When is the economy good or bad?

The great "vibecession debate" continues. Is the economy good or bad?

Y'all know I think it's bad, bad now under Trump, bad even when people with my politics touted its virtues as Bidenomics.

Whatever.

My purpose here isn't to take sides, but to offer a bit of a historical perspective on how we perceive the economy. Let's take a look at a chart:

(Click to enlarge.)

This chart is of the census bureau's real median family income series.

I didn't choose the series because I think it is a uniquely valuable indicator, but because the "data guys" on the Panglossian side of the debate like to point to real median household income. Unfortunately, the household series only goes back to 1984, and I wanted to go back farther. The family series goes back to 1953.

The two are qualitatively very similar. The main difference is that single-member households are excluded from the family series, pushing the median income upward. (Income from unrelated individuals who cohabitate with a family gets excluded from the measure, pushing incomes down, but this effect is small.)

The first thing to point out, as I've tried to emphasize before, is there's very little information in the level of these series.

In 1978, real median income was apparently higher in absolute terms than it ever had been! It was true then, as it is now, that the typical American was living a lifestyle that the greatest kings and emperors of prior centuries could never afford. GDP per capita was growing faster than 5%, a rate we've only rarely achieved since. Yet very few people, then or now, would argue that the 1978 economy was great.

You can see trouble in the data, just other data. Arthur Okun, an influential economist of the era, invented the "misery index", and that was pretty high in 1978, although it had come down dramatically from prior years. But much more than then-current conditions, commentators at the time felt — with foresight ultimately vindicated — that the relatively benign moment was a lull in a continuing storm. Despite good data, it was a bad time.

You'll notice that I've divided my chart into eras, and the era to which 1978 belongs is shaded not red for "bad", but yellow for "meh". That is because I am data-driven. Most commentators would say that the period I've marked the "great inflation" was the worst economy of the postwar US except for the great financial crisis. But by the numbers, by these numbers, the period is only a slow-growth stagnation, comparable to the aughts.

The key point is that the goodness or badness of the economy is never an objective fact about the world. It is a thing we construct, in a project often contentious in real time, but that usually hardens into consensus ex post. During soft patches, we debate whether we are in or near a recession. After a while some committee at the National Bureau of Economic Research gives a verdict that becomes authoritative, if not necessarily true.

There are an infinity of potential data sources, many of which will tell conflicting stories. There is no consistent and universal set of weightings we can apply in order to evaluate the economy. Data sources exist today that could not have existed even a few years ago, like AI-generated evaluations of media sentiment. The stories we tell, the evaluations we give, are never science. They are pastiche. The quality of the economy exists in the netherworld between subjective and objective, in the purgatory of the judgment call, where we humans also reside.

Although there is almost no information in the level of this median real family income series, there is a ton of information in the changes. Despite a misclassification of the great-recession era, turning points in this series do a really good job of segmenting the economy into recognizable eras and marking them as "good" or "bad". That's how I've parsed out the eras on the chart.

I've graphed the series on a log-scale, so that slopes correspond to growth rates. Part of the apparent virtue of the graph results from a kind of look-ahead bias. I define a turning point at 1969 rather than a similar break in 1971 or the more defensible turning point at 1973 because I know that inflation begins to draw up its skirts in the late 1960s. I arbitrarily chose four years as my minimum length of an era, because I didn't want to segment out every leg of every business cycle. This forced my Volcker Shock to extend a year longer than a mechanistic segmenting of turning points would have yielded. Still, the graph is pretty close to a mechanistic segmentation, and marks out eras commonly understood and discussed surprisingly well.¹

A few points are worth discussing. Just on the basis of this little exercise, it's unsurprising that we have nostalgia for the period from 1954 to 1969. In the history of the economy as represented by this one, imperfect, measure, there is no other period with the length and slope — the persistence and growth — of this period. "Make America Great Again" gets some of its rhetorical force from the fact that there really was, in living memory, a kind of golden age to yearn backwards toward.

(Much of this era was still under Jim Crow. But it was also when the civil rights movement bore its long-sown fruit. The period was riven by unrest and political assassinations. It was a time of remarkable music. Whether on balance it is reasonable to describe the era as a golden age is a judgment call that belongs to you, dear reader.)

I was very surprised to see that the only period that comes close to the postwar boom, in growth but not duration terms, is the Obama-Trump era, from 2012 to 2019. I will confess to you, dear reader, that my own subjective evaluation of this period would be "meh" to "bad". But by the numbers, by these numbers at least, 2012–2019 was the best economy since the postwar boom. Nevertheless, in 2016, in the middle of this apparently great period, the American people wanted a revolution. When the possibility of a benign revolution lost a Democratic primary, they chose a malignant one. Data is only useful if we remember to discuss what it omits as carefully and extensively as we describe what it seems to say.

Although I understood (and wrote ad nauseum about) the deficiencies of the Biden economy, I was surprised that restoring the Trump economy was compelling as a pitch to the American electorate. This data suggests I should not have been surprised.

Still. Was 2012-2019 a better economy than the 1990s boom? For those of you old enough to remember, I'll leave it to you to decide. We were all so much younger then. Is that the reason why it seems so wrong to me? I don't think so. I don't think a 28-year-old in 2018 had the kind of optimism about the economy that I had in 1998.

Today's "vibecession" debate is about how we will, or should, categorize the rightmost five dots on the chart. Based on this dataset alone, it's not surprising that we are arguing. If I were to draw a box around it, my mechanistic classification scheme would have it yellow, "meh". Sometimes we end up deciding the yellow periods are actually bad (the "great inflation"), sometimes we look back at them as decent if not amazing times (the "iffy aughts"). For now, at least through the lens of this dataset, it really is a "faces or vase" kind of question.

The period those last five data points most resembles is the late 1970s. The shape of the series is a similar, shallow "V", and for similar reasons, an inflation spike had eaten away at family purchasing power.

If we pair the series with Michigan consumer sentiment, we find the current pattern looks pretty similar to 1979 and 2007, peaks in real median family income followed by collapses in sentiment:

However good you think the economy is or recently was, if it's not just a vibecession, or if a vibecession can be self-fulfilling, the current set-up looks a bit ominous.

