Jobless claims, seasonal adjustments and Tim Harford

Thursday is the day of when the initial jobless claims data is released in US. As the data pertains to the previous week, it is one of the most-timely indicators of the labour market condition in the US. The numbers essentially represent how many people lost their jobs during the week. One needs to understand that job losses are a natural part of a capitalistic economy, businesses close because of purely business reasons too. Only when the numbers show a trend then the policy makers need to get involved. Now apart from initial jobless claims, every Thursday also carries the release of continuous claims number. This is a number which the jobless person files after one week of unemployment. The data is a good proxy for the number of unemployed people in the economy.

These are weekly data points, their more popular cousin NFP comes at a monthly frequency hence becomes a bit stale by the time it comes. But readers should note that the high frequency data can also be a lot noisier. It is for the markets to sieve that noise out. Data providers themselves try to assist in this by providing seasonally adjusted data by correcting data for regularly occurring fluctuations. These changes might occur because of change in weather, major holidays, the opening and closing of events etc. As these seasonal events follow a regular pattern every year their impact on the published data needs to be tempered to give a clear picture. There is no point in screaming that joblessness has increased if we know that the schools closed out last week and all the allied activities came to a stand-still.

Now coming to the data point itself, the initial jobless claims for the week ending May 15^th declined to 444k (seasonally adjusted) and 456k (non-seasonally adjusted). The SA number was lesser by 34k from the previous week. For the context the seasonally adjusted initial jobless claims number was close to 2.1 million in the week ending May 16^th, 2020. That was obviously the time when the pandemic was raging hot and the lockdowns were getting implemented. President Trump was in charge in the US and medical fraternity was still finding out the cause and cure of the virus without any broad agreement. The vaccines were not there on the horizon. The world has changed a lot since then.

On the continuous claims number, the number stood at 15.98 million for the week ending May 1^st against 16.86 million for April 24^th. These are the people who are claiming some form of federal or state unemployment benefit. Simply put 16 million new jobs need to be created to put all these claimants back to work. Readers know that there is a separate debate going on whether the benefits are the cause which are dissuading people from joining the jobs back. Many are arguing that these largesse schemes should be wound down quickly for the natural equilibrium of jobs and wage market to emerge.

British economist Tim Harford in his recent book How to Make the World Add Up discusses in detail about the data points which are either collected or churned out by the modern economies. The unemployment data, wage data, consumer confidence surveys, inflation expectation surveys and the data collection organisations like DLO, BLS, CBO etc are covered extensively in the book. Harford writes that ample thoughts are not given even in the policy circles about how these data pools are collected and disseminated. Neither their market impact is assessed, and little thought is given on maintaining the sanctity of the data. One would do well to read the book if they want to grasp the intricacies of the context behind the data. Harford recounts an instance where the last US President tweeted about the incoming Job report one hour prior to its official release on his twitter handle and markets as usual panicked. We will continue this discussion in a future note as there are multiple angles to cover. Pausing now for the paucity of space and time.