The call came shortly before midnight, waking the Economic Detective.

"Hello?" mumbled the investigator.

"It's Doc," the voice said. "You wanted to know when the numbers for 2008 came out," rasped the voice on the other end of the line. "They're out. Now."

"I'll get on it," said the Economic Detective, as the line clicked to disconnect.

It had been a while since the Economic Detective had closed the books on the case of what killed so many jobs for teens in 2007. He'd found that the Public Policy Gang was behind the massive teen job kill, but more than that, their fingerprints were all over the disappearance of nearly one million jobs from the U.S. economy that year.

But they were still out there. Still sticking their power-hungry hands into every place where they thought they might get even more for themselves, no matter the cost to anyone else. How many teen jobs might have gotten whacked in 2008 because of their schemes? That's why the Economic Detective now rose out of bed to go to work, because the Data Jocks at the BLS had finally accounted for all the minimum wage jobs in the U.S. for that year.

In 2008, the federal minimum wage had been increased from $5.85 per hour to $6.55 per hour on 24 July 2008. By all rights, that should increase the number of people counted as earning the federal minimum wage or less, since that would sweep up all the people who had been earning more than the old minimum wage but less than the new one.

That had certainly been the case between 2006 and 2007, when the number of people identified by the BLS' data jocks as earning the federal minimum wage or less had increased from 1,692,000 in 2006, when the minimum wage was set at $5.15 per hour, to 1,729,000 in 2007, when the minimum wage was increased to $5.85 per hour.

Sure enough, it was true again. The number of U.S. minimum wage earners had increased to 2,226,000 in 2008. The Economic Detective took the BLS data and broke down that figure by age group.

In 2008, teens accounted for 545,000 of the 2,226,000 minimum wage earning jobs, or 24.5% Meanwhile, young adults between the ages of 20 and 24 held 577,000 of these jobs, or 25.9%. Together, these two age groups account for just over half, or 50.4%, of all the jobs in the U.S. that earn the federal minimum wage of $6.55 or less.

But how does that compare to the number of teens earning $6.55 or less from before the minimum wage began being increased? To find out, the Economic Detective went back to his tool that modeled the distribution of hourly wage earners in the U.S. for 2005, which turned out to be remarkably accurate for 2006. Here, he found that in 2006, there were approximately some 1,058,366 teens employed at the level to where the minimum wage had been set in 2008. Taking the difference with the 545,000 teens actually counted as earning that level or less in 2008, that's some 513,366 jobs that have disappeared from the U.S. economy. Subtracting out the 261,717 teen jobs that disappeared between 2006 and 2007, that meant that between 2007 and 2008, some 251,649 additional teen jobs were now missing.

513,366 minimum wage jobs for teens is a lot to disappear over a two year period for a group that's dropped from 4.3% of the U.S. workforce to just 3.6% today. The Economic Detective wondered how many of the jobs for teens that disappeared betwene 2006 and 2008 might be accounted for by the disappearance of jobs paying the federal minimum wage of $6.55 per hour. The following chart gives the change in the average number of teens for each year from 2006 through 2008:

Here, we see that the number of employed teens dropped from an average monthly level of 6,161,250 in 2006 to an average monthly level of 5,576,000 for 2008, a decrease of some 585,250. With the decrease in the number of jobs paying teens $6.55 or less of 513,366 over the same time, that means that the disappearance of minimum wage jobs for teens accounts for 87.7% of the entire job loss for teens between these two years.

"At least," mused the Economic Detective, "the kids at the very bottom of the U.S. workforce aren't completely alone in taking the brunt of job losses in the U.S. anymore. Misery certainly loves company...."

The phone rang again. "The numbers are looking pretty brutal," said Doc. "There's so many more kids out there who aren't getting that entry level work experience that'll pay off for them as they get older. Not to mention the reality check on life that comes with that first job.

"But did you ever wonder what happens to the people at the places where these jobs aren't anymore? The ones left behind but still working? Think about that for a while." And once more, Doc hung up on the Economic Detective.

"Great," said the Economic Detective to the man on the phone who wasn't there anymore, before putting any more thinking on hold for the night and going back to bed.

Previously on Political Calculations

The Economic Detective series actually began as a fun way for us to revisit and re-create a lot of data and analysis we had done, but lost nearly a year ago. Here is the ongoing saga as it stands today:

• Natural Causes, or Not?: The BLS revised employment data going back to 2004, which provided the launch point for the Economic Detective, who discovered that "something broke" for teens seeking jobs between 2006 and 2007.

• Victim Autopsy: You can't have a good mystery unless you've got a corpse to analyze. Or a lot of corpses to analyze.

• Lining Up the Suspects: The potential culprits behind the crime are introduced and interrogated.

• Prime Suspect Revealed: Suspects are eliminated and the Economic Detective homes in on the guilty party.

• A Final, Gruesome Discovery: The case threatens to blow up and the crime gets much, much larger.

• A Late Night Call: The Economic Detective, and the case, is re-awakened as the teen job killing spree continues into 2008.

Labels: economics, minimum wage

- posted by Ironman at 1:44 PM | Permalink | Home |

"Your whole case against the Public Policy Gang on the teen job disappearances is going to fall apart," the voice on the other end of the line said. "How soon can you come to my office?"

"As soon as you want," the Economic Detective replied. It was, after all, the Paymaster. The man behind the money. All of it. Some say he's more powerful than the President. And now he had the Economic Detective's full attention.

"Good. I'll expect you shortly."

That's something that doesn't happen every day. If the man responsible for keeping the cash flowing through the economy is paying attention, the stakes must be pretty big.

The Paymaster's office resembles a bank. Classic architecture, complete with massive columns, elegant appointments, solid construction and armed guards. Everything is aimed at projecting an air of confidence, stability and control. The Paymaster fits his office, or rather, tries to fit. It doesn't always work.

"I've been following your case with interest," the Paymaster said. "There's a pretty big hole in your analysis that the Policy Gang could use to avoid responsibility."

"How do you mean?" the Economic Detective asked. "We've got their motive, we know how they pulled it off and the timing all fits. Hell, they even brag about what they did in jacking up the federal minimum wage. They're definitely the ones behind all those teen jobs going missing."