But let's end this with something cheerful!

A series not typically taken as an economic indicator, but that I like, is the suicide rate. I think the suicide rate is useful because it gets at welfare, which is properly the concern of normative economics, rather than "stuff", which is ultimately meaningless except to the degree that it contributes to welfare. King Midas could buy anything in the world, but he was not wealthy.

Welfare is a normative, rather than a positive or "scientific" concept. It is a thing we define, rather than measure. Nevertheless, it is what we care about, the ultimate object of our investigations, and in that sense welfare is more real than all of the things we can measure.

Once we have imposed a definition, we can try to measure proxies for welfare under our definition. I am not going to impose a specific definition, but I'll assert that "people not wishing to kill themselves" correlates with a wide variety of plausible conceptions of welfare, so the suicide rate might be a proxy, no doubt dirty and imperfect, for diswelfare.

Here's a chart:

(Click to enlarge.)

One thing that's fairly obvious is that the suicide rate does in fact have a pretty strong relationship with things we consider "economic". Both the Panic of 1907 and the Great Depression are very visible on the chart.²

I've included on this chart all the periods we teased out of the real median family income chart above, and bounded them with the same dates. I think that the two charts complement one another. In our earlier discussion, classifying the Great Inflation as a "meh" rather than "bad" period seemed wrong. We see a sharp rise in the suicide rate during the period, which suggests that "meh" should be bad. The suicide rate during the "iffy aughts" is much more restrained, suggesting a more benign period, as I think we mostly remember it.

There's a surprising spike in the suicide rate during the "Morning in America" period. I don't have an explanation for, or a story to tell, about that. I am eager entertain yours!

But for the period following, I think layering the suicide data on top of the purchasing power data helps us reconcile the numbers with our intuition. On the real median family income chart, the George-HW-Bush-era recession is an event comparable to the Volcker Shock and Great Financial Crisis. Those of use who lived through these events know that's wrong, the early 1990s recession wasn't fun, but it was a milder downturn than what the nation had gone through a decade earlier, and would go through nearly two decades later.

The suicide rate tells this tale: It rises during the Volcker Shock and rises sharply in the wake of the Great Financial Crisis. It is declining during the GHWB recession.

The suicide rate data also helps us reconcile discrepancies surrounding the late 1990s boom and the Obama-Trump expansion. My dark view of the Obama-Trump period may be idiosyncratic, but I think nearly all observers would agree that the late 1990s expansion was a bigger boom, a better economy, than the late 2010s. We see the suicide rate decline quite sharply during the late 1990s boom. It rises through most of the Obama-Trump expansion.

In general, there has been a strong upward trend in the suicide rate since the Great Financial Crisis, corresponding I think with the sense of upheaval and malaise also visible in social instability and the rise of outsider political movements. There's a kind of dark matter. Something is amiss, something is wrong that more conventional economic measures aren't capturing.

Perhaps the explanation is something not conventionally economic, negativity bias in social media for example. After all, social media and smartphones become prominent at approximately the same time as the Great Financial Crisis, so maybe they are the cause of the deadly gloom.

That is not my view. Long-time readers will know that I think our secular decline in welfare is related to the bifurcating class structure of American society and the unfairness implicit in how it must be sustained. In my view, the Great Financial Crisis both exacerbated these class divides and unmasked them, made them impossible to paper-over as we had managed previously.

I could be wrong! Maybe it really is just down to social media.

All I'll say is whatever it is, if there's a widely distributed product that makes us want to destroy one another and kill ourselves, that strikes me as very much an economic problem, an object that demands economic regulation. "Tech" is a big part of our economy!

Economics is either a positive science of predicting human behavior (in which case it has nothing to say about policy, except to inform other people who decide what we want), or it is a normative enterprise and its object is human welfare. If you are an economist and you say we should do X or Y or Z because it would increase GDP, but you have no reason to think the increase in GDP will make us better rather than worse off other than loose historical correlations, then you are much worse than useless. If it is plausible to you that big chunks of our economy are products so harmful they amount to half a Great Depression in suicide rates, then "maximizing GDP" cannot be a remotely decent proxy for the human welfare that it is your vocation to improve.

If I hadn't cheated at all, if I'd just placed boundaries at sharp turning points, the main difference would be taking the Postwar Boom as continuing through 1973, and then a shorter Great Inflation. The Volcker Shock would have been one year shorter, Morning in America one year longer.
↩
I marked the Great Depression as beginning in 1929, and ending in 1941, when the US joins the war. The Panic of 1907 begins in 1907, and I take its ending to be 1915, the last year of high unemployment. For more on the relationship between suicide rates and the business cycle, see Luo, Florence, Quispe-Agnoli, Ouyang, and Crosby. The suicide rate chart is a composite of three data sets, stitched together to cover the period, CDC Health, CDC (recent) Suicide Data, and CDC HIST290. Crude rather than age-adjusted measures are shown. Age-adjustment seems not to alter the qualitative picture. I want to thank Matt Darling for first showing me suicide rate data from 1900, further peaking my curiosity, and the late, irreplaceable Kevin Drum, who put together the chart Matt posted.
↩

draft by Steve Randy Waldman
2025-12-08 @ 02:45 PM EST

The qualitative is the foundation of the quantitative

People often use quantitative data to try to debunk other people's qualitative impressions.

But when the data at issue are putative welfare measures (e.g. real income, GDP-per-capita, etc), the qualitative measures are foundational, and the quantitative mere proxies.

GDP-per-capita is widely used as a welfare measure, not because it conceptually maps well to welfare — for all kinds of reasons it does not! — but because from the mid-20th Century to the first years of the 21st, it mapped pretty well to our qualitative intuitions about relative welfare.

The consensus that there is a good correlation between GDP per capita and qualitative welfare has broken down more recently. (Is Mississippi really "richer", in human welfare terms, than Spain?)

