From Mark Mandula: Thoughts on the COVID-19 Pandemic Part IV


Has COVID-19 (coronavirus disease 2019) impacted your business? Invoice financing may help ease the cash flow strain on your business. Contact us or apply now (no cost to apply) to discuss your options. 

Update to April 2, 2020 Weekly Unemployment Claims Data

My note discussed the stunning data that was released April 2, 2020 on the weekly jobless claims for unemployment in the United States. As mentioned the data for the week was 6.6 million, nearly double the anticipated estimates.

I also looked at some interesting data published before the claims data was announced by Pantheon Macroeconomics. The research looked at Google Trends data on the average daily number of searches for “file for employment”, looking at the number and trend line for the keywords. It found that actual searches rose by 60% last week, compared to the week ended March 21, 2020. The key takeaway from their research was this “If the relationship between the Google data and the claims is linear – we have zero experience of claims at their current level, so this is a guess – it implies that claims jumped last week to more than 5 million.” And the research concludes that because the Google searches are national, nor regional or local, that this made sense to them.

And it turns out, even a prediction of 5 million was way too low.

So their research accurately predicted that the consensus estimates were too low as determined by looking at the Google Search results. And even though the search results have trailed off marginally, they remain at levels never seen before in recent history. So if the relationship between search results and claims is indeed linear, a lot more jobless claims are right around the corner for the foreseeable future.

This is yesterday’s news. So today I decided to update this information by looking at real time data on the same Google search term, “file for unemployment” over a much longer time horizon. As you can see above, the data we looked at yesterday covered only the month of March, 2020. By doing this, this would allow us to compare current search results on Google in prior recessions and economic downturns that have occurred in the United States and see if what is happening now is similar or different to the past results.

I selected off Google Search the longest time duration possible for the search results; stretching back more than 15 years. In that time frame only one recession occurred in the United States for December 2007 to June 2009. Peak unemployment occurred in October 2009 at 10.0%. The decline from peak to trough in GDP in the United States was -5.1%.

As an aside I also learned that since records were kept with some validity in the United States (1790) there have been as many as 47 recessions that have occurred. A recession is commonly defined as two quarters of negative growth in GDP. The average duration of the 11 recessions between 1945 and 2001 is 11 months, compared an 18 month average from ones occurring between 1919 and 1945.

So what did the Google search data reveal over the 15+ years of analysis? The result is below, and I believe that a picture in this case does say a thousand words.

One other startling development that happened this morning also merits a quick mention and something to think about/ be concerned about. The United States Labor Department this morning reported that for the very first time in almost a decade that the United States economy recorded a net job loss for the week of 701,000 jobs. This ends an unprecedented streak of 113 months in a row of positive weekly job creation in the United States; a streak neatly twice the previous record.

But what is not in this negative number is the caveat that these job losses occurred in the first half of March, before the current swath of stay at home orders, layoffs and closing even began to occur. As a result the current unemployment rate in the United States of 4.4 is likely to double or even triple in the very near future. The CBO [Congressional Budget Office] announced yesterday that the official forecast for unemployment would exceed 10% by June 2020 and remain elevated at 9% until 2021.

Only time will tell and the ability to slow down the COVID-19 pandemic quickly and completely will determine how bad the job losses will be in the United States for the near future.


As the COVID-19 pandemic continues at full strength in much of the United States, it is becoming clearer that the economic damage done by the virus will be widespread and significant. The first inkling of this unsettling fact appeared last week when an unprecedented surge in jobless claims was announced. From the prior weeks, data of 282,000 claims came the stunning news that the current weeks total was about 3.3 million. At that point, equity and other markets in the United States nearly came unglued and the optmisitic chatter was that this would be the highest claim that would occur in the COVID-19 period. Most economists and experts had predicted that the number of claims would be nowhere as high as the 3.2+ million that wa actually reported.

To make a long story short, the optimists were wrong that the worst had come and gone. But perhaps what the optimists [and just about everyone else] forgot to take into consideration is that after the tsumani of closings, lock downs and other restictions became effective, the inftrastruture [websites, 800 numbers, etc.] for people to apply for unemployment for the first time were overwhelmed and in some cases, crashed. So everyone who was unable to get aid from last week’s stunning total were rolled forward to this week and was added to the new pile of unemployed workers impacted by additional stay at home rules, layoffs, firing and lockdowns that occurred.

