Fee spikes between blocks raise short-term Bitcoin risks

A July 2026 working paper from the National Bureau of Economic Research by Fabian Schär, Dario Thürkauf and David Yermack reports that large differences in transaction fees between adjacent Bitcoin blocks are associated with more instances of competing blocks at the same height and a longer wait for the next block.

The authors analyzed block-level data from 2017 through 2025 and identify a network-level association between spikes in block fees and brief contests over prior blocks. The paper notes the result is observational and does not establish miner intent or explain the cause of any individual contest.

When one block contains substantially higher fees than its neighbors, the network shows a higher frequency of competing blocks produced at the same block height. The authors also find that after a high-fee block, the next block is less likely to appear in the first few seconds compared with a typical block. Co-author Dario Thürkauf defined a block race as “competing blocks at the same height.”

The paper explains the economic mechanism behind the pattern. Miners receive a fixed subsidy — currently 3.125 BTC per block — plus the transaction fees included in the block they produce. If a prior block carries much larger fees than the expected reward from extending the current tip, some miners may attempt to recreate that earlier block and extend from it to capture those fees, a tactic often called fee-sniping. Whether an attempt succeeds depends on hash-rate distribution, network propagation times and other miners’ reactions.

Block-level examples show how sharply fees can vary between consecutive blocks. On Aug. 26, block 964,120 included 0.0077 BTC in fees and the next block, 964,121, included 0.0536 BTC, an almost seven-fold difference. Aggregate measures put transaction fees at roughly 0.7% of miner revenue on the same date, but those averages do not capture short-lived adjacent-block extremes.

The paper describes technical and behavioural mitigations. Transactions can include lock fields that make them mineable only after a specified block height, which excludes them from a replacement of the current tip and reduces revenue available to a miner rebuilding an earlier block. The authors discuss a draft proposal, BIP 326, that would use nLockTime or nSequence to make Taproot transactions resistant to fee-sniping. The paper also reports inconsistent defaults across transaction-creation paths in Bitcoin Core: some interfaces set nLockTime near the current block height while other interfaces default to zero, leaving coverage of protective settings uneven.

The authors list specific signals to monitor before the next subsidy reduction: the size and persistence of adjacent-block fee gaps; the frequency of competing blocks at the same height; the timing of the immediate next block after a high-fee block; and the share of transactions created with protective lock-field settings. The paper recommends tracking block-level indicators and wallet adoption of lock fields to better observe periods of elevated short-term contesting.