they can cover up the NEET problem by arguing “these days the social norm is for people to go to uni, so on a like for like basis we should now remove 18-21s from the data”.
If there was no unemployment and every man and every woman entered the jobs market on leaving school and women retired at 60 at 60 but men retired at 65 the employment rate for women on your basis would be 91% and that for men would be 100%. To base anything, let alone economic policy on a completely artificially-constructed difference in *reported* employment rates would be stupid.
Do you *want* to encourage stupidity?
…deliberately created..?
Surely not: it results from ignorance of higher-order effects, rather than intention.
True… in ‘Cockup v Conspiracy’ it’s nearly always the former (especially in government).
When many economists tell you what tends to happen if you make things hard to do and you say it’s gonna be different this time it goes beyond cockup into malice. Base your attribution of malice on what you can see. not Hanlon’s razor.
Surely not: it results from ignorance of higher-order effects, rather than intention.
They were warned but went ahead anyway, so I don’t think we can attribute ignorance. Arrogance possibly – this time it will be different – or indifference because their union paymasters are more interested in those in work and paying dues.
Genuinely a harsh world for new graduates. My daughter graduated last year with an actual genuine skillset and emerged into a profession that’s being gutted by AI and has shed 25000 jobs over the last year or so. She’s retraining
What rotten luck. It’s always been a gamble, training for any demanding trade on the hope it’ll thrive until you retire, but one year is awfully hard. How has she chosen what to retrain in?
I have a feeling that more tax would be the answer to this little problem wouldn’t it ?
It’s MOAR tax!!!!
Isn’t that the answer to EVERY problem????
Him being made emperor of the world comes up a lot too.
But not anybody else, because they can’t be trusted to administer everything in exactly the right way. Might be neoliberals, or populists, or fascists, or well-meaning socialists who wrongthink about MMT, or the purely incompetent.
Nevertheless centralising power into the hands of government is obviously a good idea and will never mean it ends up in the wrong hands, with the possible exception of all times since 1979 when we are repeatedly told it’s ended up in the wrong hands.
Yup. And when for Murphy even Gordon Brown is a ghastly neoliberal, he’s got problems.
It’s almost a pity Corbyn didn’t become PM, just so we could see Murphy complaining about him being a right-wing neoliberal propped up by the Jewish press.
What an excellent illustration of the difference between
1) observing and acknowledging reality (“NAIRU exists” or “minimum wage causes unemployment”)
2) trying to pretend the world looks like you want it to (“I don’t want NAIRU to exist / minimum wage to cause unemployment and anyone who talks about it is evil)
Accountants and lawyers will soon cease to be necessary unless the lawyers make it illegal for lawyers to be replaced. That won’t need to happen in the UK if Labour are in charge much longer. There won’t be enough leccie for AI and Sharia Law doesn’t need lawyers…
Two of mine are lawyers and they seem to think AI will help rather than hinder them. But who the fuck knows?
Currently AI has a nasty habit of making up cases to support its argument, which makes it less than useful. But presumably it’ll pretty quickly learn not to do that.
I don’t think an LLM like ChatGPT or Claude exactly *can* learn to “not make stuff up”, the way they’re designed. Whacking plausible-looking strings of text together is basically the whole idea of how an LLM works. Its factual accuracy is a whole other thing.
Rather than “teaching” the AI not to hallucinate, I think the direction of travel is building scaffolding that reins the AI in a bit, gives it a structure to work with, and uses different types of AI, doing different specific jobs, to work together. Then selling the thing as a specialist product. So just asking an LLM for legal advice is an iffy idea. Using an LLM to read and summarise a case will generally work better. Perhaps you use a different kind of AI to recommend which laws or cases are most relevant to a given situation, then those recommendations get sent to the summariser. Then the thing you “talk to” on the front end is also an LLM, but it translates the situation you describe into a carefully crafted search query for the recommendation AI.
From my admittedly poor understanding of AI and how it is trained a model that is specifically designed for a task and only trained on relevant material if far better than general AI that is trained on anything they can get their hands on. So an AI model only trained on legal cases is a lot less likely to hallucinate cases.
I don’t understand why the Guardian doesn’t train AI solely on its contributors’ columns and produce the same nonsense day after day without needing to pay anybody. The readers would never notice so long as the correct tone of whinging remained.
