Archive
‘Threading the Needle’ Without Vision…or a Needle
From time to time, you will read that the Federal Reserve is trying to “thread the needle” on rates. That is, they don’t want to keep rates too high for too long, lest they sink the economy, nor lower them too soon or too far, lest they reignite (or merely fail to finish off) inflation.
To me, that is a very odd use of the phrase. “Threading the needle” tends to describe a delicate operation that must be conducted with deft precision, since a small error represents a failure. Not to put too fine a point on it (Ha! Ha! Needle pun), but that seems as wholly unlike the problem the Fed is confronting, and the characteristics of the error function.
Let’s think way, way back to 2022 when the Fed began hiking rates aggressively.
The Fed’s first 475bps of rate hikes corresponded with a rise in median inflation from 4.56% to 6.94%, over a period of more than a year, along with accelerating economic growth. The next 50bps must have been the magic number, because Median CPI has dropped 2.6% since March of 2023 and growth has slowed a bit.
Some of this – if you believe that changes in overnight interest rates have an important impact on inflation – might be written off to lags. But historically, the correlation of changes in interest rates with changes in core inflation 12 months later is approximately zero (as I illustrated in this nifty article almost exactly a year ago: Enough With Interest Rates Already).
How is it reasonable to worry about ‘threading the needle’ when the economy does not evidently respond to 50bps, 100bps, or 200bps of rate hikes? Methinks that this is not an apt metaphor.
Part of the problem is that the FOMC doesn’t really know what the target looks like that it is trying to hit. If you fervently believe that interest rates matter, but you don’t know what the lags are or the significance to the economy of 25bps, then how can you be confident that you have the precise control implied by ‘threading the needle?’ Fed speakers have admitted, although they don’t phrase it this way, that they don’t have much control over any of the variables that determine whether or not the target is hit. They’ve specifically noted that monetary policy is not efficacious in bringing down energy prices or moderating food prices, and some have opined that the rapid rate increases might have artificially increased shelter inflation by restricting the supply of homes on offer thanks to trapped mortgagees. Did the Fed rate hikes cause Used Car prices – one of the largest sources of the disinflation to date – to drop precipitously? It seems unlikely.
To thread a needle, you need fine motor control and the Fed doesn’t have it. It was difficult to even get inflation headed in the right direction, and it didn’t happen because of what the Fed did. If monetary policy makers looked at their record objectively, they would have to admit that microadjustments in a policy rate that directly affects very little of the economy do not seem to have much efficacy. If that is true, then the correct policy response is to do nothing unless you are absolutely certain that the direction you need to move policy is unambiguous, and the magnitude you need to adjust policy is substantial. Otherwise, you’re just adding unnecessary volatility to markets.
One thing you learn as a trader is that most of your money is made on very few trades, and so if you are trading frequently the majority of your trading is just noise. Similarly, for most investors it is more important to get the position scaled for proper risk rather than to choose the ‘right’ securities. The odds that you have chosen the best-performing stocks, if you choose just (say) three, are vanishingly small. Market prices may not be completely efficient, but they’re too efficient for most people to beat on a regular basis. Similarly, the Fed ought to make few moves. Most of its activity is just noise…at best. The market is pretty good at moving interest rates to where supply and demand balance, when it is left alone, and market interest rates have a better forecasting record than the Fed by quite a lot. Stop trying to thread the needle.
Inflation Deceleration Continues – But Not Enough
[A version of this article first appeared in last month’s Quarterly Inflation Outlook, available at https://inflationguy.blog/shop/]
Core and Median inflation continue to decline. This is not really a surprise; since early 2023 the clear direction has been to lower inflation. The debate has not been about whether inflation was heading higher or lower. The debate has been about whether the downtrend was going to converge on 2% as the Fed’s target, or fall short of that level. For at least that long, my position has been that median inflation would settle in the “high 3s/low 4s.” To date, nothing has happened to change that view.
