January 22nd, 2024
A recent video claimed that the middle class isn’t just dying, but is already dead. What is the evidence presented and is it true?
While the current economy has many problems and lots of room for betterment, we currently have a far higher median standard of living today than in 1950. The original claim relies on comparing apples to oranges and other problematic data jumps. Doing historical economic comparisons requires tremendous patients and care to make sure we’re actually comparing the same two things across time. This involves sorting through definitions and, often, original documentation.
Real Median Family Income
The best measure of cost of living is the CPI. Why? Because it is the most comprehensive, careful, transparent, holistic measure of prices we have that cover such a long period. The longer period of time between things, the harder it becomes to compare prices. The Bureau of Labor Statistics spends a tremendous amount of resources, hiring thousands of agents to survey prices in person and online across the country. They presently track over 80,000 different goods or services that folks buy. They have thousands of pages of documentation, research, and methodologies. Their work has been verified and replicated by independent researchers. And trust me when I say this: if an economist professor found a serious flaw in an official economic measure – you bet they’d get a great publication out of it and a promotion to go with it. Heck, if anybody has one, send it my way, it’ll help my career 😉
https://www.bls.gov/cpi/methods-overview.htm
No measure is perfect, and no single measure can capture the whole story, but the CPI is pretty dang good for what it is trying to accomplish. Like all scholars, economists are constantly seeking ways to criticize and improve our various measures of inflation. But in the end, most of these criticisms amount to a few changes here or there, but never modifying the entire story.
In short, if we look at all prices, the purchasing power of median family income, we find an astounding growth from $38,410 in 1953 to $92,750 in 2022 (both numbers using cost-of-living adjusted 2022 dollars). There’s been a bit of a decline due to the Covid recession and inflation, but I imagine when the 2023 numbers come out we’ll see significant real growth again, as other measures show us wages grew faster than inflation this past year.
If there is one graph to tell the story of family economic well being over time: this is it.
https://fred.stlouisfed.org/series/MEFAINUSA672N
Controlling for Quality Improvements
One of the main challenges comparing goods over time is quality differences. Sometimes quality deteriorates, other times quality improves. There have been numerous and dramatic quality improvements in homes since 1950.
The 1950 Housing Census has a lot of excellent data on this.
A full 30% of homes did not have a private bathtub or shower. 29% didn’t have a flushing toilet – they used an outhouse. [See Table P]. 22.2% of homes did not even have running water or had dilapidated plumbing. [Table 7]
49.6% of homes did not even have central heating [Table Q]. 19.8% did not have a refrigerator. [Table R]
More data is found in the New Housing and Its Materials 1940-56 BLS Report. Now note, this only reflects the newly built homes that year and not a census of the current stock of homes like the previous source.
We find that only 4% of single-family homes built that year had more than 1.5 bathrooms. [Table 1]. Today that figure is reversed. Only 3% of single-family homes today have 1.5 bathrooms or less.
https://www.census.gov/construction/chars/current.html
And again, only looking at newly constructed single-family homes from that year, the average square footage that year was 983. Less than half of the average sized home is built today.
Also note that there were an average of 3.5 persons per occupied dwelling unit.
[See Table J]
This compares to around 2.5 persons per household today.
https://www.census.gov/quickfacts/fact/table/US/VET605222
So, the improvements in housing living conditions are extraordinary. If we want to calculate standard of living across time, we need to control for these quality improvements. For example, if a new house the next year has one more bathroom compared to the year before, but the nominal price stays the same, how should CPI reflect that? It should fall – that would be deflation. Likewise, if quality deteriorates (think shrinkflation), then CPI should rise even though the nominal price remained the same. The BLS takes all of this into account.
Homeownership Rates
But also, when you’re working with data and find a result that is shocking, one should take a moment and double check all of the sources. Are you accurately measuring what you think you are? For example, if we truly needed $175k today to live the same lifestyle as the median family from 1950 then why is the homeownership rate so much higher than it was back then?
The homeownership rate was 55% in 1950.
https://www2.census.gov/programs-surveys/decennial/tables/time-series/coh-owner/owner-tab.txt
Today it is 66%
https://fred.stlouisfed.org/series/RSAHORUSQ156S
This jump was largely due to the G.I. and other pro-housing reforms. It is also due to a tremendous housing construction boom that was starting in the late 40’s/50’s.
A similar story can be said for cars. If cars were more affordable back in 1950, why were there so fewer of them compared to today? We had 323.71 vehicles per thousand people in 1950.
In 2022 we had 850.32 vehicles per thousand people. That’s more than double the amount of vehicles per person!
https://www.fhwa.dot.gov/policyinformation/statistics/2022/mv1.cfm
[note, we take the total vehicles and divide it by the population measured in thousands: 283,400,986/333,287.557=850.3197]
And the quality improvement has also been nothing short of miraculous. Higher miles per gallon, far safer, extremely more reliable, and way more features from air conditioning, power windows, anti-lock brakes, automatic transmission, and many more. And did I mention how much more reliable they are now?
