Okay, so that was a big number - 3.3 million initial claims filed for unemployment insurance (Initial Jobless Claims, the topic of my last post) were filed the week ended March 21. If you look at a graph of claims historically, it looks like ... well, it looks like the face of El Capitan, as I predicted in that post.
The reporting the morning of the release was nothing short of comical. Some "business news" network's talking heads couldn't even get the terminology right. One network anchor wondered why the stock market was up so much after such a bad number. (Answer A: The Senate finally got their thumbs out of their backsides and passed the stimulus bill. Answer B: The House moved up the timing of their vote on the package and Speaker Pelosi indicated that it would pass. Answer C: The market had already priced in a really bad claims number. Estimates ranged from 1.5 million to more than 4 million.)
One national cable network anchor started an interview with one of their sister "business news" network's anchors by saying, "Even though this number was expected, it's still a shocker, right?" Yeah, I'm always shocked by news I expected. And the "business news" anchor then proceeded to conflate initial claims with ongoing (continued) claims in saying "this number" is only going to go higher next week because this is just the first time these people filed for benefits.
FYI, that "business news" anchor has a degree in Art History.
One local news channel - I won't name names but their initials are KSHB - seemed to conflate jobless claims with the unemployment rate: "The Labor Department saw an additional 3 million people seeking unemployment claims last week — the highest increase of unemployment claims the Labor Department has recorded since it began measuring seasonal unemployment. It also marked the highest level of insured unemployment since April 2018, when the unemployment rate was at 3.9 percent."
For one thing, unemployment claims are not "seasonal unemployment." And today's number wasn't due to seasonal factors (okay, so the virus is almost certainly seasonal, but I mean like seasonal layoffs in auto manufacturing as they shut down plants to re-tool). It was event-driven. And it isn't "insured unemployment," it's the number of people who filed claims to receive unemployment insurance benefits. Continued claims would be more akin to "insured unemployment." And finally, it's not the same as the unemployment rate.
But wait, it gets better: "According to figures released Thursday morning, 3.2 million people sought unemployment between March 14 and March 21." News flash: nobody seeks unemployment. Well, maybe those about to retire.
Okay, enough about the media dunderheads. To clarify things, let's look at the unemployment rate. And I chose this topic not only because of the dim-witted reporting, but because -
They're going to try to scare you again, this time when the next unemployment rate is released.
By definition, the unemployment rate is the number of unemployed, divided by the Civilian Labor Force. It is based on a survey of U.S. households. It is released on the first Friday of each month by the Bureau of Labor Statistics (with rare exceptions for Good Friday), and it is the unemployment rate as defined above, as of the end of the previous month. So on Friday, April 3, the March unemployment rate will be released. It will be considerably higher than the February rate of 3.5%, which was released on Friday, March 6.
And the media will try to scare you with it. Once again because they do not understand it in context, and they think it's their job to scare you. So a little history is in order.
The February 2020 unemployment rate of 3.5% - the same rate reported in September, November and December 2019 - is the lowest since 1969. So as was the case with initial jobless claims before the March 26 release, the unemployment rate recently has been at historically low levels, the lowest seen in a very, very long time.
More historical context: The record low unemployment rate was 2.5%, recorded in May and June of 1953 (at which point in time the Curmudgeon would not yet be unleashed on the world for more than five years). Thus the February 2020 rate is just a point above the record low. The record high of 10.8% was reached in November and December of 1982. The peak of the Great Recession was 10.0%, reached in October 2009.
And as is the case with Initial Jobless Claims, cyclical peaks in the Unemployment Rate tend to coincide with the ends of recessions - in fact, with the Unemployment Rate, it has been the case in every recession since the end of WWII that it has peaked after the recession officially ended. Usually just after, which again means that when the unemployment rate peaks, the worst is behind us.
Now that we've looked at record highs and lows, let's look at some trend data. But before we do, here's what the media will try to scare you with on Friday, April 3, when the March unemployment rate is released:
It's possible that it will be over 5%. And if not in March, it certainly could be in April.
So the headlines will read:
"Highest unemployment rate since (depends on how high it is; 6% would be 2014)!"
"Biggest one-month jump in the unemployment rate since (again depends on how high it is; 6% would be the biggest one-month jump ever, while 5.8% would be the biggest one-month jump since 1949)!"
And again, depending on how high the number is, "Unemployment rate nearly doubles!!!"
Okay. The average unemployment rate, going back to when the data was first recorded in 1948, was 5.73%. It has been below 5% only about 37% of the time, and of those 318 occurrences, 53 have come in the last 54 months. When Janet Yellen was Fed Chair (a dark time in economic history), her target unemployment rate was 5%. That was considered "good" - in fact, good enough to start raising interest rates to stave off inflation resulting from rising wages.
"Full employment" has long been considered to be 6%. (Technically speaking, "full employment" is one-half of the Fed's dual policy mandate: to promote stable prices, i.e. maintain low inflation, and to promote "full employment", which basically means "get as many people working as you can, but don't worry about the small percentage that might be more or less unemployable." Plus, there is friction in the number in that people in the survey may be in and out of a job at a given point in time, even in a good year.)
The unemployment rate has been above 6% about a third of the time, and we haven't been in recession a third of the time since 1948 - in fact, we've only been in recession about 14% of the months since January of that year. So the unemployment rate has been above 6% through about 20% of the non-recessionary months since just after WWII.
