The answer to that question is, of course, “no.” No one number can alone tell us the whole story, whether we are talking about the economy, health, education, population, or any other social statistic. But when you look at other measures of the health of the labor market, you usually find that they tell a similar story to the unemployment rate.
My goal in this post is to dive a little deeper into the data on the labor market, but really the goal is broader: to give you a little insight about how to interpret data. Some rules of thumb, perhaps. But really there is One Big Rule: numbers need context. A number on its own doesn’t tell us much of anything. How does it compare to the past? How does it compare to other places?
With the unemployment rate at historic lows for both the US and many states, I’ve started to see many people saying that, not only doesn’t the unemployment rate give us the full story, but many other indicators point in the opposite direction. Is this true? Let’s dig into the data. Here’s one example of someone saying this for Arkansas. I’ll focus on Arkansas, since that’s where I live and I pay attention to the economic data here pretty closely, but I’ll also refer to national data where appropriate.
You may have heard that there is a new viral song which deals with a few economic issues. Noah Smith has a good analysis of “Rich Men North of Richmond,” which he mostly finds to be incorrect in its analysis (for example, of welfare policy). But Smith does say that the song has a point: manufacturing wages haven’t performed well in recent years. Not only has pay for factory workers “[lagged] the national average in recent years,” for those workers in Virginia, it’s lower in real terms than in 2010.
Well that all doesn’t sound good! Smith is only going back to about 2000 with the data he shows. What if we took a longer run perspective? What if we took a really long-run perspecitive?
Here’s wages for blue-collar factor workers that goes back to 1939 in the US:
The wage data (for manufacturing production workers) is from BLS and the PCE price index is from the BEA. What do you notice as you look at the data?
First, it is true that the last 20 years or so hasn’t been great. Only about 8% cumulative growth since 2002. That’s not great!
But as you look back further, you’ll notice that gains are substantial. Compared to what some might consider the “golden age” of manufacturing wages, the early 1950s, real wages have roughly doubled. It’s true, the growth rate from 1939-1973 is much, much better than the following 50 years. Wouldn’t it be nice if that growth rate had continued! But no doubt you’ve seen many memes saying something like “in the 1950s you could support a family on one high-school graduate income, but not today!” This data suggests that view of the 1950s is a little distorted by nostalgia.
One final thing to note: we might think that one big change in recent decades is that a lot more compensation goes to benefits, rather than wages. There’s actually a total compensation series for blue-collar workers going all the way back to 1790:
The total compensation data, as well as the CPI data that I used to inflation-adjust the figures (to 2022 dollars), comes from the fantastic resource Measuring Worth. This is a total compensation measurement, so it includes benefits, but the source data tells us that up until the late 1930s, it’s really just a wage measure. So potentially we could splice this together with the above chart, to get a “wage only” series covering the entire history of the US.
However, when we look at total compensation, we still see the post-1970s stagnation. Real compensation is roughly the same as about 1977. Yikes! Note here that we’re using the CPI, since the PCE index only goes back to 1929, and the CPI tends to overstate inflation (yes, that’s right, sorry CPI truthers). Still, it’s not the most optimistic picture.
Or isn’t it? With all of the automation and global competition in manufacturing coming on board in the past 50 years, perhaps our baseline is that things could have been much worse. In any case, if we look at total compensation, it’s currently about double what it was in the post-WW2 era. That’s even with the dip in 2022 due to high CPI inflation.
Wages and compensation of blue-collar productions workers have indeed been growing slowly for the past few decades. That much is true. On the other hand, they are still among the highest they have ever been in history, over 50 times (not 50%, 50 times!) higher than at the birth of this nation. This ranks them as probably the highest wages anywhere in world history for an occupation that doesn’t require an advanced degree. That history is worth knowing.
I was reading “The Ultimate Guide to Barbie” the other day, and I noticed an interesting piece of data towards the end of the magazine: the original Barbie doll in 1959 retailed for $3. Today, according to the magazine, a Barbie costs around $14-19. And they further told us that adjusted for inflation, that $3 original Barbie is about $24 today.
