The U.S. economy continually expanded over the past 76 months. Goldman Sachs economist Zach Pandl tells us there is a 60% chance the economy will continue to expand for another 44 more.
How does Pandl arrive at a 60% probability the economic expansion will last 10 years? I can't say for sure, but Bloomberg reports that Pandl analyzed 355 expansions across 14 economies since 1850. Pandl found that expansions have lengthen in duration since 1950. Through observation and statistical calculation, Pandl produced a 60% probability that the economy will continue to expand. What's more, Pandl's black box calculation tells us there is only a 10-to-15% chance of recession next year.
It's all nonsense, of course. It's numbers masquerading as science. Class probability has once again been confused with case probability.
Class probability involves endlessly repeatable standardized events. Games of chance come to mind: cards, dice, coins, roulette. If these games are fair, we can derive precise probabilities. We can quantify the likelihood of rolling a seven in a dice game or flipping a coin for tails 10-consecurives times.
Case probability, in contrast, concerns unique events, which we can have no information about probability distributions. Such events include elections, sporting-events, and, yes, economic predictions. To offer probabilities on unique events is logically incoherent.
The futility in offering a probability on an economic event is easy to understand. A market-based economy is dynamic, capricious, and ever-changing. Isolating variables (which we can do with class-probability events) is impossible. Many meaningful economic variables in 1850 are meaningless today. To attempt to identify all economic variables is a Quixotic undertaking at best. To measure their influence and the infinite relationships among variables is impossible.
Therefore, Pandl's 60% probability isn't fact, it's opinion, and that's all it can be when case probability is involved. Unfortunately, people too frequently believe otherwise. Look no further than the pervasive belief in the validity of VaR models.
How does Pandl arrive at a 60% probability the economic expansion will last 10 years? I can't say for sure, but Bloomberg reports that Pandl analyzed 355 expansions across 14 economies since 1850. Pandl found that expansions have lengthen in duration since 1950. Through observation and statistical calculation, Pandl produced a 60% probability that the economy will continue to expand. What's more, Pandl's black box calculation tells us there is only a 10-to-15% chance of recession next year.
It's all nonsense, of course. It's numbers masquerading as science. Class probability has once again been confused with case probability.
Class probability involves endlessly repeatable standardized events. Games of chance come to mind: cards, dice, coins, roulette. If these games are fair, we can derive precise probabilities. We can quantify the likelihood of rolling a seven in a dice game or flipping a coin for tails 10-consecurives times.
Case probability, in contrast, concerns unique events, which we can have no information about probability distributions. Such events include elections, sporting-events, and, yes, economic predictions. To offer probabilities on unique events is logically incoherent.
The futility in offering a probability on an economic event is easy to understand. A market-based economy is dynamic, capricious, and ever-changing. Isolating variables (which we can do with class-probability events) is impossible. Many meaningful economic variables in 1850 are meaningless today. To attempt to identify all economic variables is a Quixotic undertaking at best. To measure their influence and the infinite relationships among variables is impossible.
Therefore, Pandl's 60% probability isn't fact, it's opinion, and that's all it can be when case probability is involved. Unfortunately, people too frequently believe otherwise. Look no further than the pervasive belief in the validity of VaR models.
RSS Feed