Macroeconomics 105
If you catch this virus, what is your likelihood of dying from it? The people want to know. Since Google is collecting your data, we can look at what the world is typing into their search bars. Here is a Coronavirus search trend analysis, courtesy of Google Trends. Number one on that list? “What is the recovery rate for Coronavirus?”
And if you try to answer that question, you’re not going to find a clear answer. Just a quick scan from news the past few weeks gives you numbers ranging from 0.6% mortality rate on the low end up to 3.4% on the high end. That’s quite a variance.
For those of you thinking “those are both small numbers…what does variance mean?” Well, basically, if you plug both of those numbers into a model to predict world deaths, one is going to be about six times bigger than the other. If I told you “you’re six times more likely to die if you go to school today”, you would probably stay home, even if the new odds were still near zero.
So, why can’t we settle on a mortality rate? We have hundreds of thousands of confirmed cases and tens of thousands of deaths, it’s simple math, right?
Here are a few variables that cause these discrepancies. We have no way of knowing how many people have the virus and just haven’t been tested. Maybe it wasn’t bad enough for them to go to the hospital, or maybe there’s a bunch of people who caught it and didn’t have any symptoms at all. Consider this.. what if the test results are flawed? There are stories of the tests producing up to a 30% “false negative” rate! That means it’s possible that up to 1 in 3 people who have been tested and told they didn’t have COVID-19 actually may have had it. How do you account for that in your mortality rate calculations?
Those are just the obvious ones. How do you account for the differences in data from country-to-country, or even city-to-city? Things like average age of population, density of living spaces, greeting culture (handshake vs hug vs kiss), and infinite other considerations, all matter. Italy, for example, has one of the oldest populations, has old and young living in the same spaces, and has a high percentage with a history of smoking, so maybe that explains their higher number of cases and deaths? But then Japan seems to have those same variables, with much less carnage so far.
Think about this. How many people are getting tested has a huge influence on how many reported cases there are, and different areas of the world have much different testing capabilities. Then there’s the problem with definitions.. if someone has a serious underlying health condition and also catches coronavirus, then dies, does that death go into the body count for the virus? The gray area leaves some things up to interpretation, which allows the possibility of death tolls getting politicized.
The New York Times published an article entitled “Teenager’s death in California is linked to Coronavirus”. In a later correction to the article, they added “The teenager has been dropped from the list of deaths from COVID-19.” Here is a local article.
It’s not hard to imagine someone along the path of that story’s publication considering that teen’s death as an opportunity to send a message to the “youth”. Whether something like that is an honest mistake, or a well-intentioned lie, or evil, doesn’t really matter, the point is the same: if we are adding and removing deaths from the master list, there is is no way we can have an accurate mortality rate.
I’ve mentioned death and mortality a bunch so far, so I’m going to pivot to a comic. Here is an xkcd on “conditional risk”: xkcd.com/795. This, in a nutshell, is why risk and probability is so complicated. If someone knows only 1 in 7,000,000 Americans die from lightning strikes each year, how does that alter their behavior? The “joke” is that it makes that person much more likely to be one of the 45 dead, due to a false sense of security.
Fivethirtyeight.com is a website dedicated to modeling. Their sage and founder, Nate Silver, rose to prominence after correctly predicting Barack Obama’s presidential run in 2008. Stats and predictions is literally what they do, but they have declined to release a COVID-19 model, because it’s so impossible to predict.
How is all of this related to GDP? Well, first I have to explain what GDP is.
Gross Domestic Product is the total (“gross”) amount of final goods and services produced (“product”) in a given country (“domestic”) in a given time period. We mainly look at GDP year-by-year and analyze it quarter-by-quarter, but with so much data, we don’t have the ability to calculate it in real time, so the Bureau of Economic Analysis releases numbers each month that are “updating” previous estimates (of previous quarters) and “estimating” the next official update. Sound confusing? Here is the last update, from March 26. Basically, what it is says is they have no updates to their previous numbers for the third and fourth quarters of 2019, so they’re fairly certain at this point both quarters displayed a 2.1% increase in GDP. You know what I mean by quarter, right? First quarter is the first three months of the year, second is the next three months, and so on.