We can have arguments about why it has broken down. Inequality, differences in how medical and social insurance are accounted in GDP, the effect of market power, and failure to account for differences in leisure time are all candidates. But fundamentally, GDP-per-capita was only ever a good measure because there was a widespread consensus that it tracked qualitative outcomes. Once that consensus has broken down, there is no reason to think it should be a welfare measure. The inventor of GDP, Simon Kuznets, explicitly argued that it should not be! (ht Marketplace for the source)

The same is true of "real" wealth or purchasing power measures! They are not inherently welfare measures. (I belabored this in a recent post.)

If people are making qualitative claims that some group's welfare is poor, and you try to debunk those claims with quantitative data, whether GDP-per-capita or real purchasing power measures, you are engaged in a kind of circular reasoning. The only reason we think these should be welfare measures is because they sometimes seem to work at capturing qualitative intuitions about relative welfare.

If qualitatively they seem to cease to work well as welfare measures, then there is no reason to think they are good welfare measures. When you debunk widespread qualitative claims about welfare with this "data", you are really debunking the quality of your measures!

That's not to say unevidenced claims about qualitative welfare must be taken as gospel, at face value. The claims could still be wrong! Welfare is unobservable, hard to measure. This is economics' foundational demon as a "science".

The moments when there is consensus that any quantitative measure maps to welfare are fleeting and precious. During those exceptional moments, it may come to seem plausible that we might maximize welfare "scientifically".

But that is only hubris. When that consensus flags, like now, we have to cop to the fact that human welfare is not a scientific observable. Welfare, by which we mean "prosperity" or "thriving", is something we experience qualitatively, we construct normatively, we strive to achieve politically. Those who think they have it as fact are only imposing elaborate fictions.

draft by Steve Randy Waldman
2025-11-19 @ 01:40 PM EST

Running on democracy hasn't been tried

I am very grateful for this post, by Matt Yglesias:

I know this is deeply unfashionable but IMO politics largely consists of two things:

— Pandering to the fickle views of the voters to try to win elections

— Handling a series of tedious technical issues that most people don't care about or understand well

Having strongly held views not helpful.

I fear readers will think I'm being sardonic, that I intend this as a back-handed diss at Yglesias. I do not. I have my differences with the gentleman. But one real virtue of an inclination toward the trollish and contrarian is sometimes you just plainly say what you think where others would not.

I don't think Eric Levitz or David Shor (both of whom I regard warmly) would put things quite this way. But I also think they'd have a hard time distinguishing their views, or at least how their views function in practice, from Yglesias'. The worldview that Yglesias' post so succinctly expresses sits at the heart of professional politics in the Democratic Party.

The party contests elections not as some saccharine exercise in representation, but as battles to win. It acts instrumentally, using the best technical and communications tools at its disposal. When it does win, it governs technocratically, understanding public opinion as a constraint to be managed.

Jean Claude Juncker is a European politician, not an American Democrat. But his famous lament — "We all know what to do, but we don’t know how to get re-elected once we have done it." — jibes pretty well with Yglesias' post and the Democratic Party's practice.

Whatever truth there is or isn't in views like Yglesias' and Juncker's, they don't express much enthusiasm for, or even aspiration towards, democracy. Elections are a problem wise technocrats must work around to govern well, rather than the beating heart of our self-government.

Over the past year, it has become conventional wisdom among professional Democrats that you can't "run on democracy". It's been tried. The electorate, frustrated over the cost of living and other "kitchen-table issues", voted for the authoritarian.

I think this analysis is dumb. I think the American public values democracy a very great deal, but doesn't perceive either of the choices on their ballot to offer it.

Democrats offer paternalistic technocracy. They presume a (genuine!) commitment to protecting the status quo institutions of US electoralism amounts to support for democracy that the public ought to reward. But the public has grown jaded about the quality of those institutions.

Republicans capitalize on the public's resentment toward precisely the attitude Yglesias expresses. They effectively ask, what do you prefer, sanctimonious deference to institutions that, while notionally "democratic", fail in practice to connect government to the people, or a bunch of assholes who you might not agree with, but who are candid with you, who could never be mistaken for the technocrats that ignore and manipulate you on the grounds that they know best?

There is a certain beauty, really, in the outcome of the 2024 election and the year that has followed. Republicans got to prove that professional Democrats are a bunch of pinheads much of the public hates and resents. Democrats got to prove that professional Republicans, while definitely not pinheads, are incompetents and crooks who should be nowhere near the levers of power. Everybody wins, everybody loses.

Except democracy. Democracy just loses. People, like me, devoted to the idea of government of the people, by the people, for the people lose. H.L. Mencken coughs out a laugh in his grave, "Democracy is the theory that the common people know what they want, and deserve to get it good and hard."

I am grateful to Yglesias for his candor. I often respect his technocratic insights. But he is wrong, and people with his take on social affairs and "democracy" are maybe not the best guides for an institution charged with representation. Democracy is dying in the West because it has been overtaken by a caste of professionals, politicians and technocrats, for whom democracy is a slogan, a contest, a source of legitimacy in an instrumental sense, but not an idea or ideal they respect or aspire to. An alternative has emerged in the form of barbarians at the gate, who are even more detestable, even more dangerous to the possibility of any improvement toward democratic ideals.

Mencken is wrong. Democracy is not the implausible "theory that the common people know what they want" with respect to policy technicalities. Democracy is the theory that it is possible to organize society in such a way that free and equal citizens participate in and come to constitute rational government on their own behalf. Democracy is not disproven when a public whose civil society consists primarily of tabloids and television, or now twitter and tik-tok, turns out not to be well versed in its own affairs. Democracy does not demand that every member of the public become expert at "tedious technical issues".

Democracy is the project of building institutions under which every citizen's values and interests, on which each of us are sole authorities, are taken into account; institutions in which any of us who wish can participate; institutions that do become expert as institutions at tedious technical issues, and then advocate coherently and capably on constituents' account.

Democracy is also an attitude, almost a religion, of valuing one another as equals, of genuine interest and curiosity into the perspectives of our neighbors, of reverence in every public institution for participation and contribution. We can work side-by-side at the school board to make sure the textbooks arrive and the teachers are hired, and argue spiritedly with one another while we do so. We disagree. Our disagreements together constitute a common enterprise. That common enterprise scales from a neighborhood block party to, yes, government by the people, for the people, of the people, at its highest levels. Our institutions must sometimes be hierarchical, but they should be fluid and permeable.

