Inflation Has Wiped Out Average Wage Gains During the Pandemic (maybe)

The latest CPI inflation report didn’t have a huge surprise in the headline number, with 8.3% being very similar to last month. But with the two most recent months of data, we can now see something very unfortunate in the data: cumulative inflation during the pandemic as measured by the CPI-U (11.6%) has now almost matched average wage growth (12.0%), as measured by the average wage for all private workers. I start in January 2020 for the pre-pandemic baseline.

What this means is that inflation-adjusted wages in the US are no greater than they were before the pandemic. They are almost identical to what they were in February 2020 (just 2 cents greater). But as regular readers will know, the CPI-U isn’t the only measure of inflation, and there’s good reason to believe it’s not the best. One alternative is the Personal Consumption Expenditures price index. Cumulative inflation for the PCE is slightly lower during the pandemic (9.0%, though we don’t have April 2022 data yet).

This chart shows average wage growth adjusted with both of these different measures of inflation, expressed as a percent of January 2020 wages. The CPI-U adjusted wages (blue line) have been falling steadily since the beginning of 2021, though the declines have accelerated in 2022. The PCE-adjusted wages (orange line) have also not performed superbly, but at least they are still 2-3% above January 2020. Still, the picture is not rosy: they’ve basically been flat since mid-2020 and have started to drop in early 2022.

Of course, average (mean) numbers can be tricky and sometimes misleading. What if instead we used median wages? Unfortunately, there is no hourly median wage data that is updated every month. The closest data that I usually look at is median weekly earnings, which is available on a quarterly basis. Here’s what that data looks like, expressed as a percent of the first quarter of 2020. I limit the data to full-time workers, since that should give us a roughly comparable number to the hourly data (hours of work may have changed, but using full-time workers should make it roughly constant).

For median weekly earnings, we can see that the picture is even less rosy. Median earnings have been declining consistently since the second quarter of 2020, regardless of which inflation adjustment we use. The decline in the PCE-adjusted measure isn’t quite as steep since early 2021, but both figures are below the pre-pandemic level, and have been for the past two quarters.

One final note: if we look at weekly earnings across the distribution, and not just at the median, we see something very interesting. Earnings at the bottom of the distribution seem to be performing better than those at the top. In fact, the 10th percentile weekly wage is the only category that is still above pre-pandemic levels. I’m only adjusting using the CPI-U here, but the patterns for the PCE-adjusted earnings would be roughly similar.

We should be cautious about interpreting this data too: if workers dropping out of the labor force are primarily at the bottom of the distribution, it will artificially push up the 10th percentile earnings level. It would be good to know how much of that is going on here. Still, I think this is an important result in the current data.

Gas Prices are High — But Don’t Adjust Them for Inflation!

Gasoline prices are high and rising. Anecdotally, they seem to be increasing at the pump by the hour. And indeed, in nominal terms they are now the highest they have ever been in the US (this is true with both the AAA daily price level and the EIA weekly price level). At over $4.10 per gallon, the price now exceeds the peaks briefly hit in 2008, 2011, and 2012. And it’s looking like this peak might not be so brief.

But we all know you can’t compare nominal dollars over long periods of time. We need some context for this price! Plenty of news stories provide what they think is the right context: adjust it for inflation! For example, USA Today reports that today’s price “would come to around $5.25 today when adjusted for inflation.”

$5.25: that’s a pretty concrete number. But it’s not really useful. OK, so clearly that’s higher than the current price, about 20% higher in fact. Still, it doesn’t really give us the right context.

As I argued in a previous post on housing costs, inflation adjustments aren’t always the best way to contextualize a historical number. Yes, when you want to compare income or wages over time, it’s good to adjust for inflation. It’s necessary, in fact. And a good economist will always do that.

However, when comparing particular prices over time, it doesn’t really make sense to adjust for other prices. All you are really saying is “if the price of gasoline increased at the same rate as the average price level, here’s what it would be.” Perhaps slightly useful, but it doesn’t really get at the thing we’re really try to address: is gasoline more or less affordable than in the past?

The best approach is to adjust the prices for changes in wages or income. Which measure of wages or income you choose is important, but it’s the best adjustment to make. No need to make any inflation adjustments, are worrying about whether the index you choose is properly accounting for quality changes, substitution effects, etc. If you want to know how affordable something is, compare it to income.