So understand that while an unemployment rate of 5% or 6% isn't as good a situation as we had before this virus hit, it's actually pretty good historically. And even if it goes higher, it's not likely to for long. Unless, again, you defy the scientists - the qualified ones like Drs. Fauci and Birx, not the self-described ones on Facebook - and believe that this thing isn't going to prove to be seasonal, and that this is going to last many months. But if that's the case, you're reading the tin-foil-hat posts on Facebook, not this, so I'm guessing that if you're still with me, you're okay.
Expect a number above 5%, and don't be "shocked" by what you expect. And know that in context, it's not the Great Depression. Far from it. In fact, most of the recessions during which unemployment peaked below 8% have been relatively mild, and short in duration.
Think of it this way: it's like the folks who have only been looking at mortgage rates since the Great Recession, and they think that a mortgage rate above 5% is "high." Historically speaking, it is not. My first mortgage carried a rate of 10.5%. It has only been during this very recent (since 2009) period of extreme and unprecedented accommodation by the Fed that mortgage rates have been below 5%. They've been as high as 18.6% - and people still bought homes. So just as sub-5% mortgage rates are not "normal," sub-5% unemployment is not "normal."
One final note: lest anyone be offended by my comments about the media: I have nothing against people with majors in Art History, Theology, or Interdisciplinary Studies. Heck, I once had an investment sales rep who worked for me who had a History degree.
Okay, bad example. I fired him.
My issue is when those folks join the media and try to apply their economic ignorance and innumeracy to the economy and the markets, and they sensationalize things to try to scare people, because if you're terrified, you're riveted, and they can sell more of their their sponsors' crap.
But look on the bright side: if you're a newly-minted grad with a degree in Etruscan Civilations facing six figures of student loans from an Ivy League college you couldn't afford because going to juco to be a paralegal was beneath you, you may be in luck: there's a promising future ahead of you. In "business journalism."
Saturday, March 28, 2020
They're Going to Try to Scare You Again. Don't Let Them. Part III.
Two posts ago, I addressed how the media would try to use what would be a significant increase in Initial Jobless Claims to try to scare you. I talked about the results of that release in my last post, which addressed the unemployment rate. This post will focus on Continued Claims, which I introduced in the post about Initial Claims.
But first - the Curmudgeon would like to offer his humble thanks to all who shared the link to the post on Initial Claims. It was the most-read post in the history of this blog. So keep the shares coming, and maybe we can keep a lot more people from being scared (for the record, I don't make a cent from doing this). And if you liked the most recent post, feel free to share it too.
As a reminder, Continued Claims is the aggregate number of people receiving unemployment insurance benefits in a given week. There is a one-week lag in reporting the number vs. Initial Jobless Claims. Thus the Continued Claims number that was released on Thursday, March 26 showed total claimants as of the week ended March 14, whereas the Initial Claims number released that morning was for first-time filings as of the week ended March 21.
Now, it stands to reason that if initial claims spike, the following week's continued claims will jump. You may recall that initial claims captures people who file for the first time, then the next week they fall out of that total and move into the continued claims number, where they stay until they no longer qualify for benefits (found a job, no longer looking, benefits expired, etc.).
The move from initial to continued claims is not a precise additive transition. For example, initial claims for the week ended Mar. 7 were 211,000. The increase in continued claims from the week ended Mar. 7 to that ended Mar. 14 was 101,000, not 211,000. This could be due to some of the first-time filers going back to work (unlikely), some of the previously reported continued claimants going back to work (thus there were people moving into and out of the continued claims pool), or delays between filing and receiving benefits.
So, the continued claims total reported on Thursday, April 9 for the week ended Mar. 28 is unlikely to show an increase equal to the 3.3 million initial filers that we saw in the initial claims release for the week ended Mar. 21, which came out last Thursday and was the topic of that original post. But there will be a very large increase, easily over 1 million. The most recent reading was just over 1.8 million, so the number will increase by more than half, and could certainly double, or worse.
And the media is going to try to scare you with it.
The headlines this time will be:
"Highest continuing jobless claims since (probably 2012 or 2013, depending on the number)!"
"Biggest one-week increase in continuing claims in history!" (It will be.)
"Continuing jobless claims doubled in just one week (if the number does double)!"
And, as they always do because they are woefully under-qualified to report on economic data, they will muck up the terminology they use and the ways they abuse the data to the point that my Curmudgeonly head explodes, and I find myself standing in front of the television and cussing at it. (I'm doing a lot of that these days - my dogs are very confused.)
However, you should not be fearful as a result of the number, for several reasons. The first is that you now understand what it means, and you expect the large increase, just as you expect another increase the following week, and further increases in the weeks after that until initial claims subside and people go back to work, at which point the continued claims number will begin to fall. And, unlike the media simpletons, you are not shocked by that which you expected.
Second, let's look at historical highs, lows and trends for perspective, as we have done in the last two posts. The most recent continued claims total of 1.8 million as of the week ended Mar. 14 was the first reading above the 1.8 million mark since April 14 of last year. Things weren't so bad back then, right? So even though the last reading was up by 101,000 from the prior week, that shouldn't alarm us - historically, 1.8 million continued claims is a very low level.