"Perhaps they are," the Paymaster evaluated. "The problem you face is showing that what they did, did in fact cause those jobs to vanish. Right now, all your case shows is that those job losses are correlated with the gang's activities. And we all know that correlation does not imply causation. Even freaks know that.

"Right now, you've shown that those Age 16 to 19 have had hundreds of thousands of job opportunities disappear, but that could be any job, paying anywhere from minimum wage on up to whatever Hannah Montana might be making these days.

"To really prove your case, you'll need to demonstrate that the jobs most affected by the change in the minimum wage are in fact the ones being lost. You need to show that it is the minimum wage jobs held by teens that are disappearing to prove causation. You have a lot of work left to do. "

And with that, the Economic Detective was back to work. But where to start?

Economic investigations are a lot like mining for gold. You have to sift through tons of crushed rocks in hopes of finding a single gold nugget. But if you can find that one, you might be able to find the vein of ore it came from and mine for more. The first nugget found in this case lay in a bed of revised data from the Bureau of Labor Statistics. Could the vein be nearby?

The Economic Detective hit pay dirt. The Data Jocks at the BLS put together a study every year on the characteristics of minimum wage workers. In it, they break out who makes the federal minimum wage or less each year. 2006 and 2007 were no different. We pulled the appropriate numbers from the reports:

Characteristics of Minimum Wage Workers
Year	Age 16-19	Age 20+	Total
2006	436,000 (25.77%)	1,256,000 (74.23%)	1,692,000 (100.00%)
2007	373,000 (21.57%)	1,356,000 (78.43%)	1,729,000 (100.00%)
Change	-63,000	100,000	37,000

From these numbers, it seemed that just 63,000 of the 245,500 jobs we knew had been lost for teen between 2006 and 2007 were minimum wage earning jobs. But something about these numbers didn't sit right. The Economic Detective kept digging, when suddenly he had a flash of insight.

The reports put together by the BLS' Data Jocks skimmed over the change in the minimum wage between 2006 and 2007. The report states that it's counting the number of people earning the minimum wage or less. It doesn't clearly identify what that minimum wage is, except in footnotes. But if the minimum wage changes from one year to the next, that invalidates comparisons between those years. In effect, the Economic Detective had only compared is the number of people making $5.15 per hour or less in 2006 with the number of people making $5.85 per hour or less in 2007.

And that's what didn't sit right in the numbers. Out of a population of over 300 million, with around 145 million people working, did it really make sense that the total number of people counted as earning the minimum wage would only go up by 37,000 people, from 1,692,000 in 2006 to 1,729,000 in 2007?

The right comparison would be between the number of people making $5.85 or less in 2006 and the number of people making $5.85 or less in 2007. But how many people were making $5.85 or less in 2006?

To find out, we turned to an old case, where we had worked out how many people there were earning any given hourly wage in the U.S. Although built using data from 2005, that same tool should provide a fairly accurate estimate of how many people there were making $5.85 or less in 2006.

Using the tool, we first used the data we had for the level of the federal minimum wage of $5.15 per hour in 2006 to see what kind of error there might be, before estimating the number of people making $5.85 per hour. The table below shows our results:

Estimated vs Actual Number of Hourly Wage Earners in 2006
Hourly Wage	Estimated Value	Actual Value	Error (Percentage of Actual)
$5.15 or Less	1,698,106	1,692,000	+6,106 (0.36%)
$5.85 or Less	2,681,439	Not Available	Unknown

Given what we know of the distribution of hourly wages in the U.S., that figure of 2,681,439 makes a lot more sense. To get to the number of teens earning $5.85 or less in 2006, we'll need to multiply this figure by their percentage representation in the workforce. Here, we'll split the difference between the percentages we found for minimum wage earning teens in 2006 (25.77%) and 2007 (21.57%) and use 23.67%, since we don't know the exact proportion of teens earning $5.85 or less in 2006. Using this approximation should put us fairly close to the actual number.

Doing that math, we would predict that there were roughly 634,717 individuals Age 16-19 earning $5.85 per hour or less in 2006. If we subtract the number of teens actually counted as earning the minimum wage or less in 2007 (373,000) from that figure, we obtain a value of 261,717, which would be the estimated number of teen jobs that disappeared for this age group between 2006 and 2007.

That would more than account for the 245,500 figure that we had found previously. We therefore find that *all* of the jobs for teens that disappeared between 2006 and 2007 were, in fact, minimum wage earning jobs. More specifically, they were positions whose hourly wage was $5.85 or less - the exact jobs directly affected by the Public Policy Gang's increase in the federal minimum wage.

And that's when the Economic Detective was startled by a final gruesome discovery. Having focused so much on the mysterious disappearance of jobs for teens in an otherwise good year for the U.S. economy, he hadn't looked beyond those numbers except in passing, but here they were now staring him right in the face. Looking at the estimated data again for 2006, there should be around 2,681,439 individuals earning an hourly wage of $5.85 or less. But in 2007, there were just 1,729,000 individuals counted as doing so.

That would put the Public Policy Gang's total job kill for their first minimum wage hiking spree in 2007 at least at 952,439, almost within spitting distance of 1 million jobs. That becomes well within spitting distance if you factor in population growth. No wonder the Paymaster was so concerned.

But for teens, the mystery had been solved. Now there was only one thing left to do in that case, so the Economic Detective did it. He closed the file.

Labels: economics, tool

- posted by Ironman at 5:34 AM | Permalink | Home |

Having ruled out "Dr. Doom" and The Demographic Kid from culpability for the massive reduction in the number of teen workers between 2006 and 2007, we had the Public Policy Gang rounded up for questioning. These gang members are about as two-faced and crooked as they come - talking fast to keep you distracted all while taking you for all you've got.

While waiting, the Economic Detective reviewed the case file, trying to work out how the gang might have pulled the caper off. From the victim autopsy, he recalled that teens weren't really the target of the hit - they were just in the way and took the blow. The real target was the source of the supply of jobs available for teens to fill, and that was the hard-working, honest and mostly mom-and-pop small business owners and operators who employ teens.

