Regressing To The Mean

Statistics don’t lie, do they? If you believe in data, then you may believe that our national economy must be in serious trouble. After all, according to a Stanford University study that was released last week, millennials who were born in 1980 only have a 50% chance of earning more money than their parents did at the same age.

The researchers compared that metric to members of the Silent Generation who were born in 1940; those individuals had a 92% chance of earning more money than their parents did. And the reaction of the national press to the study was dire. The Los Angeles Times, for instance, cried that the “American dream slips out of reach for millennials.”

The New York Times was a bit more optimistic, concluding that “if the American dream could survive the Depression, and then thrive in a way few people imagined, it can survive our current troubles.” But it similarly bemoaned the reality of the “fading American dream” for most American citizens.

At first glance, the 50% metric does look pretty dire, doesn’t it? Especially when we compare it to the 92% metric for individuals born in 1940! But it may be helpful to keep in mind an important fact regarding such data.

Here’s the fact. The 50% metric doesn’t merely seem relatively low in comparison to the earlier generation’s 92% metric. It is actually relatively low because of the earlier group’s 92% metric. Or, more precisely, it is relatively low because of the conditions that were temporarily extant at the earlier time.

Here’s an example to help illustrate the point. Let’s assume that a good football team plays three games in a row against other equally good teams. The first and third games are played outdoors in December, in relatively normal (i.e. uncomfortably chilly, but certainly not frigid) late fall or early winter conditions. The second game, however, is played in an indoor domed stadium, in perfect “68 degrees and dry” conditions.

The quarterback plays the first game and does relatively well. Then he plays the second game. What are the odds that he’ll pass for more yardage than in the first game?

It’ll be very high, perhaps as high as 92%. Why? It’s not because he is an inherently better player than he was during the first game. It’s just that playing conditions have turned temporarily better.

Now he plays the third game. What are the odds that he’ll pass for more yardage than in the second game?

It’ll be very low, perhaps as low as 50%. Why? It’s not because he is an inherently worse player than he was during the second game. It’s just that playing conditions have returned to a normal level.

In fact, in any strong, stable, and non-volatile system, you might expect any one’s odds of performing better than any one else to be an average 50%. That’s just as true of football teams as it is of American generations.

And in our football example, if the second game had been played in normal outdoor conditions, the percentages may have remained constant at an average 50% across all three games. We may not have witnessed any volatile variations at all.

Indeed, the reason why the third game’s 50% metric is so low is because the second game’s 92% metric is so high. This “high followed by low” effect is called “regression to the mean.” Unusually high performance metrics, when caused by temporary fluctuations, are usually followed by low performance metrics. And the reverse, i.e. a “low followed by high” effect, often occurs as well.

So how should we interpret the Stanford study data? Well, those American babies who were born in 1940 were about to live their lives under temporarily ideal economic conditions. The Second World War had begun, and by the time it ended just a few years later, each major developed nation in the world but one — the United States — saw its industrial economy obliterated and its society decimated. China, France, Germany, Great Britain, Japan, Russia: all were laid low by the conflict.

The great rebuilding of these societies required massive amounts of production supply and financial capital. And the American economy was the only nation on earth that could provide such support.

Those pro-American conditions couldn’t possibly last forever, given that the major economies of the world all eventually recovered. Subsequent American generations thus saw their probabilities of economic success, relative to their parents, regress towards the natural mean of 50%.

As long as we avoid any more cataclysmic global wars, future generations of Americans may find themselves bouncing along at that 50% metric as well. And they shouldn’t necessarily consider their nation in decline in comparison to the 1940 generation’s 92% metric. Instead, they should consider themselves fortunate to be members of a relatively strong, stable, and non-volatile society.