Comments on economics, mystery fiction, drama, and art.

Friday, October 29, 2004

A Partial Explanation For Teen Labor Force Participation

As I discovered (see “Why Have Teen Labor Force Participation Rates Declined?”), I found two recent attempts to explain declining labor force participation for teens. The first, in a Bureau of Labor Statistics publication ["Issues in Labor Statistics Series" (http://www.bls.gov/opub/ils/pdf/opbils49.pdf), September 2002], argued that the explanation might be found in rising enrollment rates for teens.

The second, in a "Monthly Economic Outlook" prepared for Raymond James and Associates , Inc., by Scott J. Brown. In the Monthly Economic Outlook" dated April 8, 2004 (
http://www.raymondjames.com/pdfs/monit/mnth0404.pdf), suggests that rising unemployment might be an important factor.

(A third possibility is that teens have been increasingly working in the “underground economy,” off the books, being paid in cash, and not reporting themselves as in the labor force or employed. The usual explanation for a growing underground economy is rising marginal tax rates. This does not seem especially plausible for the 2000 – 2004 period.)

Both these hypotheses can be examined statistically. I extracted data on the labor force participation rates for teens, for unemployment rates for teen, and for school enrollment rates for teens from the BLS web site (
www.bls.gov) (both genders, all ethnicities, ages 16 – 19). Because the enrollment data is available only in the October Current Population Survey, I had a data set covering the years 1948 – 2003.*

I regressed the October LFPR on the October enrollment rate and current and nine monthly lags (of which only the first, second, fourth, sixth, and ninth were significant) in the unemployment rate. The resulting regression has an R-sq. of 0.655, and, in fact, higher enrollment rates and higher unemployment rates are associated with lower labor force participation. A one percentage point increase in enrollment rates (about a 1.25% increase) seems to be associated with a 0.159 percentage point reduction in labor force participation (about 0.4%).

The results track actual labor force participation pretty well—declining from the late 1940s through the mid-1960s, then rising until about 1980. After 1980, there was a slow, and erratic decline until the 1990s, when the teen labor force participation rate (at least in October) stabilized at around 48 – 50% (both for actual and predicted.And both the actual and predicted LFPR have declined since 2000. Including a couple of earlier years, the following table shows actual and predicted LFPRs for all teens:

…………….Actual……..Predicted
Year………..LFPR………..LFPR
1990………..53.9%.......... 50.7%
1995………..50.4%.......... 49.2%
2000………..49.6%.......... 49.1%
2001………..47.6%.......... 48.6%
2002………..45.6%.......... 47.0%
2003………..42.2%.......... 46.0%

In 1990, 1995, and 2000, the predicted LFPR is less than the actual, but only slightly. But in 2001, 2002, and 2003, the actual LFPR is not only less than predicted, the gap is steadily widening. Since 2000, (October) teen LFPRs have declined by 7.4 percentage points (about 15%). But predicted LFPR has declined (both because enrollment rates have increased and because unemployment rates have also increased) fell much less—by 3.1 percentage points (about 6.5%). And since October 2003, monthly teen LFPRs have been below their year-earlier level by about 1.4 percentage points.

So part of the puzzle of declining teen labor force participation is, in fact, not a puzzle. The recent declines, however, are larger than we would have expected, steeper than in any earlier period since World War II. And they appear to be continuing.

*I discovered, when I looked at the monthly LFPRs over this period that I had, in fact, missed some significant changes over time. For all teens, LFPR declined from a little more than 40% in the late 1940s to a little less than 40% by 1965. This decline was fairly consistent. From 1965 until about 1980, LFPR rose, again fairly consistently, to about 55%, after which if feel to about 50% in 1990, where it stabilized through that decade.

Tuesday, October 26, 2004

The Overtime Rules--They Haven't Gone Away

I wrote this in August, before I started this, and, since the proposed overtime rules are still around, I wanted to post it here:

The Department of Labor put in place new rules governing eligibility for overtime pay on August 23, 2004. In implementing these rules, the Secretary of Labor, Elaine Chao, emphasized that some additional workers would receive overtime coverage as a result of increasing the threshold below which overtime pay is required. The Economic Policy Institute, among other independent research organizations, focused on changes in the rules that will make it easier for employers to classify workers’ duties in ways that will make them ineligible for overtime pay.

What has so far gone without comment is the ways in which the new rules change the incentives for workers and for employers.

Workers will now have an incentive to want to work fewer hours per week.

Employers will now have greater incentives to increase the length of the work week, and to employ fewer workers.