Neither US political party has a great track record on democracy. Democracy is really my only issue at this point. I spend my time thinking about policy, coming up with clever mechanisms I think might make the world a better place. But the world cannot be a better place without democracy. The Biden Administration was kind of great on domestic economic policy. The Inflation Reduction Act, CHIPS Act, muscular antitrust enforcement, and other interventions constituted what could have been the start of an American renaissance, from my perspective as a technocrat. All of it has been scribbled over by a toddler fingerpainting with shit.

The usual critique is that Biden did a poor job of selling his accomplishments, he wasn't a great communicator, he was too old. That's all true. But it's too narrow. Democracy is not about any one man, no president, no king. The problem is Democrats governed in an institutional context insulated and alienated from any meaningful civil society, a context in which arguably there was no meaningful civil society. Whether you are selling out to lobbyists or erecting the foundations of a brilliant economy, the public won't preserve what it has no means of perceiving, understanding, and evaluating. Our job is to build an institutional context that renders the public capable. Nothing will "work" until we do.

draft by Steve Randy Waldman
2025-11-18 @ 04:40 PM EST

Real purchasing power over time is not economic welfare over time

Will Stancil posted the following over the weekend:

What drives me absolutely insane about this debate is how claims like 'there has been a generational stagnation of purchasing power' have taken on the tenor of religious faith among progressives, and seemingly cannot be disproven with any evidence, no matter how overwhelming. It simply isn’t true! (a FRED graph of real median household income follows)

I want to offer a straightforward point that a lot of economic controversializing should better take into account.

Over long periods of time, you cannot treat changes in measured "real purchasing power" as equivalent to changes in economic welfare.

This is confusing, because over short periods of time, you can and should! Change in real purchasing power is what you want to look at, when evaluating growth on a year-to-year basis.

But there are slippages in the measure, so that as years accumulate into decades, comparisons become effectively meaningless.

Let's talk about some of those slippages.

Purchasing power is distinct from welfare, in a manner time varying
Only the most quantifiable aspects of changes in quality are captured in inflation measures
Inflation measures abstract away the bundling of goods and services
Changes in inequality of quality are overlooked
The composition of price changes has distributional effects

We end with a short conclusion.

1. Purchasing power is distinct from welfare, in a manner time varying

The object of the economy is not dollars and cents, not goods and services. If we built a paperclip maximizer and the paperclip maximizer consumed us all but produced paperclips in quantities such that they would (well would have) sold for ten times current GDP at their final price, that wouldn't actually count as growth.

The object of economics is welfare. We are trying organize production and distribution of goods and services in whatever manner would make human beings "better off".

"Better off" is inherently a normative question. Human welfare per se is not observable. Economics, with its pretensions toward being a "science" has struggled with this for a century.

Nevertheless, you have not said anything of interest if you have said people can buy more "stuff". It has to be the stuff they need, in a manner constitutive of well being.

Economists model human behavior in terms of utility maximization. We claim, at any given time, people face a budget constraint. Within that budget constraint they put together a bundle of goods and services that maximizes their "utility". That utility then translates to "welfare", the unobservable, unscientific, well-being that we actually care about, but cannot quantitatively measure.

Measures of real purchasing power can tell us something about Transition A, from a budget constraint to a bundle of goods and services. They can tell us nothing, unfortunately, about Transition B, from a bundle of goods and services to a person's welfare.

Nothing suggests that the relationship between a bundle of goods and the welfare a person derives from it stays the same over time. Obviously it does not. A parent derives well-being from diapers in her consumption basket when her baby is 1 year old, but would not when her child is 11 years old.

So the relationship between the composition of the consumption bundle and welfare is time varying. But so too can the relationship between the value of the consumption bundle and welfare. A twenty year old without children can thrive on a salary that would leave her immiserated when she has one-year-old twins and needs to fund the diapers and formula and child care.

The value of goods and services in welfare terms depends not only on their amount, but also upon the circumstances and obligations of their purchasers.

As individuals we all understand this, but as technocrats, we often fail to aggregate the insight.

Transportation is a large fraction of what we spend money on, roughly 10% of the CPI-U consumption bundle if we include fuel costs. Consider a typical suburban commuter, who drives thirty minutes and twenty miles to and from work every day. A century ago, his doppelganger might have lived in an urban neighborhood, and walked or taken a streetcar to a workplace much closer to his home.

The quantity of transportation our modern commuter purchases with his transportation budget, in terms of miles traveled, is extraordinary, even miraculous, from his antecedent's perspective. Less than ten percent of his wages now purchase miles and miles of point-to-point, off-street-line transportion, which might easily have consumed his entire salary in 1925!

So, in a certain sense, our modern commuter is a whole lot richer. Economic statisticians accurately measure his real consumption bundle as much, much larger than his predecessor's due to the collapse in cost, increase in speed and comfort, of transportation.

But in another respect, our modern commuter is not richer at all. The world has changed. He no longer has the option of taking a job whose quality and pay would support his typical commuter lifestyle absent accepting an obligation to purchase all of those transportation miles. He can purchase a much larger bundle of goods and services than his predecessor could, but he also has to purchase a much larger bundle of goods and services just to live a life that will be welfare comparable.

One can argue that, by the preferences revealed in America's historical suburbanization, our contemporary sprawling life is better than the more compact urban life that preceded it, and actually commuters are gaining welfare roughly in proportion to the vast expansion of transportation miles purchased. One can argue that America's suburbanization reflects political, economic, and social factors beyond individual preference, that are in any case locked into the built environment in ways now exempt from personal choice, and that in fact suburban life is alienating and undesirable, as revealed by the very high cost of housing in compact urban areas that offer similar levels of safety and amenity under a pedestrian lifestyle.

These are live questions! No purchasing power measure can answer them for us.

We get to choose the bundle of goods and services we purchase from our income. Real purchasing power measures can tell us something about "how much" we can buy, although collapsing a very heterogenous mix of goods and services into a single scalar value is always fraught.

