Yesterday, the Department of Labor released their weekly unemployment claims figures. This is the number of people filing for unemployment insurance that hadn’t filed in the previous weeks. The New York Times, the LA times, CNN, BBC, the Guardian, and the Economist all report that 3.3 million new people filed for unemployment insurance. This reporting is WRONG!!! The real number is 2.9 million, which is a staggering number in itself. However, the difference between the two is 400,000, roughly twice the number of a normal week.
The 3.3 million figure is the seasonally adjusted number, which the department of labor releases along with the raw figure. Normally, the seasonally adjusted figure is what should be reported, as it allows for better week-to-week comparisons. However, this is not a normal week. Seasonal adjustments work by calculating how a dataset correlates with a seasonal variable, such as the week of the year, and then it adjusts by that correlation to eliminate the seasonal component. Thus, this week is normally 14% lower than the average week, so to adjust for seasonality, the Department of Labor multiplied the raw figure by 114%, and it will compensate by multiplying a high week by a number that decreases the total.
The week being reported, however, is not a normal week. For one, the figure itself is an outlier, roughly 12 times the normal week or 4.5 times the previous record high. In addition, it is a week where the underlying data generating process, people becoming unemployed, was different from any point since the Department of Labor started recording this data. As a result, the seasonal correlations probably aren’t the same as in a normal week.
To reiterate, this isn’t a normal week. The week being reported is a week in which a large portion of the American population was told to stay at home for public health reasons. It is the week in which 4 of the largest metropolitan areas in the country were placed in lockdown. California, New York, and Illinois alone comprise over 70 million people, or 20% of the US population. In this context, 2.9 million people, less than 1% of the American population, filing for unemployment insurance is not an extreme number. Many of these people might potentially return to work once the pandemic has subsided, something that is not the case during a normal week of unemployment insurance claims.
So why have all the news outlets reported the higher number? I see two options: incompetence or deliberate misrepresentation. One could potentially claim that these same news outlets normally report the seasonally adjusted numbers and wanted to keep with that figure for consistency, but that would either imply that they didn’t understand that this week was different or didn’t think that they should report that fact. At a time in which the news itself is alarming, we do not need news outlets altering the facts to make them seem more alarming.