Normally, unemployment insurance is a form of insurance. Employers pay premiums for their employees, and those premiums are used to pay benefits should those employees become unemployed. Both premiums and benefits are based on wages, covering some fraction of income for some limited period of time - typically up to 26 weeks - while the employee finds a new position similar to work he has done in the past. During a recession, the federal government often funds an extension to unemployment benefits on the theory that it takes longer to find a similar new position during a recession.
Sometimes, however, getting a similar position may no longer be realistic. Not as many human market makers are needed when there are computers to help with the job. There will probably never again be as many pre-IPO PR jobs available as there were during the dot com bubble. In these cases, extended unemployment benefits can simply facilitate denial: they'll encourage people to turn down realistic job offers while they hold out for their dream job that no longer exists.
There is data available to quantify this effect. Studies regarding extension of benefits indicate that it causes the average unemployment period to increase by between 4.2 and 10.6 weeks - 8% to 20% of the additional benefit time.
That's why Congress' recent refusal to continue extending unemployment benefits beyond the basic period is likely to be the best jobs creation program of this year. Unemployment benefits had been extended to 99 weeks - nearly two years - and now they will be going back down to the regular 26 week range. Doing the arithmetic, we should expect a reduction in the unemployment rate of about a percentage point - from 9.5% to 8.5% - as the change takes hold.
[edit: Unfortunately, Congress caved and renewed emergency unemployment shortly after this post was written. It was finally allowed to expire at the end of 2013. The first half of 2014 then saw the fastest job growth in over a decade.]
Data and anecdotes on the effect of insurance on unemployment
Another place unemployment benefits go:
Yeah, I agree that at some point the unemployment benefits have to go back to their pre-recession mode and we may certainly be at that point, if not past it. But to be fair to the large number of unemployed people, and I think your 8% to 20% figure backs this up, the majority of unemployed people are not holding out for their dream job.
I agree that the majority probably look for the best that's available. They're also likely finding something within six months or a year, so two years is overkill, even in 'recession mode'.
I don't know that it's time to go back to pre-recession mode yet. Rather, I think that at some point we should recognize that the economy doesn't contain the type of job that was lost any more, and people should be transitioned to a safety net that provides subsistence level benefits, rather than one that scales with former salary.
Tapering the benefits also might work better than a fixed cutoff - and might not require special tinkering during a recession.
Now a paper shows that the cutoff of unemployment benefits at the end of 2013 "can explain nearly all of the observed aggregate employment growth in 2014". That this is the likely causal factor is shown by the fact that, based on county level data, the places where the reduction in benefits was greatest were also the places with the greatest job gains. According to the summary at fivethirtyeight, the cut in unemployment benefits is estimated to have "led to the creation of 1.8 million jobs in 2014, and that nearly 1 million of those jobs were filled by workers who otherwise would have stayed out of the labor market."
Fivethirtyeight summary (middle of three papers discussed):