But real purchasing power measures do not tell us anything about the circumstances under which we purchase or deploy that bundle of goods and services, how much we require to stay above some waterline of welfare.

And our requirements changes, quite radically. A person with the income, in real terms, of say a successful physician in 1925 would find themselves living in a mobile home park today, and incapable of practicing their profession, even presuming updated skills, for want of any ability to afford prerequisites to participation in terms of transportation, wardrobe, liability insurance, etc.

At an individual level, we understand that an easy mistake a person can make is to let their cost structure increase faster than their means, even as their means increase. If your salary doubles, but you trade up for a more expensive car, a more expensive home, the kids in private schools, you may find yourself in a worse situation than you had been before your raise.

But our economic measures don't consider that we might do this as a society, in ways that override the choices that we might have made individually. Most of us can't really opt out of the cost structures we've collectively established as typical, absent very radical breaks from society that would themselves negatively affect our welfare. Even as our correctly measured, median-not-average real income increases, our actual choice set can grow worse.

2. Only the most quantifiable aspects of changes in quality are captured in inflation measures

Inflation is not correctly measured. In saying that, I cast no shade on the fine econometricians at BLS, who do (and I hope will be permitted to continue to do) great work.

But when constructing measures of the price level, if we fail to take into account adjustments in quality, we'll get things very wrong. BLS does adjust for quality! But they can only do so for characteristics that can be straightforwardly quantified, and their relationship to price modeled. Plus, there are lots of characteristics that plainly affect the quality of our consumption of a good or service that would, for a variety of reasons, be controversial to incorporate into a price measure, and so aren't.

Suppose we faced a new national crime wave that led to a broad-based increase in ones risk of getting mugged or ones house burgled. The risk level has increased similarly basically everywhere. For all of us, then, the value in real terms of our housing will have declined. Nearby crime is definitely a component of the lived quality of housing!

In theory, a price index could take this into account. To hold welfare roughly constant, a person living in what was once a modest crime neighborhood but is now very dangerous could move to a comparable home in what was once a practically crime-free neighborhood but now experiences modest crime. The difference in price between these two otherwise comparable homes would be a measure of how much the change in crime rate had impaired the value of the original home, even though observed home prices might not change much, since since the relative rankings of safety have not changed and everybody still has to live somewhere.

Safety is a big part of what we pay a premium for when choosing a home. The national crime wave makes us all poorer, in terms of qualities of housing we actually pay for. From a conceptual standpoint, if aggregate prices are unchanged, we should treat that as housing inflation. You have to pay more to get the same safety-adjusted home.

But we don't do that. Our thought experiment about how we might quantify the impairment depends upon accurate comparables that would be difficult for a statistical agency to put together. We don't think crime rates are measured very well or consistently across localities.

Crime obviously and profoundly affects the real value of housing services we purchase. Home prices obviously depend, quite a lot, on nearby criminality. But we exclude it from the quality adjustment exercise, and we're unlikely to do anything about that.

This goes both ways! The past few decades have in fact been a tale of decreasing average criminality, despite a post-COVID bump. That has amounted to a large increase in the quality of American housing. It's been a source of overstating inflation.

The point, however, is that aspects of the quality of goods and services unaccounted for by statistical agencies introduce noise, error, into our measures. Over short periods of time this error is likely to be small, but unless we assume without evidence the positive and negative omissions cancel, we should expect our mismeasurement to accumulate and grow over time. Comparisons, then, of "real" purchasing power become more and more surreal, the longer the interval over which they are compared.

3. Inflation measures abstract away the bundling of goods and services

Aspects of housing quality that we can and do adjust for include measurables like square footage. If, over a decade, observed home prices rise by 30%, but the size of the typical home sold also increases by 30%, should that count as home inflation at all? Mostly it doesn't, because BLS adjusts for home sizes. ¹

But housing is not in fact a good one buys by the square foot. When one buys a house, or rents an apartment, one must buy or rent all of its square feet. And the square footage of homes available is correlated to other characteristics, like recency of construction, quality of school, nearby crime rate, etc.

A stupid point widely made in the punditocracy is that, if one wanted to live like a middle-class 1950s family on a single middle-class wage, one absolutely could, if one bought that same 1950s home with its 1950s square footage. But one would not actually be buying the same home at all, because the amenities that are the primary source of value for most homes are related to the affluence of ones neighbors. The small single-family home you could still buy at 1950s prices relative to a single middle-class wage earners' wages is very likely in a neighborhood with not-so-great schools, dollar stores rather than a Safeway, some issues with crime.

Suppose a young couple, two incomes but not huge incomes, wants to buy a home in a new neighborhood with great schools. They could afford to buy in the neighborhood the thousand square feet that was typical of a single family home in the 1950s, at the current going rate per square foot. Unfortunately, our couple would likely find that there are no homes smaller than 2000 square feet in an upscale new single family home development.

Our real purchasing power measures presume that they could buy such a home. But in practice, they have to buy twice the home they would have preferred in order to obtain the other goods bundled with housing. They are significantly poorer than their headline CPI adjusted wage suggests, because the methodologies behind inflation indices assume purchasers can independently set the quantities of goods purchased over a continual range, and often that they can substitute at will.

In unequal and class stratified societies, it is often the case that a few choices — in what kind of home one lives and near whom, what means of transportation you employ — segments whole classes of typical consumption bundles, in ways that sometimes seem perverse. For example, in the US, one can eat very cheaply if one stocks up on bulk goods from Costco. But to do so, one wants to live not so far from a Costco, have something like an SUV, have storage for bulk and also refrigerated goods, have financial liquidity one can afford to tie up in inventory. The low cost of bulk Costco purchases is correctly factored into the average that forms CPI grocery prices, but is inaccessible in practice to people who've selected (through lower cost housing and transportation) into a different consumption bundle.

Again, over short periods of time, the bundles are what they are, they remain pretty constant so measured changes in the price level effectively control for this.

But over longer periods of time, the whole structure of these bundles shift around, introducing error into comparisons of "real purchasing power".