But, of course, the more thorough way to compare affordability of a vehicle is comparing the prices of a vehicle with the nominal median wages of families. And, as we can see for both used and newer vehicles, vehicles are significantly more affordable today than they were in 1950 relative to wages.
https://fred.stlouisfed.org/graph/?g=1ejC7
Income Definitions
The video cites from 1950 is the “average family income” to get the $3,300. Specifically, this is a median, and the precise number is $3,319 from Table 1.
https://www.census.gov/library/publications/1952/demo/p60-009.html
As previously mentioned, in 2022 (the latest date we have data for median family income), the number is $92,750.
https://fred.stlouisfed.org/series/MEFAINUSA646N
Your modern number you cite in your video is for median household income, not median family income. Because many households have just one individual, the median household income is usually lower.
As you can see looking back at Table 1 from the 1950 report, the median incomes of “unrelated” individuals are lower than “family” incomes. This is another reason the household incomes are lower than the family incomes.
Housing Price Definitions
Now to housing. Unfortunately, housing price data going back that far takes a lot of care to get accurate. The 1950 number you cite is for “median home value.” This is looking at the estimated value of every single-family home structure. Which will be far more houses than just the brand new constructions sold that year. The distinction is crucial. Brand new homes will always be worth more than pre-existing homes. Especially in 1950 when modern conveniences such as plumbing, laundry machines, and central heating are all growing.
https://www2.census.gov/programs-surveys/decennial/tables/time-series/coh-values/values-unadj.txt
The definition of the measure is found here:
https://www.census.gov/quickfacts/fact/note/US/HSG495222
Now, the difficulty is getting the equivalent measure today. Unfortunately, the Census survey that collects this information only reports it on a 5-year average. So our most recent comparable measure is between 2022-2018, inclusive. That figure is $281,900. This is the best apples-to-apples measurement. The only problem is – it’s a little outdated as home prices grew substantially in the past 3 years.
The figure you cited for modern times is Median sale price for a new house – that is for brand new construction. Most homes bought and sold every year are used homes. One can very much be middle class and purchase a home built 20 years ago. So, yes, $434,700 in Nov 2023 makes a bit more sense in that context. Also note that the average square foot today is 2,383.
https://fred.stlouisfed.org/series/MSPNHSUS
So, maybe we could get some apples-to-apples comparison from 1950. What did it cost to purchase a brand new construction home then? Unfortunately, we don’t have the data. The closest we have (that I could find) is from 1954. See Table 8, below.
We see that in 1954, for a 1,140 square foot new construction home, the median selling price was $12,300.
In that same year, the median family income was $4,167. Here is a table summarizing this:
| median family income | Median selling price for new single family | years worth of income | average square footage | |
| 1954 | $4,167 | $12,300 | 3.0 | 1,140 |
| 2022 | $92,750 | $434,000 | 4.7 | 2,383 |
So, we have more than doubled the square footage of new construction homes, significantly increased the quality (better windows, insulation, appliances, # of bathrooms, etc.) and yet have less than doubled the amount of annual family income to acquire it.
Now, I hope you see where I’m coming from. If you wanted to purchase an 1140 square foot home today with old used and relatively unreliable appliances and lack of modern insulation or other remodeling, you’ll get one for far cheaper than $434,000.
I don’t have a nationally representative sample, but a quick search for a smaller home in a medium priced city of Las Vegas finds many new and upgraded homes in the $200-300k range. All of these will be a higher quality than the new constructions from 1954.
(It’s a big URL, so here is the link)
This price range puts you squarely in the 2-3 year of annual salary that we saw in 1954.
Conclusion
Now that we’re at the end of this, I want to emphasize that we are definitely not as good as we ought to be. While the overall home ownership rate has basically been the same for decades (more or less), millennials are somewhat behind what boomers were at a similar age. (although note that the majority of millenials now own their home). Source
And, the average age of first time home buyers is now 35 instead of 29, as it was 40 years ago.
https://www.chartr.co/stories/2023-12-06-3-home-buyers-have-been-getting-older
But this is more of a housing issue, rather than an overall inflation or income issue.
While housing is the largest component of our budget (it accounts for around 30% on average), there are many other things that have gotten cheaper since the 1950’s. Groceries are far cheaper, vehicles are cheaper relative to income, clothing and furniture, appliances, recreational goods and services, and of course communication and information goods and services are dramatically cheaper. Once we add all of this up and take into account wages, that’s where we get the graph I showed at the beginning of this long email. The purchasing power for the median worker, the median family, and many other measures is multiple times larger than what it was 70 years ago.
Inequality has risen. The middle class, while not generally worse, certainly hasn’t grown at the same level as they could have had the gains in society been spread at similar rates as they were in generations past. That’s where we need to focus our energies and rhetoric if we want to solve this problem. 1. Build more housing for everyone to make it more affordable. 2. Make sure everyone (especially capital owners) pay their fair share in taxation. 3. Invest in children by eliminating child poverty and increasing the quality of public education universally across neighborhoods.
Accurate historical data comparison isn’t an easy quick google kind of a thing. It takes a lot of attention to details. I hope this helps.
Econ Chris Clarke 