Reduced Work Incentives

One of the overlooked aspects of overtime pay is that it provides an incentive for workers to want to work additional hours. What we have learned about the decisions workers make about their preferred weekly hours tells us that it pay for the additional hours has a significant effect on the length of the desired work week.

Put simply, if workers receive higher pay for additional hours than for the basic work week, they will be more willing to work additional hours. This premium pay helps compensate them for the real additional costs of spending more time at work, and less time with their families or in leisure activities. Overtime pay does just that.

Removing the overtime premium will have exactly the opposite effect. Consider a group of workers currently earning straight-time pay of $15 per hour, and working, in a typical week, 45 hours per week. The current overtime rules make additional hours worth $22.50 per hour. Under the new overtime rules, if their employers reclassify their work so that these workers lose their overtime eligibility, their pay for additional hours will be zero. Furthermore, with overtime eligibility, a worker choosing to work a little less would lose $22.50 for each hour not worked. That same worker, reclassified as ineligible for overtime pay, faces a cost of zero for working less.
Clearly, workers who are reclassified will now want shorter work weeks.

Fewer Jobs

When Congress enacted the Fair Labor Standards Act in the 1930s, one objective was to increase the cost to firms of using workers for long hours, and to provide employers with a greater incentive to hire additional workers (by reducing the length of the work week).

Employers have always had reasons for preferring extending the work week to expanding employment. There are fixed costs of hiring additional workers. These additional workers must be recruited, screened, and trained. These hiring costs are about the same whether someone works 30 or 50 hours per week. In addition, some benefits, such as employer-paid health insurance, are also per-worker, rather than per-hour, costs. An overtime premium raises the costs of additional weekly hours, and at least encourages employers to consider expanding employment rather than extending the work week.

Now, consider what happens when we remove the overtime premium, by making it easier for employers to reclassify workers, making them ineligible for overtime pay. In booms, the incentive employers have to extend the work week will be greater. In recessions, the incentive employers have to lay workers off, rather than reduce the length of the work week, will be greater. In normal times, the incentive employers have to extend the normal work week and reduce employment will be strengthened.

In booms, employers’ demand for labor will increase. If employers have reclassified workers, making them exempt from overtime pay, they will now find that the costs of extending the work week, relative to the costs of hiring new workers, will be lower than it was under the old rules. Indeed, the ease of reclassification into salaried jobs, rather than hourly jobs, adds even more to this outcome. So, in booms, extending the work week becomes, under the new rules, an even more cost-effective way of increasing use of labor. So there will be fewer jobs in booms.

In recessions, employers must chose between reducing the work week (work-sharing) and layoffs. If a firm’s employees, before the recession, are receiving overtime pay because of regularly-scheduled overtime, then reducing the work week can lead to relatively large savings. For employers, however, who have reclassified workers, the cost savings of reducing the work week are essentially zero. Their incentive is even stronger than it would have been to use layoffs to adjust to the recession. So there will be fewer jobs in recessions.

In “normal times,” the new rules will have a similar effect. By making it easier to reclassify workers, making them exempt from overtime pay, the new rules will reduce the costs of extending the work week, even if workers do not normally receive overtime. Workers who were formerly paid on an hourly basis at least received additional pay for any additional hours if the work week were to be extended. Once reclassified, they will not. The cost to the employer of extending the work week falls. At least some employers will now give additional consideration to reclassifying workers, extending the work week, and reducing the number of people they employ. So there will be fewer jobs in “normal times.”

Conclusions

There may be some debate about that will happen to the number of workers eligible for overtime pay under the new overtime rules. But no one who is familiar both with the economic theory and with the facts about how workers and employers make decisions would argue anything other than that these three propositions are true:

(1) Workers will want to work shorter work weeks.

(2) Employers will want to extend the work week and hire fewer workers.

(3) And tensions in the work place will increase.

Monday, October 25, 2004

Why Have Teen Labor Force Participation Rates Declined?

I argued earlier that a large percentage of the decline in overall labor force participation rates can be accounted for by the decline in teenage labor force participation. This, of course, raises the question of why teen LFPRs have declined.
I did what an academic does--I looked to see what someone else might have concluded, and I found only two recent attempts to explain the declines in teen LFPR.
One is in the Bureau of Labor Statistics "Issues in Labor Statistics Series" (http://www.bls.gov/opub/ils/pdf/opbils49.pdf), from September 2002. The analyst (Katie Kirkland) examines teen LFPR in July from 1948 through 2002, and examines the relationship between labor force participation and (summer) school enrollment from 1994 through 2000. Noting that LFPRs have declined and school enrollment has increased in the 1994 - 2000 period, the analysis concludes, "...the increasing proportion of teens enrolled in school during the summer and a drop in student's labor force participation contributed to the overall decline in teen summer labor force participation during the recent expansion. Data for October of each year indicate that labor force participation among high school students also dropped during the school year, although nonstudents were increasingly likely to participate in the labor force. Together, these facts suggest that, among teens, an increased emphasis was placed on school rather than work during the summer and during the school year."