Then he came across a bit of information from The Demographic Kid that we had passed over previously. The Kid had said that employed teens were, compared to their older counterparts, inexperienced, unskilled and uneducated. That combination makes them pretty unattractive as employees, but the Kid said the teens made up for it by working cheap. So cheap, in fact, that teens made up 373,000 (21.6%) of the 1,729,000 U.S. hourly wage earners who made the minimum wage or less in 2007, as the chart on the left shows.

So there it is. The Public Policy Gang controls the minimum wage in the U.S. By raising the minimum wage, they could get good marks by appearing to care about low income earning people, with the most greedy and gullible mindlessly giving them their support in return for crumbs from the table, while taking massive "contributions" for themselves and their "pet" front organizations from big business and big unions, who can't stand the competition. And what better way to go after the competition than by making it too costly for them to compete?

It all fit - they had the motive and the opportunity. But the method was missing - how did they do it?

The interrogators slowly pieced together a coherent timeline from the various gang members we had rounded up. The timeline is given in the chart below, which shows the relative size of the Age 16-19, Age 20+ and the Combined number of employed individuals has changed as a percentage of their respective values from November 2006:

The Public Policy Gang had new leadership take over in November 2006. They made it clear in January 2007 that they were going to jack up the minimum wage to what they called a "living wage" level, and put enough pressure on the White House that they got them to go along with the scheme in May 2007. Employers with substantial numbers of employees who would see the mandated wage increase, knowing their costs were going to go up and not able to increase their prices to compensate because of competition, began cutting back on the number of jobs they created in step with the gang's progress toward their "living wage" goal. The employers had to if they were going to stay in business.

Teens, being disproportionately more likely to earn the minimum wage, were simply in the wrong place at the wrong time.

After May 2007, teens got a bit of a breather, but many got a nasty shock when the first increase in the mandated minimum wage took effect in the last week in July 2007. They over-reacted, letting lots of teens go in August, then brought a good number back the following month before continuing to reduce the number of employed teens at a more measured pace. By December 2007, at the peak of the expansion phase of the business cycle, the teen employment level was just 93.7% of its November 2006 value.

Teens got another break from the downward trajectory in April 2008, as the economy briefly picked up steam, but with another minimum wage hike scheduled for July 2008, it didn't last long. It wasn't until October 2008 that Dr. Doom's recession finally saw older workers take a larger share of job losses, as economic contraction took hold in the U.S. By the end of 2008, the teen employment level had dropped to 83.6% of its November 2006 value.

The Economic Detective was all set to wrap up the case, when a surprise call came in. It would seem that those 245,500 fewer employed teens in 2007, the difference between the average monthly number of those Age 16-19 counted as being employed in 2006 and 2007 during a time when the economy was still growing, were just a drop in the bucket as a gruesome discovery was made....

Labels: economics

- posted by Ironman at 1:28 PM | Permalink | Home |

Who, or what, could have inflicted the blunt force trauma visited upon America's teenagers that resulted in so many fewer teens with jobs?

That's the question the Economic Detective is left with after first establishing that the act was not the result of natural causes and that the first blow was struck on the victim sometime from November 2006 through January 2007. Meanwhile, the victim's autopsy revealed that the blunt force trauma wasn't aimed at teens so much as it was aimed at their employers. The teens, it would seem, were just the sorry saps who were in the wrong place at the wrong time.

Let's identify the likely suspects, or rather, those whose rap sheets show that they could pull off this kind of caper:

1. The Demographic Kid
2. "Dr. Doom"
3. The Public Policy Gang

Now, let's interrogate them in order to see what they'll tell us.

The Demographic Kid

This whole investigation got started because teens were going disappearing from the U.S. job market in numbers that were way out of sync with their representation within the workforce. We wanted to first find out if demographics were behind the change in employment figures. After all, if the proportion of teens in the working population at large took a dive, so would their percentage representation in the employment pool.

The chart to the right summarizes what we found when we dug into the numbers. From January 2005 through December 2008, the percentage of individuals Age 16-19 with respect to the working age (Age 16+) population of the U.S. stayed nearly constant, ranging within six hundredths of a percent of an average 7.29% throughout the period. Meanwhile, the percentage of employeed Age 16-19 year olds went from averaging 4.24% during 2005 and 2006, beginning their roller-coaster ride descent below that level in January 2007.

So, demographics didn't do it, but could The Demographic Kid explain why teens were so affected? We applied our unorthodox methods to get an answer.

The chart to the left confirms that the employment picture for everyone older than 19 throughout 2007 was pretty good. By the peak of the business cycle in December 2007, preceeding the current recession, the total Age 20+ workforce had grown by nearly 1.1 million. Meanwhile, the job market for those between the ages of 16 and 19 had turned south, with the U.S. economy shedding some 389,000 jobs over the same time, pulling the total change in total employment from November 2006 down to 707,000.

So, what makes the Age 16-19 workforce so different from their older counterparts in the job market? The Demographic Kid 'volunteered' that compared to the Age 20+ workforce, teen workers are inexperienced, unskilled and uneducated. Plus, the Kid thinks a lot of them smell really bad.

It was pretty clear that we weren't going to get any more useful information from The Demographic Kid, so we turned our attention to Dr. Doom....

"Dr. Doom"

Despite his resemblance to Nouriel Roubini, "Dr. Doom" turns out to be more of a what than a who. "What" would be negative economic forces that lead to recession, or more simply, a worsening economy.

Our first question: Could a worsening economy throughout 2007 have caused teen job losses? After all, history has shown that when an economy goes into recession and contracts, people lose their jobs all over.

We started in with the big picture. The chart to the right modifies our earlier chart showing the change in the number of Age 16-19, Age 20+ and Combined total employment levels by adding the inflation-adjusted GDP growth rate measured by the Bureau of Economic Analysis for each quarter covering the period from November 2006 through December 2008.

Here, we see that total employment levels tend to grow when the real GDP growth rate is flat-to-positive, while employment levels tend to fall when the economic growth rate turns decidedly negative, which we confirm by noting that economic expansion peaked in December 2007, and fell thereafter. We see the combined total employment level generally rise and fall in step with the growth recorded in the U.S. economy.

