Common Statistical Measures 29:50
Antony Davies discusses a few popular statistics and clears up common misconceptions about how they are measured. We focus on unemployment rates and GDP.
Antony Davies: We’ve talked a little bit about statistics and how we can use them. One of the important things to always keep in mind is that statistics don’t always measure that you think they’re measuring. Good example is unemployment. What could be easier than calculating how much unemployment there is? You say, [00:00:30] the unemployment rate is 4%, that means 4% of the people are unemployed, right? It’s a little more complicated than that.
First off, when we talk about unemployment the number applies only to people who we call workers, that is people who are in the labor force. People who are in the labor force, or people either who have jobs or don’t have jobs but would like them. So people who are retired, they’re not in the labor force. People who are full time students, they’re not in the labor force. People who [00:01:00] are chronically ill, they’re not in the labor force. The labor force is a subset of the population, so when we say for example unemployment rate is 4%, we mean 4% of the labor force does not have jobs.
Now that seems pretty straight forward. Except it isn’t. Consider this case. Suppose you had five million people in your labor force, and of these five million people, [00:01:30] 4.7 million of them had jobs and the other 300,000 did not have jobs. Your unemployment rate is the number of unemployed people, the 300,000, divided by the labor force of five million. So, 300,000 divided by five million is 6%, it’s a 6% unemployment rate.
Now, one of the problems with defining the labor force as people who have jobs or people who don’t have jobs but would like jobs, is [00:02:00] it’s unclear what it means, “someone who doesn’t have a job but would like a job.” If we could call all of the people who didn’t have jobs and ask them, “Would you like a job if there were one available?” That’s okay, but we can’t do that. We have to attempt to estimate the number of people who don’t have jobs but would like to have them.
So one of the things that we do is we ask questions like, “Have you [00:02:30] actively looked for a job in the past few weeks?” That is, you’ve responded to an ad or you’ve gone down to the local unemployment office, whatever it is. Anybody who says yes to that, we say, “Alright, well you’re a person who doesn’t have a job but would like one, so you’re part of the unemployed, that you’re our 300,000 number.” Right?
But now things start to get sticky. What happens if you’ve got somebody who would really like a job but [00:03:00] this person just kept looking and looking and looking, and not found any job at all. He’s come to the conclusion that it’s not worth his time to look for a job, so he stops looking. This sort of person is someone who does not have a job but isn’t looking for one either, but would take one if one came along. So he’s rationally stopped looking because it’s not worth his effort. He’s no longer counted on our unemployment rolls as unemployed. He is what we call non- [00:03:30] employed.
So interestingly we’ve got now three categories of people, the employed, the unemployed, the non-employed. The non-employed are people who don’t have jobs but we have determined, using some metrics that we use, that these people don’t want jobs either. We understand we might be wrong about that but given that we can’t call up everybody and ask them their particulars as to why they don’t have jobs, we’ve got to draw a line in the sand somewhere. So we say, “Look, if you’ve been out of work a certain period [00:04:00] of time, or you haven’t looked for a job in a certain period of time, we’re going to put you off the rolls, count you as non-employed.”
Now, here’s where the statistics get very messy. Let’s suppose of our five million people, 4.7 million of them have jobs, 300,000 of them don’t. 100,000 of these 300,000 have been unemployed for long enough that we move them off the rolls. We now consider them non-employed. When we move 100,000 of these [00:04:30] unemployed people off the rolls or reclassify them as non-employed two things happen. First the number of unemployed people has declined. It was 300,000, it’s now 200,000. Second, our labor force has declined it was five million, the employed and the unemployed people. Now it’s 4.9 million, because these 100,000 become unemployed, or sorry, they become non-employed. We no longer count them as part of the labor force.
So when I go to calculate my unemployment numbers, [00:05:00] I have 200,000 people who are unemployed out of a labor force of 4.9 million. That’s an unemployed rate of 4.1%. So there’s something fascinating, simply by these 100,000 people being unemployed for so long that we no longer count them, my official unemployment rate has dropped from 6% to 4.1%. Even though I have the same number of jobs I had before. I had 4.7 million people working before, I have 4.7 million [00:05:30] people working now.
This is one of the problems that we have to keep in mind as we think about statistics, in this case specifically the unemployment rate, is be very careful that you understand how these things are measured. Now, that’s a hypothetical case, but it becomes real when you start to look at unemployment numbers in the US starting with the great recession.
