Read Part I
Cap & Trade & Tax & Spend is not dead. It was never even ailing. It’s healthy and alert and standing in the corner, unnoticed. It has on its face a grin of the type we can’t mention on a family blog.
Senate Democrats worked with Al Gore, and he cobbled together a list of 2000 people (some scientists, some not) to sign a letter demanding action on “climate change”, a.k.a. global warming. In Part I, we saw that many of the letter’s signers were economists.
These economists are being shifted to the front line to mask the climatologists behind them. This is because climatologists have, over the past four months, received a series of black eyes; right now, they’re not too pretty to see.
Make no mistake: the main arguments you will hear shortly in support of Cap & Trade & Tax & Spend will be economic and not scientific ones. We therefore must understand the nature of these arguments.
Yesterday, we examined briefly how conditional probability should be used in unwinding economic predictions so that we can estimate their true likelihood. That is, usually economic statements are said to be true with a certainty far greater than warranted.
An economist might say, “If we don’t do something about global warming, then we’ll lose over three million jobs a year starting in 2020.” This sounds like certainty: we will lose at least that many jobs etc. It is not certain.
At the least, this prediction is conditional on the belief that harmful man-made global warming (HGW) is true. It is not. Even the IPCC says that HGW is only 90% certain. That simple fact means that our economist’s prediction can be no more than 90% certain, and is probably far less certain than that.
That prediction also relied on a string of assumptions of how job loss is related to temperature (and perhaps other climatological variables). None of these assumptions is certain. In other words, each assumption is only probable.
If each assumption is logically independent from each other assumption, then the probability that each assumption is true can be multiplied together. The resulting product will say how likely the entire prediction is.
For example, a typical economic model might have a dozen assumptions, plus the assumption that HGW is true. If we generously suppose each assumption is 50% likely, and since we’re nice people, we’ll only suppose there are four assumptions, then there is only a 5% chance that the prediction is true. This comes from 0.5 multiplied by itself four times and then multiplying that by 0.9.
The prediction, therefore, has gone from scarily true to probably wrong.
The situation is more complicated when each model assumption is not logically independent; which they usually are not. When they are dependent, the string of them is more likely than when they are independent. How much more so depends on the assumptions themselves. There is no general formula, of course, so each prediction must be handled individually.
However, since any economic prediction conditional on HGW will contain a large number of assumptions, the chance that any prediction is true is probably still small.
But what if there are dozens of predictions, each equally dire? Shouldn’t the sheer quantity of doomsday scenarios cause us to worry?
No. Uncle Mike in a comment to Part I presented an example, which I’ll modify here. We have a die which we’ll toss once. Given the usual premises, the chance that it shows a six is 1/6. Thus, if I predict a six will show, there is a 17% chance I’ll be right.
Now suppose that you and I both predict a six will show. What is the chance that at least one of us is right? It is still 17%. We are predicting the same thing and we are both using the same theory, by which I mean the same premises. You could equivalently say that we are using the same model. Or—yet one more way to say it, and one that uses the language of both parts of this article—we are both conditioning on identical information.
Our predictions are not logically independent: they are the same thing. The chance that at least one of us guesses correctly is still 17% if you and I and your mom also predicts a six. Or if a million people predict a six, or if every person on the planet does. The chance that at least one of us is right is never better than 17%.
So when we hear hair-raising predictions coming from all quarters, we can’t assume that the probability that at least one or more of them is true is high just because there are so many of them. While each horror has slightly different information than its brother, at base they are all conditioning on much the same information.
The voices in the choir are all singing the same song.
Read Part I