I’m not sure exactly where they got that number. Using the BLS CPI tool, it’s more like $31.50. And while I appreciate the attempt to give us historical context, I think for the typical reader will still be a bit perplexed. What does it mean to say $3 in 1959 is equal to $24 (or $31.50) today? Well, it means that the price of Barbie dolls has risen more slowly than other goods and services (quality adjusted). But I think we can do better on the context.
Here’s my best attempt to give context:
The chart shows the number of minutes of work that the median woman would need to work to purchase a Barbie doll for her daughter. In 1959, it took almost 2 hours of work. Today, it takes only about a half hour (I’m using the lower range from the magazine, $14 for a Barbie today, although there are plenty of $10-11 Barbies on Amazon).
Another way of thinking about it: with the same amount of work, a working mother today could buy her daughter 3-4 times as many Barbies as her counterpart in 1959.
I deliberately used median female wages here to make another historical comparison. Women’s earnings have increased much more than men’s since 1959. Back then, median female earnings for full-time, year round workers was only 61% of male earnings. Today, it is close to 85%. True, that’s still not parity. And for those that know the history, you will also know that the closing of that gap has stagnated in recent years. But this is still some major progress during the Barbie Era.
Finally, as I have emphasized before, looking too much at the cost of one product over time has limits. What about other goods and services? A toy, even a well-known brand like Barbie, is a tradable good that can be manufactured anywhere in the world (it looks like Indonesia is where many Barbies are made today). So it wouldn’t be surprising that it has got cheaper over time. But what about all goods and services?
Here’s where inflation adjustments are most useful. Not for individual goods and services, but for looking at incomes over time. How much stuff can a given income purchase compared to the past? That’s what inflation adjustments are for. And this chart shows male and female median earnings in 1959 and 2023, with the 1959 figures adjusted to 2023 dollars using the PCE price index.
When we adjust for changes in all prices, not just Barbies, we can see that median female earnings have roughly doubled between 1959 and 2023. That’s not quite as robust as the “Barbie standard of living,” which allows you to purchase 3-4 times as many dolls. But 2 times as much stuff is pretty good. It’s especially good when compared with male earnings growth, which grew about 44 percent.
It should be obvious here that these are just the raw medians, not controlling for anything like education, experience, or occupational choice. Controlling for those will shrink the gap a bit more. But the gains for women in the labor market since the introduction of Barbie are large and worth celebrating.
About one year ago, I wrote a post with the title “Are We in A Recession?” At the time there was much talk, both in the popular media and among economists, about whether we were in a recession or not, and what “technically” counts as a recession. Now with hindsight, I think we can pretty clearly say that we were not in a recession last summer, nor at any point in 2022.
One thing is true: GDP did decline for two quarters in the first half of 2022. In fact, even the more nuanced “real average of GDP and GDI” declined for two quarters. But as I explained in that July 2022 post, that’s not how the NBER defines a recession. It often coincides with their defined recession, but they used a separate set of indicators. And while some economics textbooks do use the two quarters of declining GDP definition, as I explained in a follow-up post, that’s not the most common textbook definition.
The first half of 2022 is a good candidate for a possible recession, but when we look at the NBER’s preferred 6 measures of economic activity, it seems pretty clear that this was not a recession. If you start the data in the last few months of 2021, you do have small declines in two measures through July 2022 (real personal income and real manufacturing sales), but this looks nothing like past recessions, which have large declines in all or most of the 6 measures.
OK, but that was then, this is now. Are we in a recession now or headed into one? You can find lots of models and surveys or different groups of economists out there. I’m not sure that any particular one is the best, so I won’t dive into those. But if we look at the average of GDP and GDI again, we do notice that 2022q4 was negative and 2023q1 was very weak. Maybe that was a recession?