Around about 2009, 2010, of course we [00:06:00] have the great recession, lots of people are out of work. The media would go and interview people on the street around about 2010 when the economy started turning around. They would say things like, “Well, the economy is turning around. Our unemployment rate numbers are down. People are getting back to work. How do you feel?” Invariably the man on the street that the media would interview would say something like, “No. I don’t think the economy is getting better.”
This persisted. It persisted so much it became [00:06:30] kind of thing in the media, of why is there this disconnect that quarter after quarter the economists are saying unemployment rate is falling, and yet quarter after quarter when we interview people in the street they say that they don’t believe the economy has improved. Why the disconnect?
Part of the disconnect is that in the great recession we had a tremendously long period of unemployment. Historically it’s the largest … the media unemployed person [00:07:00] was unemployed for a longer period than had ever been the case before. Unemployed people were unemployed for so long that census bureau stop counting, the bureau of labor statistics stopped counting them as unemployed. So we were seeing the unemployment rate drop in part because more people were getting jobs, but also largely in part because some people were unemployed for so long they stopped looking and we stopped counting them.
That was the disconnect. In fact, the people in the street were right. Their gut [00:07:30] reaction was that the economy wasn’t improving, and largely speaking, there are exceptions, but largely speaking it was about the same as it was before. The drop in the unemployment numbers was due to this statistical abnormality, if you want to call it that, of this oddness of how we calculate unemployment.
If you look at the numbers, and here you have them, the blue line on this graph is the unemployment rate, which you can see rising 2008. It starts first quarter 2008, you can see [00:08:00] it go up with the start of the great recession and kind of tail off a bit. And then the employment rate, which you see on the red, it starts up high and then drops down low. The interesting thing here is you can see the unemployment rate starting around fourth quarter of 2009 through 2010, the unemployment rate falling but the employment rate not rising. It’s this phenomenon that you would [00:08:30] only come to understand if you’re looking not simply at this raw statistics of what’s the unemployment numbers, but rather also understanding how they’re defined.
Some other things to be careful about. We’re talking about unemployment, be careful how you quote unemployment statistics, because you can find contradictory numbers coming from the same entity. Bureau of labor statistics for example measure on [00:09:00] unemployment numbers how many people have jobs, and they do it two ways. They will survey employers and ask, “Have you taken on new employees? Have you laid anybody off? How many workers do you have?” Those numbers all put together give us a picture of how many jobs there are.
Bureau of labor statistics will will also ask workers, surveying workers. “Do you have a job? Are you employed full time?” These sorts of things. And compile those [00:09:30] statistics. So on the one hand we have employment numbers coming from employers, and then we also have employment numbers coming from workers. The two don’t necessarily agree, in fact they almost always disagree. There’s some noise in the thing.
So one of the things we have to be careful of is when you hear politicians for example touting the latest job reports and saying, “Well, according to the statistics we’ve create 100,000 jobs in the last [00:10:00] month. This is good news, my policies are working.” Be careful that you understand what’s going on. Because of oddness in how we measure employment, because of differences in not just how we define unemployment, but in how we’re measuring it, are we asking employers? Are we asking employees? And because of random noise, that I can’t ask all the employees, I can’t ask all of the employers, I have to take samples. [00:10:30] Because of these three things, oddness of how we define the thing, how we measure the thing and randomness in the measurement itself, there’s lots of noise in the data.
There’s enough noise that 100,000 jobs added in a month is not very distinguishable from just the background fluctuations. In order words, for the politician to come out and say, “Look, clearly my policies work. We’ve added 100,000 jobs.” Statistically speaking that 100, [00:11:00] 000 is not distinguishable from background randomness. It’s like my saying, I can command the laws of physics. I say that when I flip a coin it will come up heads, and I flip the coin and it comes up heads. And I say, “See, my policies work.” What’s going on is by random chance it happened. You’ve got to flip it lots of times before you see that actually I have no control over the coin. Similarly here, 100,000 is like [00:11:30] the flip of the coin. It’s not a large enough number to really be saying much about. It’s randomness.
Student: When the economy is affected and people choose to go to law school maybe, or when the economy is getting healthier and more people are entering the workforce, how does that affect non-employed and unemployment figures?
