The costs of unemployment — individual and economy

The cost that continues after the new job starts

A worker loses a job and finds a new one within a year. The worker has lost a few months of wages. That seems to be the whole cost.

The evidence says it is not. Economists Steven Davis and Till von Wachter followed men in the United States who lost jobs they had held for at least three years, in mass layoffs. They used official social security earnings records from 1974 to 2008 (Brookings Papers on Economic Activity, 2011). They added up everything these men lost over the next twenty years. When the national unemployment rate was below 6% at the time of the layoff, the loss was about 1.4 years of their old earnings. When the rate was above 8%, it was about 2.8 years.

Twenty years is far longer than almost any spell of unemployment. So much of the loss must come from what happened after these men found new work: lower pay that lasted for years. And losing a job in a bad economy cost about twice as much as losing it in a good one.

This lecture asks who pays the cost of unemployment, how large it is, and why it can last long after the recession that caused it.

Who pays: four places to look

The cost of unemployment spreads outwards from the person who loses the job. Look in four places.

• The person: lost income, lost skills, and often worse health.
• The economy: goods and services that are never produced.
• The government's budget: more spending on benefits and less tax coming in.
• Society: effects on families, on crime, and on whole towns.

These four are connected: a cost in one place often causes a cost in another.

The cost to the person

Lost income

The wage stops. In many countries, unemployment benefits replace part of it, but usually only a small part. Savings run down, and spending has to fall.

The scar

The second cost is the one in the hook. Economists call it scarring: a lasting fall in a person's earnings after a spell of unemployment. Several reasons add up.

• Skills that were valuable mainly in the old firm are lost with the job.
• The longer the gap, the more other skills fade.
• Employers may read a long gap as a warning sign, fairly or not.
• In a recession, people take whatever job they can find, often lower paid, and it takes years to climb back.

Scarring reaches people who were never employed at all. Oreopoulos, von Wachter and Heisz studied Canadian university graduates (American Economic Journal: Applied Economics, 2012). Those who started work in a typical recession earned about 9% less at first. The gap halved within about five years and closed after about ten. The least advantaged graduates lost the most.

Health

Job loss is also linked to worse health. Sullivan and von Wachter matched the earnings records of workers in Pennsylvania, USA, to death records (Quarterly Journal of Economics, 2009). For men with long job tenure, death rates in the year after a mass layoff were 50% to 100% higher than expected. Twenty years later they were still about 10% to 15% higher. The authors estimate that, if this lasted, it would mean a loss of about 1 to 1.5 years of life expectancy for a worker displaced at age 40.

Part of the reason is money and stress. Part is what work gives besides a wage: a routine, other people, and a sense of being useful.

The cost to the economy: output that is never made

An unemployed worker is a resource the economy is not using. A worker who could have built a house, taught a class or repaired a machine did not do it. So the economy produces less than it could.

The production possibility frontier (PPF) shows this. Every point on the curve uses all the economy's resources. A point inside the curve leaves some resources idle.

This output is lost for ever. A factory that stood half-empty last year cannot make last year's missing cars as well as this year's. So unemployment is a waste for the whole economy, not only a private problem.

How much is lost? In 1962 the American economist Arthur Okun noticed that each point of unemployment above its normal level went with output about 3% below what the economy could produce. Later estimates are closer to 2%. This is Okun's rule (some books say "Okun's law"). It is a pattern in the data, not an exact law.

The cost to the government's budget

Unemployment hits the budget from two sides. Spending rises, because more people claim benefits. Tax revenue falls, because people without jobs pay no income tax and spend less. Nobody has to decide either change: both are built into the tax and benefit rules. Economists call them automatic stabilisers.

It is easy to count this cost wrongly. Find the mistake in the working below.

The benefits themselves are not lost to the economy. They are a transfer: money moved from taxpayers to the unemployed. They protect the people who lost their jobs, and they keep some spending going in the shops. That softens the recession for everyone, which is why they are called stabilisers.

Are there any gains? A fair evaluation names them. When many people are looking for work, firms hire more easily, and wages and prices rise more slowly. Some frictional unemployment is useful too: it is time spent finding a job that fits. But these gains come from a small amount of unemployment, not a large one.

