Unemployment- Let’s add it up one more time!
The post and courier reported the December monthly unemployment data for South Carolina. Unemployment is the most often quoted number and also the most misunderstood of the economic data floating around. Even for a seasoned professional, it is easy to get crossed up. This particular article referenced two unrelated data sets while discussing South Carolina unemployment.
…
When citing unemployment detail, employment numbers need to be culled from the same data set. The recession started in December of 2007. Employment since then decreased by 113,820, not the 109,900 in the article. However, what is most interesting, and left out, is persons on unemployment increased by 150,693 over this same time period! (Employment pdf) So not only are there less employed, but also more unemployed. It is likely this number will creep higher over the next couple of months to 14 percent (Feb 09), partly as a result of re-benchmarking and partly due to the South Carolina economy. But where did the 109,900 come from? That number is Unadjusted Current Employment Statistics (CES) employment data from the “establishment survey”, a different program. I really like these data and often use them in conjunction with unemployment data thus providing a better picture of where jobs may have been gained or lost. CES data in this situation, should not be compared to unemployment data. Unemployment data is derived from the Current Population Survey (CPS), “the household survey”. This survey is the labor force measure for the nation.
CES Data- the poser
CES is reported two ways. Over longer timelines, Seasonally Adjusted CES data is a better estimate of employment, while Unadjusted Seasonal data is more acccurate in real time. I have to admit I tend to default to Unadjusted Seasonal data, but there is a time and place for Adjusted and this may be that time. My default to unadjusted is a result of statistical variations with Seasonally Adjusted data, most could give a hoot about- but they are there.
“CES data are a coincident economic indicator and are often cited in national and local newspapers, magazines, and reports. This press generates enthusiasm, curiosity and a wealth of outside material for supplementary reading. The College of Business Administration at the University of South Carolina uses seasonally adjusted employment as an indicator of current employment trends in South Carolina. The regional Federal Reserve Banks use CES data in easy-to-understand economic applications….”. Source: Bureau of labor Statistics
The CES table shows the difference in data sets for the state of South Carolina. (CES pdf) The unadjusted is 109,900 with adjusted coming in at a loss of 102,200. 102 seems a little soft in this economy and, consequently, it is another reason for not quoting the sum of change in the face of unemployment data from the household survey. The bigger picture is which direction is employment going in the future. I will try to address that by looking at migration.
United States – South Carolina Unemployment Comparison
I had a recent question as to why South Carolina (SC) unemployment rates appear to be lower than the United States (U.S.) rates before 2001 and higher after 2001. With this question we need to back up a few steps. There are three factors which, when taken into account, explain unemployment rate expected variation.
First, the U.S. unemployment rate is calculated using the Current Population Survey or CPS. State and local unemployment data is calculated using Local Area Unemployment Statistics or LAUS. These are different programs with different methodologies, although related. Many times, as reported in the press, these data do not corroborate.
Second, compatibility of reference data sets. Many forget that SC is a state with a population of only about 5 million persons. The U.S. has 300 plus million persons. In other words it is more likely than not that SC is not a representative sample of the U.S. So there is no reason to believe that what happens either in the U.S. or in SC will affect the other by the same proportion at any time, historical or otherwise! This is an issue not only in this case, but with many data comparisons- the old apple and oranges problem. Because they appear to be related, we make a big deal out of it when in fact the two are independent events.
Alternatively, it is more likely the economies of Georgia or North Carolina, our trading partners, will have a greater impact on SC, in both the short and long run.
A third factor, and a complicated one at that, is the change in survey method over time. There have been significant survey updates in 1996, 2000 (Census), and 2003 to mention a few. Often data is labeled with codes indicating a methodological change in the series. The Bureau of Labor Statistics often describes these changes in great detail with formula- yikes! It is safe to say, the changes do impact the data, sometimes with sudden jumps up or down, considered by some to be either favorable or unfavorable. The fact is, it was a change in method and all data still sums to 1. Unfortunately, the next day you will still be either employed, unemployed or looking for work in this data set. Here is a link to the summary of differences.
