The Phillips Curve of crypto: token inflation, staking yields, and network activity

In 1958, A.W. Phillips observed an inverse relationship between wage inflation and unemployment in the United Kingdom. Higher inflation coincided with lower unemployment. Crypto ecosystems face a strikingly parallel question. Can token inflation (new emissions) buy network activity? The data from multiple blockchain cycles suggests a nuanced answer: yes, temporarily; no, permanently. Exactly the conclusion Milton Friedman and Edmund Phelps reached about the original Phillips Curve in the late 1960s.

The short-run relationship

When a new DeFi protocol launches with aggressive token emissions, activity reliably follows. Compound's COMP distribution in 2020 kicked off "DeFi Summer" by paying users to borrow and lend. TVL surged from millions to billions. Users who would never have interacted with lending protocols suddenly appeared, attracted by token rewards that subsidized the cost of participation.

The pattern has repeated across hundreds of protocols and multiple chains. Avalanche's Rush incentive program in 2021 attracted billions in TVL. Arbitrum's ecosystem saw explosive growth post-airdrop. In each case, token emissions preceded activity growth.

This mirrors the short-run Phillips Curve mechanism. Just as inflationary monetary policy stimulates real economic activity by reducing real wages and increasing hiring profitability, token emissions stimulate onchain activity by subsidizing participation and reducing the effective cost of transacting.

The mechanism works through two channels. Direct incentives pay users in tokens to provide liquidity, stake assets, or execute transactions. Speculative anticipation attracts mercenary capital from users who expect future token distributions. Both channels produce measurable activity. The protocol's dashboard metrics look excellent.

The long-run breakdown

Friedman and Phelps argued that the short-run Phillips Curve tradeoff was illusory. Workers eventually adjust their inflation expectations. Once they demand wages that account for expected inflation, the stimulative effect disappears. The economy returns to its natural rate of unemployment, now with higher inflation. The long-run Phillips Curve is vertical.

Crypto has its own expectations-adjusted breakdown. Yield farmers and airdrop hunters are sophisticated actors who adjust rapidly. When Compound distributed COMP, early participants earned outsized returns. Within weeks, competition drove real yields down. Within months, "yield tourism" became established: capital moved from protocol to protocol chasing the highest subsidized returns, staying only until incentives declined.

The data on post-incentive retention is sobering. Studies of DeFi protocols that reduced or eliminated token emissions consistently show sharp TVL declines. The "natural rate" of activity, the amount sustained by genuine utility without emission subsidies, is frequently 70% to 90% lower than the incentivized peak.

Layer 2 networks show the same pattern at chain level. Chains that attracted billions through incentive programs saw significant capital outflows as programs ended and users migrated to the next subsidized opportunity. The activity was real but rented.

Measuring the crypto Phillips Curve

Constructing a crypto analog to the Phillips Curve requires defining the axes. The vertical axis maps token emission rate (annual inflation percentage). The horizontal axis maps some measure of real network activity: daily active addresses, non-incentivized transaction count, or organic protocol revenue.

In the short run (0 to 6 months): higher emissions correlate strongly with higher activity. The slope is steep and consistent. More inflation buys more activity.

In the medium run (6 to 18 months): the relationship weakens. Users internalize the inflation cost (dilution) and demand higher rewards to maintain participation. Protocols enter a treadmill: they must increase emissions just to maintain current activity levels.

In the long run (18+ months): the relationship inverts for protocols without genuine utility. Protocols that emit heavily but lack sticky use cases see activity decline despite continued emissions. The market has priced in the dilution.

Natural rate of network activity

Friedman introduced the concept of a "natural rate of unemployment" determined by real structural factors. Crypto protocols have an analogous concept: the natural rate of network activity, the level sustainable without token emission subsidies.

For Ethereum, it's the transaction volume driven by genuine DeFi activity, stablecoin transfers, and application usage independent of ETH staking rewards. For a DeFi protocol, it's the TVL and fee revenue generated by users who find the product useful regardless of token rewards.

Estimating the natural rate requires isolating organic activity from incentivized activity. One approach: measure the activity that persists after emissions drop to their long-term steady-state rate. Protocols that maintain 50%+ of peak activity after emission reductions likely have genuine product-market fit. Those that retain less than 20% were running on subsidy.

For investors, the gap between current activity and estimated natural rate is a critical metric. A protocol trading at a valuation based on incentivized metrics is priced for activity that will not persist. The valuation premium above natural-rate-justified levels represents emission-dependent value that evaporates when subsidies decline.

Adaptive emission strategies

Front-loaded emissions followed by sharp reduction accept the short-run tradeoff: use high emissions to bootstrap a critical mass of users, then cut emissions sharply and retain those who find genuine value.

Performance-based emissions tie token distribution to metrics that correlate with genuine usage rather than pure capital deployment. Rewarding transaction count, unique user growth, or protocol revenue rather than TVL alone selects for active participation over passive farming.

Retroactive rewards (like Optimism's RetroPGF) wait until value is demonstrated before distributing tokens. Rather than running expansionary policy and hoping activity follows, they observe the natural rate first and reward it after the fact.

Investment implications

Discount incentivized metrics. A protocol with $1 billion in TVL running at 30% annualized token inflation is in a very different position than one with $500 million in TVL running at 3% inflation. The second protocol's activity is more likely to persist.

Watch for the expectations adjustment. Early movers in new emission programs earn outsized returns. By the time high yields are widely known, yield-adjusted returns have already compressed.

Evaluate emission reduction plans. Protocols with published, declining emission schedules have committed to eventually revealing their natural activity rate. Those running open-ended programs are deferring this reckoning indefinitely.

Related Concepts

• Emissions vs revenue: Separating subsidized growth from organic demand
• Real users vs subsidized activity: Distinguishing genuine adoption
• DeFi liquidity trap: When yield farming incentives stop working entirely
• Airdrop economics: Who pays when tokens are distributed
• Monetary policy of Layer 1s: How emission schedules shape token economics
• Payback period for token incentives: Measuring ROI on emission spending
• Ponzinomics vs real yield: Identifying sustainable vs unsustainable models