One common misconception is that quarter growth should be in comparison to the previous quarter (quarter-over-quarter, aka QOQ), but it’s actually more informative to measure against that same quarter the previous year (YOY). So, saying the fourth quarter of 2019 demonstrated 2.1% growth year-over-year means Gross Domestic Product for October/November/December was 2.1% higher than that same stretch in 2018. If that doesn’t make sense, consider that if you were a Christmas tree company calculating revenues, it would be more helpful to compare one holiday season to the previous one, rather than the three months before or after the holiday season (it would look like your revenues “decreased” in January/Feb/March, but that is normal and expected!).
The first quarter of 2020 ended on the last day of March, but we won’t have the first Q1 estimate by the BEA until April 29, which they call an “advance estimate” (which means it could be way off). Oh, and for the record, the most common definition of a “recession” is negative GDP growth for two consecutive quarters. So, if you do the math, we can’t even *know* if we’re in a recession until six months or more after it started! This lag is an argument against the usefulness of these sorts of macroeconomics numbers, the equivalent of Karen Smith’s weather report in Mean Girls: “It’s 68 degrees, and there’s a 30% chance that it’s already raining.“
So, what is counted in GDP?
There are four categories: consumption (consumer spending), (business) investment, government spending, and exports minus imports. GDP=C+I+G+(x-m).
I should go into defining each of these, but it gets tricky and I’m nearing my word count limit. If you read the previous article I linked to, you’ll have a good basic understanding. And if that’s not incentive enough, I’m also putting a bonus question on the quiz…
The latest estimate for United States GDP in 2019 was $23.429 Trillion, the highest in the world. If you divide that by the U.S. population, called “per capita GDP”, you get $65,000, which is roughly top-10 in the world, depending on which source you consider.
Hopefully that is enough information for you to follow along when the news is talking about GDP, for now. The bad news is that serious economists don’t really use GDP much anymore (sorry). It was invented when our methods of calculation and collection of data were much worse, and it has all sorts of shortcomings and weaknesses. Talking about GDP today is similar to talking about a MLB player’s batting average or debating which NBA player is better just by looking at how many points they score per game. In short, we’ve come up with much better ways to measure the things that matter. So, why do we still talk about it? We’ve been using it for decades, and it’s still helpful for the general public as a starting point in the conversation.
But I promised to connect GDP to COVID mortality rates, and I haven’t done that yet.
One connection is the challenge of collecting massive amounts of data and drawing accurate conclusions in real time. It’s fairly likely to be wrong enough to hurt as much as help.
A second takeaway is thinking twice about how much it helps the average person to know things like global death rates from a virus, or your nation’s gross domestic product. How does that affect your daily life, when it comes to making decisions?
For example, if the global death rate from COVID-19 ends up being 3.4%, but in some communities it traveled through it was 0.2% and in your community it was 20%, then doesn’t that imply the global death rate is a pretty worthless statistic to people who were trying to figure out how to live through the pandemic?
The same thing applies to any macroeconomic data, whether it’s unemployment rates or housing markets, or whatever else. During the Great Recession, the nationwide unemployment rate peaked at 9%, but in some states it was well over 15%, and in communities hit hard you can imagine it being over 50%. So, if you’ve living in a town where real estate prices have bottomed out and you are underwater on your mortgage and you can’t find a job, what good is it to you to watch the news and see them report that GDP growth is up 2.1%! The macro statistics actually create a disconnect.
I’m not saying we shouldn’t collect that data, or that it’s worthless, I’m just suggesting we need to be careful not to conflate big data with real people making decisions specific to their lives. Knowledge of particular time and place is important.
That’s all.
Oh, I made a Mean Girls reference, but if you know me that’s not a new recommendation sooo… my movie recommendation for this week is actually a TV show!
Community. I like to say it’s my favorite TV show ever, mostly because I’m the only person I know who likes it enough to say it’s my favorite TV show ever, but also because it’s simultaneously cynical/self-aware and empathetic/redemptive. Also, I went to community college. Also, after I transferred to Hillsdale College, I would host weekly gatherings to watch the new episodes as it came out each week, so it’s got that nostalgic thing for me, too. Community is what matters, see?