We also see that the teen employment levels throughout 2007 would appear to defy the economic growth trend. For example, with the real GDP growth rate clocking in at a high 4.8%, we would expect that teen employment levels would rise in response. Instead, we see a general decline in these levels throughout 2007, with an unexpected and unusually large step downward in August 2007, some five months into the period with the highest rate of economic growth during the year.

But, Dr. Doom makes a good point: not every part of the economy was growing in 2007. Some were, in fact, contracting pretty severely. One case in point: the U.S. housing sector. Here, after booming for several years, the U.S. housing market had begun slowing in mid-2005 and by mid-2006, sales of new and existing homes were slowing dramatically and the housing sector was, in Nouriel Roubini's words, "in free fall."

So, we would therefore expect employment levels to be falling in the residential construction industry and other economic sectors closely tied to U.S. housing markets, such as mortgage lenders, home furnishings, etc. during this period. The question for us is: "how many teen jobs are in these sectors?"

The answer is not many. In 2006, the average number of individuals Age 16-19 working in construction and extraction related fields was 296,000. That number dropped by 13,000 in 2007 to 283,000, a year in which the total number of jobs in all construction and extraction occupations increased by 28,000 from 9,507,000 to 9,535,000. That decline accounts for just 5.3% of an average decline of 245,500 jobs for teens between 2006 and 2007.

There's little doubt that the housing sector was declining in 2007, however that decline accounts for very little of the decline in the teen employment situation. No, far more teens are employed in other sectors of the economy, as the chart to the left demonstrates, and with just 5.3% of the average decrease in the number of teen jobs from 2006 to 2007, that percentage is pretty much in line with the percentage share of teens in the construction sector.

So we find that there's nothing really disproportionate happening here - Dr. Doom is off the hook.

That leaves just the Public Policy gang as the most likely culprit behind teen job losses. We'll put them under the hot lights of economic interrogation in short order....

Labels: economics

- posted by Ironman at 3:15 PM | Permalink | Home |

Our economic detective work continues today, after having determined that where teenagers are concerned, something "broke" in the U.S. economy sometime between November 2006 and January 2007. Whatever it was that broke led to hundreds of thousands of teenagers leaving the ranks of employed individuals in the U.S.

Since we've already shown that the number of employed teens as a percentage of the total number of individuals counted as being employed in the U.S. has declined since that time, today we'll take a closer look at the victim, performing an economic autopsy so to speak, to consider what that decline in the percentage representation of teens in the U.S. means in terms of actual numbers.

We'll first consider the overall trend in the size of the U.S. Age 16-19 population. The chart to the left indicates that the number of working-age teens in the U.S. has increased steadily over the period from January 2005 through December 2008, although the BLS' measure of the population isn't so steady. Here, we see that the BLS apparently adjusts its data each year in January, evident in the shifts upward and downward in the number of teens the government agency records as existing in each year.

As you might expect, with a steadily increasing population over time, would tend to lead to an increasing number of Age 16-19 year olds becoming members of the U.S. workforce. In fact, in the chart to the right, that's exactly what we see for both 2005 and 2006. But then we see an unexpected break in this expected pattern, beginning roughly around November 2006, when the number of members of the Age 16-19 workforce abruptly begins to decline.

Here, we see that decline accelerate beginning in January 2007, before adjusting to a slower rate of decline after August 2007. Remarkably, that sharp decline occurred well before the expansion of the business cycle peaked, as recently determined by the National Bureau of Economic Research when they pegged December 2007 as the beginning of the current recession.

In effect, this decline might be considered evidence of a blunt force trauma acting upon the employment status of U.S. teens!

Since the Age 16-19 Civilian Labor Force includes the number of teens both counted as being employed and those counted as being unemployed, we'll next look at just those teens counted as being employed. The chart to the left confirms the observation of a "blunt force trauma" occurring around November 2006 to January 2007, after which, the number of employed teens bled out of the U.S. workforce.

The chart to the left gives the number of teens who have "bled" out of the U.S. economy since the proximate November 2006 "break" point. Here, we see that after the initial break, 567,000 fewer teens were counted as being employed in August 2007. After this low point however, the number of working teens rebounded back up to where only 294,000 fewer than November 2006 were employed in October 2007. The overall decline then resumed, ending 2007 at a level 389,000 below the November 2006 level. All this decline in the number of working teens occurred when the U.S. economy was still expanding, which according to the NBER, it did all through 2007.

As we can see, some 1.02 million fewer teens are employed as of December 2008 compared to November 2006.

But are these individuals counted as being unemployed? We noted earlier that the size of the Age 16-19 Civilian Labor Force includes both the number of employed and unemployed members of this group. We wondered how the unemployment rate might reflect the changes we observe since November 2006. The chart below gives both the change in the number of 16 to 19 year olds counted as being unemployed since November 2008 (on the left scale) and the recorded Age 16-19 unemployment rate (right scale). The total number of Age 16-19 members of the Civilian Labor Force counted as being unemployed in November 2006, which provides the base against which the change in number of unemployed is measured in the chart below, is 1,089,000:

Since the BLS' unemployment data only counts those who were dismissed from their jobs by their employers, such as through layoffs, the data in this chart combined with the observed decline in the number of working teens in the period from November 2006 onward tells us something very important about the nature of the "blunt force trauma" that affected working teens in the U.S. What it tells is is that the decline in teen employment levels is largely the result of a disappearance of jobs for teens.

We can confirm this in the chart above by observing that even when the number of employed teens began decreasing sharply after November 2006, the change in the number of teens counted as being unemployed actually dropped for several months and didn't begin growing in earnest until May 2008, after the economy had entered into recession.

What this tells us is that whatever it was that caused the job market for teens to break down between November 2006 and January 2007, the blunt force trauma applied was directed at the supply of jobs created or available for teens to fill, rather than jobs that already existed.

And just who, or what, could do that? We'll soon have to line up the potential suspects....

Labels: economics

- posted by Ironman at 1:25 PM | Permalink | Home |

When multi-billion dollar firms seek to merge, the temptation to use every means possible to attain approval from regulatory authorities and realize tremendous financial gains can lead to some really questionable claims being advanced to achieve that end.

