Policy Sketch · April 2026

The Automation Dividend.

A city-level levy on autonomous ride revenue, large enough to pay 4 million drivers out of the surplus their replacement creates. The numbers work. Here they are.

~12 min read Interactive model Policy sketch
§ 01The precedent problem

The precedent problem

Driverless fleets arrive in the next decade. The first city that handles the displacement well will set the template for every automation wave after it. The first city that bungles it will do the same.

The United States has about 4 million professional drivers. The UK has 360,000 licensed taxi and private hire operators. Their skill is about to become worthless, on a schedule they didn't pick.

4.0M ~0.3M 2026 2046
Fig. 01 · Professional drivers in the US, projected. ~92% erosion over the transition window.

Slowing the rollout isn't the answer. Autonomous vehicles, even on conservative estimates, could prevent tens of thousands of road deaths a year. Every year of delay is paid for in funerals.

The problem is where the surplus lands. Lower costs, higher utilisation, fewer crashes: the value flows to the fleet operators and the riders. Drivers built the market the fleets are colonising, and they eat a concentrated loss while the gains diffuse across millions of people who will never notice.

Diffuse gains, concentrated pain. That's the standard shape of technological displacement, and if we let it play out on default settings for drivers, we've written the script for warehouse workers, paralegals, and radiologists.

So don't slow the cars. Build a financing mechanism that moves a slice of the new surplus to the people losing their jobs, fast enough and visibly enough that the first displacement wave becomes the argument for the next one instead of the warning against it.


§ 02The expanding pie

The pie gets bigger

The usual frame treats every autonomous ride as a human ride stolen. Fixed pie, smaller slice for the driver. That frame is wrong, and the whole financing argument hinges on it being wrong.

When prices drop and availability hits 24/7, demand doesn't stay put. The 2am trip no one took because surge pushed it to £40. The grandmother who stopped going to bridge night when she lost her license. The teenager stranded in a suburb with no bus. The household that sells its second car. The delivery that only pencils out at an AV price point. None of these rides exist today.

Yr 0 Yr 8 Yr 20 inflection
Fig. 02 · Total ride market vs. today, logistic. 2–5× expansion is the revenue base the levy taps.

Reasonable estimates put the AV-era ride market at 2–5x the current human-driven one. That's the revenue base the levy draws from: not the taxi market as it stands, but a much larger market that only exists because the technology exists.

the key move
The transition fund is paid for out of value that wouldn't exist without AVs. Nobody's current wallet is raided. It's a redistribution of surplus the displaced workers helped create the market for.

§ 03The mechanism

Do it at the city

A national automation tax is politically dead on arrival and administratively miserable. Cities are the right jurisdiction. They already license taxi medallions, run congestion charging, and grant utility franchises. Adding AV fleets to that list is not an imaginative leap.

The rule: any company running an autonomous fleet inside the city pays a percentage of gross ride revenue into a ringfenced Driver Transition Fund. Percentage, not per-ride and not flat. A percentage scales with the market automatically, it doesn't plant a regressive floor under the per-trip price (the price drop is the whole point), and it audits itself: one number on one P&L line.

AV surplus, allocated Operators · 54% Riders · 36% Levy · 5% → drivers
Fig. 03 · Where the new surplus lands. A 5% slice funds the transition without touching operator margin or consumer surplus much.

Somewhere between 3% and 8% of gross does the work. Below 3% the fund is cosmetic. Above 8% you start eating the consumer surplus that made AVs attractive in the first place. The model further down lets you push on the number yourself.


§ 04The displacement credit

The displacement credit

Any driver who can show prior earnings in the affected market gets a monthly Displacement Credit. Direct cash. No retraining requirement, no jobs board, no caseworker.

Set the credit at 60–70% of the driver's prior monthly earnings. Full replacement is tempting and wrong: a check that matches your old paycheck dulls the incentive to find the next thing, and most drivers will want to work. The gap between the credit and a normal wage is where the rest of the economy recruits them back in.

The credit tapers. For a defined window — say eight years at the full rate, then a five-year linear taper — payments land regardless of how fast AVs actually roll out in the driver's city. The drivers get planning certainty; the city gets a bounded liability.

credit formula
monthly credit = prior earnings × replacement rate × phase factor

Phase factor is 1.0 during the full-payment window, then decays linearly to 0 across the taper. Everything else — the replacement rate, the window length, the taper — is a policy dial you can turn in the model below.

Drivers take the credit as cash, or spend it on accredited retraining, or convert it to a small-business grant. Their call. Mandatory retraining programs have a long and embarrassing track record; unconditional cash is the version that respects the recipient.


§ 05Pulling the future forward

Borrow against the future

Here's the timing problem. The levy revenue grows with the AV market over twenty years, but the pain is worst in the first five, when the fleets are small, the levy is small, and drivers are losing shifts anyway. A proportional fund pays drivers least exactly when they need it most.

Fix it with transition bonds. The city issues municipal bonds secured by projected future levy revenue. The bond cash pays generous credits in years 1–10. The levy revenue in years 10–30 services the debt.

This is the plainest kind of municipal finance. It's how cities already pay for tunnels, stadiums, and water systems. The future AV levy is arguably a better-secured revenue stream than most of those, because AVs' cost advantage over human drivers makes sustained market growth close to inevitable once the fleets are legal.

