Middle Class Stagnation – or not?


First thing first: Solitude Challenge Results!!!

So finally back to a nice boring economics podcast again. This is a bit like a nerd alert on steroids but is kind of true to my theme of bringing an alternate view, backed by data, to some conventional wisdom. This time, I want to tackle the alleged stagnation of the middle class.

Everyone knows the media spin on this. Articles like this have been a dime a dozen in the last several years. The basic theme is that over the last several decades, while America has grown a lot economically (Think of where we were 30 years ago), the vast majority of those gains have gone only to the super-rich. And those in the middle class have ‘stagnated.’ That is their real wages have either stayed the same or declined.

So to examine this, I’m going to use for the bulk of this podcast, an incredible working paper put out by an economist at Cornell named Richard Burkhauser. I actually heard him interviewed on one of my favorite podcasts; Econtalk. (Episode here).  What I really like about his work, is that not only is it a great perspective on kind of a hot button issue, but it is also a great example about how important it is to define our assumptions and exactly what we are measuring when doing this type of analysis.

So let’s dive in and see what Burkhauser and his team did. At the root of everything, they are basically trying to answer the question; ‘How has the middle class done over the past 30 years or so, and is it true that they have stagnated?’ So the first question we need to answer, is how do we measure that? Personally I think the best way to measure well being would be to take all sorts of stats about how much leisure time people have, what kind of cool electronics they have in their house, combined with their income etc. But that type of data is super hard to get, so we are a bit limited by the data that is out there.

The simple answer is to just measure their income (inflation adjusted of course). So we just go to the IRS, get everyone’s tax returns, take the average each year, and whammo, we’ve got a nice looking graph that definitely trends up and to the right. Case closed, everything’s good right?

Turns out that when we are talking about income, just simply taking the average to find the “middle-class” has some problems. The main problem is outliers. With all of the economic growth over the past few decades, there have been some really super rich people created, and those outliers can artificially bring up the average and make the middle class look a little better than they probably are.

So the generally accepted way to measure income, or at least to find the middle class, is to look at median income. Quick review on what a median is, you just line everyone up in a gigantic line, and pluck out the person in the exact middle. No calculations necessary, you’ve found your median income.

Our next question is what are we lining up? Should we line up every single person? Man, woman, child etc? Obviously then you’re going to wind up with a ton of individuals that look like they have no income. That might be valuable for answering certain questions, but remember, we are asking ‘How has the middle class done?” So for our question, a stay at home mom of 2 whose husband makes $100K should probably not be seen as living in poverty because her personal income is zero. So instead of lining up individual people to find the median (middle class) income, we need to use some type of sharing unit.

This is where the Burkhauser paper really gets awesome, because his whole thesis is that this next decision we can make, has big implications on what he resulting data looks like.

So there are basically two sharing units that he addresses; the tax unit, and the household. It’s very possible that these are exactly the same, and actually, in most cases they are, but let’s define what they are and what each means:

Tax Unit: This one’s pretty easy. A tax unit is whatever is filed on your income tax return. So you and all your dependents. So in the traditional family example, that man, woman, and two kids would be considered one tax unit. These people might not necessarily all live under the same roof; think a college student whose parents still claim them as a dependent. The data on tax unit income comes from IRS records. So when we are lining things up to find the median middle class income, think about a bunch of tax returns sitting in a gigantic line, and we pick out the middle one.

Household: This one gets a little trickier. A household contains everyone who shares living expenses. The important nuance to note here, is that a household could have multiple tax units. Think of when we first graduated college. When we were living in Minneapolis together right after college, we had 4 separate tax units in one household. So as we decide what to put in a line as we are trying to figure out median income, picture actual houses being put in a line, and we pluck out the median house.

So are you starting to see how it’s possible to get a very different measure of median income based on what sharing unit you decide to choose? Think of where we would have fallen in that big line when we were all living together. So let’s say that the four of us were each making $30K. If you searched for us in the tax unit line, we’d all probably be below the median and pulling that median income down. But if you combine our incomes and looked at us as a household, all of a sudden our household income is now $120K and it doesn’t look like we’re struggling quite so much.

So in general, most researchers have used the IRS data to measure ‘well being’ of the middle class. This means they are using the median tax unit as their sharing unit. So they looked at data that took them across the last 3 business cycles. The business cycle is probably a podcast pillar in of itself, so I won’t get into it right now. But over the last 3 business cycles, starting in 1979 and going to 2007, the median income has only gained 3.2%.

Now that is legitimately stagnation. That is a horrible figure. Only gaining 3.2% over 30 years is really bad. But let’s first dissect exactly what that is measuring. For one, that is something called Pre-tax, pre-transfer income. Burkhauser actually used data from the census that happens every year instead of the traditionally used IRS tax return data. This is a survey that happens every year that asks people about their income. And the designation of “pre-tax, pre-transfer income” is important.