How low? Well, in October of last year, continued claims were the lowest since 1973 at just under 1.65 million - pretty darn close to where we are today. The average from that low to the most recent reading is right at 1.7 million. So a trend between 1.65 million and 1.8 million is about as low as we're gonna get.
And if we go back to the early days of this data series, we find that the record low was 988,000 in May 1969. (The U.S. population was more than a third smaller, too - remember, denominators matter.)
Now let's look at highs. The record for continued claims was 6.635 million in May 2009, just after the end of the Great Recession. (By the way, for anyone unfamiliar with what I'm referring to when I reference the Great Recession, it's a name commonly given the most recent recession, from 2008-09, which was brought on by the housing collapse and subsequent financial crisis.) Next highest was just after the 1981-82 recession, in November 1982, at about 4.7 million. And the third highest total followed the 1973-75 recession, at about 4.64 million in May 1975. No other recession since the claims data was first recorded in 1967 has seen continued claims reach 4 million.
What's significant about these highs and lows? First - and this is very important in terms of mitigating your risk of the numbers scaring you, so pay close attention - these spikes resulted from the three longest recessions in the history of the data. The Great Recession lasted 18 months, the longest downturn since the Great Depression of the 1930s. The recessions of 1973-75 and 1981-82 lasted 16 months each. All three of these recessions were followed by slow recoveries. The next-longest recession since WWII lasted 11 months, and the average duration of recessions since WWII, excluding the three noted above, has been nine months.
Generally speaking, shorter recessions see shorter peaks in continued claims. However, it does not hold true that the shorter the recession, the lower the peak in continued claims. So if we see a peak above 4 million, as I strongly believe we will, do NOT assume that means we're in for a 16-month downturn. The 1980 recession lasted only six months, yet continued claims peaked shortly after it ended at slightly more than 3.9 million. I do not anticipate a prolonged downturn or a slow recovery, for reasons I have explained in previous posts.
Second, denominators matter (okay, so I'm sounding like a skipping record - younger readers will need to google that reference). So let's divide each of the three most severe peaks by the Civilian Labor Force at the time.
To recap, the peak of the Great Recession was 6.635 million continued claims. As a percent of the Civilian Labor Force at that time, continued claims were 4.29%. At the 1982 peak, continued claims were 4.19% of the labor force. And at the 1975 peak, they were 4.90% of the labor force - higher than in 1982 or 2009.
So, to exceed those levels on a percentage of the labor force basis, we'd have to see continued claims of more than 8 million - exceeding the most recent reading by more than 6 million. And yet, even if that happens, it's not the most significant indicator. Why?
Because the duration of unemployment matters most. Weeks unemployed peaked in July 2011 at 40.7. (That was 25 months after the recession ended.) The recovery from that recession was the slowest ever, due in part to fiscal policy and the causal factors of the recession. That record was far and away higher than any previous duration of unemployment. After the 1982 recession, weeks unemployed peaked at about 17. No other recession even came close. And as I've noted, I don't expect the duration of unemployment to reach the "worst-since" levels we've seen previously, for a variety of fundamental reasons.
Finally, the peak in continued claims always occurs after a recession officially ends, as is the case with the unemployment rate. So when that peak arrives, the worst is over.
The consensus forecast for the initial jobless claims release scheduled for Thursday, April 2 (for the week ended March 28) is around 3.5 million, which would exceed last week's record level by about 200,000, and set a new record. I'll go out on a limb and say that'll probably be the highest level we'll see, though there will still be new filings each week well in excess of the 210,000 or so trend that we were seeing before the pandemic response shut things down. So we probably will see a peak above 8 million continued claims, and the peak will probably occur within a few weeks. But that still doesn't foretell a prolonged downturn.
The duration of unemployment this time should be short. If we look at the model data from the University of Washington's Institute for Health Metrics and Evaluation (the same model that has been referenced by Drs. Fauci and Birx - and more on models in a subsequent post), we see that in most states, the curve flattens within a couple of months. The modelers and experts seem to agree that this virus is likely to be seasonal. So if we assume that things will open back up within a couple of months, either on a rolling basis state-by-state based on curve flattening projections, or on a wider basis, then the duration of unemployment is likely to peak at a few weeks over and above the levels we were seeing before the virus hit.
A caveat about that. Those pre-pandemic levels were already at about 21 weeks, which was above the peaks that followed every downturn before the Great Recession. These levels were unusual given how strong the labor market was through February, and resulted from the government's response to the Great Recession. As part of that response, the amount of time someone could draw unemployment was extended to account for the extremely slow job growth at the beginning of the recovery.
However, that amount of time was never reduced in many large states when things normalized, and thus more recently was well beyond the time it would have taken to find a job in the prevailing labor market. So many people had a disincentive to work, if they could live off their unemployment checks. As a result, if the duration of unemployment resulting solely from the pandemic is about 11 to 16 weeks - as the curve data suggest - adding that to the "base" levels we saw in February could result in a reported duration of unemployment of more than 30 weeks, which would be historically pretty high. And would, of course, be distorted by the media to try to scare you.
One last point. My numbers could be wrong in terms of the level of the peak in claims or in the unemployment rate. Those peaks could be much higher. It doesn't matter this time - what's critical is the duration of unemployment. As noted before, there will (and has already been) a sharp, cliff-like spike in the unemployment metrics, but every indication from the medical data is that the duration will not be as long as seen in many recessions.
