The July jobs report was much more encouraging than June's. It now seems that May's much-improved report gave a "false start" on the recovery, June's much-worse report was a correction, and we are now getting back on track. The economy is still shedding jobs, but the pace of decline is slowing, consistent with the view that output has hit bottom and growth is now resuming. In addition to the slower pace of jobs decline this month (247,000 compared with 443,000 in June), the May and June declines were revised down (i.e., made less severe) by a combined 43,000.

There was also good news in that total hours worked showed no change this month—the first month without a decline since August 2008. And the length of the workweek edged higher, a good leading indicator of an economic upturn. Hours in manufacturing rose, signaling that July will show the first increase in manufacturing production since October 2008 (and probably a strong one). There will be a particularly large jump in vehicles production—as there was in vehicles employment—since the usual seasonal slump in production in July did not happen.

The unemployment rate fell, but it is hard to believe that it has peaked already. Unemployment declined this month because the labor force fell faster than employment. Some prospective workers have just given up looking.

The report showed hourly earnings up 0.2% this month, which is helpful for consumer incomes, although since earnings growth is slower than a year ago, year-on-year gains are continuing to fade.

In the payroll details, manufacturing cut 52,000 jobs in July, compared with 131,000 in June. This is the smallest manufacturing job loss since July 2008. The big swing this month came in motor vehicles and parts, where employment rose 28,000 this month, after falling 22,000 in June. There is normally a sharp decline in vehicles employment during July, due to summer shutdowns. But shutdowns came early this year, so the seasonally unadjusted drop in vehicles employment this July was only 9,000. Once seasonally adjusted, that translated into a 28,000 increase. But even if motor vehicles and parts are excluded, the decline in manufacturing employment slowed to 80,000, from 109,000 in June. Almost all sectors still showed declines, but less severe ones than in prior months.

The manufacturing workweek rose sharply, from 39.4 to 39.8 hours. Here, too, the increase was led by motor vehicles and parts (jumping from 38.9 to 40.5 hours). But it is notable that of the 10 major categories within durables manufacturing, nine showed an increase in the workweek. Since overtime hours barely moved (up 0.1), the message is that workers previously on short-time were getting more hours. Manufacturing output probably hit bottom for this cycle in June, and is now on the way up.

Construction jobs fell by 76,000 in July, compared with an 86,000 loss in June. The job losses were heavier in nonresidential construction (39,000) than in residential construction (27,000), not surprising since nonresidential construction still has much further to fall, while residential construction is beginning to flatten out. There were 10,000 jobs lost in heavy and civil engineering construction, where one might hope to see the impact from fiscal stimulus coming through—but one cannot know how many jobs would have been lost without the stimulus.

The private service sector lost 126,000 jobs in July, compared with 172,000 in June. The biggest shift in the labor market over the past three months has been in this sector. Since April, private services have shed an average 126,000 jobs per month (the same loss as in July). In the previous three months, they lost an average 351,000 jobs per month.

The big improvement in July came in the broad professional and business services category, which lost 38,000 jobs, after a 106,000 decline in June. The improvement here was broad-based. This sector includes temporary help, where jobs declined 10,000, compared with 31,000 lost in June. And it also includes high-end professional and technical services, where the job decline slowed to 38,000, from 106,000 in June. Leisure and hospitality also saw an improvement, adding 9,000 jobs, after losing 18,000 in June, while financial job losses slowed from 29,000 to 13,000.

Other service categories were more mixed, with some actually losing more jobs in July than in June, including retail trade (down 44,000 instead of 21,000), wholesale trade (down 19,000 instead of 14,000) and transport and warehousing (down 22,000 instead of 12,000). Healthcare showed a similar increase relative to June (up 20,000 rather than 25,000), but education lost 1,000 jobs.

Government employment rose by 7,000, after a 48,000 decline in June that was exaggerated by the loss of temporary Census workers. The federal government added 12,000 jobs, possibly stimulus-related, while state and local government lost 5,000. The state and local losses would probably have been more severe but for stimulus funds coming from the federal government.

The headline unemployment rate fell from 9.5% to 9.4%, its first decline since April 2008. But before concluding that the rate has peaked, we must note that it declined not because employment rose but because the labor force fell. The household survey, which determines the jobless rate, showed employment down 155,000 and the labor force down 422,000, so unemployment fell by 267,000. Over the past year, there has been no net increase in the labor force, which has helped dampen the rise in unemployment. We will need to see sustained employment gains before concluding that unemployment has peaked, and that probably will not be until the first half of 2010, with the unemployment rate peaking above 10%. But today's report brought the light at the end of the tunnel a bit closer.

The most comprehensive measure of underemployment—which includes workers who would like a job but are not currently looking, plus those working part time who would rather work full time—also fell, from 16.5% to 16.3%.

Overall hours worked began the third quarter with no change in July—the first month without a decline since August 2008. Hours were down heavily in the second quarter (falling 7.8%), and since we know that GDP fell only 1.0%, that implies a huge increase in productivity. This suggests that if output is now beginning to increase (as we believe), with GDP probably rising at least 1.5–2.0% in the third quarter, firms are unlikely to be able to sustain such rapid productivity growth from their remaining staff, and will have to increase hours worked per week and slow the pace of job losses substantially in the second half of the year. The stabilization in hours during July is the start of that process.