The game is a toy, but it is not arbitrary. Every relationship in it corresponds to something the Bank of England actually worries about. Here is the whole thing, with the numbers.
You set Bank Rate once a quarter for twenty quarters — five years, the length of a Governor's term. Two numbers judge you: CPI inflation, which the Bank is legally required to hold at 2%, and unemployment, where the target band in the game is a simplification of the Bank's secondary objective to support growth and employment.
The game starts in August 2026 with Bank Rate at 3.75%, inflation at 2.6% and unemployment at 4.9%. You score for quarters where both sit inside their bands — inflation 1–3%, unemployment 3–5% — which is much harder than either alone, because the two pull against each other.
The central tension is the one A.W. Phillips identified in 1958: push unemployment down and inflation tends to rise, squeeze inflation out and unemployment rises. In the game, raising Bank Rate above the neutral rate cools prices and costs jobs; cutting below it does the reverse. There is no setting that improves both at once, only settings that balance them.
The modern refinement matters too. The trade-off is only exploitable in the short run — over time the economy returns toward its natural rate of unemployment (4.4% here) regardless of what you do to prices. Try to hold unemployment permanently below it and you buy nothing but inflation.
Bank Rate does not act on unemployment directly. It acts on output — how much the economy is producing relative to what it could — and unemployment follows from that, by the relationship Arthur Okun identified in 1962. Roughly: every 2% of output lost costs a percentage point of employment.
gap = 0.84 × gap − 1.24 × squeeze − 0.32 × |shock tail| − 2.0 × demand shock
unemployment = natural rate − 0.5 × gap
The natural rate is where unemployment settles when output is at potential. It is not a constant of nature and it is not the same in every era — it was near 3% in the post-war decades and around 7.5% after the 1980s, which is why the historical scenarios each carry their own.
Here is the part that should change how you play. A slump deep enough and long enough does permanent damage: skills go stale, people leave the workforce and do not return, and firms that would have invested never do. The economy does not get that capacity back when policy eases. Economists call it hysteresis.
if gap < −2%: scar += 0.02 × (|gap| − 2) ·
natural rate = base + scar
The scar never decays. It means over-tightening is not a mistake you can simply undo next quarter by cutting again — some of the cost is already banked, and it raises the floor under unemployment for the rest of your term. It is the reason the growth line under the two tiles is worth watching even though you are not scored on it directly.
This is not a theoretical flourish. It is the standard explanation for why UK output never returned to its pre-2008 trend.
A rate change in the game affects the economy the following quarter, never the current one. This is the single most important thing about running monetary policy and the thing players find hardest. The Bank's own estimate is that policy takes somewhere between one and two years to have its full effect, which is why the MPC forecasts rather than reacts. If you wait until inflation is visibly out of control before moving, you are already too late; the medicine arrives after the illness has spread.
Inflation in the game does not revert toward 2%. It reverts toward what people expect, and expectations drift toward whatever inflation has actually been running:
chase = 0.10 + 0.16 × clamp((inflation − 5) / 10, 0, 1)
expectations += chase × (inflation − expectations)
Below about 5%, expectations move slowly — the target still has credibility, and inflation gravitates home on its own. Above it, the coefficient rises and the public starts pricing off what it sees rather than what the Bank promises. That is a wage-price spiral, and it is the mechanism that took UK inflation to 24.2% in 1975. Expectations only re-anchor toward 2% while you are actually holding inflation inside the band, so the way out of a spiral is to get inflation down and keep it down long enough to be believed.
Below zero, falling prices raise the real value of debt and give people a reason to postpone spending. Demand weakens, unemployment climbs, and prices fall further. Japan spent the best part of three decades in that trap, which is why the game's deflation medal is named after it. Deflation is not a safe place to be; it is the other cliff.
There is a neutral rate of interest — economists write it r* — at which policy is neither stimulating nor restraining. It is unobservable in real life, and it is hidden from you in the game too. It also moves: world interest rates wander through the term and drag the UK's neutral rate with them, so the Bank Rate that was right two years ago can be quietly wrong today.