Remember that when the media tries to scare you with things like a recent St. Louis Fed blog post (which is already being sensationalized) that estimated that 47 million jobs would be lost in the second quarter and the unemployment rate could hit 32%, a record high. Even that blog post acknowledged that it is not the level but the duration of unemployment that is important in this environment. And their estimate also assumes that no part of the economy opens back up until after Q2, which contradicts the data regarding when the curve flattens throughout the U.S.
Stay calm, stay safe and stay sane.
But first - the Curmudgeon would like to offer his humble thanks to all who shared the link to the post on Initial Claims. It was the most-read post in the history of this blog. So keep the shares coming, and maybe we can keep a lot more people from being scared (for the record, I don't make a cent from doing this). And if you liked the most recent post, feel free to share it too.
As a reminder, Continued Claims is the aggregate number of people receiving unemployment insurance benefits in a given week. There is a one-week lag in reporting the number vs. Initial Jobless Claims. Thus the Continued Claims number that was released on Thursday, March 26 showed total claimants as of the week ended March 14, whereas the Initial Claims number released that morning was for first-time filings as of the week ended March 21.
Now, it stands to reason that if initial claims spike, the following week's continued claims will jump. You may recall that initial claims captures people who file for the first time, then the next week they fall out of that total and move into the continued claims number, where they stay until they no longer qualify for benefits (found a job, no longer looking, benefits expired, etc.).
The move from initial to continued claims is not a precise additive transition. For example, initial claims for the week ended Mar. 7 were 211,000. The increase in continued claims from the week ended Mar. 7 to that ended Mar. 14 was 101,000, not 211,000. This could be due to some of the first-time filers going back to work (unlikely), some of the previously reported continued claimants going back to work (thus there were people moving into and out of the continued claims pool), or delays between filing and receiving benefits.
So, the continued claims total reported on Thursday, April 9 for the week ended Mar. 28 is unlikely to show an increase equal to the 3.3 million initial filers that we saw in the initial claims release for the week ended Mar. 21, which came out last Thursday and was the topic of that original post. But there will be a very large increase, easily over 1 million. The most recent reading was just over 1.8 million, so the number will increase by more than half, and could certainly double, or worse.
And the media is going to try to scare you with it.
The headlines this time will be:
"Highest continuing jobless claims since (probably 2012 or 2013, depending on the number)!"
"Biggest one-week increase in continuing claims in history!" (It will be.)
"Continuing jobless claims doubled in just one week (if the number does double)!"
And, as they always do because they are woefully under-qualified to report on economic data, they will muck up the terminology they use and the ways they abuse the data to the point that my Curmudgeonly head explodes, and I find myself standing in front of the television and cussing at it. (I'm doing a lot of that these days - my dogs are very confused.)
However, you should not be fearful as a result of the number, for several reasons. The first is that you now understand what it means, and you expect the large increase, just as you expect another increase the following week, and further increases in the weeks after that until initial claims subside and people go back to work, at which point the continued claims number will begin to fall. And, unlike the media simpletons, you are not shocked by that which you expected.
Second, let's look at historical highs, lows and trends for perspective, as we have done in the last two posts. The most recent continued claims total of 1.8 million as of the week ended Mar. 14 was the first reading above the 1.8 million mark since April 14 of last year. Things weren't so bad back then, right? So even though the last reading was up by 101,000 from the prior week, that shouldn't alarm us - historically, 1.8 million continued claims is a very low level.
How low? Well, in October of last year, continued claims were the lowest since 1973 at just under 1.65 million - pretty darn close to where we are today. The average from that low to the most recent reading is right at 1.7 million. So a trend between 1.65 million and 1.8 million is about as low as we're gonna get.
And if we go back to the early days of this data series, we find that the record low was 988,000 in May 1969. (The U.S. population was more than a third smaller, too - remember, denominators matter.)
Now let's look at highs. The record for continued claims was 6.635 million in May 2009, just after the end of the Great Recession. (By the way, for anyone unfamiliar with what I'm referring to when I reference the Great Recession, it's a name commonly given the most recent recession, from 2008-09, which was brought on by the housing collapse and subsequent financial crisis.) Next highest was just after the 1981-82 recession, in November 1982, at about 4.7 million. And the third highest total followed the 1973-75 recession, at about 4.64 million in May 1975. No other recession since the claims data was first recorded in 1967 has seen continued claims reach 4 million.
What's significant about these highs and lows? First - and this is very important in terms of mitigating your risk of the numbers scaring you, so pay close attention - these spikes resulted from the three longest recessions in the history of the data. The Great Recession lasted 18 months, the longest downturn since the Great Depression of the 1930s. The recessions of 1973-75 and 1981-82 lasted 16 months each. All three of these recessions were followed by slow recoveries. The next-longest recession since WWII lasted 11 months, and the average duration of recessions since WWII, excluding the three noted above, has been nine months.
Generally speaking, shorter recessions see shorter peaks in continued claims. However, it does not hold true that the shorter the recession, the lower the peak in continued claims. So if we see a peak above 4 million, as I strongly believe we will, do NOT assume that means we're in for a 16-month downturn. The 1980 recession lasted only six months, yet continued claims peaked shortly after it ended at slightly more than 3.9 million. I do not anticipate a prolonged downturn or a slow recovery, for reasons I have explained in previous posts.