NOTE: The BLS does make mistakes, most recently with the Current Employment Statistics. However, they are a group with integrity who fix mistakes, correct data, and describe the impact. What I like most about the BLS is their drive continually to improve data through refining their techniques. Businesses and individuals who use these data have benefited significantly.
How to Think About Unemployment Data-
Example: As a former labor market analyst, I emphasize the big picture, the trend. Here is a link to a spreadsheet that compares, for the fun of it, SC, MI, US and SD back to 1976. Note how differently these states labor forces have reacted to this and other booms and recessions. Recently, the MI labor force has crashed, likely driven by the employment outlook, thus creating a high unemployment rate. SC, on the the other hand, has strong labor force growth, maybe too strong (over new employment), creating higher unemployment. This is better than a labor force crash, however! Have you ever been to SD? I am poking fun at a former neighboring state of one where I previously lived. The point is that each state has its own economy, and it is best to understand that economy and focus on what it can and can’t do. In SC, we have been hit hard in the textile industry (another plant announced a closure today), missed the finance crash, but took it on the chin in the auto parts industry (about 2000 jobs since 2007). We are a rural state where persons have a difficult time migrating to new jobs. We are also a state where people go to spend money- tourism. Persons have been tight with that money recently. Most employment and unemployment data makes sense when we take a look at a regions industry make-up (mix), and how those industries manage over time in up and down economies.
February 09 Employment
The Post and Courier had a nice article on the February 09 unemployment rate. The data, as one can imagine, is complex ,and therefore it can be helpful to step back and take a look at the big picture as to the root causes of unemployment and where it might be going.
Currently SC unemployment is 11.5 percent. Unemployment in the Low-Country (Charleston MSA) is 9.1 percent. The unemployment rate is calculated by looking at labor force, employed and unemployed workers. I have captured the graphs from the Bureau of Labor Statistics that display how these different data affect each other.
Note that labor force numbers continue to rise. In other words, there continues to be a net gain of new persons entering the workforce. However, at the same time, there is a decrease in the number of available jobs. This increasing gap is one reason for the spike in the unemployment rate. The other item that one notices is the irregularity of the graphs. The reason is that some industries are seasonal, like hospitality and construction. Therefore, in SC it is not uncommon to see large jumps in employment and unemployment depending on the time of year. I originally thought that the recent one percent jump month to month was unusual. However, after a little research, I found an example of a similar jump in 2007. Only this time it was a drop of .6 percent from March to April.
Unemployment also varies across our state by region. One reason is the variation in industry distribution. The linked table calculates industry employment percentages of SC versus the MSA. Note that 53 percent of SC employment is in industries hard hit by the downturn versus a subset of 47 percent locally. The Charleston region also has a slightly higher percentage of government employment (including education) that adds slightly more stability to the region. This industry distribution is, in fact, born out in the numbers.
The other factor which affects unemployment is type of job and skills required for employment. Many of the jobs that have been brought into SC are low-skill jobs, including warehousing, retail, and some manufacturing jobs. Unfortunately, in a poor economy those are typically the first to see layoffs, regardless of the industry – easy come easy go. As a result of the lower skill level, job characteristics, and industries affected, SC as a whole has taken a harder hit then what would be predicted.
It actually could be significantly worse if we had an uncontrolled construction industry or a financial sector with a larger presence. Fortunately both of those industries were only a part of our overall economy. I believe it is possible for SC’s unemployment to hit 14 percent for reasons mentioned above and then stabilize. It will, of course, take government, educational institutions, and private industry all working closely together to decrease that rate. Time to roll up the sleeves again.
November 2008 Unemployment
Here are some numbers to digest. These are from Dean Bakers BLOG Center for Economic and Policy Research.
