Unemployment numbers

In short

A friend of mine once frowned at the evening news and asked me why the jobless rate went down when so many folks he knew were still looking for work. His question is a fair one. If you are watching the unemployment number, you should know it counts only people who searched for a job in the last four weeks and could not find one. It does not count everyone without a paycheck. The Bureau of Labor Statistics, part of the U.S. Department of Labor, publishes the headline rate along with several wider measures. Look at more than one of them before you decide what the economy is telling you. Keep an eye on the trend over several months, since any single month can mislead. And remember that these figures describe the whole country, not your street or your trade.

The whole of it

What it is

I once watched a man at a county fair guess the weight of a pig, and he came within two pounds. He did it by looking carefully, not by weighing every hog in the state. The unemployment rate works much the same way. Nobody counts every worker in America by hand. Instead, the Census Bureau surveys about sixty thousand households each month for the Bureau of Labor Statistics. That survey is called the Current Population Survey. The answers from those homes stand in for the whole nation.

Here is how a person gets sorted. If you are sixteen or older and have a job, you are employed. If you have no job, you looked for work in the past four weeks, and you could start if one were offered, you are unemployed. If you have no job and have not looked, you are neither one. You are outside the labor force. That group includes students, retirees, and parents at home with young children. It also includes folks who gave up looking.

The unemployment rate is the number of unemployed people divided by the labor force. The labor force is the employed and the unemployed added together. Simple enough on paper.

How it works

You have probably heard that the number comes out on the first Friday of the month. That is about right, though the Bureau of Labor Statistics posts its release dates ahead of time on its own calendar. The report is often called the Employment Situation. It carries two different counts. One comes from the household survey I described. The other comes from a survey of employers, which counts jobs on payrolls. The two can disagree for a month or two, and that is normal. They ask different people different questions.

The headline rate is called U 3 in the Bureau's own tables. The Bureau also publishes a wider measure called U 6. That one adds people who want work but stopped searching lately. It also adds those working part time who would rather have full time hours. Because it casts a wider net, U 6 runs higher than U 3. Neither is wrong. They answer different questions.

Numbers also get adjusted for the season. Retailers hire in November and construction slows in January, and the Bureau smooths out those regular swings so you can compare one month to the next. Early figures get revised later, too. A first report is a best guess, not a final verdict.

The numbers, and where to find yours

If you want the real figures, go to the source. The Bureau of Labor Statistics keeps them at bls.gov, and you do not need a login. Look for the Employment Situation summary and the table that lists the alternative measures of labor underutilization. It shows U 1 through U 6 side by side. Table numbers can change over time, so look for that title rather than a number. For your own state, the Bureau runs a program called Local Area Unemployment Statistics. It covers states, metro areas, and counties.

One thing deserves a plain word. The Bureau also sets a margin of error on the household survey. The Bureau explains on its own site that a small monthly change in the rate may not be a real change at all. It may be statistical noise. So a move of one or two tenths of a point is a reason to wait and watch, not to worry or cheer.

Some benchmarks are set by law or by official policy rather than by the data. The Federal Reserve has a goal it calls maximum employment, and it does not pin that to one number. If you want a figure for what economists consider a normal long run rate, the Congressional Budget Office publishes its own estimate of the natural rate of unemployment, currently the current figure, which the official source publishes each year. And if you are asking about benefits, the weekly amount a person can collect is set by each state, so check your own state workforce agency. The federal rule on how many weeks regular benefits generally last is the current figure, which the official source publishes each year.

A worked example

A woman named Dolores lives in a town of 10,000 working age adults. Last month, 6,000 of them held jobs. Another 300 had no job but had looked hard and were ready to start. The remaining 3,700 had no job and were not looking. Some were in school, some were retired, and a few had quit searching in the spring.

Start with the labor force. Add the employed and the unemployed. That is 6,000 plus 300, which equals 6,300.

Now divide the unemployed by the labor force. That is 300 divided by 6,300, which equals about 0.0476. Written as a percent, that is 4.8 percent after rounding.

Dolores reads the number and feels fine. But she has a cousin, Walter, who gave up looking in the spring. Walter is not in the 300. He sits in the 3,700. Suppose 200 of those 3,700 actually want a job and stopped searching only because they were discouraged. Count them as unemployed and the math changes. The unemployed become 300 plus 200, which equals 500. The labor force becomes 6,000 plus 500, which equals 6,500. Then 500 divided by 6,500 equals about 0.0769, or 7.7 percent.

Same town. Same people. Two honest answers. The gap between them is why the wider measures exist. It does not mean anyone cooked the books. The headline rate is built to count one specific thing, and Walter falls outside it.

Where it goes wrong

I have made a mistake or two in my own life by reading one number and thinking I knew the whole story. The unemployment rate invites that mistake. A falling rate can mean more people found jobs. It can also mean discouraged people left the labor force, and that is a very different story. Look at the labor force participation rate, which the Bureau publishes right alongside, to tell the two apart.

Another trap is the national average. It hides a great deal. A rate for the whole country can sit calm while one industry, one age group, or one county is hurting. The Bureau breaks the numbers out by age, race, education, and state, and those breakdowns often tell you more about your own world than the headline does.

A third trap is treating a first estimate as settled. Revisions happen. So do survey errors. Wait for a trend of a few months before you draw a conclusion.

Last, remember what the rate cannot tell you. It says nothing about pay, hours, or whether a job is any good. A person working two part time jobs to cover rent counts as employed. That is a real limit. It is not a flaw you caused by misunderstanding it.

Questions to answer before you leave this page

Are you looking at the headline rate alone, or have you checked the wider U 6 measure too? Do you know whether the labor force participation rate rose or fell in the same month? Have you found your own state's figure on the Bureau's local page, rather than leaning only on the national one? Is the change you are reading larger than the margin of error the Bureau lists? Have you waited for a few months of data before deciding what a single report means? And when you hear a number on the news, can you say in one plain sentence who it counts and who it leaves out?