Second, denominators matter (okay, so I'm sounding like a skipping record - younger readers will need to google that reference). So let's divide each of the three most severe peaks by the Civilian Labor Force at the time.
To recap, the peak of the Great Recession was 6.635 million continued claims. As a percent of the Civilian Labor Force at that time, continued claims were 4.29%. At the 1982 peak, continued claims were 4.19% of the labor force. And at the 1975 peak, they were 4.90% of the labor force - higher than in 1982 or 2009.
So, to exceed those levels on a percentage of the labor force basis, we'd have to see continued claims of more than 8 million - exceeding the most recent reading by more than 6 million. And yet, even if that happens, it's not the most significant indicator. Why?
Because the duration of unemployment matters most. Weeks unemployed peaked in July 2011 at 40.7. (That was 25 months after the recession ended.) The recovery from that recession was the slowest ever, due in part to fiscal policy and the causal factors of the recession. That record was far and away higher than any previous duration of unemployment. After the 1982 recession, weeks unemployed peaked at about 17. No other recession even came close. And as I've noted, I don't expect the duration of unemployment to reach the "worst-since" levels we've seen previously, for a variety of fundamental reasons.
Finally, the peak in continued claims always occurs after a recession officially ends, as is the case with the unemployment rate. So when that peak arrives, the worst is over.
The consensus forecast for the initial jobless claims release scheduled for Thursday, April 2 (for the week ended March 28) is around 3.5 million, which would exceed last week's record level by about 200,000, and set a new record. I'll go out on a limb and say that'll probably be the highest level we'll see, though there will still be new filings each week well in excess of the 210,000 or so trend that we were seeing before the pandemic response shut things down. So we probably will see a peak above 8 million continued claims, and the peak will probably occur within a few weeks. But that still doesn't foretell a prolonged downturn.
The duration of unemployment this time should be short. If we look at the model data from the University of Washington's Institute for Health Metrics and Evaluation (the same model that has been referenced by Drs. Fauci and Birx - and more on models in a subsequent post), we see that in most states, the curve flattens within a couple of months. The modelers and experts seem to agree that this virus is likely to be seasonal. So if we assume that things will open back up within a couple of months, either on a rolling basis state-by-state based on curve flattening projections, or on a wider basis, then the duration of unemployment is likely to peak at a few weeks over and above the levels we were seeing before the virus hit.
A caveat about that. Those pre-pandemic levels were already at about 21 weeks, which was above the peaks that followed every downturn before the Great Recession. These levels were unusual given how strong the labor market was through February, and resulted from the government's response to the Great Recession. As part of that response, the amount of time someone could draw unemployment was extended to account for the extremely slow job growth at the beginning of the recovery.
However, that amount of time was never reduced in many large states when things normalized, and thus more recently was well beyond the time it would have taken to find a job in the prevailing labor market. So many people had a disincentive to work, if they could live off their unemployment checks. As a result, if the duration of unemployment resulting solely from the pandemic is about 11 to 16 weeks - as the curve data suggest - adding that to the "base" levels we saw in February could result in a reported duration of unemployment of more than 30 weeks, which would be historically pretty high. And would, of course, be distorted by the media to try to scare you.
One last point. My numbers could be wrong in terms of the level of the peak in claims or in the unemployment rate. Those peaks could be much higher. It doesn't matter this time - what's critical is the duration of unemployment. As noted before, there will (and has already been) a sharp, cliff-like spike in the unemployment metrics, but every indication from the medical data is that the duration will not be as long as seen in many recessions.
Remember that when the media tries to scare you with things like a recent St. Louis Fed blog post (which is already being sensationalized) that estimated that 47 million jobs would be lost in the second quarter and the unemployment rate could hit 32%, a record high. Even that blog post acknowledged that it is not the level but the duration of unemployment that is important in this environment. And their estimate also assumes that no part of the economy opens back up until after Q2, which contradicts the data regarding when the curve flattens throughout the U.S.
Stay calm, stay safe and stay sane.
Wednesday, March 25, 2020
They're Going to Try to Scare You. Don't Let Them.
In case you've been living under a rock for the last several weeks ...
Okay, wait. For many of us, it feels like that. Let me start over.
In case you haven't watched or read the news or been on social media for the last several weeks, I have some important information for you:
The news media believes that their job is to scare the bejeebers out of you, make you an anxious wreck, increase your blood pressure, and paralyze you with abject fear.
Yes, it's despicable. It is beneath contempt, and if they were capable of empathy, they would be ashamed. But they're not.
So on Thursday morning, March 26, 2020, in the middle of this coronavirus pandemic and the ensuing government-mandated shutdown of our economy, when they've already made you anxious enough -
They are going to try their best to scare the crap out of you. Don't let them. Read on, and you'll be far more informed than they will ever be.
Every Thursday morning, the U.S. Employment and Training Administration (ETA), a division of the Department of Labor, releases Initial Jobless Claims for the week ended the prior Saturday. So on Thursday, March 26, they will release the number for the week ended Saturday, March 21.
And it's going to be a bad number. The media is going to go into a frenetic Chicken Little dance over it.
Because they have no idea what the number means. It doesn't mean what they think it means. It doesn't mean what you think it means. Let me explain.