You’re making a big assumption there…
Assuming you mean LLMs specifically: training on a relevant data set helps but doesn’t cure the problem. It’s a text prediction engine, not a knowledge engine. It doesn’t even have the concepts of “memories of true facts” so has no idea that it is hallucinating when it strays off the reservation. It just outputs a string of text tokens that seems “most likely” to follow the input prompt.
Ask it about a case that exists and it has been trained on, and it’s likely the output will be reasonable. Ask it about one that simply doesn’t exist, and it will still write the most plausible text output it can come up with, regardless of whether the whole thing is a hallucination. Even if there wasn’t such a case it may feel like there “should” have been, and it will be given the most “likely sounding” name. Can even generate a complete citation which turns out to be rubbish when you look it up.
This is why forcing the AI to actually perform a search is currently the better strategy, especially if they can search a well curated database, then get them to summarise the findings from that. At some point we will get something that superficially feels like the current LLMs but has much more capacity for introspection about what it’s generating and whether it matches known facts or not, but this would represent a significant technological change.
I ought to hedge this slightly, that some LLMs have been coded to have a degree of self awareness that something might be wrong when they detect the output probabilities for the choice of next word become low. In that situation they can even be trained to respond that “that information isn’t my training data” and depending on the model even ask if you want to do a search instead. But that reply is actually a bit misleading.
The AI doesn’t really know what’s in the training data or not, it doesn’t have a “memory bank”. It may have access to certain facts about the training data, eg it only covers a certain date range and you’ve asked it for something more recent. But in general it cannot know whether a specific piece of data was included in its training or not. It has no database of training data available to refer to. Everything it “remembers” about the training data is just the result of training reshaping the neural network. When LLMs were first discovered, researchers were surprised by just how much the LLMs can recall (in fact even in recent months there have been surprising discoveries about this) – it shows the neural network is acting as a really extreme form of data compression. But the compression isn’t perfect, so many facts were given in training data but cannot be recalled, and topics about which little data was provided often get recalled in a somewhat distorted way. But how to tell that apart from another topic with a minor weighting in the training data but which is recalled with greater accuracy?
There are some fairly accessible papers about this. Good search terms include hallucinations, communication of uncertainty, self awareness of uncertainty (not the same thing as whether the AI communicates this uncertainty to the user), and whether AIs “know what they do not know”.
Incidentally, Rumsfeld got absolutely slaughtered in the popular press for his “unknown unknowns” speech. Was given the Foot In Mouth Award for it, even. But the “Rumsfeld matrix” of known knowns, known unknowns, unknown knowns, and unknown unknowns is the frame that a lot of analysis of AI self-awareness of limitations of its knowledge is based on. https://en.wikipedia.org/wiki/There_are_unknown_unknowns
Unknown knowns weren’t in the speech but they’re obviously a real thing – stuff the AI is capable of recalling but cannot vouch for, so doesn’t “know” that it knows it. And while I hate to credit this guy, Slavoj Žižek pointed out the logical omission in the original context: “If Rumsfeld thinks that the main dangers in the confrontation with Iraq were the ‘unknown unknowns’, that is, the threats from Saddam whose nature we cannot even suspect, then the Abu Ghraib scandal shows that the main dangers lie in the ‘unknown knowns’—the disavowed beliefs, suppositions and obscene practices we pretend not to know about, even though they form the background of our public values.” Wouldn’t agree it’s always the ‘main danger’ but obvs there’s stuff people ‘know’ but don’t feel able to bring to the discussion table, even if they really should. Perhaps because if asked to justify it they don’t have anything other than gut instinct, or they’re estopped because admitting it is taboo, offensive or reveals something unseemly about the speaker or their audience. One of the dangers of rendering certain things, important but sensitive, into the category of ‘unsayables’ is that at least in terms of public debate they become ‘unknown knowns’.
The lawyers will make it illegal for lawyers to be replaced with AI. Also they fight very hard to keep archives of case law out of the clutches of AI and keep it unavailable for an AI (other than theirs) to trawl by whatever legal means they can devise.
“The economy must always contain a pool of people who cannot find work.”
With any luck,.the entire political class will be joining this pool soon.
Pool? For these creatures cesspool surely?