In fact, it cannot escape notice that inflation has been coming down a lot more slowly than it went up. When the initial spike happened, certainly the ‘transitory’ crowd expected inflation would fall at least as rapidly as it went up, and even many of those who correctly understood that the underlying dynamics were not accidents of fate but the results of terrible policy thought that the return round-trip would take roughly the same amount of time as the outbound leg. But that hasn’t happened. The softening of inflation has been more reluctant than was the upward thrust. This is partly because, since the initial move in prices was not transitory, it kicked off a feedback loop: so wages went up to reflect the pressures that workers were feeling, and that fed back into inflation.
For Median CPI, the sharp acceleration took off from August 2021 at 2.4% and extended 18 months until it reached 7.1% in February 2023. In the 15 months since then, median has declined only to 4.3%, and this rate of improvement appears to be flattening out rather than accelerating.
On Core CPI, the difference has been more striking. The jump from 1.6% y/y to 6.5% y/y took 12 months, from March 2021 to March 2022. Since the actual 6.6% high in September 2022, we have had 20 months of declining inflation and core is only back to 3.4%.
The optimistic view is that we have had more months of decelerating inflation than we had of accelerating inflation. The more realistic view, especially considering that Median CPI hadn’t been above 3.33% for 28 years prior to COVID (and Core, not above 3.1%) is that inflation is converging to the mean…but to a different mean. This is what I have argued (for a long time) was happening: the perturbation to the former equilibrium displaced the whole distribution to a new equilibrium (“high 3s, low 4s”). We are now getting data that seems to support this notion.
One important characteristic of mean-reverting series is that the amount of mean-reversion “pressure” is related to the distance of the current point from the mean. That is, when inflation is far away from the mean, it tends to revert more quickly and when it is closer to the mean the pressure to converge is less. The general form of a mean-reverting series1 is:
In this equation, the economic variable is represented by the time series S, the long-term mean is μ, and the mean reversion rate is k.2 Because there is also random noise, and because many economic series don’t tend to see large perturbations on a regular basis, it is not a trivial thing to pick out the long-term mean and the reversion coefficient from the noise. But the point is that such series, when they are strongly perturbed, initially spring back rapidly but then gradually slow how much they are rebounding, until they approach the mean. That certainly looks like what we have here. The chart below shows core and median CPI, but from the point of the shock to new highs I have added ‘mean reversion lines’ where the long-term mean is taken to be 4% for Median CPI and 3.5% for Core CPI, and the mean reversion coefficient is taken to be 0.12 in each case.3
There are lots of different combinations that can produce plausible dynamics, and my point isn’t to claim that these are the right parameters. I am merely trying to illustrate that the recent behavior looks like a series that is mean reverting to new, higher means.
(For what it’s worth, if you want to see why most economists last year thought that we would be back at target inflation in late 2023/early 2024, use 2% for μ. In that case, inflation starts down much more steeply than we actually saw, and doesn’t flatten out until lower levels of inflation.)
Why is the rate of improvement slowing? It is slowing because the easiest improvements have already happened. For example, core goods inflation has declined from over 12% to -1.7% y/y. That’s great news – but the first 14% of disinflation is surely the easiest! Other, stickier parts of the CPI, such as shelter and ‘supercore’, are coming down more slowly (shelter) or not at all (supercore, which is at the same level it first reached in March 2022). In the conventional view, this is “improvement that is waiting to happen.” But if overall core/median inflation is converging to a higher mean, then these improvements will be mostly offset by an increase in core goods inflation from -1.7% to, say, 0%.
The road gets harder from here, and that’s what the decelerating deceleration is telling us!
- I’ve excised the complicated-looking, but irrelevant for this discussion, symbology for the noise term so as not to perturb readers too far from their means. ↩︎
- Worth pointing out, since I have used the ‘spring’ analogy to explain the behavior of money velocity, is that the ‘pressure’ part of this equation is identical to the physics of a spring, where F=-kx and x is displacement. ↩︎
- Actually, I’ve also removed the recursion – that is, the dotted line isn’t based on the most-recent S, but on the starting S and then thereafter on the calculated S. It would be what your mean-reversion-inspired forecast would look like, from the initial point. ↩︎
Inflation Guy’s CPI Summary (May 2024)
The CPI report for May was definitely good news. In April, core CPI was +0.29% and Median CPI was +0.35%; this month those figures were +0.16% for core and +0.25% (est) for median. That would be the best median CPI print since last summer and this was the best m/m core CPI print since 2021.
