When a country holds dollar reserves, it holds a claim on an asset controlled by another sovereign. That means sanctions exposure, policy risk, and the long-run risk that the issuer's fiscal decisions erode purchasing power. Gold has no issuer. No sovereign controls it. For institutions managing national balance sheets, that's a meaningful structural tradeoff. This is what makes the current data worth taking seriously. A World Gold Council survey found 95% of central banks expect to grow their gold reserves in the next 12 months. China's PBoC has added to its holdings for 16 consecutive months. These are the same institutions that built and still publicly defend the dollar reserve system. Russia's $300 billion in frozen reserves after 2022 was a live demonstration of what that counterparty risk looks like in practice. The gold accumulation trend accelerated notably after that point. Whether that's causation or coincidence is worth sitting with — but either way, you now have institutions quietly building a parallel position that performs better if confidence in the system they manage weakens. What does it mean that the behavior and the official posture have diverged this cleanly?

A strong dollar isn't just good news for Americans. It's a crisis signal for half the global economy. When the DXY — the dollar index — surges, countries that borrowed in dollars face a mechanical problem. Their local currency revenue shrinks relative to fixed debt obligations. This isn't new. It's happened at least three times in modern history, with nearly identical mechanics each time. 1982. 1997. 2018. Same pattern. Same mechanism. Different countries.

When institutional credibility is the primary asset, deferral becomes the dominant policy tool. Not delay out of incompetence — structured delay, signaled through mechanisms that preserve optionality: quarterly dot plots that can be revised, legislation with sunset clauses, guidance frameworks that can be recalibrated on short notice. This week offered a clean illustration. Supercore services inflation printed near 4%, a number that forecloses Fed easing without a credibility cost. Congress passed a CBDC ban — framed as strong opposition to a digital dollar — that expires in 2030. Both moves follow the same logic: commit to a position that's politically defensible today, and push resolution to a future date when conditions might be different. The question worth asking is whether deferred decisions actually improve conditions — or compound the pressure they were designed to avoid. Monetary institutions that have relied heavily on credibility management over structural adjustment have generally found the cost of eventual resolution higher than the cost of earlier action. Which pattern is this one following?

Most arguments about dollar dominance focus on reserve holdings and Treasury demand. There is a deeper layer: the Fed's swap line network. When dollar funding seizes in a crisis, foreign central banks cannot print dollars. They can draw down reserves, but reserves run out. What they can do is call the Fed. The Fed creates dollars on demand and swaps them temporarily for the requesting central bank's local currency. No other institution can do this at global scale. The euro, yen, and yuan have no equivalent backstop. This means the US is not just the issuer of the world's preferred reserve currency. It is the lender of last resort for the entire global dollar system. In 2020, when dollar funding markets seized, the Fed activated swap lines with 14 central banks. Markets stabilized almost immediately. Not because traders suddenly trusted America more, but because there was only one institution capable of supplying dollars without limit. If a country or bloc wanted to meaningfully reduce dollar dependence, they would not just need an alternative reserve asset. They would need an alternative emergency dollar supplier. What would that even look like?

In 1960, economist Robert Triffin told Congress the dollar's global dominance contained a structural guarantee of self-defeat. Not a warning. A guarantee. The US would have to run persistent deficits to supply the world with dollars — and those deficits would eventually erode the very credibility that made the dollar worth holding. By 1971, US gold had fallen from $17.8 billion to $10.5 billion backing over $65 billion in foreign claims. Nixon suspended convertibility. Today: $900 billion annual current account deficit, $7.5 trillion in foreign Treasury holdings, dollar reserve share down from 72% in 2001 to 59%. The dilemma Triffin identified in 1960 is still running — just without the gold floor. Read the full analysis:

The Triffin Dilemma is Still Running

Most CPI analysis anchors on the headline number. But the component the Fed actually uses to determine whether underlying demand pressure has broken is services ex-shelter — supercore. It strips out goods deflation, energy volatility, and lagged shelter surveys to isolate what wage-driven service inflation is doing in real time. February's reading: 4.0% year over year. Supercore is sticky because it reflects domestic wages and spending rather than global supply chains or rental contract cycles. Services — healthcare, restaurants, insurance, personal care — move with employment conditions. A 4% reading suggests those conditions haven't shifted in any meaningful way, even as goods and energy have moderated. The market repriced June cut odds down to 60% after this morning's print. That's a reasonable short-run adjustment. But the more important question is whether 4% supercore is a temporary stall in a longer disinflationary trend — or something closer to the structural floor of where this economy runs given current labor conditions. If it's the latter, the Fed's optionality for the rest of 2026 is narrower than current pricing implies. What would it take — which reading, over how many months — for you to conclude the structure itself has changed?

