The Types of Unemployment, and What Each One Signals

Unemployment is not one condition. Somebody between jobs, somebody whose skills no longer match the work on offer, and somebody idle because less is being bought are three different situations that read three different ways. The headline rate adds them into one number, and one number can sit perfectly still while the situation underneath it changes completely.

The rate itself, the boundary of the labour force and the participation measure are settled separately, under employment growth against economic growth. Take those as given and a different question opens: once the number is in hand, what is it made of? Everything here runs on the Republic of Sankhya, an invented country. In its third year Sankhya had 8.32 crore people at work, 0.48 crore without work and looking, a labour force of 8.80 crore, and a rate of 5.45 per cent. Those 5.45 per cent turn out to be four or five quite different stories wearing the same jacket.

Why does the type matter more than the rate?

Put two economies side by side. Both report that 0.48 crore people in a labour force of 8.80 crore are without work and looking. Both report 5.45 per cent. In the first, employers are hiring and people are switching, so almost everybody in that 0.48 crore had a job three months ago and will have another one within the quarter. In the second, almost everybody in that 0.48 crore has been without work for more than a year, and the jobs being advertised call for skills they do not have or sit in towns they cannot move to. The rate is a headcount and the type is a diagnosis, and the headcount is identical in two economies that share nothing else.

Here is the everyday version. A doctor is told the patient has a temperature of 38.5 degrees. The reading is a real, measured, useful fact, and the reading is completely silent about whether the patient has a chest infection, food poisoning or heat exhaustion. Nobody treats the thermometer reading as the diagnosis. An unemployment rate works the same way: it is a reliable count of a condition and it says nothing at all about the cause. The word for what is inside the number is compositionThe internal make up of a total. Two totals that match exactly can still be built from completely different parts, and the total alone does not say which parts., and the composition is where the information is.

What is frictional unemployment, and why would a reading of zero be a warning?

Some people are without work because they are moving from one job to another. A weaver leaves a loom shed in one town for better piece rates in the next and spends five weeks arranging the move. A graduate finishes a course in April and starts work in July. A shop assistant walks out on a Monday because the owner has been shouting for a year, and takes six weeks to find a counter she prefers. Every one of those people is genuinely without work and genuinely looking, so every one of them is inside the 0.48 crore. None of them is stuck.

People only move between jobs when there are jobs to move between, so some part of this is evidence that a labour market is working rather than a malfunction of one. Follow the logic to its end and it becomes obvious. If this number went to zero, it would not mean everybody had found the right job. A reading of zero would mean nobody was changing jobs at all, and nobody changes jobs when there is nothing to change to. A market with no churnThe steady flow of people leaving jobs and starting others. High churn is not the same as high unemployment; it describes movement rather than idleness. is not a market that has been fixed but a market that has stopped.

What is structural unemployment, and why does waiting not clear it?

Now change the reason. Here the work exists and the people exist, and the two cannot be joined. A district built around hand block printing loses its buyers to a mill four hundred kilometres away. The printers are still there, still willing, still skilled. The skill has simply stopped being one anybody pays for in that district. Or the jobs appear, but three states away, and moving means giving up a house, a school place and every relative who helps with childcare. Or the vacancies are all for people who can operate a machine that arrived last year, and nobody in the town has ever seen one.

What breaks in a mismatch is the match between what people can do and where they are, not how much anybody is spending, so this type does not clear itself when buying picks up again. No distinction on the subject is missed more often, or costs more when it is missed. A recovery in demand puts more orders into the mill, and the mill hires people who can run mill machines. The recovery does not turn a block printer into a mill operator, and it does not move a house. The mismatchA gap between the work that is available and the skills or locations of the people available. Nothing about a mismatch changes just because total spending rises. sits in a different place from the shortfall in orders, so an increase in orders passes straight over it.

What is cyclical unemployment, and what exactly does it wait for?