So that brings us to the April 2 and the highly anticipated announcement of new unemployment claims for the week just ended. There was a wide range of estimates being bantered about what the new jobless claims numbers would be; but most seemed to agree that the most likely range was in the 2 to 4 million new jobless claims total. A median number seemed to be about 3.1 million +/-.

Prior to revealing the actual April 2, 2020 results it might a good idea to quickly review and discuss why weekly jobless numbers are carefully watched and why they are very important to provide us some insight into the current and near term state of the economy in the United States. Almost all economists, whether in the private sector or in government agree that weekly jobless claims are one of, if not the most important data series to determine how severe a shock is when a pandemic or other event [hurricance, etc.] occurs. But there hasn’t been an event of the magnitude, scope or size of COVID-19 in recent United States history and this makes relying on past behavior to predict future results iffy at best.

When the United States Department of Labor announced that the new number of U.S. workers filing claims for jobless benefits rose to 6.65 million there was shock and disbelief. This meant that not only had the number increased from the stunning total of 3.3 million from the prior week, it has more than doubled. Doubled in a single week.

This also needs to be looked at in a historical context to grasp the gravity of the number. In the worst week of the prior United States recession (2009) the single largest weekly increase was only 665,000 or 10% of the most recent claims number. Again why these numbers are very important to assess the overall health [or sickness] of the United States economy is because weekly jobless claims are the most comprehenisve and accurate data about what is happening on the ground all across the United States. Estimates by ivory tower academics and policy newsletters are just that; guesses not grounded in hard facts.

Weekly jobless claims data provide insight into key metrics like:

  • How many businesses are left open and how many have simply closed up shop due the spread of the virus
  • How many businesses have laid off employees due to a crash in the demand for the services they used to provide or the products they manufactured or sold
  • How many businesses are laying off workers to hoard cash and pay what bills they are able to pay
  • How all of these daily events will then impact the supply chain networks for months or longer to come

So as important as it is to understand today’s very unsettling information is to try to estimate/guess what will happen over the next couple weeks or months to the jobless claims data and overall unemployment in the United States. Anyone who categorically announces that they have 100% clarity and accuracy in predicitng future data is either clarvoiant or delusional. There simply has never been an event of this scope to help us predict future results with any real accuracy.

However, one very interesting tool that we do have to help us simply didn’t exist 20 or 40 or 100 years ago; namely the internet and more spcifically Google keyword search results. A fascinating research article I read [among many others] authored by Pantheon Macroeconomics dated April 2, 2020 had some very compelling analysis and data if a correlation between key words searched on Google and actual job claims does exist.

The research looked at Google Trends data on the average daily number of searches for “file for employment”, looking at the number and trend line for the key words. It found [see graph below from their excellent work] that actual searches rose by 60% last week, compared to the week ended March 21, 2020. The key takeaway from their research was this “If the relationship between the Google data and the claims is linear – we have zero experience of claims at their current level, so this is a guess – it implies that claims jumped last week to more than 5 million.” And the research concludes that because the Google searches are national, nor regional or local, that this made sense to them.

And it turns out, even a prediction of 5 million was way too low.

So their research accurately predicted that the consensus estimates were too low as determined by looking at the Google Search results. And even though the search results have trailed off marginally, they remain at levels never seen before in recent history. So if the relationship between search results and claims is indeed linear, a lot more jobless claims are right around the corner for the foreseeable future.

So where might this end? What could the total unemployment numbers look like later this spring and summer? Recall that the total is already a scary 10+ million. If what we read online and hear via the media, many states remain ill prepared and are overwhelmed to handle the current let alone future people seeking unemployment benefits. This leads to additional backlogs and eventually higher claims results on a weekly basis. While this could change, it does not bode well for the total unemployment rate in the United States to be as high as 12% to 20% in less than a month.

Only time will tell and the ability to slow down the COVID-19 pandemic quickly and completely will determine how bad the job losses will be in the United States for the near future.