When talking about Lean and Six Sigma, people often associate them with improved productivity. Other times they associate them with their possible consequences: high profitability or loss of jobs. In today’s economic environment, many people are concerned with the potential trade-off between productivity improvement and employment. But is there such a trade-off relationship?
A recent report by the McKinsey Global Institute (Growth and renewal in the United States: Retooling America’s economic engine) explores how higher productivity can help renew the economy, including long-term job growth.
The report has a lot of useful data and analyses, most of which I won’t discuss here. What interested me was how they analyzed the relationship between productivity and employment growth.
In their Executive Summary, the authors state
“Since 1929, every ten-year rolling period except one has recorded increases in both US productivity and employment. And even on a rolling annual basis, 69 percent of periods have delivered both productivity and jobs growth (Exhibit E6).” [Page 5]
On page 27, “Box 2. Long-term economic growth through higher productivity and more jobs” the authors provide 3 reasons why higher productivity can help create jobs.
1. Lower prices and more consumer savings boost demand for goods and services
2. Higher quality and customer value boost demand
3. Global competitiveness necessary to attract and maintain jobs
They also argue that “the perceived trade-off between productivity growth and employment growth is a temporary phenomenon. More than two-thirds of the years since 1929 have seen positive gains in both productivity and employment (Exhibit 15). If we look at quarters, employment growth followed gains in productivity 71 percent of quarters since 1947 (Exhibit 16).”
What does their data analysis tell you?
Not much, I have to say. Why? See the figure below.
I could draw the same conclusion with this figure as they did with Exhibit 6 or 15. But I won’t. The reason is that the 2 series of data (Pro and Emp) used to generate this figure were completely independent of each other — there is no relationship at all. In other words, the figure shows what we would expect from random data of 2 independent variables.
To create 80 random values of Pro, I simply set a 15% probability of being negative (down) and 85% positive (up) and let the Excel function [=Rand()-0.15] do the rest. Similarly, 80 values of Emp were randomly generated with a 25% probability of negative and 75% of positive. The two lists are paired as if they came from the same period. The figure shows how often they are both up or down or different in each period. Although the figure shows one sample of randomly generated numbers, other simulated results more or less show the same pattern.
Now I hope you see the analysis presented in the report is as valuable as a relationship I could draw from a random sample of data of two independent variables — not much.
The same lesson can be learned from their Exhibit 16. What ADDITIONAL knowledge do we gain by knowing that 71% of the time both productivity and employment are gains, without knowing what is expected by chance alone based on individual probability? Anytime I have two independent variables with a probability of 0.8-0.9 being positive, I would expect them both being positive about 70% of the time.
Again, we learn by asking the right questions. Before we draw any conclusion from data, ask “how likely is the result due to random chance?”