4. Changes in inequality of quality are overlooked

Charles Murray, whatever else one might have to say about him, wrote a good book in "Coming Apart". Among other observations, Murray pointed out that American success was increasingly concentrating into "super zips". A few concentrated geographies were becoming enclaves of the well-to-do, while the rest of the country languished.

Let's consider how this dynamic might affect inflation and therefore "real purchasing power" measures, versus how it might affect welfare.

Everyone must be housed, and to a first approximation, the housing stock of the country matches the number of households. Houses are a game of musical chairs: Absent a serious surplus — which is very far from our problem — roughly all the houses will be occupied. (Yes, second homes exist, and some homes fall into disused vacancy, but these categories are small, and if anything they exacerbate the dynamic we'll describe.)

So, if a small fraction of the houses, for whatever reason, become much more desirable than the rest, their prices will be bid up. However, there is no reason to think that the average price of housing will increase. The purchasing power devoted to paying up for housing in the "super zips" is purchasing power no longer directed to housing elsewhere. If we hold the aggregate housing budget constant, the effect would be home prices in the desirable areas would spike, while housing prices in the rest of the country would fall a bit, so the average cost of housing remains unchanged.

In practice, the aggregate housing budget, even for existing homes, is not held constant. Soaring top inequality helps ensure prices of the most desirable homes are continually bid upward. Our financial system lends purchasing power to home buyers at an increasing rate, a rate that for the past four decades has exceeded the rate of inflation. Housing prices in aggregate rise.

But the effect we've described still holds: Housing prices in "super zips" rise much faster than elsewhere, housing prices elsewhere increase more slowly, as so much of the aggregate bid on housing flows to premium enclaves. An average ends up in our inflation index.

If we hold the quality of housing constant, this distributional oddity translates to an increase in most homebuyers purchasing power. Rich peoples' odd fetish with "super zips" means that housing is cheaper everywhere else for everyone else than otherwise it would have been. Noah Smith famously made this point with his "yuppy fishtank" metaphor. Normal people should delight as yuppies and tech bros and other people with more money than sense waste their housing dollars in a few overpriced districts. It prevents the rest of us from having to compete with their vast pocketbooks if we want a bit of space.

However, as Murray observed, we don't in practice hold the quality of housing constant. The quality of housing is largely a function of who your neighbors are and how that shapes the parks, schools, commercial amenities, social and employment opportunities that surround you. Well-to-do people are not in fact concentrating themselves into these few districts because they are fucking weirdoes who like to be fucking weirdoes. They are doing so because there is a network effect. The more people with resources you put together in a tight geography, the more likely it's going to be a safe, appealing place to live with good schools and good jobs. When those resources withdraw themselves from the rest of the country to enjoy this network effect, the rest of the country suffers a loss in quality. Safeways get replaced by Whole Foods on one side of the divide, by Dollar Trees on the other.

Housing price increases come to have a bimodal distribution, a high rate for coveted geographies, a lower rate everywhere else. The measured average rate of housing inflation is a correct average, but matches lived experience on neither side of the divide. The median household is likely to experience a lower rate of housing price increases than is included in the measure.

However, before we pronounce on the correctness of the measure for any household, we also have to adjust for quality. It could be the case that the relative decline in quality of non-super-zip areas exactly matches the slower-than-average price appreciation, so the inflation measure is in fact correct for the median household!

But it could also be the case that the decline in quality of households in less coveted locales more than offsets the lower-than-average rate of home price increases, in which case headline real purchasing power for the median household would be overstated.

The same is true for wealthy areas. It could be that the increase in quality due to rich-people network effects exactly matches the higher-than-average rate of price increase, and so the headline inflation measure is correct for these households. It could be that rich people getting together and sharing lattes and nice parks is so awesome that the quality difference outpaces the above average home price appreciation, and so measured inflation is overstated for these regions.

It could also be the case that there are diminishing returns to rich-people network effects, so that over time the quality increase fails to match the high rate of home price appreciation, so their true rate of housing inflation is higher than the headline measure.

Note that there is no necessary relationship between quality changes in the two regions. Nothing compels a loss in quality in less coveted geographies to be in any sense "equal to" a gain in quality in the special places. It could be that rich people leaving affects the quality of life in less pricey places not at all, so rich people self-segregating amounts to a positive sum change, they enjoy nicer parks at no cost to anyone. It could, however, be the case that better-resourced people segregating away has a very large negative effect on the places they abandon, which more than offsets, in aggregate terms, the benefits the wealthy enjoy in their enclaves. If this is the case, the dynamic could make us in aggregate much poorer, while our real median purchasing power measures remain entirely oblivious.

5. The composition of price changes has distributional effects

Among other responses to Stancil's post above was this (in)famous graph, taken from the Baumol effect Wikipedia page:

"Real" purchasing power is based on the average price of some bundle of goods, usually allowing for some degree of substitution between similar goods. Even while the average stays relatively stable, prices of the components of the bundle can vary widely. As the graph shows, some things have grown much more expensive, while others have grown much cheaper, even while the overall price level ("CPI for All Items" in the graph) changed less dramatically.

Further, the composition of the bundle replicates what an average rather than median purchaser might buy, which has important distributional consequences.

It's worth unpacking this a bit. In the econo-micro-blogosphere, there's a typical dialog that goes something like this:

Apologist: [posts graph like Stancil's at the top of this piece]

Agonist: You say that incomes have grown, but that's just because, like, Elon Musk and Jeff Bezos have gotten so rich.

Apologist: It's median income. Do you even know what a median is, ya fuckin' commie?

[Suddenly in His Infinite Patience, Matt Darling appears to explain what a median is.]

"Apologist" is right that the income measure she has chosen is a median, and so is unaffected by Musk and Bezos' hypertrophied extraction.

But there is a wrinkle! The very units of that measure are based on average purchasing habits, which mean they are skewed to the purchasing habits of the relatively rich who, unsuprisingly, spend a disproportionate share of the money spent on consumer goods!

(See Nathan Tankus, Eduardo Ley.)