Today's example of junk science is our second where pseudoscientific practices have had real world legal implications. In this case, the discussion revolves around the situations that arose when pseudoscientific econometric analysis was presented by firms seeking the U.S. Department of Justice's Antitrust Division's approval to merge their businesses, but which was instead detected by the division's staff economists and subsequently challenged by the division's attorneys in legal proceedings.

Here are the relevant items that apply from our checklist for detecting junk science that come to play in today's example.

How to Distinguish "Good" Science from "Junk" or "Pseudo" Science
Aspect	Science	Pseudoscience	Comments
Goals	The primary goal of science is to achieve a more complete and more unified understanding of the physical world.	Pseudosciences are more likely to be driven by ideological, cultural or commercial (money-making) goals.	Some examples of pseudosciences include: astrology, UFOlogy, Creation Science and aspects of legitimate fields, such as climate science, nutrition, etc.
Inconsistencies	Observations or data that are not consistent with current scientific understanding generate intense interest for additional study among scientists. Original observations and data are made accessible to all interested parties to support this effort.	Observations of data that are not consistent with established beliefs tend to be ignored or actively suppressed. Original observations and data are often difficult to obtain from pseudoscience practitioners, and is often just anecdotal.	Providing access to all available data allows others to independently reproduce and confirm findings. Failing to make all collected data and analysis available for independent review undermines the validity of any claimed finding. Here's a recent example of the misuse of statistics where contradictory data that would have avoided a pseudoscientific conclusion was improperly screened out, which was found after all the data was made available for independent review.
Models	Using observations backed by experimental results, scientists create models that may be used to anticipate outcomes in the real world. The success of these models is continually challenged with new observations and their effectiveness in anticipating outcomes is thoroughly documented.	Pseudosciences create models to anticipate real world outcomes, but place little emphasis on documenting the forecasting performance of their models, or even in making the methodology used in the models accessible to others.	Have you ever noticed how pseudoscience practitioners always seem eager to announce their new predictions or findings, but never like to talk about how many of their previous predictions or findings were confirmed or found to be valid?

While the goal of commercial gain clearly identifies the motive behind the pseudoscience that will be discussed, the actual mechanisms by which it was carried out involve the cherry picking of data favorable to the parties seeking to merge their companies and the models they selectively presented to claim the proposed mergers would not negatively impact consumers.

Like good detective stories, it can make for fascinating reading. Today's story picks up below, with the recognition of the contributions that economists are making to the DOJ's antitrust division, before getting into how attorneys consider the data and analysis contributed by economists. We've emphasized a number of key points in boldface type....

---

Economists play a vital role in every significant merger investigation and litigation we handle. They have the expertise and facility to extract from mountains of raw data a variety of useful measures such as sales volumes, revenues, margins, market shares and others. They review bid data to determine the frequency with which companies encounter each other in the market, the closeness of competition and whether bidding behavior is characterized by coordination. They help us understand how competition works and how data can be used in assessing competitive constraints and predicting the likely impact of a merger. They develop theories of harm that not only guide their own quantitative work but also inform the work of lawyers who pursue the non-quantitative evidence on which cases rely. They also have a seemingly endless supply of lawyer jokes.

It would be impossible for me to list the many times and the many ways in which economists have helped me better understand the matters I have worked on at the division. They have predicted price increases in transactions across a variety of industries. They have modeled competitive effects for commodity products where price is determined by output decisions of suppliers. They have shown how mergers can give parties increased bargaining leverage, in industries ranging from cable and internet to healthcare. They have assessed party arguments about efficiencies and alleged mitigating effects in two-sided markets. To name just a few.

But I am not here simply to congratulate the profession for all its fine work. If hourly rates are any indication, you already know how valuable you are. I would also like to share a few candid thoughts about some of the economic evidence I have seen presented to the division. I am providing these observations from the perspective of a lawyer involved in merger decisions and litigation. My colleagues in the Economics Analysis Group undoubtedly have additional thoughts on these topics.

To begin with, you should understand that your primary audience when presenting economic evidence to the division consists of our economists. You need to demonstrate to them that your work is valid, that the data are reliable and that your conclusions are supported by accepted methods of economics. I can assure you that if our economists are not convinced, it is extremely unlikely that the legal team will second guess them on the economic work. I have been in front office meetings when arguments have broken out between the parties’ economists and ours. That is not where you want to be.

Second, when presenting to the lawyers at the agency, you need to be able to explain in simple terms the nature of the work you have done, the source of the information that feeds into that work and why it demonstrates the point you are trying to make. It is not persuasive to rattle off technical jargon and expect us to take on faith that a conclusion is valid simply because you obtained some value for a statistic. We need to understand the intuition behind the work and the basis for the results. Note also that if we can’t understand it at the agency, you should not expect to be able to explain it to a generalist federal judge who might have little or no experience with antitrust cases or merger challenges.

While we are on the topic of simple explanations, I believe the economic literature needs to become more accessible to non-economists. There is a problem that some in the profession have referred to as “mathiness” in economics. You write articles for each other, not for the rest of us. I can’t tell you how many times I have picked up a paper written by a leading economist on an important antitrust topic, generally absorbed the gist of the abstract and the introduction, then turned the page only to be confronted with complicated equations and a discussion I could not follow. Obviously scientists must devote considerable time explaining their work to each other and debating the merits of that work in terms the rest of us can’t be expected to grasp. But if the science is to be influential with lawyers who make enforcement decisions and with courts, there need to be sources to which we can turn, written professionally and by respected experts in the field, that provide clear explanations.

Third, I cannot emphasize strongly enough how important candor and credibility are in presenting economic evidence, just as it is with all interactions parties might have with the government or the courts. If your methodology is sound and the data support your conclusion, you shouldn't need to pull a fast one on us. Conversely, if you try to pull a fast one, we assume either the methodology is unsound or your conclusion is not supported. I offer two examples, one small and one large.

I recall sitting in a front office meeting receiving a presentation from the parties on a merger that was a close call. One of the pieces of economic evidence that was presented was a before-and-after study. The results were shown in the form of graphics to demonstrate that a particular event had not impacted a variable of interest. It was an impressive point. But then something caught my eye. In very small font beneath each graph was a listed data source. The data source for the “before” graph was different from the data source for the “after” graph. One was an objective third-party source and the other was described simply as the economist’s estimate. Of course I became immediately suspicious of the validity of this study. The parties had an explanation when asked and maybe it would have been credible if it had been volunteered upfront, but it raised a significant doubt in my mind. We ultimately cleared the transaction but, at least speaking for myself, that part of the presentation played no role in my thinking. This is the small example.