The third reason is the simplest and the one most people already have in mind. Less is being bought, so less is being made, so fewer hands are needed to make it. A furniture workshop that shipped forty beds a month is shipping twenty, and it lets four of its twelve carpenters go. Nothing has changed about the carpenters, the workshop, the tools or the town. The number of people placing orders has changed. How a fall in buying spreads through prices and quantities is set out separately under demand, and only the conclusion is needed here: when the orders come back, the workshop needs twelve carpenters again.

Cyclical unemployment is the one type that genuinely reverses when buying returns, and mistaking a mismatch for a shortfall in orders therefore leads to waiting for a recovery that cannot arrive. The two look identical in the headline. The two behave in opposite ways when demand comes back. Watch the two bars in the drawing below across a full fall and recovery in buying: one of them climbs, peaks and comes all the way home, and the other one never moves at any point in the story.

What is seasonal unemployment, and why does an unadjusted series repeat itself every year?

The fourth reason is the calendar. Some work exists only at some times of year. Cane cutting runs for a few months and stops. A hill station that fills every room from April to June empties in July, and the porters, drivers and kitchen staff who fed off those rooms have nothing to do until the next season. A firecracker unit hires heavily before the festival and stands almost empty after it. Nobody involved is surprised, including the workers, who have usually built the gap into how they plan the year.

Predictable is not the same as painless. A series that has not been corrected for the calendar shows the same rise and fall every single year, and means nothing new on any of them. That correction has a name: seasonal adjustmentA statistical correction that removes the part of a series which repeats at the same point every year, so that what is left can be compared with the month before rather than only with the same month last year.. Without it, a reader meets a jump every winter and a fall every spring and is tempted to explain each one, when the honest explanation is that it is winter again. With it, a movement that survives the correction is worth asking about. Which correction has been applied, and whether one has been applied at all, is a property of the particular series rather than of the calendar it corrects for.

What is underemployment, and why does the rate not see it at all?

Every type so far sits inside the 0.48 crore. Underemployment does not. Think of a tailor who wants six days of work a week and gets one and a half. Think of an engineering graduate delivering parcels because that was the offer that came. Think of a household in which four adults share the work of a plot that one adult could manage, so all four are busy all day and the plot yields what it always yielded. Nobody in those examples is without work. Every one of them is counted as employed, and none of them is anywhere near the 0.48 crore.

The rate was built to count people without work rather than to measure how much work the people with work are actually getting, so underemployment is a boundary of the rate rather than a flaw in it. Ask it the second question and it will answer the first one, confidently and wrongly. The word for the gap is labour underutilisationA wider idea than unemployment: it takes in people who want more hours than they get and people whose work sits well below what they are able to do, both of whom count as employed.. Measuring that gap needs its own questions in a survey, so it comes out as a separate measure rather than as a correction to the rate. Where a large share of work is informalWork carried on outside registered payrolls and formal contracts, often in very small units or on own account, where hours and earnings are irregular by nature., spread across many small jobs and many small holdings, the gap between people who are employed and people who are fully occupied gets wider, and what informal work is and why its share matters is taken up separately under formalisation.

What happens to the rate when somebody stops looking?

Here is the other thing standing outside the number. A weaver spends nine months answering every advertisement, gets nowhere, and stops. She still wants work. She has simply concluded that looking is not producing anything, so she stops looking. In the arithmetic of the rate, a person who is not looking is not in the labour force at all. She leaves the top of the fraction and she leaves the bottom of it too. The word for her is discouraged workerSomebody who wants work but has given up actively searching, usually after a long unsuccessful stretch. Because the search is what puts a person in the count, they drop out of it entirely..

A discouraged worker leaves both the count of the unemployed and the labour force. People giving up is therefore one of the ways an unemployment rate improves, and it is the worst reason available. The two ways a rate can fall, one where people found work and one where people stopped searching, are worked through in detail under employment growth against economic growth. The second route exists, it looks exactly like the first from the outside, and it is the reason the number of people at work has to be read next to the rate rather than after it.