The less rich spend a greater much fraction of their wealth on necessities and inferior goods than the wealthy do. A quick glance at the graph above confirms that prices have tended to increase quickly for the necessities that consume a large share of the median consumer's budget, while we've enjoyed outright deflation in more discretionary goods. So "median household real income" is overstated for the median consumer, because its very units are biased toward the experience of the better off.

This bias would disappear if variability in the rate of price changes happened to be orthogonal to differences in the consumption bundle between the rich and less rich, but that is unlikely ever to be the case. Independence of variables in social affairs is the rare exception, not the rule.

The direction of the bias could invert if high-budget-share necessity prices grow more slowly than the prices of discretionary goods. But for the past decades, the pattern of facts means that "median real household income" measures are overstated for the median household.

The size of this bias will be related to the level of consumption inequality. We have interminable arguments over whether inequality is increasing or decreasing, and what kinds of inequality count. Under Biden, wage inequality compressed for a while, but comprehensive income inequality tends to be driven more by asset valuation changes, and home prices boomed. Although the recent claim that the top 10% account for 50% of consumer spending is almost certainly overstated, consumption inequality in the US is high, and has likely been growing over the last few decades. If consumption inequality increases while inflation hits necessities more than discretionary items, both trends will reinforce one another in exaggerating measures of "real median household income".

Short Conclusion

Stancil writes

The American economy is much larger than a few decades ago and much of those gains (not enough! But a lot) have been spread across the income distribution. Economically speaking, you’d almost certainly rather live today than any prior time.

That's his point of view. It might be right! What we measure as GDP is, factually, larger than it ever has been.

But any claim that things are "better" for the typical American is an opinion, one you can try to muster various sorts of evidence to support, but on which you can't pull rank — because science, data — and coronate as fact.

A lot of people — yes, a lot of us, I number among them — disagree that these are materially the best of times in the good ol' US of A. There is evidence on our side too. Suicide rates have been on a consistent uptrend. The political upheavals we've experienced might have other-than-material causes. But it does suggest a lot of us might not perceive our circumstances as too wonderful if more of us want to kill ourselves and the public seems willing to risk burning it all down just for the possibility of a change.

Whatever we are doing seems not to be conducive of welfare, and economics ultimately is the science of organizing production and consumption in the service of human welfare. If our measures say everything is awesome, we might hope for more discerning measures.

I guess you can argue that things really are awesome and we are last men, blowing it all up out of ennui, because we are too comfortable. I'd gently suggest you are mistaken. From my perspective, if this is your view, you don't live in the world. In the world I inhabit, even the objectively affluent suffer astonishing levels of financial distress. Lots of people aren't objectively affluent, perceive themselves as losers in the richest country ever, do consider suicide. The cost structure of contemporary American life pushes us all to become grifters. Whether we succeed or fail, what is surely unaffordable is to be good people.

Whichever side of the argument you take — on this as on most social questions — what we can observe definitively in factual terms is unlikely to be dispositive.

This is why we have democracy in the first place. Ultimately human welfare is a matter of subjectivities. We require our subjectivities to be incorporated in the policymaking process if the polity is to thrive. You are a bad technocrat if you imagine you can substitute your technocratic measures for those subjectivities. Your job is to grope for a reconciliation between the objective constraints about which you may in fact be expert and the subjective outcomes that are ultimately policy's purpose.

I'm happy to concede that most Americans can buy "more stuff" in some sense than ever before. The typical household can buy a lot more televisions. But it is perfectly possible — I think probable, I think obvious — that matching the stuff you can buy with the prerequisites of a materially secure, comfortable life has become much harder for the typical person than it was even in the relatively recent past.

Stancil — and you dear reader! — are free to disagree. I certainly will not claim that I can prove you factually wrong.

I do think I can reject Stancil's assertion that "claims like 'there has been a generational stagnation of purchasing power'" are somehow a shibboleth of the progressive side of economic debates. Let's hear from Rod Dreher, a "post-liberal" on the Orbanist right, speaking with a young conservative:

I asked one astute Zoomer what the Groypers actually wanted (meaning, what were their demands). He said, “They don’t have any. They just want to tear everything down.”

Then he went on to explain in calm, rational detail why his generation is so utterly screwed. The problems are mostly economic and material, in his view (and this is something echoed by other conversations). They don’t have good career prospects, they’ll probably never be able to buy a home, many are heavily indebted with student loans that they were advised by authorities to take out, and the idea that they are likely to marry and start families seems increasingly remote.

If our economic miseries are only mass delusion, it is delusion more widespread than the progressives with whom Stancil so remarkably squabbles online.

It's more complicated than this, BLS adjusts for number of rooms, rather than square feet directly, and the cost of owned homes is imputed as rents foregone rather than based on sales price. But conceptually, BLS tries to control for changes in home size.
↩

draft by Steve Randy Waldman
2025-11-12 @ 03:40 PM EST

A fertile corpse

The Democratic Party should campaign and run on a single issue: Blowing up the two-party system.

The Democratic Party, as presently constituted, cannot succeed under contemporary circumstances. The best thing it can do is become a fertile corpse from which more capable successors can grow. The party's infamous "fecklessness" is structural.

To understand this, I recommend a recent paper by Tom Pepinsky, which usefully distinguishes between states that structure or referee pluralistic conflict and states that purport to express an overriding collective will:

[S]hould the state aggregate, channel, or otherwise structure political organization and contestation among the units of society, or should it express the popular will? In the former approach to the state, political institutions mediate conflict and channel popular demands by establishing formal and informal rules that define public and private spheres and structure political competition. Countermajoritarian institutions, for example, restrain the majority from enacting its preferences, and representative institutions organize the politics of lawmaking, policy advocacy, voting, and other forms of claims-making in ways that prevent disordered, disorganized, or extreme political demands from becoming law or policy.

An alternative view of the state sees any such mediating institutions as an obstacle to the full expression of the popular will. In this approach, the appropriate function of the state is to express or manifest a collective will, rather than to direct or constrain that will.

The United States' electoral system, built on top of single-winner, first-past-the-post election strongly encourages the emergence of two and only two "major" political parties. Americans have generally conceived of these as necessarily "big tent" parties, each the product of fractious internal negotiations yielding broad coalitions which fully satisfy no one, but which represent — or at least better-represent than the other party — something like 50% of the electorate.