Core goods decelerated to -1.7% from -1.3%, y/y. I have long admonished that we are running out of room for deceleration in inflation to be driven by core goods as it’s hard to imagine goods deflation of a couple percent continuing for very long. Yet, so far, that is what we have gotten! Core services, meanwhile, was steady at +5.3% y/y.
But while there’s optimism in some quarters that we have seen the light at the end of the tunnel, this data was not unequivocally good news. The disinflation going forward cannot be all about goods, but in this report it mostly was. New car prices declined (although Used Car prices rose). Apparel declined, with some of the largest m/m declines in the CPI this month for its subcategories. Durable goods declined, and ‘education and communication commodities’ (things like computers, software and accessories, telephone hardware, etc) was a measurable drag. Those are all good things, but while Bullard today was talking about ‘immaculate disinflation’ (which is an idiotic term) there wasn’t really any sign of broad immaculateness. It was mostly in core goods.
As I mentioned, core services was steady year over year. But medical care – both goods and services, actually – both accelerated. I have been watching hospital services, within medical care, and it actually decelerated (7.2% y/y from 7.7%, chart below). Yay! On the other hand, the long-suffering Doctors’ Services accelerated to 1.4% y/y from 0.9%, and Medicinal Drugs rose to 3.4% from 2.6%. Boo. The +1.3% m/m rise in Medical Care Commodities was actually one of the month’s biggest gainers in the CPI.
Airfares dropped -3.6% m/m! And motor vehicle insurance -0.25% in a welcome respite. And car/truck rental -1.2% m/m. Thus “supercore”, which is core services ex-housing, actually declined m/m for the first time in a very long time even with medical care services going up, and the y/y number took a very small deceleration on the following chart.
That is welcome news, to be sure. But if goods prices were down and core services ex-housing were down (collectively), then obviously the fact that the overall inflation number was positive means rents are still percolating. Primary rents rose +0.39% m/m, and Owners’ Equivalent Rent rose +0.43% m/m. Those are both accelerations compared to the prior month, which is not expected! Y/Y, the numbers are still slowing, but not as fast as anyone would like.
This has led some people this morning to say that inflation right now is still ‘all about rents,’ and dismiss the 40% of the consumption basket that ensures people don’t get wet when it rains. What’s funny about that is that a few months ago, economists were pointing to rents as being the main reason to be optimistic about inflation because it would soon be in deflation! Remember?
Rents are decelerating y/y, but they’re not even decelerating as fast as I thought they would (and I was on the side of ‘they’ll go down a lot slower than you think, and not as far’).
The optimist here will say that the part we don’t have a long lead time to forecast – core goods and to some extent core services ex-rents – are looking good and ‘we know’ that rents will get better so ring the bell, the Fed’s job is basically done. That would be valid, if there was reason to think that core goods would continue to contribute the deflation that we have seen recently while rents continue to decelerate. But rents are sticky, and goods are not. To that point, consider the story of Wal-Mart, which announced last week that they will be replacing paper shelf labels with electronic labels over the next couple of years. You don’t do that to make it easier to lower prices. https://finance.yahoo.com/news/walmart-replace-paper-shelf-labels-221637323.html Typically, sellers try to raise prices quickly and lower them slowly. If you think goods prices are going to go back to the old regime of basically flat, with a small downward tilt, you’d keep using a slow pricing gun.
On the goods side, we also have to deal with the rising tide of global protectionism over the last few years (see picture, source Global Trade Alert), and the mass immigration to the US which puts pressure on demand long before the new source of labor contributes to supply (as with: housing). So far, a dollar which has generally risen over the last decade has helped to blunt those effects. But that won’t be the case forever.