What does each type actually signal?

Each cause carries a different message and a different set of things that follow from it, so sorting by cause pays off. The translations below rank nothing. Which type is worse, and what anybody should do about it, are policy questions covered separately.

Can one rate of 5.45 per cent be three different economies?

Sankhya in its third year, with the headline held completely still. Every economy below has 0.48 crore people without work and looking, in a labour force of 8.80 crore. 0.48 divided by 8.80 and multiplied by a hundred gives 5.4545 in all three, and 5.4545 rounds to 5.45 per cent. Each composition is held in hundredths of a crore, so the parts always add back to 0.48 exactly.

The headline is identical in all three economies, and no reading of that headline alone could separate them. The last row shows how far apart they really are. A full recovery in buying, run through each one, lands economy C at 2.16 per cent while A and B are still at 4.89. And that last row still does not separate A from B. Both land on exactly the same number by different routes: A because there was almost nothing there for a recovery to reach, B for the same reason and with a very different set of people left behind. A third question, about how long people have been waiting, is needed before A and B come apart at all.

Then take a fourth economy, where nobody finds work and the number improves

Start again from the same year three position. Employment 8.32 crore, without work and looking 0.48 crore, labour force 8.80 crore, rate 5.45 per cent. Now let 0.05 crore of those people stop searching. Not one of them found a job. Not one employer hired anybody. The count of people at work is 8.32 crore before and 8.32 crore after, unchanged to the last decimal. But those 0.05 crore have left the labour force. The count without work is 0.48 minus 0.05, or 0.43 crore, and the labour force is 8.80 minus 0.05, or 8.75 crore. Divide 0.43 by 8.75 and multiply by a hundred to get 4.9143, and 4.9143 rounds to 4.91 per cent.

The best looking number of the four is the worst outcome of the four, and the improvement from 5.45 to 4.91 per cent was produced entirely by people giving up. There are still 0.48 crore people in Sankhya who want work and do not have it. Exactly as many as before. The only thing that changed is how many of them the rate is willing to look at.

How should an unemployment figure be read?

Three questions, in order, and the order matters. First, what is the rate. The number itself has already answered that one. Second, what is inside it, meaning how those people split between movement, mismatch, weak buying and the calendar, and how long they have been waiting. Third, what is outside it, meaning how many people are working far less than they want and how many have stopped searching altogether.

A rate quoted on its own has answered the first question and has said nothing whatsoever about the other two. The rate does its own job precisely. The limit is the size of that job. Where a labour market is summarised in one number and one adjective, the adjective is doing work the number cannot support, and the breakdown is worth finding before the summary is accepted.

Why does a lender care which type it is?

Picture a co operative bank with Rs 40 crore lent into one district, of which Rs 26 crore sits in small loans to shopkeepers, transporters and repair workshops whose customers are the people who work at the two large units on the edge of town. Unemployment in that district rises. The lender's useful question now is not the rate but the composition of the rise. Two compositions lead to two completely different repayment stories.

If the rise came from weak buying, the lender is looking at a timing problem, and if it came from a mismatch, the lender is looking at a permanent change in who lives in that district and what they can afford. In the first case, orders at the two units fall, hours get cut, the shopkeepers see thinner takings, and when orders return the takings return with them. ArrearsAmounts a borrower has fallen behind on. A rise in arrears says money has not arrived on the date it was due, and says nothing yet about whether it eventually will. rise and then unwind. In the second case, one of the two units has closed because what it made is now made elsewhere, and the people it employed will not be re employed by any recovery in output. The lender who read only the headline sees the same rise in both cases and provisions for a dip. The lender who asked what the rise was made of provisions for a dip in one district and for a smaller local economy in the other. What either lender ought to do about it is a separate question. Two situations that are not the same situation show up as identical in the rate alone.

The Republic of Sankhya is invented.
Educational material. Not advice on any investment, tax, budget or market position.