The other example I have in mind involves a regression model that purported to show that a prior transaction in the same industry but involving different products had no effect on prices. The model suffered from a number of flaws but one of the things that most bothered me about it was that the underlying data the expert used included the actual products at issue in the transaction under review and the expert’s model yielded statistically significant results showing that prices of those products had gone up. These results were not presented in the expert’s report but were discovered only after some dogged detective work by economists at the division who reviewed the backup data. The explanation for omitting these results from the report: the economist thought they were not robust because they varied when he did alternative runs of the model, which themselves were not disclosed. Whatever debate might have been available to the parties about the robustness of the results or the significance of the parties’ own model showing an increase in prices is of no moment; this should have been disclosed and presented with candor. The transaction was ultimately abandoned.

This brings me to my fourth point. You have to give us the information we need to test and verify your economic work and you have to give us time to do the work. It is our job to be skeptical of models that are constructed by experts employed by the parties. To be sure, the work that is presented to us is often immensely helpful. I have seen it play a big role in our decisions. But we will not take it at face value. So, to take a real world example that I recall, don’t present us with a model that purports to show the market is broader than we have always viewed it in a particular line of commerce, with only one week to go before the deal has to close. It’s not realistic to expect us to take that kind of risk on behalf of consumers without having the time and the information we need to fully vet the economic evidence you are presenting. This transaction also was abandoned.

Fifth, use lots of charts and pictures when presenting to lawyers. Text slides, not so much.

Sixth, recognize that the farther you stray from known methods of studying markets and the more customized and unusual the model you present, the more skeptical we will be that it provides meaningful information. Moreover, the fact that an expert can produce a model to achieve a certain result does not necessarily help us or a court decide the case. In this regard, while I believe the quality of economic work in the field of merger enforcement is generally first rate, we do see junk science from time to time. When data are mined and variables mixed and matched in such a way that a facially helpful conclusion can be drawn from a regression, even though an alternative result can easily be obtained by tweaking the inputs, we will likely view it as biased and unpersuasive.

I recently learned a term I hadn’t heard before from an episode of Last Week Tonight with John Oliver. This episode explored how scientific studies often have little value and yet are widely reported by the news media. The term Oliver used is P-hacking and if I understand it correctly, it describes how researchers can sift through data looking for relationships that will show up as statistically significant, rather than study a hypothesis objectively. This process can produce absurd results that appear to be valid but are not. Although I never knew the word for it, this seems to capture what happens with regression models presented to us on some occasions. That is, the expert runs ten regressions, discards nine unfavorable or insignificant results, and keeps the one that is favorable and significant. Don’t p-hack. It wastes time, it will be vulnerable to cross-examination and it undermines the legitimacy of the valuable contributions that economics can make to a case. As an aside, it also provides material for us lawyers to make economist jokes.

Finally, I believe economic evidence is best viewed as a complement to the other evidence in a case, not as a substitute for it. If the non-economic evidence points strongly in one direction, it is highly unlikely that a regression model pointing in the other will save the day. I don’t believe I have seen it happen in my three years at the division.

Moreover, it is wise to bear in mind that courts will be strongly influenced by the non-economic evidence. Opposing economic experts are both likely to be highly qualified professionals with many years of experience, a long list of impressive publications and an ability to recite various statistics that support their respective points of view. Confronted with such a battle of the experts, judges tend to look to other sources of evidence in deciding merger challenges. This is borne out by recent merger decisions.

In FTC v. Sysco, for example, the court extensively analyzed evidence relevant to the qualitative factors in Brown Shoe Co. v. United States in determining that broadline foodservice distribution was a market. Among other things, it had distinct product characteristics, distinct customers and the industry recognized the uniqueness of these services. The FTC’s economist also presented the results of an aggregate diversion analysis that showed a hypothetical monopolist over these services could profitably raise prices by a small but substantial and non-transitory amount. Rather than relying on these precise findings, the judge viewed them together with the rest of the evidence with which they were consistent. The defendant’s economist’s testimony, on the other hand, was not accepted by the court because it was contradicted by the weight of the qualitative evidence. In other words, it was the qualitative evidence that did the heavy lifting.

Similarly, in United States v. Bazaarvoice, the Antitrust Division introduced extensive evidence – including the parties’ own documents and lay witness testimony – showing that the market was limited to product rating and review (R&R) platforms, a market in which the merging firms were the two largest players. The division’s expert also presented the results of a hypothetical monopolist test on this question, but the court simply found that this “confirm[ed] what was apparent from the non-expert testimony,” i.e., that “R & R platforms in the United States for retailers and manufacturers are the relevant product and geographic markets.”

Sysco and Bazaarvoice are not unusual cases. One more example: in United States v. H & R Block Inc., the district court defined the market with reference to documentary and lay witness testimony, finding the economics testimony merely “tend[ed] to confirm” the market.

I will close by saying that this is an exciting time to be a lawyer or an economist practicing in the field of merger control. The agencies have been bringing cases across a variety of industries that have presented a range of important issues. With six years of experience under the 2010 Horizontal Merger Guidelines, the principles that are embedded in that document have been extensively debated and applied by the agencies and in some instances by the courts. Concepts like upward pricing pressure have become familiar to everyone practicing in the area. The economics profession should be rightly proud of its important role in bringing us to this place but must not rest on its laurels. There is endless work to be done to invent newer and better tools to study markets, develop consensus around the things that work and the things that don’t and figure out how to help the agencies and the courts make good decisions.

---

It is surprising to see the extent to which trust plays an important role in the antitrust arena, or for that matter, any legal setting, where once the validity of an economic analysis is called into question as described above, so is the ethical integrity of the analyst who produced it. Analysts would do very well to remember that one single lesson from today's example of junk science.