Initial Jobless Claims is a point-in-time number. It means something that week, then it means nothing the next week. It is a trend number, in that it is only important in context of the trend. And as the Curmudgeon has long been fond of pointing out, one week doth not a trend make.
The claims number represents the number of Americans who filed for unemployment insurance benefits in a given week. If those Americans are still unemployed the following week, they fall out of the number, because at that point, their claims are no longer initial. Another way to label it is "first-time unemployment filings." And there can only be one first time.
So it isn't cumulative (we have a couple of other numbers to capture that, but more on them later). For perspective, initial claims peaked at 665,000 in March 2009. The record was 695,000 in October 1982. This is incredibly significant - stay with me.
First, as my alter ego has been posting on social media pretty consistently lately, when it comes to statistics, the denominator matters. For example, the number of COVID-19 cases in the U.S. is north of 50,000 - third in the world behind China and Italy. However, dividing the number of cases by a country's population, we find that the number of U.S. cases is .0054%, or 54 per 1 million population. China's is 59 per 1 million, South Korea's is 176 per 1 million, and Italy's, tragically, is 1,144 per 1 million, as of this writing (the evening of March 24).
In the case of initial jobless claims, we have to look at some denominator - either the Civilian Labor Force (those unemployed plus those employed), or the All Employees (Nonfarm) published by the Bureau of Labor Statistics. I prefer the former as it is more comprehensive.
So in March 2009 during the Great Recession, when initial claims peaked at 665,000, the number was 0.43% of the Civilian Labor Force (CLF). And at the peak of 695,000 in October 1982, when the U.S. population was much smaller, that number was .63% of the CLF. Get it? Larger numerator, smaller denominator. (And remember those dates - we'll come back to them.)
Now, that data would tell you that the recession of 1981-1982, which lasted 16 months and saw a trough in GDP growth of -6.69% year-over-year, was worse than the Great Recession of 2007-2009, which lasted 18 months and saw a GDP trough of -8.45%. The unemployment rate peaked in 1982 at 10.8%, vs. 10.0% in 2009.
As someone who lived through both recessions, I can tell you that '81-82 was bad. It was nearly a double-dip recession, coming on the heels of a much shorter and milder downturn in 1980. It was driven by extremely high energy prices, high inflation and high interest rates.
But 2007-09 was worse. It was driven by the most massive housing bubble the U.S., and the rest of the world, have ever seen. The fallout was Armageddon-like. And bubble-driven recessions tend to be more severe and far-reaching than those driven by other factors.
Back to jobless claims, and back to the present. For the week ended Feb. 1, 2020, initial claims totaled 201,000 - the lowest level since Nov. 1969 (when the Curmudgeon was but a lad, not the senior citizen he is today). That's a long time. The upshot is that claims of late have been near all-time historic lows.
So last Thursday, when the ETA reported that claims had increased 70,000, from 211,000 the week before to 281,000, the media's collective hair, real or otherwise, caught fire. At this point, they use their own irrelevant comparisons and misleading numerators to try and scare you, the unwitting:
Bonus reading: lest you doubt my assertion that the media pundits don't know jack about economics, consider this:
Okay, wait. For many of us, it feels like that. Let me start over.
In case you haven't watched or read the news or been on social media for the last several weeks, I have some important information for you:
The news media believes that their job is to scare the bejeebers out of you, make you an anxious wreck, increase your blood pressure, and paralyze you with abject fear.
Yes, it's despicable. It is beneath contempt, and if they were capable of empathy, they would be ashamed. But they're not.
So on Thursday morning, March 26, 2020, in the middle of this coronavirus pandemic and the ensuing government-mandated shutdown of our economy, when they've already made you anxious enough -
They are going to try their best to scare the crap out of you. Don't let them. Read on, and you'll be far more informed than they will ever be.
Every Thursday morning, the U.S. Employment and Training Administration (ETA), a division of the Department of Labor, releases Initial Jobless Claims for the week ended the prior Saturday. So on Thursday, March 26, they will release the number for the week ended Saturday, March 21.
And it's going to be a bad number. The media is going to go into a frenetic Chicken Little dance over it.
Because they have no idea what the number means. It doesn't mean what they think it means. It doesn't mean what you think it means. Let me explain.
Initial Jobless Claims is a point-in-time number. It means something that week, then it means nothing the next week. It is a trend number, in that it is only important in context of the trend. And as the Curmudgeon has long been fond of pointing out, one week doth not a trend make.
The claims number represents the number of Americans who filed for unemployment insurance benefits in a given week. If those Americans are still unemployed the following week, they fall out of the number, because at that point, their claims are no longer initial. Another way to label it is "first-time unemployment filings." And there can only be one first time.
So it isn't cumulative (we have a couple of other numbers to capture that, but more on them later). For perspective, initial claims peaked at 665,000 in March 2009. The record was 695,000 in October 1982. This is incredibly significant - stay with me.
First, as my alter ego has been posting on social media pretty consistently lately, when it comes to statistics, the denominator matters. For example, the number of COVID-19 cases in the U.S. is north of 50,000 - third in the world behind China and Italy. However, dividing the number of cases by a country's population, we find that the number of U.S. cases is .0054%, or 54 per 1 million population. China's is 59 per 1 million, South Korea's is 176 per 1 million, and Italy's, tragically, is 1,144 per 1 million, as of this writing (the evening of March 24).