It’s not “must”, it’s “does”. Again, he’s confusing observation with intent.
I’m struck by his tendency to assume that everything is planned & coordinated by who, exactly?
Somehow, in his vision, an employer decides to lay off a young worker with no dependents, not the person who is least productive – srsly?
We also frequently told that it’s the older, experienced & more expensive workers who are first to get dumped in favor of the cheap ones.
Pick a lane mate
Again, confusing a measurement with a control. “This thermometer reads 38C, we must enforce a change in thermometers”
Why does UK have macroeconomic policy?
It is remarkable. In the greatest period of our greatest growth and, frankly, greatest everything else, and we did not have government departments for transport or housing or health or culture (lol). In fact we had government departments for not much domestically and those departments we did have were staffed by blokes called Sebastian and their nepotistic uncles and all they did was worry about theft by Germany of the plans for our latest submarine.
And as for the idea that the finance minister ought to ‘run’ the economy, or that the government should have a ‘plan’ for the country, most people would quite rightly have thought it was mad to suggest such things.
Fuck me, when I consider what we have given up …
Attlee. And JMK.
Two excellent Speccie articles about this:
https://archive.ph/Ei2Zl
https://archive.ph/iSzxa
From the first:
There has never been any democratic consent for mass migration. Indeed, the British electorate have voted against it again and again. Now, after decades of being ignored, we face becoming a minority, marginalised and ultimately replaced in our own country. That is wrong. The Scots, Welsh and English, have been here for at least 1,500 years. It is perfectly right to wish to remain a majority in our home. To that end, merely slowing the pace of replacement is not enough. Recent arrivals like the Boriswave must return home.
There it is. In the MSM.
Interesting. As you say, in the MSM. About time too.
However, welfare cannot be reformed (even if the other, hostile parts of the British state can be reprogrammed).
We should know this by now. Every few years a new ministerial broom comes along to tackle whatever welfare outrages have found their way into the popular press, and he introduces some new change with a new moniker (‘working credits’, or some such) and for a year or two it stymies many of the dole bludgers. But they find hacks and before you know it we’re back to where we were. Because it’s free money and of course there is a thing called moral hazard.
There is no eliminating the central moral problem of welfare. It’s been tried and tried and tried and it cannot be done.
All we can do is decentralise welfare so as to increase stigma, improve accountability with local knowledge and, ultimately, ensure that bludgers bludging is limited by lack of funds to hand out.
A return to the poor house, and parish charity in other words. Some deserving cases will of course slip through the net. But that’s already happening even with runaway welfare spend.
The benighted are always with us, and they’re far worse than people claiming to be poor.
Let Them Crowdsource.
Charity is not a legitimate function of government.
A return to the poor house, and parish charity in other words. Some deserving cases will of course slip through the net.
Agreed. ‘Cept that “net” stuff. No one “deserves” government money, unless they provide adequate consideration.
I was about to say I’m not sure I’d describe the Spectator as “mainstream”.
But I checked, and its circulation is 114,911. The Guardian last reported its circulation as 105,134, some years ago, and is thought to have dropped well below 100,000.
This guy.
The first people to be shown the door are the *really experience older guys*. Because they’re expensive. They will be replaced with some recent college grads – or more likely a call center in Bangladesh.
NEETs aren’t NEETs because they got fired. They’re NEETs because the government is so generous it gives young people all they need to eke out an existence without working – and young people don’t have a lot of needs so they take it. This trains them to constrain their desires and ambitions . . . so they *stay* NEETs.
If you train people that all they can expect is weed and porn, that’s all they’ll ever want.
The government gives *everybody* (except the self-employed) all they need to eke out an existence without working – that is the purpose of the welfare state: to look after those unable to work, whether through illness, injury or economic disaster.
My pensions provide enough for me to eke out an existence without working: that is the purpose of pensions..
NEETs aren’t suffering through injury, illness, nor economic disaster.
Of course it’s a bloody economic disaster!
Agammamon’s point is that people adjust to a particular lifestyle or habit. If you’re 50 and lose your job, with a certain expectation about what you should be able to spend in a year and a mortgage hanging over your head, you’re going to feel a lot of difference if you start relying on the state to eke out an existence versus if you find a new job. If you can’t find work you may need to sell your home and car and swap them for cheaper ones, and give up holidays, hobbies and life’s little luxuries.