There was broad consensus in the United States that role of government was to referee a pluralistic society. The major parties took on a kind of fractal, subsidiary role in the system, each refereeing contestation within its own half of the electorate.

This consensus on the role of government has been shattered now, but asymmetrically. One of America's two major parties has reconceived itself as striving to realize a true collective will against obstacles imposed by internal enemies. Republicans now claim a "mandate" that cannot be justified on quantitative electoral grounds, but which they nevertheless proclaim represents the true and correct will of the American people. It is not yet clear, under Pepinsky's taxonomy, whether the new Republican Party will ultimately prove to be "populist" — channeling a popular will conceived as an aggregation of what individual citizens want, or "integralist" — organized to serve social categories and hierarchies conceived as prior to and overriding of individual caprices. (My take is the Republican Party sells itself, while it still must sell itself, as populist, but its leaders mean to rule on authoritarian integralist terms.)

While the Republicans monomaniacally express a will they fabricate on behalf of "real Americans", the Democratic Party remains a referee of its very fractious coalition. It seems as if "we" have no core, no underlying principle, because "we" constituted as the Democratic Party do not in fact all share similar values. For example, social democrats (like me) and the still very neoliberal "centrists" agree that we want a pluralistic society under rule of law. But in fact we hope for very different outcomes, very different kinds of societies, to emerge under that rule of law.

When that was also true of the Republican Party — when their religious conservatives and their neoliberals were locked in analogous internecine struggles — there was a kind of symmetry. Both of the two parties faced the same kind of handicap, and the electoral system could function giving factions across the two parties more or less equally compromised forms of representation.

But the Republican Party is something different now. Its internal differences are overridden by shared allegiance to the person of Donald Trump. It is united by struggle against internal enemies, a "far left" which those of us who allegedly constitute it perceive as farcical exaggeration, as caricature, but whose social reality lies not in representational accuracy, but in how it motivates and organizes the people who present it.

Against this kind of party, the Democratic Party is a nullity. It can win elections sometimes, after the party which claims to express the collective will governs in ways that prove unpopular. But it subsists only in that party's negative space. A divided self-refereeing big-tent can exercise no real agency. It reacts. It must hope for and rely upon the other party's incompetence. Politics becomes a game played out between one clear protagonist and an inchoate cast of minor characters.

The internal diversity of the everybody-but party is not, alas, that party's strength. Almost anything that would really excite one faction in the grab-bag party becomes a wedge issue alienating other factions. Those who identify with the populist or integralist or perhaps fascist party find themselves unified, excited, thrilled by their leaders' bold choices. Those who identify with the residual party find themselves constantly frustrated, betrayed, asked to embrace compromises that satisfy no one in the name of "strategy" that amounts to little more than holding onto our best shot if the other guy fucks up badly enough.

This just isn't a winning hand. Even when the residual party wins elections, it disappoints by its compromised governance, rendering every reprieve fragile and temporary.

Human affairs are not mere numbers games. Successful political struggle relies upon political passions. Perhaps a system in which nearly everyone's passions were frustrated or attenuated by the institutions of political contestation was once moderating, virtuous. Perhaps it was never virtuous, and that's why it failed.

Regardless, it has failed. It is no longer on offer. We now require political parties that organize political communities with which people passionately identify, within which and on behalf of which people energetically act.

The Republican Party is that. The Democratic Party cannot be that, not without alienating constituencies it requires in order to have any hope of victory. We can't all passionately identify as "Democrats", and share a common leadership, while that leadership necessarily sells out this group or that in the name of avoiding wedge issues and preserving the prospects of electoral incumbents we can't afford to lose. We would be more effective as distinct passionate communities who work in coalition, as strange bedfellows, on the pluralistic commitments we share without having to pretend we share views that in fact we do not on how a good society would be organized.

But the factions of the Democratic Party cannot productively divorce so long as the electoral system overwhelmingly punishes defection from one of what must be only two structurally "major" political parties. The Democratic Party's monomaniacal challenge right now should be to insist that the House of Representatives is chosen by some form of proportional representation, and that inherently single-winner positions like Senator and President are chosen by approval voting.

Opening up the electoral system would shatter both of the legacy major parties, not just the Democrats. The MAGA wing that now controls the Republican Party has cowed and silenced the rest of its coalition, but Republican electoral success still relies upon people who have misgivings, but who fear "the left" more, or who prioritize low taxes, or who just vote their cultural affinity. These people too could find homes and champions outside the ominous conformity enforced by Donald Trump.

Most voters detest the two party system. Nearly all of us understand we are poorly represented by the choices we are offered. But we have been trained — correctly, in our current system! — that voting for a third-party amounts to childish self-harm, spoiling the prospects of the less awful major party, handing victory to the very worst or us.

Once we educate the public about how and why seemingly technical changes to how we vote would in fact blow up the two-party system, electoral reform will become very popular, a great issue to run on. It's a promise we can deliver on if we win. You don't need a Constitutional amendment to make these changes. Just an act of Congress. A President who vetoes these reforms would make himself very unpopular, and we'd just do it again in two years.

To compete with a party that passionately expresses its (very ugly) version of a collective will, the rest of us will require political parties that can passionately channel our own collective aspirations. Forming and joining such parties doesn't mean giving up on the pluralistic society most of us want to inhabit and nurture. It simply means removing pluralistic contestation from the level of big-tent political parties, where it once accidentally found itself, into the institution of the United States Congress, where it Constitutionally belongs.

The Democratic Party cannot win and govern as a party defined primarily as "not those bastards". It can't commit to a clear identity (Bernie-style populist social democrats? Warren-style technocrat social democrats? Obama/Clinton-style progressive neoliberals?) without catastrophically demoralizing a large fraction of its coalition.

But the Democratic Party can organize a murder/suicide of both legacy political parties, and of America's two-party system. Sweeter fruit will grow from its fertile corpse.

draft by Steve Randy Waldman
2025-10-04 @ 01:30 PM EDT