In this edition of Fun With Statistics we take a look at the effect of a point of reference on data. How old are you? Do you feel old? How is that related to that all-important point of data: your age. From Tyler Cowen at Marginal Revolution comes a study by John Shoven:
The current practice of measuring age as years-since-birth, both in common practice and in the law, rather than alternative measures reflecting a person’s stage in the lifecycle distorts important behavior such as retirement, saving, and the discussion of dependency ratios. Two alternative measures of age are explored: mortality risk and remaining life expectancy. With these alternative measures, the huge wave of elderly forecast for the first half of this century doesn’t look like a huge wave at all. By conventional 65+ standards, the fraction of the population that is elderly will grow by about 66 percent. However, the fraction of the population that is above a mortality rate that corresponds to 65+ today will grow by only 20 percent. Needless to say, the aging of the society is a lot less dramatic with the alternative mortality-based age measures. In a separate application of age measurement…GDP would be between seven and ten percent higher by 2050 if retirement lengths stabilize.
Perhaps this is analogous to push vs pull statistics? Tyler comments,
Note that the entire increase of life expectancy of the twentieth century has been taken in the form of retirement, rather than extra work. Of course our social security and Medicare policies have encouraged early retirement, and we have not adjusted age eligibilities for longer life spans and better health. For fiscal reasons, we will likely have to increase eligibility ages; not only will we spend less money but it will encourage more work. If you have been thinking that a demographically-based American economic collapse is virtually inevitable, this paper gives some grounds for hope.
Let’s forget the sociopolitical implications and move on, briefly as I presume we’re all fairly brain dead by this point, to an application to business data.
Should sales, margin, return-on-whatever, first pass yield, on-time delivery, and inventory be measured from zero? Or relative to the starting point of certain levels. Achieving $100,000 in sales is a completely different effort and activity than achieving $1,000,000 or $1,000,000,000. Moving from $1,000,000,000 to $1,000,100,000 is a considerably different than moving from $0 to $100,000 (I know that, painfully, from experience creating my own startup after having worked in a Fortune-25). The incremental, or even logarithmic incremental, has a lot of meaning. The delta from zero not so much. Similarly the effort required to get down to 5 inventory turns, then 10, then 20, then 50. It’s a completely different organizational, leadership, planning, and execution dynamic to achieve each of those intervals.
Does your data really reflect the underlying activity?