It's worth noting that the causation here might run in the other direction--teens left the labor force for other reasons, and, as their alternative to labor force participation, chose school enrollment.

The second mention I could find occurs in a "Monthly Economic Outlook" prepared for Raymond James and Associates , Inc., by Scott J. Brown. In the Monthly Economic Outlook" dated April 8, 2004 (http://www.raymondjames.com/pdfs/monit/mnth0404.pdf), he writes: "Teenage labor force participation has fallen sharply, suggesting that older workers have taken a greater share of low-end jobs." This would suggest that the declining teen LFPR is a consequence of a deteriorating labor market for teen labor, which might show up (as well) in higher unemployment rates.

This is not terribly satisfactory, but whatever academic/statistical analysis is being done has presumably not yet been concluded, or published. I'll come back to this issue.

Tuesday, October 19, 2004

On Robert Lucas on Ed Prescott

Ed Prescott recently was the co-recipient of the Bank of Sweden Noble Memorial Prize in Economics. Some time ago, Robert Lucas had this to say about his work with Prescott, on search theory in labor markets:

“Ed had provided an equation that claimed to explain how labor markets work. It was my job to understand it and decide whether I agreed with this claim. This took me a while, but I saw that Ed had replaced an assumption of mine that workers who leave any one location hit on a new location at random -- maybe a worse location than the one they had left -- with the alternative assumption that searching workers were fully informed about options elsewhere and bee-lined for the best destination.”

(From David Warsh, “A Day in the Life of Ed Prescott,”
http://www.economicprincipals.com/issues/04.10.17.html

Unquestionably the approach Prescott suggested and that he and Lucas developed was interesting and exciting and provided many issues to develop. (One of the implications, of course, is that all unemployment is voluntary. If workers are fully informed and can “bee-line” elsewhere, then any unemployment is the consequence of choices made by workers NOT to do that.) It’s just too bad that Lucas’s initial assumption—random job search—and Prescott’s replacement of that assumption—perfectly informed job search—are both divorced from reality. Much more difficult than either of these, but also a much closer fit to what workers actually know, is imperfect knowledge of where to search, search that’s neither random nor fully informed. Of course, this is much harder to model.

Forensic science

The Chicago Tribune is running a series this week on forensic "science," with some excellent analyses of what amounts to yet more pseudo-science used to send people to jail. You can find the first three installments:

The introduction:
http://www.chicagotribune.com/news/specials/chi-0410170393oct17,1,5880839.story?coll=chi-newsspecials-hed

On arson investigations:
http://www.chicagotribune.com/news/specials/chi-0410180222oct18,1,2472960.story?coll=chi-newsspecials-hed

On identifying bite marks:
http://www.chicagotribune.com/news/specials/chi-0410190150oct19,1,3259394.story?coll=chi-news-hed

Monday, October 18, 2004

The Declining LFPR

We’ve seen that the recent apparent improvement in the labor market situation (as shown by the decline in the unemployment rate in the household survey) is mostly a consequence of a declining labor force participation rate. There’s an obvious question: What accounts for the decline in the LFPR? There are several possible answers.

First, it might be that enough households find themselves enough better off that they can achieve their desired income levels with only one person working. This does not, however, appear to be consistent with the data on median household and on real earnings.

Second, it might be that the aging of the population has led to a lower overall LFPR as a larger percentage of the population is now in age groups with historically low LFPRs. There’s probably some truth to this, but there is also an awkward fact. If LFPRs had remained unchanged since 2000, the growing fraction of the population age 55 and over would have pushed the overall LFPR down by about 0.1 percentage points (of a total 1 percentage point decline. The awkward fact is that LFPRs of older workers have been rising since 2000 (for the 55-64 age group, up from 56.7% in 1994, to 59.3% in 2000 to 62.1% in 2004; for the 65 and over group, up from 12.5% in 1994 to 13.1% in 2000 to 14.8% in 2004; LFPRs for September of each year). Taking this into account, we would expect an increase of 0.1 percentage points in the overall LFPR.

There is another demographic explanation however. Labor force participation of teenagers (ages 16 – 19) have declined. A lot. Look at the following table (LFPRs for September in each year).