The bottom line is that while this is a good CPI report – in some ways, one of the best reports we have had in some time – it is not an unvarnished positive. The failure of rents to decelerate according to plan, and the stickiness of wages so far at a fairly high level, is the underlying story. Goods and airfares are what painted the pretty picture this month. But if the picture keeps getting pretty over the balance of this year, it will be using paints from a different palette. I continue to expect housing costs to decelerate some (before re-accelerating), but I am not sanguine that goods and airfares will continue to drop at the pace which made today’s report so pleasant. Indeed, I expect that next month some of these categories will likely have some give-back so unless rents start to drop faster we could have a surprise in the other direction.
Naturally, as I always admonish, it is wise to not make major investing decisions based on one data point. One month’s figure should never cause you to change your medium-term forecast, unless it represents an accumulation of data that causes you to reject your prior hypothesis. This data point does not do that, since after all it is really the first really positive data point we have had in a while. I continue to expect median inflation to settle in the high 3s, low 4s. And as I said in our Quarterly, and in the podcast recently, I think that while the FOMC has no real reason to ease they likely will lower rates a token amount, at least once over the next few months prior to the Presidential election.
Talkin’ ‘Bout the China Gold (Whoa Oh)
I ran this chart in the Quarterly Inflation Outlook released 3 weeks ago or so.
Here’s what I wrote:
In general, gold behaves like a very-long-duration inflation-linked bond with a zero coupon. This makes sense – if we were to issue a bond that, in exchange for the current gold price, offered to pay the bearer no coupons but redeem for 1 ounce of gold in 100 years, it would have the same payoff as holding one ounce of physical gold for 100 years. If gold is a true inflation hedge over time, which means its price rises with the price level, then that bond would have the same payoff if we defined the payoff not in terms of an ounce of gold, but in terms of the change in the price level over that 100 years. And that would be a 100-year zero-coupon TIPS bond.
So, we tend to see over time that the spot gold price tends to track pretty well to the implied price of a zero-coupon TIPS bond. The chart above (Source: Bloomberg and Enduring calculations) illustrates the stark divergence which started roughly when the Fed began its tightening campaign. We do not have a very good explanation for this divergence, other than to postulate a clientele effect in that perhaps gold investors are more animated by inflation and TIPS investors tend to be less-excitable institutional owners. Whatever the cause, at the moment gold represents a TIPS bond that is yielding about -2.25% real yield, or roughly 435bps expensive. Unfortunately, relative value observations like this have no mechanism to force them to close, so we cannot recommend selling gold and buying TIPS as an arbitrage. However, we are comfortable saying that investors could create a significantly better-performing commodity index by leaving gold out of the index, or by replacing gold in the index with a TIPS bond. Call Enduring Investments if you are interested in creating such an investment!
At the time, I wasn’t aware (because I don’t track gold flows – gold to me is just another commodity, albeit one that has a very high real duration and a pretty low inflation duration) that China has been buying gold consistently for a year and a half. That only became apparent to me recently when news stories highlighted that China has stopped the accumulation for now. Here is a chart from Bloomberg of China’s monthly gold reserves. It certainly seems as if the timing of the Chinese purchases correspond reasonably well with the divergence in the first chart above.
However, I am not sure that’s the true reason although it is probably a contributor. If you back up and look at China’s reported reserves over the entire period covered by the first chart above, you can see that the recent increase is the largest since 2015…but certainly not the largest on record. Even if the jumps on the chart are due to less-regular reporting updates, the overall rate of increase prior to 2016 was not dissimilar to that of the last 18 months. And yet, that buying did not cause a divergence of any meaningful amount on the first chart above.
So I am back to thinking that this is a broader clientele effect, of people who responded to the biggest spike in inflation in 40 years by buying an asset that historically has sort of a meh history of protecting against inflation over short or medium periods, and a much clearer history of large yield sensitivity. If that’s the case, then while there’s no trigger for closing the gap we should expect that the gap will, eventually, close. Which preserves the implication I mentioned in the Quarterly Inflation Outlook: prudent investors should consider lightening the allocation to gold in their commodity allocations.
As an aside, the title of this piece comes from a song by the Doobie Brothers that I can’t get out of my head now, and hopefully neither will you!
