HUNDREDTHS OF A GOAL AND FAR-OUT FORECASTING

Fast is fine, but accuracy is everything. ~ Wyatt Earp 

Here are 2 examples of published numbers that for me cross an important line of credulity (in the mental sand of my mind). These examples illustrate a bothersome and unfounded inflation of perceived accuracy in printed statistics.

Hundredths of a Goal.  First, I’ve enjoyed watching the Women’s World Cup soccer matches being played in Canada, and remain hopeful that the US team will again capture the Cup. There’s been much to enjoy as favorites like the US and Germany, as well as “Cinderella” teams like Canada, have navigated FIFA’s very strangely-composed brackets towards the July 5 championship game. One thing that’s distressed me, however, is related to one of my eccentric pet peeves, unwarranted and misplaced accuracy.

Throughout this tournament, the media has highlighted the lack of goals scored in the games. This isn’t surprising because the US has scored a total of just 7 goals through its first 5 winning games. So much for being an offensive powerhouse (so far). The latest example is the US 1-0 victory over China on June 26.

However, media stories about this lack of scoring have routinely transgressed my trust. A piece in The Guardian said, “The average goals per game at the end of the group stage was 2.97…” A New York Times article stated, “The tournament was averaging 3.08 goals a game entering Monday (June 15)… The first-round scoring average was only 2.24 goals a game entering Monday. The 2011 event finished with an average of 2.69 goals a game.”

Let’s take a minute of stoppage time to think about these stated averages. Soccer goals only come in whole numbers e.g., 1, 2, (or 10 if you’re Germany). Thus there’s only 1 or at most 2 significant digits in a soccer team’s goal total for a game. To display the average number of goals in hundredths of a goal with 3 digits (e.g., 3.08) is absurd and actually no more accurate than stating the average in tenths of a goal (e.g., 3.1). Come on; can Abby Wambauch score a hundredth of a goal? Never; so why add 2 decimal places to this average? It’s done so these averages appear to be more precise and accurate than simply saying “about 3” or “3.1.” But they’re not more accurate. With only 1 or at most 2 significant digits as inputs, the quotient’s average cannot meaningfully include more than 1 or 2 numbers. This escalation of false “precision” is readily-obtainable to anyone – including a reporter – who has a calculator in their smartphone; where you are shown 1.66666667 as the result of dividing 5 by 3.

Will this trend continue; so the average number of goals scored in the 2016 Olympic soccer tournament be presented with 3 decimal places – 3.141? I hope not.  False precision is more widespread than just reporting soccer scores.But the next time you read any news article that offers numerical results, see if the number of decimals printed exceed the number of significant digits being referred to. My bet is the resultant’s decimals will exceed the relevant input digits, illustrating unfounded accuracy. That’s bad numerical form.

Bridges to the Future.  Second, precision is an issue that forecasters constantly face. It’s the unstated underpinning for the apocryphal quote, “if you have to forecast, forecast often.” The longer the forecast period, the more likely the prediction is going to be erroneous. We economists have difficulty accurately forecasting economic changes over the next 12 months, let alone several years. Weather forecasters feel lucky when they get their 5-day forecasts right.

In The Signal and the Noise, Nate Silver states that despite the challenges, short-term weather forecasts are overall more reliable than other types of predictions, including economic and financial forecasts. Yet complex environmental models are being used to somehow predict impacts during the next 85 years. That’s right, forecasts over 85-years.

A recent New York Times story cited an EPA report that provides long-term quantitative measures of the consequences of not abating fossil-fuel consumption. The report states that by 2100 (85 years in the future), there will be 12,000 deaths from extreme heat and cold, and 720 to 2,200 bridges would become structural vulnerable and cost $1.1 billion (B) to $1.6B to fix. Forecasting deaths, bridge collapses and costs 85 years in the future? Really?

As many people do, I believe global climate change (né global warming) has been underway for a considerable time increasingly due to humanity’s actions. These induced and detrimental effects need to be mitigated now by a series of specific public policies, such as imposing a meaningful tax on all carbon use.

Ironically, justification for implementing such policies does not require forecasting at all. Environmental policies can be rationalized based on measured changes in what’s already happened to our environment, and continues to happen. Nevertheless, I find it highly doubtful that predicted really long-term effects of climate change have much quantitative credence. Yet I haven’t seen much discussion about these far-out forecasts’ precision. I consider these multi-decade predicted deaths and costs to be decent examples of imaginary numbers.

Think about it; who in 1930 (85 years ago) could have foreseen key details of our present economy such as its size, the composition of our economic output, let alone the number of bridges that are now in operation, and how many may now need repair. For some perspective, the US real GDP last year was more than 16 times as large as it was in 1930. In 1930, 21% of our labor force worked on farms (10 times what it is now).

The use of such far-out forecasts to substantiate needed environmental policies unfortunately gives climate change deniers an opportunity to be listened to. I don’t understand how any scientist – environmentalist, climatologist, economist or engineer – can place much credibility in such truly long-term forecasts. And yet, these far-out forecasts are presented and discussed as if they accurately represent what will actually be happening nearly 100 years in the future. It’s all quite puzzling, and unfortunate.