CPI dropped this morning at 8:30 AM. Markets reacted to the headline. But one third of that number is telling you a story about rent prices from late 2024. Shelter makes up 33% of the Consumer Price Index. The Bureau of Labor Statistics doesn't track current rent. It tracks rent agreements signed 12 to 18 months ago. That lag is structural, not a bug. And it means headline CPI is measuring inflation that already happened.

The Fed and financial markets watch the same CPI data and often reach different conclusions. That's not a communication problem — it's a measurement one. Markets have anchored on headline year-over-year CPI as the primary signal. The Fed's attention has shifted increasingly to core services ex-shelter, sometimes called supercore — the component that strips out food, energy, and lagged housing costs to isolate wage-driven inflation in labor-intensive service sectors. Supercore moves slowly, responds poorly to rate hikes, and rarely makes headlines. When the two measures diverge — headline falling while supercore stays elevated — markets and the Fed are effectively pricing two different economies from the same release. The market reads a soft print as confirmation that rate cuts are coming. The Fed reads the same data as persistent wage pressure that forecloses them. Both interpretations are internally consistent. The more durable question isn't what any single print shows. It's whether supercore is structurally converging with headline as goods deflation fades — or whether the gap has become a permanent feature of how post-pandemic inflation gets measured and interpreted.

In 1960, economist Robert Triffin identified a structural contradiction at the heart of the Bretton Woods system. The country that issues the global reserve currency must run persistent deficits to supply the world with its currency. But those same deficits gradually erode the credibility that made the currency worth holding. He called it a dilemma because there was no clean solution, only a slow accumulation of tension until the system resets. Nixon closing the gold window in 1971 was that reset. The dollar's gold convertibility broke precisely because the US had supplied too many dollars relative to its gold reserves. The mechanism Triffin described played out eleven years after he published it. The current debate about dollar dominance follows the same structural logic. The US still supplies the world's reserve currency. It still runs persistent deficits to do so. The dollar is no longer backed by gold, but it is backed by something just as fragile: confidence. And confidence, unlike gold, doesn't have a fixed price. If Triffin's framework holds, the question isn't whether this tension resolves. It's what the resolution looks like this time, and whether it happens through gradual diversification, a sharp break, or something the current consensus isn't pricing in.

The Fed's rate decision framework and the headline CPI number it follows are measuring different things. Headline CPI captures food, energy, and shelter — all of which have structural reasons to lag or distort the current picture. The Fed's actual focus is services inflation excluding shelter, called super-core: what remains after stripping out the volatile and the slow-moving. That is the real-time read on labor-intensive service inflation. Super-core is sticky because wages drive it. Service businesses — healthcare, personal care, financial services — do not have commodity input exposure the way goods producers do. Their primary cost is labor, and labor costs move slowly. A tight employment market keeps those costs elevated even as goods prices fall. Headline and super-core can, and often do, move in opposite directions. The implication is structural: a soft headline print can produce a market reaction that prices in rate cuts the Fed has no mechanism to deliver yet. The Fed's rate path follows super-core, not the headline. Until super-core actually softens, the rate cut timeline does not move — regardless of what the top-line number shows. Most post-release analysis will track the wrong variable.

Sovereign reserve management has a structural incentive problem. Physical gold carries no counterparty risk — it settles without permission from any government. Dollar-denominated reserves do. After 2022, when Russia's reserves were frozen, that distinction became impossible to ignore for central banks globally. That explains why central banks averaged roughly 27 tonnes of gold purchases per month throughout 2025. Not a gold thesis — a counterparty risk hedge. The structural incentive was real, and the buying pace reflected it. But buying at scale has a natural ceiling. Gold hit record highs heading into 2026. Institutions with fiduciary mandates don't chase all-time highs. They buy on weakness, or when the cost of inaction outweighs price risk. January 2026: purchases dropped to 5 tonnes. The harder question isn't whether the pace will resume. It's whether January marked price discipline at work — in which case buying resumes on a correction — or whether the diversification push has hit operational, legislative, or political constraints that make 27 tonnes monthly unsustainable regardless of price. Those are two different stories with very different implications for what sovereign reserve diversification actually looks like over the next decade. Which explanation do you think fits better?

The petrodollar isn't a formal treaty requiring oil sales in dollars. It never was. The actual 1974 agreement was simpler and more structural: Saudi Arabia would price oil in dollars and invest the proceeds in U.S. Treasuries. The mechanism mattered more than the mandate.

Rate cuts are reactive policy, not proactive generosity. The Fed cuts when conditions require it — either the economy is weakening or inflation has fallen enough to create room. Neither scenario is inherently a green light. The mechanism most people skip: the underlying condition that triggers the cut often matters more than the cut itself. In 2001 and 2007-08, cuts came steadily while asset prices kept falling, because the deterioration outran the policy response. The cuts were real. So was the pain. There's a structural reason markets celebrate cuts regardless. Asset holders genuinely benefit from lower rates — higher valuations, cheaper financing, multiple expansion. That's a rational preference. But it creates pressure to interpret the cut as unambiguously good news, when the condition producing the cut is the more important signal. The question isn't whether cuts are coming. It's what conditions are producing them — and whether markets are pricing the cut or the cause.

