In the previous three articles in this series (click here for part 2) the definitions of recession and depression have been discussed based on the NBER (National Bureau of Economic Analysis) definition of recession and a modified definition of the business cycle. The previous articles have had an analysis of GDP (Gross Domestic Product) that explicitly separates the rebound from the bottom of the business cycle into two parts: the recovery, which is the gain necessary to reach the previous peak, and the subsequent expansion as GDP goes to new heights. In this article the business cycle behavior of employment will be examined
Cyclic Behavior of Employment Data
Just as GDP rises and falls going through the business cycle, so does employment. The following graph shows employment during the past 60 years. (Click to enlarge) 
The declines in the number of employed seems quite inconsequential for many of the past recessions. As we shall see in the following section, the apparent mild nature of job loss in recessions is an illusion of scale. The “minor” dips of several million lost jobs in a recession are actually very significant. That can be seen when the variations in employment are seen on a different graphic scale.
Normalized Employment
Just as GDP was normalized to population in the preceding articles (GDP per capita), the same has been done for employment. The following graph shows the ratio of civilian employment to civilian population age 16 and older. 
The graph below shows the ratio of civilian employment to total population.
The two previous graphs each have a particular purpose. The first graph describes the percentage of the population of working age that is employed. This is a measure of the efficiency of the economy in utilization of the potential labor force. The actual labor force is smaller. It is defined by the U.S. Dept. of Labor as only containing those people who are employed plus unemployed persons who are actively looking for work. The ratio of employment to the actual labor force (called the civilian labor force) is known as the employment rate. That graph is not shown here. The complement of the employment rate is the official (U-3) unemployment rate, currently at 9.7% (as of January 2010).
The two previous graphs each have a particular purpose. The first graph describes the percentage of the population of working age that is employed. This is a measure of the efficiency of the economy in utilization of the potential labor force. The actual labor force is smaller. It is defined by the U.S. Dept. of Labor as only containing those people who are employed plus unemployed persons who are actively looking for work. The ratio of employment to the actual labor force (called the civilian labor force) is known as the employment rate. That graph is not shown here. The complement of the employment rate is the official (U-3) unemployment rate, currently at 9.7% (as of January 2010).
The second graph describes the ratio of the number employed to the entire population. It represents a measure of the support of the society by those who are employed. When the ratio is low, a larger proportion of the total population is supported by a smaller proportion of the population that is working. The steep decline in the 15 years following World War II is largely the result of the baby boom influx to the population. It was also affected to a lesser degree by higher military draft rates during the Korean and Viet Nam wars. The steep decline in the past three years is a result of job losses on a scale not seen in the post World War II era.
Analysis of the Raw Number Employed
For this we show a graph associated with the 1980 and the 1981-82 recessions. 
For the 1980 recession, the NBER start date occurred before the peak in employment and the NBER end date occurred after. The employment recession started three months earlier and ended one month later in the case of the NBER recession of 1981-82. So much for the old canard that employment is a lagging indicator; sometimes it lags and sometime it leads.
The start and end of employment recessions, recoveries and depressions since 1948 are listed in the following table, as well as the corresponding durations.
This recession far exceeds the duration of any other employment recession since 1948. This employment recession is approaching the duration of any previous depression over the past 60 years, and the recovery has not started yet.
Analysis of the Ratio of Employment to Total Population
The third example shows employment normalized to the total population for the time associated with the business cycle around the recession of 1973-74.
In this recession, the decline in employment occurred in the second half of the NBER recession. This is a case where employment was a lagging indicator for the start of the recession, but a coincident indicator for the end.
Another example shows an employment depression lasting 25 years from 1948 into 1973. In this period, employment behaved as a leading indicator, a coincident indicator and a lagging indicator at different times. The two most notable things about this time period:
1. There was an employment recession in 1951-52, but no NBER recession.
2. There was a much longer employment recession associated with the NBER recession of 1957-58.
Normalizing employment to total population during this time period may be over stating the decline in normalized employment for two reasons:
1. The baby boom spanned time period from 1946 to 1964, so there was above normal percentage of children in the population.
2. There was a draft during most of this time period, with a large military population. The Korean War was fought during the 1948-53 years and the Vietnam War was waged from 1963 to 1973.
The dates and durations of all employment recessions, recoveries and depressions (based on employment normalized to total population) since 1948 are listed in the following table.
Analysis of the Ratio of Employment to the Civilian Population 16 and Over
The first example here is for the 1980s and shows employment normalized to the civilian employment age population. This graph can be compared to the graph in the raw employment section (two sections back). In the non-normalized employment graph there are two closely spaced depressions. When normalized to account for population growth, the entire time between August 1979 and March 1985 (more than 5-1/2 years) is in an employment depression.
The period from 1948 to 1973 was seen to have experienced a 25-year employment depression when employment data was normalized to the entire population. In the preceding discussion it was suggested that normalization to total population might be distorting the depression data due to the baby boom and the military draft (with two wars) during that era. The same period is shown in the following graph with normalization to the adult civilian population. This removes any bias due to (1) baby boomers under the age of 16 and (2) military population.
There is much better correspondence between employment recessions and NBER recessions when employment is normalized to the civilian adult population than to the total population for the years from 1948-73. Instead of one 25-year employment depression we now see three employment depressions of 2-1/2 years, 16 years and a little over 4 years. There were two breaks in the depression totaling about 2-1/2 years. This may not seem like much of a difference, but the burden of unemployment was only about half by the calculation in this section than it would be if we used the total population for normalization.
The final time period to be examined in this section is the most recent years, from 1998 through 2009. That graph is shown below.
The durations of all employment recessions and depressions (based on employment normalized to the adult civilian population) since 1948 are listed in the following table.
The employment recession of 2000-03 is far longer in duration than any previous recession since 1948. The current recession is very close to equaling the duration of 2000-03.
Summary
A number of conclusions can be drawn from this study:
1. The old canard that employment is a lagging indicator is worse than wrong – it is irresponsible.
2. Normalizing the number of employed to the adult civilian population gives an appropriate measurement of relative employment. After all, one million employed for a working age population of 100 million is a far more positive economically than it would be for a population of 200 million.
3. Normalized employment indicates that there have been extended depressions in employment over the past 60 years.
The next article in this series will address the question of how employment depressions and GDP depressions compare. The question will be addressed: Are there different business cycles for Wall Street and Main Street?
Disclosure: No stocks mentioned
