Nobel-prize and shrewd economist Krugman released a somewhat controversial tweet today, which on the face of it looks fairly detrimental for the incoming Trump Administration. But a careful analysis of the historical data doesn't necessarily portend that bad times are before us, and in fact (based on the same historical probabilities from archived data) there is maybe a 50/50 chance that we'll continue to see a smartly decent labor market, for years to come. This is not a forecast, and Donald Trump hasn't even been sworn into office yet. In any event, let's root for the President-elect to succeed, since that is equivalent to rooting for all fellow Americans to also flourish.
This sort of regression analysis on unemployment data is nothing new, and has been around nearly as long as labor statisticians have officially collected unemployment data, and using regressions themselves. All of the famous academic and private sector economists have used it in and out of government. But the success of it has been mixed. For example, former economics adviser Romer famously used it recently to justify the $800 billion ARRA stimulus program that she promised would top the rise in the financial crisis unemployment (at 8%!) Mathematics professor Nassim Taleb may refer to economists such as Professor Romer, as IYI.
In order to start thinking about this topic, let's take it layer by layer and first see my version of Professor Krugman's chart. It doesn't at all matter which dates within the period one sub-selects and -unlike in the tweet above- we show all of them below. We focus on 5% unemployment rate, just north of where we currently are. This is the 30th percentile for unemployment since official records have been kept in 1948. Suffice to say being below this round number would generally be considered "good". Additionally, we show the 90% confidence interval about the data, as well as a horizontal orange line separating the 4-year forward unemployment rate rising versus falling.
Of course the horizontal orange line is arbitrary in the sense that simply being a little above or a little below zero is less relevant versus knowing what the actual rate is. Nonetheless we can see that 5% unemployment has a respectable probability of sticking, based on the amount of confidence interval that is below 0 on the vertical axis (implying reduced unemployment or equally stated as improving jobs!)
What is more important than simply the unemployment, 4-years down the road, is a measure of the path of unemployment over that period. So the first thing we'll look at is the average unemployment over a 4-year period, instead of the ending unemployment rate. And using this important approach, it doesn't appear that joblessness appears "bad" over the next four years, if unemployment is (as it is now) just below 5%. The darker levels of the mapping below show a higher probability density, and those contours are concretely at or below 5% unemployment.
Besides 4-year forward economic data, what's just as pertinent for all of us to gauge is to look at the duration in which the low unemployment can endure. Consider this a reverse stress-test, except we're not stressing anything! Simply stating low unemployment "reverts" is not an indication that there is a high probability such reversion is anytime soon. For any given unemployment, see how many additional months, which unemployment continuously remained below 5%. While we find the chart on the left a little easier, both charts below show the same information but through two different distribution charting styles.
The bottom line here is that even with our sub-5% unemployment rate, we have about a 50/50 chance of continuously remaining <5% for another 2 years! So anything can happen. But it should be considered enormous news that there is an additional 25% chance this low unemployment ("good" jobs environment) remains for more than the next 3 years.
The point of all this is that surely Professor Krugman and others are correct that joblessness will eventually rise above the median rate of 5.7% (which recently has been in the 6% range). And when it does we will of course be in a recession. But we simply don't know when it will later surge, and can't say with a high degree of confidence that the Trump Administration (and as an extension Americans) are suddenly doomed by a simple/historic 4-year forecast based on the current low levels of unemployment. Such ideas have proven difficult for elite economists, many of whom assumed that the current business cycle would have concluded a while back.
With some things that are random, it is simply not that straightforward. Let's be optimistic when possible. Just a note that the Trump Administration starts in a couple weeks, and he'll have a few years to influence all of these economic statistics (similar to how we did in the Obama Administration during the early years)!
Do people realize that good jobs news is bad for Trump? A low unemployment rate now means it probably rises on his watch. Data 1948-2016 pic.twitter.com/wBt40aQ078— Paul Krugman (@paulkrugman) January 6, 2017
In order to start thinking about this topic, let's take it layer by layer and first see my version of Professor Krugman's chart. It doesn't at all matter which dates within the period one sub-selects and -unlike in the tweet above- we show all of them below. We focus on 5% unemployment rate, just north of where we currently are. This is the 30th percentile for unemployment since official records have been kept in 1948. Suffice to say being below this round number would generally be considered "good". Additionally, we show the 90% confidence interval about the data, as well as a horizontal orange line separating the 4-year forward unemployment rate rising versus falling.
Of course the horizontal orange line is arbitrary in the sense that simply being a little above or a little below zero is less relevant versus knowing what the actual rate is. Nonetheless we can see that 5% unemployment has a respectable probability of sticking, based on the amount of confidence interval that is below 0 on the vertical axis (implying reduced unemployment or equally stated as improving jobs!)
What is more important than simply the unemployment, 4-years down the road, is a measure of the path of unemployment over that period. So the first thing we'll look at is the average unemployment over a 4-year period, instead of the ending unemployment rate. And using this important approach, it doesn't appear that joblessness appears "bad" over the next four years, if unemployment is (as it is now) just below 5%. The darker levels of the mapping below show a higher probability density, and those contours are concretely at or below 5% unemployment.
Besides 4-year forward economic data, what's just as pertinent for all of us to gauge is to look at the duration in which the low unemployment can endure. Consider this a reverse stress-test, except we're not stressing anything! Simply stating low unemployment "reverts" is not an indication that there is a high probability such reversion is anytime soon. For any given unemployment, see how many additional months, which unemployment continuously remained below 5%. While we find the chart on the left a little easier, both charts below show the same information but through two different distribution charting styles.
The point of all this is that surely Professor Krugman and others are correct that joblessness will eventually rise above the median rate of 5.7% (which recently has been in the 6% range). And when it does we will of course be in a recession. But we simply don't know when it will later surge, and can't say with a high degree of confidence that the Trump Administration (and as an extension Americans) are suddenly doomed by a simple/historic 4-year forecast based on the current low levels of unemployment. Such ideas have proven difficult for elite economists, many of whom assumed that the current business cycle would have concluded a while back.
With some things that are random, it is simply not that straightforward. Let's be optimistic when possible. Just a note that the Trump Administration starts in a couple weeks, and he'll have a few years to influence all of these economic statistics (similar to how we did in the Obama Administration during the early years)!
Austrian Business Cycle Theory (ABCT) holds that the more the interest rate curve is suppressed by government, the more mal-investment distorts future economic performance. Simply put, low cost money allows people to think they can afford to approve long term capital-funded projects. When the government can no longer sustain its suppression of interest rates, and rates jump up, suddenly these projects no longer pay for themselves. So a recession is backed into the cake that might have Trump's name on it.
ReplyDeleteOf course progressives will instantly call it the Trump Recession. They will make him own it. You heard it here first.
DeleteThanks thebuckwheat. Math, and positivity makes the world go 'round! Also wanted to cite everyone's darling billionaire, Warren Buffet (who had read and discussed my work): "We've long felt that the only value of stock forecasters is to make fortune tellers look good.", or "We will continue to ignore political and economic forecasts which are an expensive distraction for many investors and businessmen."