The Institutional Series: Staking Rewards and the Rate Cycle
Why the direction of travel matters more than the level
Last week I said this piece would look at staking rewards against the fixed income alternatives in a rate cycle. We now have a situation where the Federal Reserve raised its target range on Wednesday by 25bps to 3.75%–4%, a unanimous vote and the first increase since 2023, with more signalled. The ECB lifted its deposit rate to 2.5% the week before. This morning the Bank of England held Bank Rate at 3.75% by a majority of six to three, with the three dissenters voting for an immediate rise to 4%. August CPI came in at 3.1%, and the committee's own language is that inflation is likely to rise further over coming quarters.
The driver is an energy shock rather than demand. Crude and refined energy prices have risen again since the last meeting, with Brent at $106 a barrel on 14 September and UK wholesale gas up 78% since July. That matters for how long this persists and how central banks respond to it. It does not change the fact that confronts an allocator: the risk-free benchmark is flat to rising in dollars, sterling and euros simultaneously.
If we look at what’s happening on the staking reward side, seven-day Ethereum staking APR has fallen to 2.46%, down from a peak of 5.06% in June 2023, a decline of more than half in three years. Six weeks ago it stood at 2.66%. Net of operator fees, an institutional allocator is realistically underwriting something around 2%, with MEV adding perhaps half a point to a point on top.
So an allocator running the standard comparison this quarter sees a benchmark near 4% and rising, against a gross staking reward near 2.5% and falling. A gap that has widened from both ends at once, in every major reporting currency.
Two rates, two generating mechanisms
A logical question is why the comparison is behaving in this way, and the answer is that the two numbers are produced by processes with nothing in common.
A policy rate is a reaction function. It is the output of a committee responding to inflation, employment and financial conditions, and it is forecastable to the extent that you can forecast those inputs and the committee's response to them. That is the entire discipline of rates strategy, and it is a mature one. The 6–3 split on the MPC this morning is a live example: the disagreement is about second-round effects in pay and prices, and it is legible to anyone who follows the data.
An Ethereum staking reward is not a reaction function. It is the output of an issuance schedule. Issuance scales inversely with the square root of total staked ETH, so the more validators join, the smaller the per-validator share becomes. Transaction fees and MEV sit on top, driven by network activity. No part of that responds to CPI, to the labour market, or to anything a central banker says. Forecasting it requires a different model with different inputs, and there is no macro overlay that helps you.
Institutions routinely put these two numbers side by side as though they were comparable quantities differing only in magnitude. They are not. One is the price of money set by a policy authority. The other is a subsidy paid by a protocol to secure itself, priced by how many people want to provide that security.
You are in your own denominator
There is a further feature of staking rewards with no equivalent in fixed income, and it deserves more attention than it gets from people selling staking services.
The compression in staking rewards over the past three years is not a market accident. It is the direct arithmetic consequence of institutional capital arriving. Staked ETH reached 43.2 million by mid-September 2026, roughly 35% of circulating supply and an all-time high, driven substantially by yield-distributing ETFs and corporate treasuries. More stake, lower per-validator reward. The curve did exactly what it was designed to do.
For an allocator this means something genuinely unusual: your own flows are an input to the return you are underwriting. If your investment thesis is that institutional adoption of staking will accelerate, then your thesis contains a prediction that your realised reward rate will fall.
No fixed income instrument behaves this way. Buying a gilt does not lower the coupon on the gilt. Any multi-year staking reward projection that does not model participation growth is an extrapolation, and it will be wrong in a direction you can already identify.
The comparison committees run is not the decision they face
I often hear discussions on the comparison of staking reward versus Treasury yield within the institutional space. It is the natural framing, because it is the one the governance process already has language for.
It is the wrong comparison, for a reason that is straightforward once stated. A staking reward is denominated in the staked asset. A Treasury yield is a nominal return in the reporting currency with principal protection. An allocator choosing between them is not choosing between two yields, they are choosing between two entirely different exposures, and the yield differential is close to irrelevant to that choice.
The decision that actually presents itself is narrower. For an institution that already holds the asset, for whatever reason it decided to hold it, the live question is staked versus unstaked. And that comparison has a different character, because not staking is not a neutral position. Proof-of-stake issuance runs at roughly 2,800 ETH per day at recent validator counts, with annualised supply growth of about 0.85%, a reversal of the burn-dominant period of 2022 to 2023. A holder who does not stake is diluted by the issuance paid to those who do.
Framed correctly, staking is closer to a dilution offset than to a yield. That framing survives a rate cycle in either direction, because it does not reference the policy rate at all.
Where fixed income wins
There is one dimension on which the fixed income alternative is straightforwardly superior right now, and a piece that omitted it would not be worth reading.
Duration. An allocator who believes rates are near a peak can extend duration, lock the level, and collect capital appreciation if they are right. The instrument exists precisely to convert a view about the path into a position.
The Bank reinforced that this morning. Alongside the rate decision, the MPC voted unanimously to unwind its stock of gilt holdings to zero through a multi-year plan, at an average annual pace of £46 billion to the end of 2034, including £20 billion of annual sales. More duration supply into the market is upward pressure on term yields. An allocator who wants to lock a level is being offered a better one to lock, and more of it.
There is no equivalent in staking. You cannot term out a staking reward and there is no five-year contract at today's rate, no mechanism to fix the level and nothing to hedge it with. A staking allocation earns whatever the protocol pays, recalculated continuously. For an allocator whose mandate is built around locking known cash flows over a known horizon, that is a real and structural disadvantage, and no amount of enthusiasm about protocol economics changes it.
What this resolves to
The level tells you very little. A gap of 150 basis points between a policy rate and a staking reward has almost no predictive content, because the two sides are produced by unrelated processes and will not move together in any reliable way.
The direction of travel tells you a great deal, but only if you track the right two paths and this week those paths are the point. Policy rates across all three major currencies are flat to rising off an energy shock, on committees whose dissent is running hawkish. Network participation is at an all-time high and compressing the protocol reward mechanically. Those two facts are entirely unrelated to one another, and they happen to be pointing the same way.
An institution that builds a staking allocation off the spread between them has underwritten a coincidence. The institutions that get this right will stop asking whether staking rewards beat the risk-free rate and start asking the more precise question: given that we hold this asset, what is the cost of not staking it, and what is the quality of the infrastructure through which we would.
The second half of that question is an underwriting exercise, and it is a more demanding one than most institutions currently treat it as.
Next: how to underwrite validator risk. Slashing, uptime, concentration, and the counterparty evaluation framework that institutional staking allocations require.