Antony Davies: Yeah, it does affect them. When you have a bunch of people who graduate college and now hit the job market, you’ll see, that would contribute upward pressure to the unemployment rate, because [00:12:00] all the sudden here are people we weren’t counting before. They were full time students and now they’ve entered into the market looking for jobs. They now get classified, they get moved from the non-employed to the unemployment category. That will give you upward pressure on unemployment, then once they get jobs it comes back down again.
A similar phenomenon is happening now actually with baby boomers hitting retirement. As baby boomers hit retirement they’re exiting the labor force. [00:12:30] Now we have a decline not only in the number of employed, because they were employed people who are now ceasing to be employed, but they’re moving from employed to non-employed.
So we talk unemployment, another measure to be careful that’s quoted frequently is inflation. When we hear inflation numbers, we often mistakenly assume that these numbers are attributable to all prices everywhere. [00:13:00] Keep in mind that inflation is an average number. So when we say something like, and the actually number is in 2014 the inflation rate in the US was just under 1%. Now what does that that mean? What it means is on average the prices of things that consumers buy rose by 1%, on average.
To give you some specific examples, in that same year when inflation was 1%, the price of gasoline fell 4%, the price [00:13:30] of computers fell 10%, the price of housing rose 5%. So we have to keep in mind the definition of inflation, it is the average change in the price level across all goods. Just because we see a low inflation rate it doesn’t for example that we should expect the cost of tuition to be rising at 1% because inflation is 1%. The individual prices will rise at different [00:14:00] rates. They’ll just average to what we call inflation.
So measures that people commonly hear, statistical measures people commonly hear, unemployment, inflation, another one is GDP. GDP, people generally are correct in saying, “Well, this is a measure of the economic activity that’s going on.” There are some details in the number but basically that’s right. As GDP rises, generally speaking that means that we’re [00:14:30] better off, because increased GDP means we have increased incomes.
Of course technically speaking you need to adjust for inflation, because GDP could go up simply because prices rise. So if I buy this coffee cup for $2 this year, that’s $2 that goes toward GDP. If I buy the same coffee cup not for $2 but for $5, that’s $5 that goes toward GDP. [00:15:00] The GDP is higher. Now it looks like we’re better off because the GDP is higher, in fact we’re not any better off. It’s the same coffee cup it was before, all that happened is I had to pay a higher price for it. One of the things we have to be careful of is when we think about GDP over time, make sure that you’re looking at GDP that’s adjusted for inflation. That’s a better measure of our wellbeing as a whole.
But even GDP [00:15:30] adjusted for inflation only tell us how well off we’re doing in the aggregate. It doesn’t tell us much about how we’re doing on an individual basis. Good case in point is China’s GDP is a little bit higher than the United States’ GDP, but their population is much higher. So on a per person basis the GDP is actually much lower in China than in the US.
[00:16:00] To look at GDP as people will quote it, as this is a measure of how well off we are, keep in the back of your mind we need to make sure we’re looking at GDP adjusted for inflation and on a per capita basis. There are problems there still, but generally that’s not a bad proxy for how well off people are on average.
One of the bigger problems with GDP as a measure of how well off we are, is that GDP [00:16:30] simply adds up the value of all the goods and services that are sold. It ignores what those goods and services are. For example, if I sell you an apple pie for $10, that $10 gets calculated into the GDP and so we see the GDP is, you know, however large it is, and this measure of GDP is an attempt to represent [00:17:00] the apple pie and all the other things that you bought.
Okay, fine. Suppose now I sell you not an apple pie for $10 but a mud pie for $10. When I calculate GDP it’s the same $10 that goes in here. So whether I sell you a mud pie for $10 or an apple pie for $10, it doesn’t alter the statistics, the GDP is the same. It’s $10 more GDP. The problem [00:17:30] of course is nobody wants a mud pie. So you are worse off, even though the GDP is the same. I sell you apple pie, GDP is here. I sell you a mud pie, GDP is here. But if sell you a mud pie you’re worse off, and all the sudden you see the GDP doesn’t necessarily always reflect how well off we are. It moves in concert with how well off we are but there are some exceptions to this.
The exceptions, one source of the exceptions is government intervention [00:18:00] in the economy. Go back to the mud pie, why would you buy a mud pie from me for $10? Generally there’s no reason. Nobody wants a mud pie. But if I have a government that taxes you and has decided, for whatever reason, that it would be good if you had mud pies. The government taxes you, uses that money and buys the mud pie and puts it on a table and says, “Here’s a mud pie.”
