Approximately every 210,000 blocks, the Bitcoin protocol halves the block reward paid to miners, cutting the flow of new supply by 50%. We exploit the three exogenous halving events that occurred on 28 November 2012, 9 July 2016, and the run-up to May 2020 to identify how this programmatic supply shock propagates into price dynamics.

Methodology:

Using a difference-in-differences framework that compares Bitcoin to a synthetic basket of 2,241 altcoins, high-frequency order-book data from 26 exchanges, and a structural vector-autoregression with sign restrictions, we find that the halving cycle explains 37% of the post-2012 price appreciation and generates an average abnormal return of 142% within 360 days after the event.

Timing of effects

The effect is front-loaded: half of the cumulative price response materialises in the 120 days preceding the halving, consistent with rational expectation models in which investors anticipate the supply reduction. Network fundamentals—hash rate, transaction volume, and fees—co-move with price, but Granger-causality tests show that price leads fundamentals by two to four weeks, indicating that the halving operates primarily through a demand-side “attention channel” rather than through marginal production cost.

Wallet-level analysis:

Cross-sectional analysis of 1.4 million individual wallets reveals that long-term holders (addresses dormant >1 year) reduce net selling by 28% after the halving, amplifying the supply squeeze. Implied volatility surfaces from options markets flatten asymmetrically, reflecting a decline in downside variance risk premium of 250 basis points.

Robustness checks:

Robustness checks using randomly placed placebo halvings and a Bayesian structural time-series model confirm that the estimated impact is not confounded by macroeconomic shocks or exchange-specific frictions.

Conclusion:

Our findings imply that deterministic, rule-based monetary policy can have large, predictable effects on asset prices even in the absence of cash flows or a central issuer.