Your read on it comes from the Committee. Each of the eight other members prices their own preferred rate off a Taylor-style rule and the true neutral rate, so when the vote comes back "8 members wanted it far higher", that is the game telling you where neutral has moved. They only advise — the decision is always yours — but ignoring a lopsided vote is usually expensive.
Every quarter carries a chance of an energy shock, in three sizes:
| Tier | Chance | Inflation | Persistence |
|---|---|---|---|
| Minor | ~11% | +0.32pp | Fades in a quarter |
| Major | ~4.7% | +0.75pp | A couple of quarters |
| Crisis | ~1.8% | +3.6pp | Years |
The crisis tier is modelled on 2022, when the invasion of Ukraine pushed UK CPI to 11.1% — its highest in forty years. A crisis in the game can do the same, and the tail keeps feeding inflation long after the headline event. Because it is a supply shock, it is the nastiest thing that can happen to you: it raises inflation and hurts employment simultaneously, so both your targets break at once and the textbook trade-off offers no comfortable answer.
Policy transmission saturates rather than scaling linearly:
squeeze = tanh((rate − neutral) / 4)
inflation += −2.00 × squeeze · unemployment += 0.62 × squeeze
This matters. A linear rule would let a 15% Bank Rate erase a decade of inflation in a single quarter, which is nonsense. With saturation, going to 15% is genuinely powerful but it takes years to break a spiral — and it puts a great many people out of work while it does. That is the actual Volcker experience: the US Federal Reserve took rates above 19% in 1981, broke the back of inflation, and caused the deepest recession since the 1930s doing it.
Bank Rate is not the only thing a central bank has, and below about 1% it stops being the useful one.
Quantitative easing appears once Bank Rate is at or below 1%. You buy gilts, the stock accumulates, and it is expressed as a shadow rate: the stance the economy responds to, below the headline you can no longer cut.
shadow = 0.75 × ln(1 + stock/120) — about 0.74pp at £200bn
output responds to (rate − shadow) ·
prices respond to (rate − 0.35 × shadow)
That asymmetry is the point. QE is not a rate cut by another name: it works mainly through portfolio balance — supporting asset prices and lending, and so output and jobs — while the evidence for its effect on prices is weaker and far less certain. So it can pull unemployment back without the full inflationary cost, which is exactly why it is reached for at the zero bound.
It is not free. Buying is fast at up to £75bn a quarter; unwinding is capped at £25bn. Overshoot and you are still stimulating an economy that no longer needs it — the live criticism of 2021. The 2022 scenario opens with £300bn already on the books for exactly that reason: the stance you inherit is looser than the rate you were handed, and shedding it is capped, quarter by quarter, while inflation runs.
How readily the public believes the 2% target scales how fast expectations re-anchor when you are hitting it:
if inflation in band: expectations += 0.20 × credibility × (2 − expectations)
It moves only when you write to the Chancellor. Blame the shock and you pay less now but are believed less later; pledge to return to target and expectations settle at once, provided you actually deliver. Getting believed is most of a disinflation, which is why the cheap answer in the letter is not always the cheap one.
Finish a modern term at par or better and you are reappointed — and you inherit the economy you left. Not a fresh 2026, but your own inflation, your own unemployment, the rate you set last, whatever balance sheet you built, the expectations you shaped, and the credibility you earned or spent.
The scar comes with you too, and that is the point. Over-tighten in your first term and the jobless floor is permanently higher in your second: the natural rate you are steering toward is no longer 4.4% but 4.4% plus whatever you destroyed. Nothing you do later brings it back. It is the clearest statement the model can make about why central bankers worry so much about going too far.
Fall short of par and the run ends there; the next term starts clean. Scenarios never carry forward — a second 1974 term would be 1979 to 1983, which is a different history — and a sandbox is unranked, so neither leaves a legacy.
No fiscal policy, no exchange rate you can see, no housing market, no banking system, no distinction between headline and core inflation, no productivity or population growth, and a single interest rate standing in for an entire yield curve. Real forecasting models run to hundreds of equations and still miss. This one has about six, and the point of it is the trade-off, not the forecast.
Play the game →