Again, we can start the NBER indicators around that time to see. Starting from September 2022, we can indeed see that there is some weakness in a lot of the measures for the next 2-3 months. But when we look out 6 months or so from then, we once again only have 2 of the 6 indicators that are below the September 2022 level, and the declines are mild (less than 1 percent). You can play around with the start date a bit, but I think September is the best candidate for a peak, and it’s still pretty weak.
OK, OK, you say, but that’s still all the past. What about the future? Sorry dear reader, I don’t have a crystal ball or the economic equivalent (a model). All I can say is what the data shows right now (which is always backward looking), and as of right now most broad measures of the economy aren’t declining. Yet!
This doesn’t mean everything is great in the economy. Inflation is bad. Poverty is bad. Inequality is, often, bad. We always have these things. But are they getting better? Or are they getting worse? A recession is a particularly bad thing, and something that is often hard to precisely define and measure (for good reason: the economy is complex and hard to measure!). All indication of the available data is that, whatever other bad things are happening right now, a recession is probably not one of those things.
Last week I wrote an optimistic take on inflation. The rate of general price inflation has fallen a lot in recent months, and wage growth is now clearly outpacing inflation. That’s all good news.
Today, the Fed will announce their latest interest rate decision. Will the good news on inflation lead the Fed to stop raising interest rates? I’m not very good at making predictions, but today I’ll give a pessimistic take on inflation which suggests the Fed (and everyone else) should still be concerned about inflation.
The pessimistic take can be summarized in two charts. First, this chart shows the year-over-year change in the core PCE inflation index. As most readers will know, core indexes take out food and energy prices. This is not a “cheat” to mask important goods, it’s done because these are particularly volatile categories of goods. If we want to see the true underlying trend in inflation, we should ignore price fluctuations that are driven largely by weather and geopolitics.
While there is some moderation in inflation in this chart, we don’t see anything like the dramatic decline in the CPI-U, which fell from about 9 to 3 percent over roughly the past year. True, there is some decline over the past year, but only about 1 percentage point, and it has been stuck at just over 4.6 percent for the past 6 months. This is not a return to normalcy, as this rate historically has stayed in the band of 1-2 percent.
The second pessimistic chart is M2, a broad measure of the money supply.
The dramatic increase in M2 during 2020 is clear. That’s a big source of the inflation issues we’ve had over the past 2 years. There is some cause for optimism in this chart: M2 has clearly shrunk from the peak in Spring 2022. In fact, using a year-over-year percentage change, M2 has been negative since last November.
But if we look very recently, there is less cause for optimism. Since late April, M2 has stopped falling. In fact, it’s up a little bit. Is this a sign that the Fed doesn’t really have inflation under control? Perhaps. The increase isn’t huge, and there’s always some seasonality and noise to this data so we shouldn’t overanalyze this small deviation from the general decline in the past year plus. But we’ll need to continue watching this data.
The latest CPI-U price data shows that the rate of inflation in the US has slowed significantly to just 3% in the past 12 months. That’s a huge improvement from the peak last June, when the annual rate of inflation was over 9%. Still, prices as a whole aren’t falling, and they clearly aren’t anywhere near where they were before the pandemic: using the CPI-U, prices are up over 17% since January 2020.
Lately I’ve heard many people asking a good question: will prices ever get back down to where they were? Usually they mean pre-pandemic prices, though sometimes they refer to a particular point-in-time (such as the start of Biden’s presidency). The only correct answer is “we don’t know,” but I think a likely answer for many goods and services is “no.” For many reasons, the nominal prices of most goods and services rise over time. Though this is not true for everything, of course (newer technologies are one example we often see).
But what about specific goods that we buy frequently? Will we ever see gasoline consistently below $3 per gallon again? Will we ever see milk consistently below $4 per gallon again? What about eggs and bread? And indeed, these prices are well above January 2020 levels: 23% higher for milk, 43% for bread, 45% for gasoline, and a whopping 52% for eggs. For the price data, I am using this convenient data on common food and energy goods from BLS.