Here’s what I think is the best simple comparison for gasoline, which I’ll explain it below. In short, it tells us how many minutes the average worker would need to work to purchase one gallon of gasoline.

Since the price of gasoline is rising sharply every day lately, my chart will surely be out of date very soon. But right now, it’s the most current data I could provide with a comparable historical series: EIA weekly data current through March 7th, 2022 (Monday). We can see that at current prices, it takes about 9 minutes of work at the average wage to purchase a gallon of gasoline. At the peak in 2008, it took over 13 minutes of work to purchase a gallon, and it fluctuated between 10 and 12 minutes of work for much of 2011-2014.

Continue reading

Home(r) Economics

Is it harder to buy a home today than in the past? Many seem to think so. Lately, some people have used the example of the fictional Simpsons family to make this claim. A recent Tweet with around 100,000 likes expressed the sentiment:

The unspoken implication is that today a “single salary from a husband who didn’t go to college” couldn’t buy a typical home in the US. Or at least, it would stretch your budget so thin that you would have to give up something else or need two incomes to support that lifestyle (famously dubbed “the two-income trap” by Elizabeth Warren).

And it’s not just a Tweet that caught fire. A December 2020 article in the Atlantic claimed “The Life in The Simpsons Is No Longer Attainable” and used housing as a prime example. And while a 2016 Vox article on Homer’s many jobs doesn’t mention the cost of housing, they draw a similar conclusion and implication: “Homer Simpson has gone nowhere in the past 27 years — and the same could be said of actual middle-class Americans.”

But is this an accurate picture of the Simpsons family over time? And does that picture accurately represent a typical family in the US? Let’s investigate. And let’s start by pointing out that as measured by the availability of consumption goods, the Simpsons do see rising prosperity over time. They have flat screen TVs now, instead of consoles with rabbit ears, as the late Steve Horwitz and Stewart Dompe point out in their contribution to the edited volume Homer Economicus. But with all due respect to my friends Steve and Stewart, I don’t think many would deny that TVs, cell phones, and computers are cheaper today than in the 1990s. The familiar refrain is “but what about housing, education, and health care?”

In this post I want to take on the question of housing, partially by using the Simpsons as an example. My main result is this chart, which I will present first and then explain.

Continue reading

This Is Not the Most Expensive Thanksgiving Ever

“Thanksgiving 2021 could be the most expensive meal in the history of the holiday.”

That’s the first sentence of a recent New York Times story. The Times and the New York Post rarely agree on editorial matters, but on this topic the Post ran a very similar story the same week. You can find many such headlines.

But is it true? In short: no. I’ll explain why, but my larger goal is to get you to think more clearly about inflation.

How should we measure the cost of a Thanksgiving meal? A widely used measure comes from the Farm Bureau, which shows that the cost of a traditional turkey-centric meal costs about 14% more than last year. In dollar terms it is $53.31 for a turkey, a pumpkin, cranberries, sweet potatoes, stuffing, etc. That’s more that it has ever been, in dollar terms. Farm Bureau has been tracking the cost of this same meal since 1986.

So in one sense, it seems like the headline claim is true. Most expensive Thanksgiving ever!

But we need to think deeper. A nominal price doesn’t actually tell us much. If a long-lost cousin from the Republic of Horpedahl told you it costs 1 million Jeremys to buy a Thanksgiving dinner, what would your reaction be? The first and best reaction is: how much do people earn in the Republic of Horpedahl?

We should ask the same question in the United States today: how do incomes today compare to incomes in the past? Which measure of income you use is important, but if we use median usual weekly earnings of full-time workers, we can make a simple comparison of how much of your weekly earnings would be needed to buy a traditional Thanksgiving meal. This chart shows exactly that. In 2021 that meal will be the second lowest it has ever been as a percent of median earnings — higher than last year, but tied with 2019 for the second lowest. And much less than in the late 1980s and early 1990s (I use third quarter data for each year, the most recent available).

Adjusting for income is the best way to look at this question. It’s not perfect — part of this depends on what income measure you use — but it’s much better than the alternative. The worst approach is to just look at nominal prices. This tells you virtually nothing.

Continue reading