AT 8:30 a.m. last Friday the Bureau of Labor Statistics reported that the August unemployment rate remained at 6.8 percent. Immediately, bond yields fell as financial markets ignored the unemployment stability.Instead, they focused on the small 30,000 rise in nonfarm payroll jobs and the large 310,000 drop in civilian employment, as measured by a different survey, as signs that the recovery was faltering. Has the economy slipped back into recession? Probably not: these figures are consistent with a slow expansion that is being sustained by productivity gains rather than jobs growth. Nonfarm jobs statistics come from a monthly survey of 300,000 establishments, ranging from aircraft plants to zoos. Employers report how many workers were on their payrolls at the middle of the preceding month - 108.8 million in August, up 34,000 from July. They also tell how many hours their employees worked per week and their pay. Unemployment and civilian employment numbers come from the world's largest public poll, the Household Survey. Each month, trained interviewers ask 60,000 randomly selected households detailed questions about the labor market activities of each person living there. The labor force is the sum of the employed plus the unemployed, i.e., those at work plus those looking for work. The 6.8 percent unemployment rate comes from dividing the number of unemployed persons by the size of the labor force. The Household Survey conducts its own different count of the employed: persons who worked at least some time during the previous month. This total (116.7 million in August, down 310,000 from July) is larger than the payroll survey because, among other things, it includes farmers and the self-employed. How does a person get counted as unemployed? Collecting unemployment compensation is not one of the criteria. Instead, he or she has to be looking for work. Visiting employment offices clearly counts, as does reading the help-wanted ads. Those who have given up looking because they think nothing is available are classified as "not in the labor force." They are sometimes called "discouraged workers." This survey is, of course, subject to a sampling error of plus or minus so many percentage points. Thus, for the national unemployment rate, it takes a monthly change of at least 0.2 percentage points to be considered statistically significant. Anything less represents no change. Unemployment rose through June and then dropped to 6.8 percent in July. The unemployment rate is a "leading indicator" of impending recession, rising a few months before the economy begins to contract. During recoveries it switches to a "lagging indicator." Thus, unemployment's recent behavior is consistent with the view that the recession ended around May. Nonfarm jobs have to rise or fall by about 120,000 to be significant. Thus, neither August's rise of 34,000 or July's decline of 72,000 are meaningful. What is important is that May's 150,000 rise marked the end of 10 straight monthly declines that averaged 150,000. Employment figures as measured by the Household Survey, which must change by at least 170,000 to be considered meaningful, paint a weaker picture. They fell an average of 230,000 in July and August. However, this followed a large rise of nearly 300,000 in May. Which is the better survey? Payroll data provide less-volatile initial estimates of monthly job gains and losses, but are subject to large subsequent revisions. Household data are never revised (except for the annual updating of the seasonal adjustment factors). However, the Household Survey is subject to large month-to-month swings. My judgment as a long-time user of both surveys is to interpret the August employment numbers as flat. Does this flatness mean that recession is still going on? Technically, a recession ends when the overall measures stop declining. The economy is growing right now - albeit slowly. That's because productivity is improving. However, to most people, recovery means more jobs and not just more efficiency. And the best of all worlds is more of both. The fact that the jobless rate stopped rising and has fallen a bit is further evidence that the recession has ended.