Dean goes on to say-
The December employment report showed the economy losing 524,000 jobs in December. It also showed sharp upward revisions to job losses in the prior two months, bringing job loss over the last three months to 1,531,000. This is the highest 3-month total since the months immediately following the end of World War II, although the job losses in the 1958 and 1974-75 recessions were larger relative to the size of the workforce.
With the length of the average workweek getting shorter, the decline in hours worked has been even more rapid than the drop in employment. From September to December, the index of hours worked for production workers fell at a 9.4 percent annual rate. This rate of decline in hours would be equivalent to losing 12.8 million jobs over the course of a year, if the length of the workweek remained constant.
We saw that today, 01/14/09, with the announcement of the Charleston Metro Chamber cutting hours.
Recenty The Post and Courier highlighted “experts” who couldn’t figure out why the unemployment rate in SC is higher than the rate in the United States. The reason is very clear; SC is NOT like the United States in many ways and should not be compared directly without looking at the details. First here is the unemployment-nov-08 from the BLS. I like these tables since they show the relationship between employment, labor force, unemployment and the unemployment rate. Note the STEEP drop-off in employment. Wow! This is a big drop. It could be a lot worse though, depending on what the labor force does. We will have to wait and see. So one can imaginewhat the unemployment rate would look like, if instead of companies cutting back hours like the Chamber, those were actual layoffs. Regardless, as Dean Baker stated these hourly cutbacks are real and will continue to affect the economy.
Back to our comparison. I reviewed a number of data sets but chose two that highlight the differences between SC and the United States. I chose Education (Census) and what is called Location Quotient (LQ) from the BLS. Education compares SC educational attainment rates with the United States. LQ compares industry concentration between the United Sates and SC. See ed_lq-data-nov-ui
The top table is education. Note the difference in higher educational attainment. We know that persons with more education on AVERAGE are less affected than persons with a lower education. The lower table is LQ or the concentration of industries. Note how SC has a higher concentration of Manufacturing, Construction (HOMES) and Retail than the United states as a whole. All three of these industries have been hit hard. But also note the lower concentration in Health Care than in the United Sates as a whole. Of course, Health Care is one of the bright spots for new jobs, thus our not rec eivingthe full benefit of that industry. So there are significant differences between SC and the United States. Therefore, I would expect to see the SC unemployment rate rise much faster than the US in the months ahead. The GOOD news! We do not have much of a finance sector!
Unemployment
The Bureau of Labor Statistics has released their November Jobs numbers . . . Yikes! There are some interesting data in this release.
In November, the average workweek for production and non-supervisory workers on private nonfarm payrolls fell by 0.1 hour to 33.5 hours, seasonally adjusted–the lowest in the history of the series, which began in 1964. Both the manufacturing workweek and factory overtime fell by 0.2 hour over the month, to 40.3 and 3.3 hours, respectively. There are also questions about how new firms are calculated into the data.
Since new firms are not included in the sample data, that number is imputed. Unfortunately this data is from a historical perspective; thus on a downturn in employment, as we have had, the number of NEW jobs is usually over estimated. In other words, the employment situation was most likely worse than expected!
Does the establishment survey account for employment from new businesses? Yes; monthly establishment survey estimates include an adjustment to account for the net employment change generated by business births and deaths. The adjustment comes from an econometric model that forecasts the monthly net jobs impact of business births and deaths based on the actual past values of the net impact that can be observed with a lag from the Quarterly Census of Employment and Wages. The establishment survey uses modeling rather than sampling for this purpose because the survey is not immediately able to bring new businesses into the sample. There is an unavoidable lag between the birth of a new firm and itsappearance on the sampling frame and availability for selection. BLS adds new businesses to the survey twice a year.
The CES Birth/Death link describes this establishment survey process along with the detailed methodology. If you review the B/D document, you will see that, by using the 2007 model, in most months construction increases, a fact that we know was not the case in 2008 – thus an over estimation of employment. The method gives a little more detail, just in case you are bored today!