But perhaps the most surprising lesson to take away from today's example of junk science is how little weight the quantitative aspects of econometric analysis carries with respect to more qualitative evidence in antitrust cases. It is not just the data and the corresponding analysis that matters so much in these cases as it is the context of the full spectrum of evidence in which such data and analysis is presented. If the econometric analysis doesn't jibe with that larger qualitative picture, it will be discarded - and in the cases where it is based on junk science, as in the cherry picking example presented in the discussion above, or as in the other examples that we've presented as part of this series, rightfully so.

References

Gelfand, David. U.S. Department of Justice. "Deputy Assistant Attorney General David Gelfand Delivers Remarks at the Bates White Antitrust Conference". Washington, D.C., United States, Monday, June 6, 2016. [PDF Document]. Updated 29 September 2016. Excerpted with permission.

Lysy, Chris. Cherry Picking Data. Fresh Spectrum. [Online Illustration]. Accessed 6 October 2016. Republished with permission.

Labels: junk science

- posted by Ironman at 4:20 AM | Permalink | Home |

The December 2008 Employment Situation report from the Bureau of Labor Statistics did a lot more than just record big jobs losses between November and December 2008, it also revised virtually every employment figure going back to January 2004.

Because it did, we went back and revised our previous analysis of the employment situation for teens, or as the government defines them, those individuals counted as being employed between the ages of 16 and 19.

We'll begin by applying our statistical dating technique to the revised teen employment figures. Here, our goal is to first identify a period of relative equilibrium in teen employment numbers. Our measure of that relative equilibrium is the percentage representation of Age 16-19 employed individuals within the total number of employed individuals of all ages in the U.S.

When such a relative equilibrium exists, the percentage of working teens with respect to the whole U.S. workforce will be generally stable over time and the percentage of employed teens in any given month will bounce around a mean value as the result of natural variation.

We assume that those natural variations are random and will therefore fall into a normal distribution, for which 99.7% of all recorded values will be within three standard deviations of the mean value.

This gives us a tool that we can use to identify when that state of relative equilibrium is holding or has been broken. That tool is called a "control chart," which was invented in 1924 by Walter Shewhart of Bell Labs.

Today, Shewhart's innovation is in widespread use throughout the industrial world to measure and adjust production processes to ensure that manufactured products are produced to desired levels of quality. While control charts are used specifically where processes are controlled, the same methodology they provide for establishing whether observed changes in measured data over time are the result of natural variation (those that have common, or chance-driven causes) or are caused by determinable factors (those that have special or assignable causes) can be applied to non-controlled processes.

So, in our case, even though the percentage of teens within the U.S. workforce is very much not something that's the result of a controlled process, we can still get a lot of insight from using a tool very much like a control chart!

Had enough discussion? Let's get to the chart showing the percentage of teens within the U.S. workforce from January 2005 through December 2008.

In considering this chart, we find a period of relative equilibrium in the percentage of teens with the U.S. workforce running from January 2005 through December 2006. During these two years, the percentage of teens within the total employment of the U.S. averaged 4.24%. Using the three standard deviation measure of the data, we can then define the normal distribution in which 99.7% of all observations would be expected to fall: between 4.10% and 4.38%.

We next note that the recorded data falls below this expected range in May 2007 with a value of 4.07%, and also that it stays below the lower end of the expected range thereafter, continuing in a largely downward trend. This observation tells us that something "broke." More than that, it tells us that the cause is not the result of natural variation, but instead, something not-natural caused the equilibrium to break. But when?

From that point in May 2007, we next go backwards in time to identify if we can clearly determine when that downward trend began. Doing so, we find the last peak in the data occurred in November 2006, at 4.27%. The data prior to this point is all within the expected range.

With the peak being above the average for the baseline equilibrium period, we next identify the first point to break below the mean, which we find to be January 2007's value of 4.21%. This tells us that whatever caused the previous equilibrium to break most likely did so between November 2006 and January 2007.

From here, the next step is to identify the potential suspects involved in causing to narrow down just what caused the percentage of teens within the U.S. workforce to begin plummeting from bouncing around an average value of 4.24% over a two year period all the way down to the most recent data for December 2008, where we find that the percentage of teens within the U.S. workforce has declined to 3.62%.

But before we can do that, we'll need to autopsy the victim....

Labels: economics

- posted by Ironman at 12:06 PM | Permalink | Home |

In previous recessions, turkey production and consumption in the U.S. has risen. At least, that was the pattern in 1991, 2001 and even in 2008. But in 2009, something very different happened: both the quantity and price of turkeys plunged as the relative demand for turkeys plummeted.

That change is all the more remarkable because it came just a year after the quantity and price of turkeys in the U.S. had soared on increased demand, even though the U.S. economy had entered into recession nearly a year earlier in December 2007.

Why then was the experience of 2009 so different from 2008? Today, we're going to get to the bottom of that economic detective story.

We've already eliminated the price of turkey itself as being a potential culprit for the massive reduction in the quantity of turkeys produced in 2009. If the price of turkeys had spiked in 2009, we could have then attributed the fall in turkey consumption and its related production to the consumer reaction to extremely high prices, just like what occurred with the price and consumption of oil in the summer of 2008.

But it didn't, so we had to look elsewhere. We decided to take a closer look at the recession itself, to see whether the job loss related to it might be the culprit. After all, nothing kills off consumer demand quite like consumers who have lost their jobs.

Adapting our chart considering job loss by age group since total U.S. employment peaked in November 2007 (aka "Thanksgiving 2007"), we find that by Thanksgiving 2008, over 2.5 million fewer people were employed compared to Thanksgiving 2007. By Thanksgiving 2009, that figure reached 8.1 million.

We see that the recession certainly worsened between Thanksgiving 2008 and Thanksgiving 2009, but that doesn't explain why turkey production and consumption would rise by 6 million turkeys in 2008 when 2.5 million fewer people were working, before falling by 26 million in 2009 after an additional 5.5 million people were no longer counted as being employed.

And then it hit us. Looking at the age-related data on the chart, we see that a major shift began taking place beginning in November 2008 in who was no longer counted as working - individuals over Age 25!

We next combined the teens and young adults into a single Age 16-24 group and charted how their numbers in the workforce have changed over time compared to the Age 25+ group. We then charted the data, going back one additional Thanksgiving for good measure to get a larger sense of the overall trends.