In the case of initial jobless claims, we have to look at some denominator - either the Civilian Labor Force (those unemployed plus those employed), or the All Employees (Nonfarm) published by the Bureau of Labor Statistics. I prefer the former as it is more comprehensive.
So in March 2009 during the Great Recession, when initial claims peaked at 665,000, the number was 0.43% of the Civilian Labor Force (CLF). And at the peak of 695,000 in October 1982, when the U.S. population was much smaller, that number was .63% of the CLF. Get it? Larger numerator, smaller denominator. (And remember those dates - we'll come back to them.)
Now, that data would tell you that the recession of 1981-1982, which lasted 16 months and saw a trough in GDP growth of -6.69% year-over-year, was worse than the Great Recession of 2007-2009, which lasted 18 months and saw a GDP trough of -8.45%. The unemployment rate peaked in 1982 at 10.8%, vs. 10.0% in 2009.
As someone who lived through both recessions, I can tell you that '81-82 was bad. It was nearly a double-dip recession, coming on the heels of a much shorter and milder downturn in 1980. It was driven by extremely high energy prices, high inflation and high interest rates.
But 2007-09 was worse. It was driven by the most massive housing bubble the U.S., and the rest of the world, have ever seen. The fallout was Armageddon-like. And bubble-driven recessions tend to be more severe and far-reaching than those driven by other factors.
Back to jobless claims, and back to the present. For the week ended Feb. 1, 2020, initial claims totaled 201,000 - the lowest level since Nov. 1969 (when the Curmudgeon was but a lad, not the senior citizen he is today). That's a long time. The upshot is that claims of late have been near all-time historic lows.
So last Thursday, when the ETA reported that claims had increased 70,000, from 211,000 the week before to 281,000, the media's collective hair, real or otherwise, caught fire. At this point, they use their own irrelevant comparisons and misleading numerators to try and scare you, the unwitting:
- "The highest total since September 2017!"
- "The largest one-week jump in claims since 2012!"
- "The largest one-week percentage jump in claims since 1992!"
And the number on Thursday, March 26, 2020 will be worse. Much worse. But again, it doesn't mean what you or they think it means. Here's why.
Back to initial claims being a point-in-time number. For it to have meaning, we'd have to sum weekly initial claims over the entire duration of a downturn. Let's look at the Great Recession.
Because claims fluctuate weekly due to seasonal and other factors, it's hard to pinpoint when they started to rise due to the Great Recession, so we'll just pick the week of Jan. 26, 2008, when claims rose from 321,000 to 366,000 (the Jan. 19 number was the low for 2008, and claims averaged about 321,000 for all of 2007). If we start there, and total claims through the week before they fell back below 366,000 - Feb. 4, 2012 - we get a cumulative total of about 98 million initial claims.
Which is a meaningless number. Why? Well over the four-plus years from Jan. 2008 to Feb. 2012, some people who filed initial claims went back to work. Some may have given up and stopped looking for work, waiting for the job market to improve further, so they didn't qualify to receive benefits anymore.
Fortunately, we have a couple of more meaningful numbers related to unemployment insurance claims: Continued Claims and Weeks Unemployed. Let's look at those.
Continued Claims - also released each week by the ETA - counts those who are past the first week of filing claims for unemployment insurance. It works like this: let's say that on Monday, having lost my job, I file for unemployment insurance for the first time. That gets counted in Initial Jobless Claims for the week ended next Saturday, which will be reported the following Thursday morning. With me so far?
Okay. After that, I will no longer be counted among the initial claims, because, as noted previously, there is only one first time. However, if I remain unemployed and am continuing to file for unemployment insurance benefits, I will now be counted - for the first time - in the Continued Claims data series. And I will continue to be counted in that series until I no longer meet the requirements to file for benefits, or until I stop filing for benefits - in other words, until I stop looking for work, my benefits expire (currently 26 weeks in most states, but that's sure to get extended in the current situation, as it does in every other downturn), I get a job, or I just stop filing for some reason.
Continued Claims peaked at 6.6 million in May 2009, which was a record. But it's far less than the cumulative initial claims during that downturn of 98 million. Far, far, far, FAR less. Because the vast majority of those people had gone back to work. So don't let the Initial Claims number scare you.
Now, as for Weeks Unemployed, or the duration of unemployment. This measures the average number of weeks that those drawing unemployment insurance benefits remain on the rolls. So it is influenced by the extension of benefits by federal or state legislators during a downturn, as well as individual states' baseline benefits expirations. Generally, in a downturn, benefits will be extended, as noted above. This series peaked in July 2011 - more than two years after the end of the Great Recession - at nearly 41 weeks. It was the slowest of the employment data metrics to normalize. It fell to just under 20 weeks in July 2019, which was about where it was in 2005. This speaks to just how ugly the bubble-driven Great Recession was, and how long it took to recover from it. (The baseline trend on this metric has risen over successive downturns, as government increasingly extends benefits, providing a disincentive for people to go back to work for some jobs.)
