Numbersense and government accountability in the new political reality

You've heard me say often, numbersense is the most important quality for good data analysts; little did I know that numbersense would become the new requirement for healthy American democracy.

From the first day in office, the new President is at war with numbers (over attendance figures at his inauguration). But I believe that getting to the bottom of data-driven claims is a bi-partisan issue: while it is obvious that Democrats want to hold President Trump responsible for delivering what he promised, such as new jobs, Republicans also want to audit the President's data to verify that he is delivering real change, not empty rhetoric.

Since his election, Trump's publicity operation has claimed responsibility for various job announcements by a host of companies, including Carrier, GM, Softbank, Alibaba, Sprint, etc. Americans would agree that it is important to create new jobs in the United States.

Having spent a large part of my career in the business world, I am familiar with the many (often legal but unethical) tricks used to dress up numbers. So let me use job numbers as an example to illustrate some of these unsavory tactics.

  1. Counting gross jobs instead of net jobs. The first job announcement came out of Carrier, a unit of the behemoth, United Techologies, which is said to invest $ 16 million in its factory in Indiana, instead of moving it to Mexico. Then, its CEO explained that the $ 16 million will purchase automation which eliminates jobs in the factory. In this case, the net jobs created are the difference between the reported number of jobs saved and the unreported number of jobs replaced by machines. The reduction of jobs from a separate action does not have to come from the same business unit. If it comes from a different division, it is much harder to track. But the point remains, the gross number doesn't matter; only the net number counts.
  2. Counting planned jobs instead of filled jobs. All of the announcements so far involve projections. Needless to say, projections are meaningless unless we are able to track their accuracy. The news media would have to turn back their short attention span in order to check whether such promises are fulfilled. In particular, should the economy experience a downturn in the next few years, would these companies go through with their hiring plans? I highly doubt it.
  3. Touting good news while burying bad news. Let's say Company XYZ has decided to create 100 new positions in a new factory in addition to laying off 100 positions at an old factory. The net jobs created is zero. But the management can play with the timing of these events. The new positions are added first, and six months later, the old positions are terminated. (In reality, this process would have started earlier, giving employees severance in exchange for keeping their mouths shut.) For the first year, the company claims it has created 100 new jobs, which is true but not quite true.
  4. Double-counting jobs. Softbank, a Japanese company, announced that it would create 50,000 jobs in the U.S. while Alibaba's CEO, Jack Ma, boasted of creating 1 million jobs in the U.S. Mr. Ma's pronouncement is curious, given that Alibaba is a Chinese company. He claimed that his trading platform would enable 1 million incremental businesses to sell wares to the Chinese consumer, and each such business would on average hire one additional staff to deal with the Chinese business. This math is seriously fishy – but let me allow it to pass for the sake of argument. So Mr. Ma is not pledging to directly hire 1 million people to work for Alibaba – he is projecting that small U.S. businesses would do the hiring. So we have to be on the lookout for disingenuous double counting. Each one of these new jobs can be claimed to be created by Alibaba (as they have done here) or by the small business itself. The point is that the same new job should not be counted twice. This double-counting is already occuring with Sprint's announcement of 5,000 jobs. Sprint is 80% owned by Softbank. Softbank is claiming the 5,000 jobs to be part of the 50,000 jobs it plans to add to the U.S. labor market. If that is the case, then one cannot simultaneously say that Sprint created 5,000 jobs.
  5. Counting new jobs that would have been created regardless. Are the new positions incremental? For most businesses, they scale their employee base with economic growth. So some, perhaps most, of the new jobs would have been created regardless of the political transition.
  6. Counting temporary, part-time positions with no benefits. As I explained in Chapter 6 of Numbersense (link), the official employment statistics do not distinguish between good and bad jobs. The bar for being counted as employed is extremely low. What Americans want are permanent, full-time positions with benefits.
  7. Counting publicly-funded jobs as private-sector jobs. In the still-developing case of Carrier, it may be the case that the company receives $ 7 million of tax credits (i.e. government subsidies) to save 500 jobs. That would be $ 14,000 per job. At a minimum hourly wage of $ 15, that would pay for 933 labor hours, or 23 work-weeks per job, assuming the new jobs are full-time positions. Readers of Numbersense (link) will recall that law schools are padding their post-graduation employment statistics by offering part-time administrative positions to their own graduates until after the employment surveys are completed.
  8. Last but not least, can businesses be held to these job announcements? That would be a question for any lawyer or accountant reading this blog.


Big Data, Plainly Spoken (aka Numbers Rule Your World)