…………All
Year…..16-19 White…..Black….Asian…..Hispanic
1994…..51.5%…55.0%…25.4%……………….40.0%
2000…..51.4%…55.0%…29.4%….34.5%…….44.8%
2003…..43.7%…47.2%…22.7%….22.4%…….34.0%
2004…..43.4%…46.7%…21.2%….22.3%…….37.6%

For purposes of historical perspective, the labor force participation rate of teens was 51.5% in September 1948, fell to 45.5% by 1964, rose erratically to 57.8% in 1978, then declined to and stabilized in the low-50s, where it remained until 2000. Since then, the teenage LFPR has declined by 8 percentage points in 4 years. As of September, it was at its lowest level in the 1948 – 2004 period, about 2 percentage points below the previous low.

The decline in teen LFPRs accounts for nearly 80% (0.8 percentage points of the 1 percentage point decline) in the overall LFPR since 2000. And, of course, it raises another question—why has labor force participation of teens declined, so much, so quickly?

Sunday, October 17, 2004

The Employment Situation

Now that the employment data for September have been reported, we can look at the changes in the labor market situation as shown in the household survey over the past decade. In the following table, I have reported data for civilian population, for the labor force, and for employment, as well as for the labor force participation rate, the employment population ratio, and the unemployment rate, for September of the indicated year.

Pop LF Emp LFPR EPR UR
1994 197,248 131,421 123,775 66.63% 62.75% 5.82%
2000 213,163 142,685 136,908 66.94% 64.23% 4.05%
2003 221,779 146,610 137,731 66.11% 62.10% 6.06%
2004 223,941 147,483 139,641 65.86% 62.36% 5.32%
2004H1 6.41%
2004H2 6.84%
2004H3 5.86%

Between September 1994 and September 2000, both the LFPR and the EPR increased, but the EPR rose much more rapidly (+2.35%, compared with 0.46%). The necessary consequence was that the UR fell dramatically.

Between 2000 and 2003, both the LFPR and the EPR fell. The LFPR continued to fall into 1994. Historically, the EPR falls during recessions, so the decline in the EPR may not be all that surprising. However, since the mid-1960s, there has never been a four-year period within which the LFPR actually fell. Again, not surprisingly, the UR increased from September 2000 to September 2003.

Between September 2003 and September 2004, the LFPR continued to fall. It's now at the same level that it was in October 1989. And the EPR rose fractionally (by 0.4%). The consequence is that the UR fell from 6.06% to 5.32%.

The decline in the UR clearly overstates the improvement in the employment situation for households. The final three rows in the table show what the UR would have been under three hypothetical situations. In the hypothetical labeled 2004H1, I calculate the effect on the UR had the LFPR remained at its September 1994 level. 2004H2 shows the UR if the LFPR had remained at its September 2004 level. Finally, 2004H3 is based on the LFPR that existed in September 2003. (A fairly simple time-series projection, taking into account the changes in the LFPR over time, suggests a "trend" LFPR of about 66.3%.)

Clearly, most of the decline in the UR in the last year is a consequence of the decline in the LFPR, not of an improvement in labor market conditions. Despite the increase in EPR in the last year, it remains significantly below its level of most of the last decade. For whatever reason, and for the first time in nearly 40 years, a smaller percentage of the population seeks to work.

The Draft

Bush has said that he does not intend to ask Congress to reinstitution a military draft. For any of a number of reasons, such a promise cannot be take seriously. If nothing else, the economics of the situation are likely to work against him--assuming he gets re-elected.

First, the current set of foreign policies seem to presume a greater use of military power than did the policies of previous administrations. So the demand for military personnel has increased.

Second, if the labor market recovers, with stronger (private-sector) job growth, then the opportunities available to potential volunteers will probably improve, causing the supply of volunteers to decrease. Additional factors causing a reduction in the supply of volunteers for military service include the rising risks of being in the military and the recent record of the administration in using stop-loss orders to extend terms of service of existing military personnel.

Third, Congress is unlikely to increase the compensation for military service, because of the large and growing federal government budget deficits.

Without a draft, what appears likely is a growing shortfall in recruiting efforts. So something will have to give. Either a re-elected Bush administration will be forced to scale back its foreign policy objectives. Or it will have to draft.

So tying your hands with a promise not to do something you might be forced to consider is at best imprudent, and at worst deceitful.

Open for Business

I have finally broken down. So welcome to SignsOfChaos (stolen from the title of a Roger Zelazny novel, if you must know). I will be commenting on economic issues (from my work life), on mystery fiction, drama, poetry, and art (from my life). I may from time to time rant about baseball as well. For tonight, a haiku I wrote on a trip to Greece:

The moon lights a path
through the water--unchanging--
but leading nowhere

posted 9:09 pm cdt