The future of QT

good morning. Yesterday, minutes from the Fed’s December meeting were released, providing further evidence that Fed directors are nervous about inflation (and Donald Trump’s tariff and immigration policies). The market had little reaction. It appears that the cautious Fed has already factored in this. Please contact us via email: robert.armstrong@ft.com and Aiden.reiter@ft.com.

Unhedged writers really enjoy Martin Sandbu’s Free Lunch Newsletter written by Tej Parikh. A Sunday edition has been added to this newsletter. I think readers of Unhedged will also enjoy it. Sign up here.

Quantitative tightening

The yield on the 10-year U.S. Treasury is about 4.7%, the highest level since April of last year. No one likes this. Government bond investors suffered losses. Stock investors will feel threatened by high stock valuations. Homebuyers are expecting higher mortgages. And for government officials, the bond market signals dissatisfaction with fiscal policy.

There are various causes. Markets appear to be absorbing recent positive economic news, suggesting that inflation may remain slightly elevated. Investors may be starting to accept the possibility that the budget deficit will widen further if Republicans take full control of the government. At least the bond market is pricing in increased uncertainty.

Structural changes in the government bond market may also be having an impact. The Ministry of Finance has increased the proportion of short-term government bonds issued compared to the past few years, making long-term government bonds increasingly rare. This may have supported prices and pushed down yields. Some observers have characterized this as a “secret” monetary policy tool, or “quantitative easing by other means.” Treasury Secretary Janet Yellen denied the allegations. Incoming Treasury Secretary Scott Bessent is said to be in favor of increasing issuance of long-term bonds. The market may be selling long-term bonds in anticipation of Mr. Bessent’s approach.

Yield increases may slow. But what happens to monetary policy if this situation continues and stress in the financial system increases?If inflation actually stagnates, the Fed will have to keep interest rates high for a longer period of time. But what about efforts to shrink banks’ balance sheets, which have become bloated due to quantitative tightening and the response to the COVID-19 pandemic? Could the Fed halt policy to reduce stress on the financial system? Could they reverse it and start buying Treasuries again, i.e. quantitative easing?

Unhedged hasn’t written about QT in a while for a simple reason. Because it’s working. After QT contributed to the 2019 repo crisis, many feared that the latest QT cycle starting in 2022 would cause a similar panic. But things are going well and there’s good reason to think they’ll stay that way.

First, QT does not appear to have had a significant impact on long yields, so ending QT is unlikely to provide much relief to the Treasury market. By buying up Treasuries, the Fed raises the price of Treasuries and lowers yields during quantitative easing (though the magnitude of the impact is debated). At the beginning of this cycle, many were concerned that QT would backfire. But the impact appears to have been more modest, with a paper by Wenxing Du, Kristen Forbes, and Matthew Ruzzetti estimating that yields were pushed up by about 8 basis points.

Second, the Fed would only end QT if it faced a major liquidity collapse in the capital markets, and a spike in Treasury yields would not cause a crisis. From Darrell Duffy of Stanford University:

The Fed only buys securities when interest rates are at zero to stimulate the economy, or when the market for Treasuries is clogged. It’s not that interest rates are high or low, it’s that the market is dysfunctional because too many investors are selling Treasuries. . . (Even if) U.S. Treasury yields become very high, the Fed will not move unless there is market dysfunction.

Either way, QT may soon come to a natural end. There is no “right place” to stop QT — the Fed is fumbling, or, in the words of Chairman Ben Bernanke, “learning by doing.” Ideally, QT would stop once bank reserves normalize, but there is still enough liquidity in the system for markets to function smoothly. One approach to finding this level is to calculate the sum of the bank reserves held by the Fed and the reverse repurchase window (RRW) balance, or the amount available to banks if they are willing to pay market interest rates on bank reserves. Pay attention to the amount including “excess reserves.” says Joseph Wang of Monetary Macro. Together, these two represent the total amount of funds available to banks at short notice. At the time of the repo crisis in 2019, this number was around 8% of GDP. Getting close to this number, but not too close, is a reasonable goal. And it probably won’t be long. Total RRW balances and reserves are approximately $3.4 trillion, or 8% of GDP in the third quarter, or $2.3 trillion.

Most analysts we spoke to expect QT to end this year, but that’s due to natural causes rather than market stress.

(writer)

Big Tech capital investment details

Several readers wrote responses to yesterday’s article about capital expenditures and depreciation expenses for major US tech companies. The gist of their comments was that the market looks through accounting and looks at cash flow. It doesn’t really matter that the big tech companies’ GAAP profits don’t represent how much they spend.

This is true, but only up to a point. The free cash flow (operating cash flow minus capital investment) of the five companies examined yesterday is as follows.

These companies (excluding Tesla) are still generating tons of free cash even after raking in billions of dollars in AI data centers. However, we note that the trend in cash generation is flat to declining. This becomes easier to understand when you look at them together.

Now, some may look at this and say, “Cash flows are cyclical, and the industry has experienced down cycles before, like 2021-2022, but no big deal.” do not have. And there’s some truth to that. But it’s worth remembering that all of these stocks underperformed the market during that part of the cash cycle (with the exception of Microsoft, whose underperformance was notable).

What matters, of course, is what happens next: how long the rampant spending on AI lasts. It’s hard to say. Meta, Amazon, and Alphabet have all indicated that their capital spending in 2025 will be higher than in 2024. More details will emerge from the company’s fourth quarter earnings report.

Looking at the future is doubly difficult because we don’t know how exactly these companies are depreciating their data center assets from an earnings per share perspective rather than cash flow.

Ravi Gomattam of Zion Research Group, who specializes in accounting issues, emphasized to me that modeling future depreciation for large tech companies is extremely complex. Companies only provide high-level information about their investments, and those high-level numbers can obscure a lot. Consider data center spending. What part are the servers? What about the non-server infrastructure? What about the buildings? Treated separately, each will be depreciated at a different rate. Or you can depreciate them together. In addition, there is the issue of how depreciation rates will change or write-downs will occur when computer equipment becomes obsolete due to technological innovation. Making reliable predictions requires many assumptions and backtesting.

What is certain is that this year will be about investing, depreciation, and cash, not just the AI ​​hype, for tech investors.

A book I read often

I got stuck in the middle.

FT Unhedged Podcast

Can’t get enough of Unhedged? Listen to our new podcast twice a week for 15 minutes of digging into the latest market news and financial headlines. Click here for past editions of the newsletter.

d template