For some of these items, I do think you might someday see prices fall back to levels consumers were used to from the recent past, since food and energy prices tend to be volatile. For others, though maybe not. But I think we as consumers can become overly focused on staples that we buy frequently and can easily recall the price in our heads. For example, while eggs, bread, and milk are items that we buy frequently (including being the staples of stocking up before a storm), in total these constitute just 0.6% of average consumer spending.
If instead of those 3 staples, your mind naturally anchors on produce prices, the trends look different: oranges are up 23%, but bananas are only up 10%, and tomatoes are, in fact, down 14% since January 2020. But again, these items are less than 0.5% of total consumer spending. Ideally, we shouldn’t anchor on any one subset of goods when doing a good analysis, even if it is natural for us to do so in our lives as consumers.
This is where the benefit of a price index, like the CPI-U, comes in.
As we prepare for the release of second quarter GDP data over the next few weeks, here is a chart showing cumulative GDP growth (inflation adjusted) and Price inflation for G7 countries. While inflation has been high everywhere (except for Japan), the US comes out looking very well relatively on GDP growth. That’s especially true compared to the UK and Germany, which have also had high price inflation, but have actually had negative economic growth since the end of 2019.
The United States has problems and always had. But the historical record of the United States as an economic powerhouse is unrivaled. The US had a bit of a head start on economic growth, being a direct descendant of the country that really kicked of the Industrial Revolution. But we took that head start and really ran with it, now being by far the highest income large country, and the highest income country that does not derive a significant part of its GDP from fossil fuels or being a tax haven.
The average American has, as best as we are able to measure it, a standard of living that is at least 20 times greater than Americans when this country began.
Last week my post was about a new article I have with Scott Winship on the “cost of thriving” today versus 1985. That paper has gotten quite a bit of coverage, including in the Wall Street Journal, which is great but also means you are going to get some pushback. Much of it comes in the form of “it just doesn’t feel like the numbers are right” (see Alex Tabarrok on this point), and that was the conclusion to the WSJ piece too.
Here’s a response of that nature from Mish Talk: “There’s no way a single person is better off today, especially a single parent with two kids based on child tax credits that will not come close to meeting daycare needs.”
He mentions daycare costs, but never comes back to it in the post (it’s mostly about housing costs). Daycare costs are undoubtedly an important cost for families with young children (though since Cass’ COTI is about married couples with one earner, they may not be as relevant). And in the CPI-U, daycare and preschool costs only getting a weight of 0.5%. Surely that’s not reality for the families that actually do pay daycare costs! If only there was an index that applied to the costs of raising children.
In fact, there already is. Since 1960, the USDA has been keeping track of the cost of raising a child. Daycare costs are definitely given much more weight: 16% of the expenditures on children got to child care and education. And much of that USDA index (recently updated by Brookings) looks similar to what COTI includes: housing, food, transportation, health care, education, but also clothing and daycare. I wrote about it in a post last year and compared that cost to various measures of income (including single-earner families and median weekly earnings). But what if we compared it to Oren Cass’ preferred measure of income, males 25 and older working full-time? Here’s the chart.
62 weeks. That’s how long the median male worker would need to work in a year to support a family in 2022, according to the calculations of Oren Cass for the American Compass Cost-of-Thriving Index released this year. Not only is 62 weeks longer than the baseline year of 1985 (when it took about 40 weeks, according to COTI), but there is a big problem: there aren’t 62 weeks in year. It is, by this calculation, impossible for a single male earner to support a family.
Is this true? In our new AEI paper, Scott Winship and I strongly disagree. First, we challenge the 62-week figure. With a few reasonable corrections to Cass’ COTI, we show that it is indeed possible for a median male earner to support a family. It takes 42 weeks, not 62 as reported in COTI.
But wait, there’s more. Much more. In our paper, we provide a range of reasonable estimates for how the cost of thriving has changed since 1985. In the COTI calculation, the standard of living for a single-earner family has fallen by 36 percent since 1985. In our most optimistic estimate, the standard of living has risen by 53 percent. The chart below summarizes our various alternative versions of COTI. How do we get such radically different results? Is this all a numbers game?
Economist Writing Every Day 