Front-loading is what makes the politics work. Drivers don't watch a payment schedule that starts at $200 a month and grows as the replacement rate kills their income. They see a real check on day one. "Technology threw me out" becomes "technology bought me five years to land somewhere." That shift is the entire point of issuing the bonds.

Get the first displaced workforce visibly taken care of and the public reframes automation from threat to expected benefit. Bungle it and every subsequent wave faces a harder fight than it should.

§ 06Why the operators should want it

Why the AV companies should like this

Obvious objection: why would Waymo or Zoox or anyone else agree to pay a revenue tax? Because it's the cheapest political insurance they can buy.

AV fleets live and die at the pleasure of city councils. One council vote bans operations. One viral story about a bankrupt driver triggers regulation in six other jurisdictions. The history of technology rollouts is crowded with cases where public backlash, fair or not, delayed deployment by years.

Against that downside, a 5% levy is a rounding error. It buys the operator an enforceable social license with a named beneficiary — the drivers receiving checks — who will defend the arrangement in public when the next moral panic starts.

The franchise structure also helps incumbents. A known 5% levy in a regulated market is strictly preferable to an unregulated market that might get wrenched out from under you after the capex is committed. First movers who accept the deal get a stable legal environment. Latecomers pay the same 5% without the goodwill.

Run the numbers yourself

Push the sliders. Break the policy. Pick a scenario to jump to a preset, or roll your own. Figures are per-city, 30-year horizon.

Market dynamicsHow big the pie gets, and at what price

Current annual ride market$4.0B
Today's ride-hail + taxi revenue
mid-city ≈ $3B
AV-era market multiplier3.0×
Long-run size vs. today, pre-elasticity (1× = fixed pie)
plausible 2-4×
Consumer fare drop at maturity45%
Headline fall in per-ride fare vs. today
est. 40-60%
Price elasticity of demand1.20
Demand growth when fares fall
urban ≈ 1.0-1.4
Peak AV share of rides85%
Equilibrium — may not be reached if operators exit
bullish 80%+
Years to peak share15 yrs
Time constant of the S-curve
analyst ≈ 12-18
S-curve steepness1.00×
Soft ramp vs. cliff-edge adoption
1.0 = default

Policy leversThe knobs a city council actually turns

Levy rate on AV gross revenue5.0%
0% = do-nothing baseline
viable 3-8%
Levy pass-through to consumers40%
% of levy pushed onto fares → shrinks the pie via elasticity
typical 30-60%
Operator margin floor12%
Below this, operators stop expanding; adoption stalls
~10-15%
Fund admin overhead8%
% of levy receipts lost to running the fund
well-run 5-10%
Jurisdictional leakage5%
Revenue routing to neighbour cities (worsens with high levy)
low ≈ 5%

Driver poolWho gets displaced, and where they go next

Licensed drivers at t=080,000
Size of the affected workforce
NYC ≈ 150K
Avg. annual driver income$38K
Pre-displacement earnings
US rideshare ≈ $38K
Baseline attrition5%/yr
Drivers leaving anyway, AV or not
turnover ≈ 5-8%
Re-employment rate10%/yr
% of displaced drivers finding ≥ prior income / year
studies 8-15%

Fund financeHow generous the credit is and how it's financed

Replacement rate65%
% of prior income the credit pays
UI analog ≈ 50-70%
Years at full replacement8 yrs
Years before the taper starts
5-10 typical
Taper years5 yrs
Years over which credit falls to zero
3-7 typical
Muni bond coupon4.50%
Cost of front-loading credits via transition bonds
AA muni ≈ 4-5%
What this run looks like Drag a slider or click a scenario to see the story.

What comes out

30-yr levy
collected
30-yr credits
paid
Per-driver
credit, yr 1
Per-driver
credit, yr 5
Per-driver
credit, yr 15
Peak fund
balance
Trough fund
balance
Operator margin
at peak adoption
Fare vs.
no-levy baseline
Adoption actually
reached
Fund insolvent
Industry stall
Broken promise
Revenue flight

30-year state of the system

AV revenue (levy base)
credits paid
cumulative fund
operator margin (right axis)
failure year
§ 08The template

Drivers first, then everyone else

Drivers are the first automation wave most people will see clearly. Warehouse pickers, line cooks, paralegals, radiologists, and call-center staff are already queued behind them. Each displaced group will face the same question: does the surplus from the technology that replaced you reach you?

If the answer for drivers is yes, in public, with real checks, the rest of the displacements inherit a working template. The question stops being "should we compensate" and starts being "at what rate, for how long." That's a much more productive argument.

If the answer for drivers is no, every subsequent wave gets harder. Backlash compounds. Regulation turns reactive and punitive. The technology arrives anyway, because it always does, but later and more chaotically, and with a much longer list of people who got nothing.

The model above is the whole case in one screen. The surplus exists. A single-digit percentage of it pays the bill. Nobody currently breathing has to take a pay cut for this to balance.

That's not charity and it isn't traditional redistribution. It's finishing a Pareto improvement that the market, left alone, stops halfway through.

The policy is easiest to pass before the first driver loses their route. It gets harder every week after.