This is a better version of that chart; the Y-axis is pinned at $0 and it’s quite remarkable to see the year-over-year shift between 2024 and 2025 presented across the full timespan.

If you’d like to explain your position on this and how either chart supports your own viewpoint on what’s going down with BNPL and debt and wages, I’m listening.

sobellian 16 hours ago | parent [-]

I'm just evaluating the claim that household income has deflated for 50 years. So I took the start of the data series at ~60k and compared it against the latest datapoint at ~80k. So I concluded that the data does not support your claim that household incomes have deflated for 50 years. I have no position on how household income influences BNPL.

▲

john01dav 15 hours ago | parent | next [-]

What basket of goods does your source use for determining income in real terms? The ratio of money on essentials versus luxuries has changed over the past few decades, and it's done so unevenly across the population. For example, a young person looking to find a place to live today is going to face astronomically higher rents while someone who bought 30 years ago in a place like California that caps property tax growth may find that their monthly (or total over a long time) housing cost is vastly lower.

Whether real wages have gone down depends on what you use as a benchmark for said wages.

The federal reserve even has an article about this topic: https://www.federalreserve.gov/econres/notes/feds-notes/diff...

▲

sobellian 15 hours ago | parent | next [-]

You can look at the linked page. It uses "Income in 2024 C-CPI-U (2000-2024) and R-CPI-U-RS (pre-2000) adjusted dollars." I would be interested in a dataset that shows deflated household income over this same period using a reasonable basket of goods. The FRED dataset shows an increase of roughly a third. It would take a very large adjustment to knock that down to significantly negative growth.

▲

kelseyfrog 15 hours ago | parent | prev [-]

While we can find extreme examples, we should be hesitant to conclude that they are a representative sample of the population. We should let outliers update our beliefs about means and medians very little to none for population level values.

What analysis that has been done of disparate inflation points out:

> The cost of medical care has more than quintupled since 1983, growing almost twice as fast as the overall price level. The BLS calculates that people 62 and older devote 11 percent of spending to medical care, while the general population devotes 8 percent. Figure 4 shows basket-share differences between older adults and the general population, and the price changes for those categories over the past 10 years. Loading figure 4...

> Older adults also spend considerably more on housing (49 percent versus 45 percent for the general population). They spend relatively less in other major categories, including food and transportation, that had lower rates of inflation than medical care and housing.

https://www.minneapolisfed.org/article/2024/breaking-down-in...

▲

expedition32 4 hours ago | parent [-]

I always believed that we should judge a country on the poorest not the Ferrari owners.

	▲	kelseyfrog 35 minutes ago \| parent [-]
		Then my comment doesn't apply to you because you're not using outliers to inform mean and median.

▲

altairprime 11 hours ago | parent | prev [-]

Perhaps your definition of 'household income' does not account for loans? Certainly loan issuers are accounting for active loan balances when evaluating income, and I see no reason to deviate from their standard. It's a sensible approach and one I wish was better incorporated into Fed reporting. While I won't try to persuade you to adopt my definition, I do owe an in-depth explanation of my reasoning for others.

Considering the % change year-over-year in revolving debt per household versus the % change in household income, it seems like growth in debt (CAGD) has been compounding more rapidly than wages for a majority of the past fifty years: https://fred.stlouisfed.org/graph/?g=1XTi9

There are two clear points that runs counter to that trend: in 2009 with subprime, and in 2020 with Covid. In both cases, revolving debt collapsed much faster than wages; but, subprime took years for the financial industry to unwind in order to be comfortable resuming positive CAGD, and it shows.

So if growth in debt has compounded more rapidly than growth in wages, even after accounting for inflation and number of households, for over 80% of the past five decades — then my understanding of compounding rates suggests that consumers are experiencing wage deflation over that period, which they are compensating for through debt inflation. While one could argue that the debt is an unknown mix of necessary and optional spending, the article we’re discussing is not unclear at all. This directly ties to my original point: if BNPL lenders are openly targeting advertising at the lowest-grade tranche of ‘pay rent and utilities with debt’ — aka subprime and largely the territory of loan sharks, payday loans, and other such usury over time — then they must have completely exhausted all other growth opportunities. A customer who is paying their rent and utilities with BNPL is much more likely to default, and if there aren't any higher-grade tranches of spending categories left to target, then the BNPL industry must be facing market saturation in all higher-grade tranches.

This is why I'm treating this as a macroeconomic warning signal: if the debt market is rather saturated from an advertising perspective, then debt per household is rather likely to stop growing. Inflation is not likely to stop growing with the levers of regulation available to The Fed. And in 2024, the last year for which data is available for all data sources, we do see household revolving debt crashing to -0.14% while household income increases by 1.25%. If that trend persists, corporations would face households demanding actual wage increases once household debt has no further room to grow; it’s no wonder they’re so desperate for AI to replace workers.

Also, I bet there are some lovely black swan options on Affirm, Klarna, et al. right now; 'no one could have predicted' their CAGR going negative when households hit their debt ceiling, etc. (I have no conflicts of interest to declare.)

	▲	sobellian 2 hours ago \| parent [-]
		I also tried to evaluate this claim that households have less real income after debt service. It is difficult because it depends on many nuances, like whether we care about all credit, or simply revolving credit. Your graph is interesting but difficult to interpret - the % change per annum is declining, but that's reasoning from some higher order derivative. I looked up a few more data series. https://fred.stlouisfed.org/series/BOGZ1FL153166006Q - consumer credit indeed rises from 1984 to present, but from ~17% to a peak of ~25% to the present ~22%. This seems to be more than adequately compensated by real household income growth over the same period. Even if we take the peak value of ~25% that leaves real household income after subtracting consumer credit higher than in 1984. But note that this series isn't debt service payments, it's credit stock. It doesn't make much sense to subtract. Unfortunately we don't have consumer credit payments going back this far. https://fred.stlouisfed.org/graph/?id=TDSP%2CMDSP%2CCDSP%2CF... - it is difficult to find debt service payments going back to 1980. Interestingly the measures relating to debt service payments appear to be basically flat or even slightly negative over this time period, though the consumer debt service series only goes back to 2005. So to my eyes it is difficult to support the notion that income-minus-debt-service has deflated over the past fifty years. I also find it difficult to take this data and state the stronger conclusion that half the population requires consumer debt to afford basic necessities but some fraction did not 50 years ago. The median appears to be better off.