What we find is teens and young adults bore the brunt of job loss in the recession up through October 2008. After October 2008, the employment situation in the U.S. significantly worsened for individuals Age 25 or older.

It's no secret why. October 2008 was the time when massive layoffs began taking place throughout the entire U.S. automotive industry production chain, which was devastated by the sudden change in consumer demand that occurred when oil prices spiked earlier that year.

But that age-based shift in job loss does explain why 2008 saw turkey consumption rise, while 2009 saw it decline dramatically.

Think about this question: what happens to a teen or young adult who loses their job? More often than not, they either continue to live at home, or in the case of young adults, they might simply move back home to live with their parents until they can find new work.

In both cases, the amount of food that their parents must provide increases. Since teens and young adults are less likely to host a traditional Thanksgiving dinner featuring a turkey as the centerpiece on their own, the amount of turkeys demanded increases as their parents host the annual holiday to feed their expanded households.

But that picture changes when it is the parents who are the ones who lose their jobs in an economic downturn, in addition to their older, working-age children. Suddenly, entire families must cut back, with U.S. turkey consumption falling as one result.

And many American families' Thanksgiving just isn't anywhere near the same.

Labels: economics, food, thanksgiving, turkey

- posted by Ironman at 7:07 AM | Permalink | Home |

We're going to start solving the mystery of why economic growth in the U.S. was so unexpectedly strong in the third quarter of 2013 in a place you wouldn't necessarily expect: the ports of China.

According to the U.S. Census Bureau, the United States' exports to China surged in October 2013, which we can see in the dramatic spike upward in the year-over-year rate of growth of U.S. exports to that nation for the month:

But that's putting it mildly. Left unstated is that the U.S. set at all-time record for the value of its exports to China:

Even Reuters took notice, but didn't really appreciate the magnitude of the month's trade, as the news agency reported from the Chinese perspective:

(Reuters) - China's imports surged in October as exports grew at their slowest rate in months, suggesting efforts to tilt the economy toward domestic demand may be offsetting the external weakness that has dragged on economic growth this year.

Customs figures showed import growth of 28.7 percent year on year in October, well ahead of the 23.0 percent forecast and far in excess of September's 20.9 percent growth rate.

Headline growth in exports meanwhile was its most sluggish in eight months, but strip out the traditionally volatile month of February and October's growth of 15.9 percent was the slowest since November 2009 when they shrank.

"We were expecting quite a deceleration as external demand continues to decline in Western economies," said Donna Kwok, an economist at HSBC in Hong Kong. "But the key thing to look at here is the strength of the domestic demand factors as imports grew nearly 29 percent."

Reuters goes on to look at where all the increase in the goods that China imported from other nations originated:

Imports from all three of China's key trading partners surged.

The rate of import growth from the United States accelerated the fastest at 20.5 percent over a year earlier, jumping by 7.6 percentage points from September's pace.

Imports from resource-rich Australia grew at 36.7 percent versus September's 33.4 percent, while European Union imports rose 28.2 percent versus 25.7 percent previously.

So most of the increase in what China imported from other nations came from the United States. But what was it that China imported in such great quantity?

We drilled down into the Census' data on what the U.S. exported to China. And what we found is that the U.S. exported a record amount of soybeans to China in October 2013, the value of which accounted for 22.6%, or nearly $1 out of every $4 of the value of all $13.060 billion worth of the U.S.' exports to that nation, accounting for nearly all of the year-over-year increase.

So where did all these soybeans come from? Because they definitely weren't there last year....

It turns out that weather conditions in 2013 across all of North America were nearly ideal for growing just about every crop grown in the U.S., with the result that U.S. farmers were harvesting bumper crops of just about everything they grow in August and September 2013.

Those bumper crops, in turn, solves a different mystery: why did the U.S.' GDP unexpectedly grow so much on increased business inventories in the third quarter of 2013?

Reuters reports on that story:

(Reuters) - The U.S. economy grew faster than initially estimated in the third quarter but weak demand and a pile-up in business inventories buoyed the case for the Federal Reserve to keep up its bond-buying stimulus for now.

Gross domestic product grew at a 3.6 percent annual rate instead of the 2.8 percent pace reported a month ago, the Commerce Department said on Thursday.

It was the biggest gain since the first quarter of 2012, but inventories accounted for almost half of the increase in growth.

Thanks to the international trade data, we now know what U.S. businesses were putting into their inventories in such great quantities! It would appear that the bumper crops produced by U.S. farmers were parked in silos and storage containers (or rather, inventories) in the third quarter of 2013. Those inventories subsequently enabled a very large surge in exports to China in October 2013, which is in the fourth quarter.

Now, what that means is that the U.S. economy isn't necessarily as strong as it would seem from its GDP data, since the vast majority of the organic growth in the economy, above and beyond the level the Fed has been able to generate through its quantitative easing programs, would appear to have occurred just in the agricultural industry. Which we note would also appear to be the result of unique factors that applied in 2013 that are well outside the control of that industry.

That in turn would explain why the stock markets are reacting so negatively to the increasing expectation that the Fed will begin to taper its QE programs in the first quarter of 2014 - they know there isn't enough sustainable organic growth in the U.S. economy to offset the loss of the Fed's stimulative QE programs.

And we've already demonstrated that increasing likelihood of a Fed QE taper is behind the recent resurgence of recessionary conditions in the U.S. economy, with the greatest impact being felt by the U.S. housing industry.

We should also note that the spending increases in the new budget deal in Congress would not meaningfully add to GDP in 2014, since it doesn't address the source of over 90% of the fiscal drag holding down the U.S. economy: President Obama's 2013 tax hikes.

Nor does it address the imposition of higher income taxes on low and middle-income earners, which will take effect in 2014 thanks to the multiple failures and mandates of the Affordable Care Act (popularly known as "Obamacare"), and which have put obtaining affordable health insurance coverage beyond their means unless President Obama agrees to address them.

On the whole, unless something dramatically changes in the very near future, 2014 won't be anywhere near as promising as what a lot of economists are currently suggesting. But then, all of the rest of them haven't figured out the soybean thing yet....

Labels: trade

- posted by Ironman at 4:11 AM | Permalink | Home |