Notice what’s happened. With the government intervention, the government through taxation and spending [00:18:30] can effectively force you to buy things you would not voluntarily buy on your own. So we can have a GDP of a certain level and say, “Well, we must be well off, because we have this nice big GDP.” What we’re missing is what comprises this GDP. Is it generally goods that people have chosen to buy for themselves? Or is it goods that the government has chosen to buy for you?
This becomes important [00:19:00] when we start to talk about the importance of, for example, stimulus spending. So people will say things like, “Well look, it’s good that the government spends money, because when it spends money it creates jobs and this creates income for other people and lots of nice things happen.” None of that is wrong, but there’s an important point that’s missing. That is, what is the government spending the money on? Simply because it’s spending money doesn’t mean you’re better off. It does mean the [00:19:30] GDP will be higher, but it doesn’t mean necessarily that you’re happier.
If I right now take $5 from each of you and buy a pizza you’ll all be happy. If I take $5 from each of you and buy more paper and throw it on the table, not necessarily so. But the GDP is the same in both cases.
Student: So given these deficiencies of GDP calculation, are there any substitutes [00:20:00] that are debatable?
Antony Davies: Yeah, are there substitutes? There are. There’s a measure called gross output, which is an interesting thing. It measures more things than GDP does. It suffers from … It’s free of some of the problems of GDP, but it suffers from some of the same problems of GDP and it has some other problems that GDP doesn’t have.
There is no perfect way to do this. We’re [00:20:30] talking about measuring people’s wellbeing. The reason there’s no perfect way to do it is because each individual is different and things that you like aren’t necessarily things that he would like. If I add up things that you like and say, “This is a measure of wellbeing,” I haven’t captured him. But if I add up things that he likes and say, “This is a measure of our wellbeing,” I haven’t captured you. So there is nor right way to do this. There are just competing semi-bad ways to do it.
[00:21:00] The important thing here is that when we look at statistics like inflation, unemployment, GDP, GDP growth, that we understand that what we’re looking at is not really the thing we want to see. It’s a proxy for the thing we want to see. It has some good points, it has some bad points. It’s just necessary to be aware of what those are.
Student: Is there a way for us to measure things, goods and services, [00:21:30] that aren’t necessarily counted as they normally would be in GDP? So things like black market stuff aren’t necessarily traded, at least in any way that we could probably measure, so is there any way that we can track that and measure that sort of goods and services are actually being traded outside of the regular market I guess?
Antony Davies: Yeah, that’s a good question. There are two points here. The first is, GDP, our attempt to measure all the goods and services [00:22:00] that are produced, is not a complete measure because it ignores anything for which there isn’t a paper trail. That includes goods that are bought and sold under the able. So you hire the neighbor kid to mow your lawn, you pay him a couple of bucks, that’s a service that’s being provided. It’s no included in GDP, there’s no paper trail. You buy something illegally, clearly the local drug dealer is not going to give you [00:22:30] a receipt. There’s no paper trail. There’s paper trail, there’s no way to measure this thing.
Also, there are products for which there’s no paper trail that have nothing to do with being under the table or being illegal. If you have a row boat and you like to go out rowing on the river on a Saturday. You go out on a Saturday and you row along, you do your thing, you’ve having a good time. This is leisure. You are producing something. You’re producing leisure for yourself and you’re consuming this. This has value, and yet it’s not included in GDP, [00:23:00] again because there’s no paper trail.
So anything that can’t be tracked through taxes or filings or this kind of thing, we don’t measure. Economists attempt to get their heads around these things, you could imagine the problem. Estimating the value of people’s leisure is hard enough given that A, it’s a hard thing to measure, but B, people are at least willing to answer the question when you say, ” [00:23:30] How do you spend your leisure time?” Measuring the drug trade you can imagine is hard, not only because there are things that are difficult to measure, but people aren’t interested in answering your question, when you say, “What’s the value of the illegal drugs that you sold last quarter?”
Economists attempt various ways to get at these, and there are fuzzy numbers as to how large those numbers might be. Our takeaway here is to understand that the GDP measure that we look at, that measures [00:24:00] the magnitude of our economy, is smaller than the actual economy because of these things we’ve left out. How much smaller? We don’t know, but it is smaller than the actual economy.