Everyone watches the March FOMC meeting for the rate decision. Cut or hold. One word, one number. But that framing misses most of what the meeting actually contains. Four times a year, the Fed publishes a Summary of Economic Projections alongside the rate decision. This is the dot plot, plus GDP estimates, inflation forecasts, and unemployment projections. Each of the 19 officials submits their own independent forecast, and the aggregated result is the closest thing markets have to the Fed's real-time model of the economy. When rates stay flat, the SEP can still move significantly. And when it does, that movement carries more information than the vote. The reason is mechanical. The rate decision answers: what did they do today? The SEP answers: what does the committee collectively believe about the next 12 to 24 months? A hawkish hold, where rates stay flat but the median dot moves up, signals less room to cut than markets had priced. A dovish hold sends the opposite message. September 2023 is the clearest recent case: rates unchanged, dots shifted higher, and yields climbed sharply after. The vote said nothing new. The SEP said everything. Most FOMC post-mortems analyze what the Fed did. The more useful read is what the committee revised — which projections moved, by how much, and in what direction. That is where the committee's current read on the economy actually lives. The vote is almost always the wrong variable to track.

When a policy gets reversed after businesses have already adapted to it, the reversal is not the same as the original policy never existing. The system moved on. The Supreme Court striking down Trump-era tariffs forces roughly $166 billion in refunds to around 330,000 importers — but the businesses receiving that money already repriced their goods, restructured their sourcing, and passed costs downstream. Those decisions are embedded. The refund returns capital. It does not return the conditions that preceded the tariff. That matters because capital injections and policy reversals behave differently. The money flows back to the firms that paid the most, not to the consumers who absorbed the markups. Prices tend to be sticky downward. Supply chain decisions tend to stay made. What looks like a correction on paper functions more like a balance sheet event in practice. The question worth tracking is where $166 billion actually redeploys — and whether it creates fresh price pressure in sectors that already repriced once.

Shelter is roughly one-third of CPI, but it is not measured from actual rents. The Bureau of Labor Statistics uses Owner's Equivalent Rent — a survey asking homeowners what they think they could charge to rent their own home. That number lags real rental market conditions by 12 to 18 months. The mechanism creates a structural timing problem. When real-world rents peaked in late 2022 and began cooling, that shift did not appear in CPI shelter readings until well into 2024. Policymakers reading shelter as a current signal were reading last year's conditions. The disinflationary trend they were crediting to policy had already been happening in the underlying market for over a year before it registered. The components that reflect conditions closer to the present are core services ex-shelter — insurance, healthcare, personal care — and core goods, which have been disinflationary for some time. Shelter is still catching up. The headline number Wednesday will move markets, but which direction it moves may tell you less about current inflation than about where rents were in 2023.

Everyone watches for yield curve inversion. Almost no one watches for the uninversion. That's the mistake.

Two surveys go into the monthly jobs report. One asks employers how many people are on payroll. The other asks workers directly whether they have a job. They often agree. When they don't, that gap is worth understanding before you interpret the headline. In the February 2026 report, these two measures are diverging. The establishment survey is the one that generates the headline number, and it runs through the birth-death model — a statistical adjustment that estimates jobs from businesses assumed to have formed since the last benchmark. Those businesses have no payroll records yet. The model adds them anyway, and the headline absorbs them as if they were counted. The model is structured this way because new firm formation is hard to track in real time, which means the adjustment is systematically optimistic during expansion phases. Markets react to the headline. The composition gets analyzed later. That lag is where the mispricing tends to live — not in the number itself, but in what the number is made of. The revision, not the release, is usually where the actual read lives.

Tariffs are usually analyzed as trade instruments. They restrict imports, protect domestic producers, adjust comparative advantage. That framing misses something. When tariffs pass through to consumer prices, they function as a monetary event. The imported goods get more expensive. That is inflation at the point of consumption. At the same time, the uncertainty created by tariff escalation tends to drive capital into dollar-denominated assets. Safe-haven demand lifts the dollar index. So you get a stronger dollar and rising consumer prices simultaneously. That looks contradictory on the surface, but it follows cleanly from the mechanism. The problem is that conventional metrics read currency strength as a health signal. A stronger dollar is supposed to mean tighter conditions, reduced import costs, and stable purchasing power. When that signal is produced by a mechanism simultaneously eroding purchasing power, central banks are working with a distorted map. The nominal and real pictures diverge. That divergence has unequal consequences. Dollar-denominated asset holders see nominal gains. Workers and consumers absorb goods inflation. Same policy event, different outcomes depending on where you sit. The question worth tracking is whether policy can be calibrated around a signal that structurally means two different things to different parts of the economy at the same time.