Retail sales or payrolls: which is right about the US economy?

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Good morning. Treasury yields at both the short and long end of the curve have been creeping up the past few days. Is the September rate cut, which seemed like a lock after the July jobs report, coming into doubt? Federal Reserve chair Jay Powell speaks in Wyoming on Friday. Email us: [email protected].

Retail sales vs jobs

Everyone knows that economic forecasting is hard. Less appreciated is how difficult it is to know what the economy is doing now, or even was doing a few months ago. The signals are often equivocal and they generally disagree. Today we have a good example of this from two of the most important US indicators: retail sales and job growth.

After the June retail sales report, Unhedged noted that while retail sales were holding up overall, there were signs of tariff pressures, for example in weak sales in import-dependent categories where prices are rising, such as furniture and electronics. The July report showed some of these unpleasant trends (electronics, negative again) but not others (a bounce in furniture). And the June report was also revised up a bit. It comes as a slight a surprise to us, but retail sales have kept on rolling:

Those are nominal numbers, but even adjusting for inflation, retail sales are growing at perhaps 2 per cent a year. The solid news is echoed in private data. Bank of America reports that spending on its debit and credit cards rose almost 2 per cent in July, the fastest rate since January.

On the other hand, as we have written, job creation has been declining: 

How to close the apparent gap between steady consumption and weakening job creation? We see four possibilities:

• The decline in jobs reflects a falling supply of workers, not weakening underlying demand that would also show up in consumer spending. We wrote about this possibility here. 

• Because consumer spending is increasingly driven by the richest cohorts of consumers, the sales data underplays weakness in the economy that shows up in the employment report. Here, for example, is Bank of America’s breakdown of credit and debit card spending by income level:

• Household balance sheets remain quite strong, so households can continue to spend normally even as the economy and job growth cools. Evidence of balance sheet resilience is contained in the New York Fed’s household debt and credit report for the second quarter, which showed declining rates of transition to delinquency across loan types and age cohorts (with student loans as a significant exception). Here, for example, are credit cards:

• Consumption will eventually slow to match the jobs trend. Oliver Allen of Pantheon Macroeconomics, for example, writes that “further weakness lies ahead, as the weak labour market and further tariff-related increases in goods prices mean that real incomes essentially flatline. Moreover, the 0.4 per cent drop in food service sales in July highlights that the consumer downturn since the start of this year is weighing on demand for discretionary services too, not just goods.” 

We’re open to other thoughts from readers.

Volatility suppression

Last week we noted that the Treasury market has been remarkably stable in recent months: yields have stayed in tight trading ranges across the curve, and implied volatility has been declining steadily. We ascribed this to an underlying tension between the growth and inflation outlooks, which counterbalance one another. But there is another interpretation: that volatility is being deliberately suppressed.

Michael Howell of CrossBorder Capital writes that the suppression of Treasury bond volatility “appears to be a deliberate policy” of the Federal Reserve and the Treasury.

His argument runs as follows. The key marginal buyer of long-term Treasuries is basis traders: funds holding highly leveraged Treasury bond positions against short positions in Treasury futures, betting that the small difference in prices between the bonds and the futures will close. With a flattish yield curve and declining interest in Treasuries from foreigners, basis traders are the only buyers for whom long Treasuries are particularly attractive. But as with any highly leveraged trade, low volatility is a must. A rise in Treasury volatility would send the basis traders running for cover.

So the monetary and fiscal authorities want to keep the basis traders in the game by keeping money markets flush with cash, which in turn keeps the Treasury market calm. Volatility suppression might be behind a number of policy choices, according to Howell. The decline of the Fed’s reverse repo facility; the pick-up in active Treasury “buybacks” (replacing off-the-run Treasuries with more liquid on-the-run Treasuries); the mooted revocation of banks’ supplementary liquidity ratio (which would allow banks to own more Treasuries)’; and the shift towards shorter duration US debt issuance are all examples.

Some readers may find Howell’s view a bit conspiratorial. But Unhedged agrees with it this far: low Treasury volatility is a major support for all corners of financial markets, because Treasuries are the most important form of trading collateral, and the value of this collateral falls as its volatility rises. Almost everyone in markets therefore has an interest in low Treasury volatility, and might engage in implicit co-ordination to protect it.

Suppressing volatility — in markets or nature — is hard to do forever. Keep a close eye on the Treasury market.

Are data centres wasting assets?

Many readers responded to yesterday’s piece about tech’s dominance of markets. Several disagreed with my claim that even if artificial intelligence turns out to be a profits bust, data centre investments will not be a deadweight loss because “all that computer power will be useful for something”. One reader, Nick Banner, correctly surmised that I was thinking back to the fibre-optic overcapacity built curing the telecom/dotcom bubble, which came in very handy after the bust. This will not be the case with the AI data centres, Banner wrote: 

A strand of glass is a strand of glass, and you achieve ever-higher data rates [over fibre] by upgrading the terminal equipment (lasers, multiplexers, amplifiers, etc).

Data centres, on the other hand, are at heart racks of high-end Nvidia chips. But those chips are in effect consumables — they’re (rightly) depreciated over 2-4 years, because they’re superseded in that timeframe. Demand for four-year-old GPUs is pretty low. And the reason why they are in demand now is almost entirely because of [AI] . . . 

If this is right, a lot is riding on the success of AI as a business (rather than a technology). If the prospects for AI profits dim, Nvidia’s $4.4tn market cap will take a major haircut, and the valuations of Microsoft, Alphabet and Meta will take a knock as well, as multibillion-dollar writedowns will be in the offing and growth expectations will have to be cut, too. Again, the problem looks less like high market concentration and more like high market expectations.

One good read 

No.