Student: How does this influence our measurement of command economies, where probably most of the stuff that’s creating value for people, or at least much of the stuff, is happening in black markets, and then a lot of the stuff that’s officially counted is produced by a mechanism of central planning that’s going to create lots of stuff that’s not producing value for people? [00:24:30] Should we just take any GDP measures from command economies with many grains of salt?
Antony Davies: Generally yes. What becomes a little bit easier with command economies is, and you’re absolutely right we can’t necessarily trust the numbers we’re looking at because they aren’t measuring things in the way that we measure them in a free market. So a dollar spent in a command economy isn’t necessarily the same kind of thing as a dollar spent in a free market.
However, some [00:25:00] things we can do is look at growth rates. To say, “Okay, here’s a command economy. They’re measuring GDP and they tell me their GDP is $2 trillion. I’m not quite sure what that means. I think it doesn’t come close to meaning what we say in a free market, where we say $2 trillion.” However, from last year to this year, their number went from two trillion to 2.1 trillion. That’s a X percent growth. Whatever [00:25:30] that growth is, they were measuring it the same way before last year and they’re measuring it the same way this year. Last year they got a 2 trillion number, now they’ve got a 2.1 trillion.
So I can conclude that there was some growth here. Whatever nastiness there is in the measure, I know that the measure rose. That rising of the measure would indicate that there’s something perhaps real going on here, assuming that you can believe the numbers at all.
Student: One more question, going back to remaining compliant. Is there any [00:26:00] measure for the GDP that’s created from having to go to H&R Block to file my taxes? That creates GDP, but then the people who buy TurboTax and spend six hours doing it themselves don’t necessarily generate that same value.
Antony Davies: Yeah. That’s an excellent question, because on the one hand the people who use TurboTax … Suppose I can use TurboTax and it’ll cost me 70 bucks [00:26:30] to buy it. That 70 bucks contributes to GDP. If I had gone to a tax accountant, it would have cost me let’s say 150. The TurboTax is only 70, but I have to add my own labor. Let’s suppose when I add my own labor I’m now up to a total of 150, 70 for TurboTax plus the other 80 bucks of my labor, that gives me 150, same as if I went to the tax accountant.
If I go to the tax accountant the full 150 counts toward [00:27:00] GDP. Over here, we don’t count the value of your labor because there’s no paper trail, so all we count is the 70 bucks for the TurboTax. That’s one issue. Doing the taxes myself will cause the GDP to appear to be less. The other thing that’s going on is, that doing the taxes at all magnifies GDP erroneously. Because what I’m doing is I’m sitting down and I’m spending this time, or I’m paying an accountant to spend the time, going [00:27:30] through all of these rules to figure out how much money I how and then cut a check to the IRS. It’s a tremendous amount of effort to do that.
That effort isn’t producing anything. Well, it’s keeping me out of jail because I’m paying the IRS, right? But it’s not producing anything in any real sense. Think of it another way, if the government had a much more simplified tax code, that resulted [00:28:00] in me owing the same amount of money I do now, but I could do it by filling out two lines on a card, I could fill out those two lines on the card, pay the government the same amount of money I’m paying them now. All of that time that I was spending doing all this nonsense is now freed up to do other things. I can be more productive, I can go rowing on the river with my canoe, or whatever it is.
What the government has done by causing me to spend a tremendous amount of time doing my taxes, or by causing [00:28:30] me to pay someone else to spend a tremendous amount of time doing my taxes. What the government has done to force me to buy a mud pie. Here’s this thing that you would not like, it doesn’t help you any, but I’m going to make you buy it.
We make jokes, people make jokes about government make-work projects. Where you hire people to dig holes and other people to fill them back in again. You look at the dollars that change hands. “We paid these people to dig the holes. We paid these other people to fill them back in.” That’s lots of GDP, but if you look at what’s produced, nothing was [00:29:00] produced here. Nothing. It’s the stuff that’s produced that we care about. The GDP measure is an attempt to get our heads around the stuff that was produced.
That’s no different than the tax code. With the tax code, we create this very complex thing. It’s like saying, “Okay, dig this hole.” And then we have to hand it over to very smart people to figure out how we can comply with it. That’s like, “Fill the hole back in.” At the end we really haven’t produced anything.