One other point about Initial Claims - remember when I said we'd get back to the peak dates from 2009 and 1982? The Great Recession ended less than two months after the peak in Initial Claims. And the '82 recession ended within four weeks of that year's peak. In fact, every recession since 1975 has ended within weeks of that downturn's peak in Initial Claims. So if we see a big spike in the next release, and maybe another two or three after that before we hit a peak, you can rest assured the worst is pretty much over, at least for the economy.
So what does all of this mean? First, it means that whatever number we see on Thursday, March 26, 2020, it doesn't tell us much. We need to know how long this thing lasts, and how we recover from it, and that will only be evident in the Continued Claims and Weeks Unemployed data.
Second, we have every reason to expect - not hope - that this time will be different. In most recessions, and especially in bubble-driven recessions, initial claims follow a pattern that, if you graph it, looks like the Matterhorn: they begin rising, they rise sharply, they peak, they begin to decline, and then at some point they level off.
There's a reason for that, and let's use the Great Recession as our lesson. First, the builders who overbuilt, the developers who over-developed, the subprime mortgage lenders who made bad loans, and the title companies who closed those loans, shed jobs. Then, the restaurants those people used to eat at, the bars where they hung out, the hotels where they had their conferences, all saw sharp declines in business, and they laid people off.
Then, those people - who had taken out subprime mortgages - stopped paying on them, and more lenders failed. The Wall Street firms that securitized and bought those mortgages in mass quantities suffered losses. The insurance companies that issued credit guarantees on the bonds that securitized the subprime mortgages ran out of money to insure the losses, and they failed. And all the restaurants, bars, movie theatres, car dealerships, jewelry stores, clothing stores, and every other piece of the economy that depended on those jobs - they all suffered. The second- and third- and fourth-order effects were catastrophic.
This time, there was no asset bubble, so this thing won't develop over time. There will likely be no second- or beyond-order effects. The first few weeks of initial claims numbers should be the worst of it, at least if we get the economy opened back up again in relatively short order (and by that I mean by summer, when seasonality suggests that this thing should have run its course for this year, and we'll just have to see what happens next year - I'm sorry, I'm not one of those doomsday theorists who says this coronavirus will be unlike ANY OTHER the world has ever seen in that it won't be seasonal). In other words, instead of looking like the Matterhorn, with a steep upward slope toward a peak, followed by a steep downward slope back to the norm, it's more likely to look like a cliff - like El Capitan in Yosemite, but with a sharper downward slope like the Matterhorn after the peak is reached.
And why do I believe that the recovery from that peak will be sharp - even sharper, in fact, than the recoveries from the last two recessions?
Again, those recessions were driven by asset bubbles, both of which were fueled by excessive accommodation (read: too-low interest rates) by the Fed - first, the dot-com bubble in 2000-01, driven by the Fed cutting rates in 1998; and the housing bubble of 2007-09, fueled by low rates in response to the dot-com bubble.
Those recessions led to the kind of business failures I noted above. In the dot-com bubble's aftermath, there were far fewer tech firms, because they had overbuilt the tech sector on cheap money, and many of those companies never came back to pick up the slack. In the case of the housing bubble, a lot of those builders and developers never came back, because we had already overbuilt housing capacity relative to demand. The subprime lenders never came back, because we've learned our lesson about subprime lending and we're not going back there - nobody wants to take out the loans, no lender wants to hold the loans, and nobody wants to securitize the loans, or buy the securitized product. Everything else - restaurants, hotels, cruise lines, casinos, retailers - all recovered.
This time, the government, in the interest of saving lives, which is the right thing to focus on, forced those businesses to shut down. This time, there is no moral hazard. This time, the business impact is through no fault of the business sector. The government mandated this, and that's why the government is bailing companies and workers out. True, we must be careful to avoid perverse incentives. But the moral hazard of 2009 isn't there today.
So where does that leave us in terms of the rebound from the effects of this pandemic on our economy?
From my own little corner of the world, my clients have postponed some of my on-site visits, but not much else has changed. We may do things via webex in the interim. But when the dust settles, they will want me on-site again. So I will need plane tickets and hotel rooms and rental cars. I will have to eat out while I'm on the road. Multiply that from my tiny sector of the economy to far larger sectors, and you will see a significant rebound in travel.
Plus, I'm going to want to go on vacation again - aren't you? We already have our next cruise booked. I'm going to want to eat in restaurants, and go to movies, and go to bars (okay, so I don't go to bars much). People are going to want to take their kids to zoos and museums that are closed. Schools are going to re-open, even if not until next year. The Olympics will take place. The NFL, MLB, NBA, NHL and college sports seasons will resume.
The demand is there. Things just have to open back up, which they will in due course. There will be spring after this winter, as there always is.
So the bottom line is, when the media tries to freak you out over Thursday's initial jobless claims number, don't let them. Now, you know better. You know that number better than they do if you've read this far. Godspeed, be safe, be sane, and be calm.
Bonus reading: lest you doubt my assertion that the media pundits don't know jack about economics, consider this:
- Ali Velshi, NBC/MSNBC's senior economic and business correspondent, and formerly CNN's Chief Business Correspondent. Has a degree in ... religious studies.
- Richard Quest, CNN's Business at Large Editor. Has a degree in ... law.
- Kelly Evans, CNBC co-anchor of Power Lunch. Has a degree in ... journalism.
'Nuff said, folks. You stand as good a chance of understanding this stuff as the media does.
Subscribe to:
Posts (Atom)