But if you’re a young person and you’ve grown accustomed to the lifestyle that can be afforded by handouts then the psychological incentive to seek work is rather smaller. Especially since you’d largely be in the market for entry level jobs whose pay isn’t much better than benefits.
I’m sure you planned for your pensions very well – if for some reason you had chosen to live off only the state pension, you could have planned for it and cut your cloth accordingly. That is effectively what an unemployed graduate is compelled to do. Right now if your private pensions suddenly stopped paying you, you’d likely feel the pain because you have normalised life with a higher level of pension. And your position would become genuinely precarious if you didn’t own your property outright. That’s a position more like an older worker losing their job. The difference is really quite substantial.
And my point is that Universal Credit is not focussed on NEETs, nor is it the cause of the problem. It is merely a sop to prevent destitution.
OTOH the actions of the Trade Unions and this government are designed to benefit those who already have jobs at the expense of the employers which inevitably makes hiring more/new people less attractive so while the initial impact is to hurt the capitalists (as intended) the secondary impact is on anyone unemployed or seeking to enter employment.
Increasing or reducing benefit levels doesn’t make any significant difference to youth unemployment..
I don’t think anybody is claiming this is all being driven primarily by the level at which benefits are paid. Obviously if businesses were better incentivised to hire young people, more of them would get hired. But a lot of businesses report a lack of interest in roles available. Now that might be their own fault, perhaps pay, hours, conditions, location, training etc are not an attractive package. Perhaps they’re low-balling in the hope they can claim there’s no native interest so can we bring over some cheap migrants on a visa please. Nevertheless, it is consistent with other evidence about NEETs claiming to be willing to work but not necessarily putting that sentiment into action when the opportunity is there.
So Agammamon has a point. Welfare dependency is a trap for the young in a worse way than it is for experienced workers, even if the welfare system is not specifically targeted at either group. Learned helplessness is a real phenomenon. If you keep paying someone enough to become dependent on it,they have no experience of the alternative, and work would hardly pay more at the level of work they could obtain, then it’s unsurprising people pursue their search less than vigorously or limit themselves only to applying for jobs they particularly like or which pay a bit more. It’s human nature.
There are other levers than the level of benefits that can help encourage NEETs to work than just cutting them outright. You can add more conditionality. Possibly up to a take-it-or-lose-the-bennies government-funded training or work scheme, though I have doubts about their effectiveness. You can take a US approach of time limited eligibility – UK politics is such that we would probably have to always pay a minimum level to people, but you might have a lifetime eligibility to, say, 18 months of higher benefits (about 3% of a working life), maybe accessible in 3 month blocks between periods of work so you don’t burn through them all at once. You can encourage people to start small by taking part time work, by structuring UC not to penalise you so much for additional hours worked (something IDS has been a big campaigner for).
I think the general public want less youth unemployment, fewer migrants taking entry level jobs, and more of those jobs going to our youth. Which sounds like wins all round so ought to be simple… But it will involve young people taking some jobs they’re not very keen on, and that’s going to need a bit of carrot and stick.
Agree with your last paragraph. I didn’t get my first choice of job because there were a lot more young graduates seeking it than there were jobs going (and I don’t interview well); training to be an Actuary was third choice of career; the carrot was the salary offered. If you offer them a decent wage most young people will take a job they’re not very keen on.
Some NEETs are willing to work but make conditions that employers are unwilling (in some extreme cases unable) to accept. [I am not saying that all conditions are unreasonable or unacceptable – e.g. in my 50s I was part of a team that included the Treasurer of a Synagogue so I used to email work to him after sunset on a Saturday and he would send the reply on Monday morning]. It is easy to say that they shouldn’t be silly but one cannot just tell them to grow up.
Your first sentence is contradicted by Agammamon’s third paragraph.
The Beeb had a sob story a day or two ago about a young woman who had applied for 400 jobs and another who had applied for over a hundred and not even been offered an interview, so it’s not just NEETs lacking interest in the jobs available: what are these jobs that no-one applies for? Do they pay enough to cover travel costs, rent and food? (I phrase it like that because rent is highly negatively correlated with travel costs).