Hysteresis: when a temporary shock leaves a permanent mark

Now put the person and the economy together. This is the hard part of the lecture, so spend time here.

A recession is meant to be temporary: demand falls, jobs are cut, and later the jobs return. But suppose it lasts a long time. Some people stay unemployed for years. Their skills fade, employers stop considering them, and some give up looking. When demand recovers, these people do not come back. The cyclical unemployment of the recession has turned into structural unemployment, as the lecture on types of unemployment warned.

So the economy has fewer usable workers than before, and its capacity has fallen. The PPF moves inwards, and long-run aggregate supply (LRAS) shifts to the left. A temporary fall in demand has caused a lasting fall in what the economy can produce.

That is it. That is the whole idea. A short recession mostly costs the output lost while it lasts. A long recession costs that, and then keeps costing every year afterwards. Try it in a model.

What should you have seen? With scarring off, the cost stops when the recession ends. With scarring on, the grey bars appear and keep going. Compare the two presets. Both have the same total of "point-years" of unemployment: 4 points for 1 year, or 1 point for 4 years. Yet the long, shallow recession costs more, $116 billion against $80 billion, because it leaves a scar.

The model leaves out a great deal. Its rules are invented. It ignores health and society, includes no policy response, and assumes scarred workers never return, when in real life some do, slowly. But its main lesson matches the research in the hook: time matters. This is the economic case for acting early when unemployment rises.

The cost to society

Some costs fall on people who never lost a job. They are harder to measure, so they need careful wording.

Crime. Steven Raphael and Rudolf Winter-Ebmer studied crime in US states from the 1970s to the 1990s (Journal of Law and Economics, 2001). Higher unemployment clearly raised property crime, such as burglary. The evidence for violent crime was much weaker. So "unemployment causes crime" is too strong; "unemployment raises property crime" is what the evidence supports.

Families. Studies link long unemployment to family stress and to worse outcomes for children. Some of this may be caused by unemployment, and some by problems that come with it, such as poverty.

Places. Unemployment is not spread evenly. When a town depends on one large employer, such as a coal mine, its closure puts the structural unemployment in one place. Shops lose customers and close. Younger workers move away. Local tax income shrinks, and services suffer. Each step makes the next more likely.

A real case: keeping workers in their jobs, Germany 2009

The policy. In the global financial crisis, demand for German exports collapsed. Germany already had short-time work (Kurzarbeit): a firm cuts hours instead of dismissing workers, and the public employment agency pays part of the wages for the hours lost. In 2008–09 the government made it easier to use, and for longer.

The prediction. Output will still fall, because people work fewer hours. But workers keep their jobs and skills. Few become long-term unemployed, so there should be little scarring or hysteresis. When demand returns, firms can raise hours at once instead of hiring and training new workers. This only works if the shock is temporary.

What happened. German output fell by about 5% in 2009, its largest fall since the Second World War (Federal Statistical Office, January 2010; later revised). Yet the unemployment rate rose by only 0.2 percentage points, to 7.5% (International Labour Organization country brief, 2010). Short-time work peaked at about 1.4 to 1.5 million workers in May 2009. The OECD later estimated that the scheme had saved more than 200 000 jobs by the third quarter of 2009. The United Kingdom had no national short-time work scheme at the time. There, unemployment rose from 1.61 million (5.2%) in January to March 2008 to 2.68 million (8.4%) in September to November 2011 (Office for National Statistics). The two economies differ in many ways, so this is a contrast, not proof.

The verdict. For a short, sharp fall in demand, the scheme worked as predicted. It did not stop output being lost. It shared the cost across many workers who each lost some hours, instead of a few who lost everything, and that avoided most of the scarring. But it costs public money. And if a shock is structural, it can keep workers in jobs that are disappearing and delay their move to new ones. ILO researchers also warned that some firms stayed on the scheme for very long periods. Short-time work suits cyclical unemployment. It is the wrong tool for structural unemployment.

Three classic mistakes