Whispers of a consumer slowdown

Good morning. Long bond yields fell sharply today, and there is a lot of chatter about the Federal Reserve being done raising rates, perhaps stimulated by a speech from Lorie Logan of the Dallas Fed. If that’s right, this is the moment many hesitant fixed-income investors have been waiting for. We’re not convinced just yet. Send us your view: [email protected] and [email protected]. 

The weakening consumer: bad, if true

Over the weekend there was a story in the Financial Times which caught our eye, but was overcome by subsequent events. Our colleagues Stephen Gandel and Colby Smith wrote that

A drop in credit card spending is raising concerns about the financial health of the US consumer . . . 

“Credit card spending was soft in September, and what was notable was that softness was across all sectors,” said Citigroup economist Robert Sockin . . . 

Credit card spending at retailers dropped nearly 11 per cent last month, Citi reported this week, based on data from the bank’s own card customers. That decline, the fifth consecutive month of “spending deceleration”, was the largest of the year so far. 

The article points to various possible causes of the card spending slowdown: the rise in interest rates making balances more expensive to carry; higher debt loads, as total card balances and delinquency rates reach pre-pandemic levels; inflationary pressure on household budgets matched with slowing wage growth. Whatever the cause, the effect is concerning. A robust consumer, backed by a strong labour market, has been the engine of surprising US economic resilience. Even a small hint that the motor might be sputtering demands attention. 

The softness in card spending is not an entirely isolated signal. We have written recently about other slightly spooky signals at the margins of a generally healthy economic picture. The shares of banks, cyclical industries and highly indebted companies are wobbling, a sign that higher rates might be biting at last. Personal consumption expenditures, while still growing, slowed in August. One or two consumer-focused companies, notably ConAgra Brands, have recently made less than totally encouraging noises about demand. 

So how worried should we be? Happily, the BEA recently began publishing a “near-real time” spending indicator based on data from the cards payment processor Fiserv. We can now look directly at credit and debit card spending estimates based a very deep data set. And, indeed, the slowdown is visible. Here is a chart of week-over-week spending growth rates over the past two years, for all retail and food services. The data is very noisy, so I’ve smoothed it with an eight-week rolling average:

A couple of observations. The data, even as smoothed here, remains volatile, so we need to pick out trends with caution. That said, a clear dip in August and September (until the final week of September) is visible. Looking at the individual sectors, spending softened accross many categories, from furniture to food. Note, however, that we had a similar period of decline in the early months of 2023, and spending bounced right back. General conclusion? There is a weak but discernible signal from card spending suggesting a consumer slowdown. Unhedged will be looking to confirm it by listening carefully to third-quarter earnings reports, which start this week. 

More on term premia

We’ve been rattling on about the term premium, the additional compensation investors receive for holding long-dated bonds over taking the prevailing short-term rate. Lately, it appears to be rising, and the Fed is taking notice. 

Vice-chair Philip Jefferson’s Monday speech cited research that term premia tend to rise “if policy tightens in response to inflationary shocks”, after which “longer-maturity bonds become riskier”. In a separate Monday talk, Dallas Fed president Lorie Logan spent considerable time walking through her thinking on the term premium. There are too many interesting bits to quote, but here is one (our emphasis):

A rise in term premiums can itself have many drivers, including increases in the stock of debt relative to investors’ demand for debt, changing correlations between the returns on different asset classes — which can influence the portfolio diversification properties of long-term bonds — and reductions in expectations for the Federal Reserve’s asset holdings . . . [implying] that other investors will need to hold more long-duration securities, which appears to be one factor among the many contributing to higher term premiums.

On Monday, we talked through several possible reasons why the term premium might be rising, including the supply-demand balance for long-dated Treasuries and the Fed’s shrinking balance sheet. But we didn’t talk about what Logan notes above: shifts in stock-bond correlation.

Many investors own long bonds as diversifiers. If, when stocks drop, bonds tend to rally, investors can profitably rebalance by selling now-expensive bonds to buy cheap stocks. That’s the beauty of negative correlation. If correlation is positive — such as if inflation ruins returns for nearly all asset classes — that diversification benefit weakens (though it need not disappear). You might still want to own bonds for another reason, but some of the upside has dissipated. In turn, investors might demand a higher term premium as compensation. 

Does this hold in practice? Steve Hou, researcher at Bloomberg Indices and a former AQR quant, thinks so. He wrote his 2018 dissertation on the relationship between term premia and Treasury supply, and sums up his findings in the chart below. It measures the relationship between one popular estimate of the term premium and how much new supply is coming online. The red line shows the relationship when the stock-bond correlation is negative (a good thing!) and the blue line shows when correlation is positive (bad!):

The story of this chart is that, all else equal, positive correlation makes bonds riskier to hold. So as more supply comes online, the marginal investor demands more term premium to take on that risk. By contrast, in a negative correlation world, bonds reduce total portfolio risk, thanks to the diversification benefits. The result: investors are happy to shoulder more supply without any term premium.

Stock-bond correlations are sensitive to how you measure them, but as a real-time measure, Hou likes using two-year correlations of weekly stock and bond returns. On that measure, a two-decade period of negative correlation may be ending:

Hou thinks the change in correlation is happening because inflation’s salience has risen. When inflation is irrelevant, changes in companies’ cash flows fuel returns: rising cash flow yields boost stocks and make bond coupons look worse by comparison. Falling cash flow yields do the opposite: stocks down, bonds up. But inflation changes the dynamic. Rates must rise to cool price growth, lifting the discount rate and hitting both stocks and bonds.

For the recent surge in 10-year yields, Hou offers a bracing explanation. Since late July, “nearly the entirety of the increase in the 10-year yield is attributable to risk compensation for making space on arbitrageurs’ balance sheets to absorb incoming Treasury supply”. In other words, against the backdrop of a positive stock-bond correlation, Treasury investors have ratcheted up the term premium something like 1 percentage point. In part, this is because the marginal bond buyer — increasingly a hedge fund, rather than a central bank — may be more price-sensitive, as we discussed on Monday. And in this new regime, the massive Treasury issuance of the last decade will begin to matter, Hou says.

Maybe. The term premium is notoriously hard to pin down (as Hou concedes), and there is a risk of rushing to conclusions too early. More importantly, the bond-stock regime change — if that is indeed what is happening — won’t last if inflation doesn’t. The unavoidable question hanging over nearly every asset price asserts itself again: is the low-inflation world over, or just on hiatus? (Ethan Wu)

One good read

How does remote work affect productivity? The results of two big studies have a 30-point spread between them.