Pre-tax, pre transfer income means any income that is basically cash coming into the sharing unit. So obviously this includes wages from jobs, farm income, interest, dividends, rents, trusts, etc.

So if we just look at pre-tax, pre-transfer income, that number looks pretty grim, but let me re-iterate the question we are trying to answer: ‘How has the middle class done over those past 30 years.” Is pre-tax, pre-transfer income really the right metric to be looking at when considering well being. At least Burkhauser thinks no.

So the next metric that we can look at, that is a little bit broader, is pre-tax, post-transfer income. This includes everything from the pre-tax pre-transfer, plus it adds in social security, workers comp, etc. So any government transfers of income. Because when we are talking about well being of the middle class, it makes sense that we should include these sources of income for families right? These programs have actually grown considerably over the last 30 years,  so when we add that transfer income in, we see the increase in median income almost double to 6% since 1979. Still not awesome, but not as bleak as most studies might have you believe.

Burkhauser argues that still our measure of median income isn’t broad enough. (When we are specifically talking about well being). We need to actually be talking about Post-Tax, Post-transfer income. This includes all the crazy stuff that happens on tax returns. Because a lot of middle class people, particularly home owners, participate in some great programs that allow them to be even better off. So maybe someone gets a homeowners credit or energy credit or something. Burkhauser says that this affects their well being as well. When these credits and liabilities are added in, the average tax unit looks even better off. 9.5% since 1979 to be exact.

Remember, we are still talking about the tax unit instead of the household, which we think is probably a better metric to use when talking about well-being of the middle class right?

So finally, Burkhauser argues that the last piece we need to add in is employer health insurance. This isn’t a figure that IRS researches were able to get to because it isn’t reported on tax returns. But employer provided health insurance definitely is something that has improved the well being of most middle class individuals over the past 30 years.

When we add in the employer provided health insurance benefits into the median tax unit’s income, we get a growth of 18.2% since 1979. So again, still not a crazy increase, but you can already see how the argument that the middle class has totally stagnated is starting to get a little shaky.

Ok, so far, we’ve still just been talking about the tax unit. Remember earlier when we said that this was maybe not the best measure any more when we are trying to answer a well-being question? Instead, we thought it would be better to use the household as the sharing unit. So just to quickly review, the tax unit is just lining up the tax returns. So you plus your dependents. The household, is everyone that you share ‘stuff’ with. So there could be multiple ‘tax units’ within a household sharing unit.

When we look at the median growth of the household, Burkhauser’s team found the following data since 1979:

Pre-tax, Pre-transfer: 12.5% (compared to 3% for tax unit)

Pre-tax, post-transfer: 15.2% (compared to 6%/tax unit)

Post-tax, post-transfer: 20.2% (compared to 9.5%)

Post-tax, Post-transfer + health insurance: 27.3% (compared to 18.2%)

So you’re starting to see that when you see a number published, in order to really understand that number, we need to ask a little bit about what the assumptions are. There was nothing wrong with the calculations that other statisticians did, but their assumptions were different and so they were fundamentally asking different questions.

Now Burkhauser takes it even one step further and adjusts all of the figures he looked at for size. This one is a little harder to grasp, but not totally and has to do with a benefit to sharing. So let me pose a question. Let’s say the four of us were single and all made $20K/year, and we are trying to assess our economic well-being. You would agree that individually, we would be better off living together under one roof right? This is because of the benefit of sharing. Going from a one bedroom to a two bedroom apartment doesn’t double the price right? So we are a little better off economically because each of our $20K goes a little further when we can share things like rent, utilities, maybe we share food so we can buy in bulk, etc.

This is actually the way they set the poverty limits as well, so it’s not a controversial idea that we should be able to adjust the numbers a bit in order to show this. For instance, the poverty line for a one person household is anything under $11,170. For a two person household, this doesn’t simply double, it goes only to $15,130.

So I won’t go into how they adjust for this, but basically the gist is that the more people that are in a household, the more economies of scale can be taken advantage of, especially if there are multiple earners in that household, so there is a calculation that researchers do to pad that a little.

So finally what we end up with, is a figure for (deep breath) – Size adjusted, Post-tax, post-transfer, median household income has increased since 1979 36.7%. So thinking hard about what we are measuring and choosing our assumptions very deliberately is supremely important when trying to answer important questions. ESPECIALLY when there are policy decisions being made on this. Now I know that probably felt like a long time, but it really is just scratching the surface of this study. I would really recommend you listen to the econtalk episode where they go through it in depth because it took me a few times through to really get a firm understanding, and I’m still not positive that it’s totally solid in my mind. Here’s the whole table that summarizes everything:













So in addition to being a cool alternative look at the well-being of the middle class, and just how badly is the middle class doing, this study really hits home about how important knowing your inputs, and thinking about what question you are trying to answer is supremely important.

Leave a Reply

Your email address will not be published.Required fields are marked *