NextFin News - An artificial-intelligence forecasting model is calling for the Bank of England to raise interest rates in February, a signal that lands squarely between the central bank's own projections and the near-uniform "hold" call from human economists. The model's timing - a quarter-point increase at the February 4, 2027 meeting - implies that the inflationary impulse from the Middle East energy shock will prove sticky enough to force policymakers' hands, even as the official forecast and most survey respondents expect rates to stay at 3.75% for the rest of this year and into next.
The prediction matters less because a machine has an opinion than because it crystallizes the tension now running through UK monetary policy: is the recent uptick in inflation a cyclical energy-driven spike that will fade on its own, or the first sign of a structural break that requires tighter policy? With the Monetary Policy Committee's next decision due September 17, the answer determines whether the next move in Bank Rate is up - or down.
The Signal and the Backdrop
Bank Rate currently sits at 3.75%, held there on a 6-3 vote at the July meeting, with three members - Megan Greene, Catherine Mann and Huw Pill - already dissenting in favor of a 25-basis-point increase to 4%. The AI model's February timing aligns closely with how money markets are positioned. Data compiled by LSEG showed roughly 36 basis points of tightening priced in by the time of the February 4 rate announcement, and about 24 basis points by the December 17 meeting. In practical terms, traders are pricing close to one quarter-point hike before February, with the September meeting itself carrying just a 15% implied probability of a move.
The inflation backdrop explains the anxiety. UK consumer price inflation rose to 2.9% in July, up from 2.6% in June and marking the first acceleration in four months. The driver was mechanical and visible: the 12-month rate for housing and household services jumped to 4.1% from 2.7%, reflecting the 13% increase in Ofgem's energy price cap that took effect in July. Gas prices surged 14.7%, the largest increase since October 2022, while electricity prices rose 3.6%.
Yet beneath the headline, disinflation is still in progress. Core CPI - excluding energy, food, alcohol and tobacco - held steady at 2.6% in July, and services CPI eased to 3.4% from 3.6%. The Bank of England's July Monetary Policy Report projects CPI inflation at 2.9% in the third quarter of 2026, falling to 2.6% a year later and 1.8% by 2028, returning to the 2% target in 2027.
"Today, we've held Bank Rate at 3.75%. Inflation has fallen faster than we'd expected, but the conflict in the Middle East continues to mean high and volatile energy prices. That will cause inflation to rise again later this year," Governor Andrew Bailey said.
Why the Model Sees a Hike When Economists See a Hold
The divergence is stark. A September poll of forecasters found a consensus for the Bank of England to hold Bank Rate at 3.75% for the rest of the year and through at least mid-2027, with inflation expected to average 3.1% in 2026 before dropping to 2.5% in 2027 and 1.9% in 2028. The AI model, by contrast, is effectively underwriting the risk scenario that the human consensus is discounting.
That is the distinctive value - and the distinctive danger - of machine-learning forecasts. Traditional econometric models and survey-based consensus tend to anchor on the central projection: energy prices spike, then revert; inflation rises, then falls back. A learning model trained on historical regime shifts is more likely to detect the conditions under which a temporary shock becomes persistent - when second-round effects take hold in wage bargaining and corporate pricing, when inflation expectations unanchor, when a central bank that looks through a commodity spike is forced to react after the fact.
The model is not predicting that energy prices stay high forever. It is predicting that the transmission from energy to services inflation - the channel the MPC watches most closely - will be strong enough that a 3.75% policy rate is insufficient to return inflation to target on a sustainable path. That is precisely the concern of the three July dissenters, and it is the risk that keeps the central projection's own Bank Rate path above current levels: the Monetary Policy Report embeds Bank Rate at 3.8% by late 2026 and 4.2% by 2027, where it remains through 2029.
In other words, the official forecast already contains a hike. The AI model is simply giving it a date.
The Core Judgment: Cyclical Shock or Structural Break?
This is the question the entire debate turns on, and the answer is that both forces are present, operating on different horizons.
On the cyclical leg, the evidence is clear. Commodity-price spikes are the textbook mean-reverting shock. Brent crude stood at $84 per barrel and the UK front-month natural gas future at 136 pence per therm at the end of July - elevated, but far below the levels that drove UK inflation above 11% in 2022. The 13% Ofgem cap increase is a one-off level shift that flows through the annual inflation rate and then drops out of the year-on-year comparison. History is littered with energy shocks that raised inflation temporarily without requiring a tightening cycle: the 2022 episode itself saw the MPC eventually cut rates as energy faded, and the 2008 oil spike produced a brief inflation pulse followed by deflationary pressure once demand collapsed.
But the structural leg is where the model earns its keep. What makes this episode different from a clean cyclical spike is the state of the labour market and the stock of inflation memory in the UK economy. Services inflation at 3.4% remains well above the level consistent with 2% target inflation, and the MPC's own minutes identify labour-market conditions as the main source of its longer-term inflation worries. After three years of inflation above target, wage-setters and price-setters have demonstrated they will resist real-income erosion. The risk is not that energy stays high; it is that one high-energy year becomes embedded in pay deals and contracts that then persist after energy falls.
A cyclical call requires demonstrated mean reversion; a structural call requires evidence of a regime change. Here, the regime change is not in energy markets - it is in inflation psychology and in the MPC's tolerance for risk. The Committee has explicitly stated that the risk of material second-round effects in price and wage-setting, against which policy needs to lean, is greater the longer higher energy prices persist. That is a conditional commitment to react, and it is the condition the model is betting will be met.
The judgment, therefore, is this: the energy impulse itself is cyclical and will fade, but the policy response it triggers may be real if second-round effects materialize in the next two wage-setting rounds. The February timing is consistent with that path - by then, the MPC will have seen the autumn wage data, the October budget, and several more inflation prints. If services inflation is still above 3% and wage growth above 4% in early 2027, a hold becomes difficult to defend.
Second-Order Consequences: What a February Hike Would Do
The first-order effect of a rate rise is mechanical: higher borrowing costs for households with variable-rate mortgages and businesses with floating-rate debt, at a time when the MPC expects unemployment to peak at 5.3% in 2027. But the second-order transmission is where the real market impact lies.
First, the gilt market would have to reprice the entire curve. The 10-year gilt yield, around 5.0% to 5.1% in late summer, already carries a term premium for fiscal and inflation risk. A confirmed hiking cycle would push two-year yields higher fastest, flattening or inverting the curve, and force a reassessment of the government's debt-servicing trajectory at a time when debt interest costs are already a leading line item in public spending.
Second, sterling would likely strengthen on a higher-rate differential. The pound was trading near $1.35 against the dollar in early September, toward the strong end of its five-year range of $1.07 to $1.40. A stronger pound would dampen imported inflation - helpful for the MPC's target - but would hit the earnings of exporters and multinationals, creating a divergence between domestically oriented companies and global earners.
Third, and most importantly, a February hike would signal that the MPC reads the energy shock as reactive rather than preventive - that it is tightening not to head off inflation but because inflation has already proved persistent. Markets distinguish sharply between the two. A preventive hold is read as confidence; a reactive hike is read as an admission that the "look-through" strategy failed. That admission would lift inflation-risk premiums across UK assets, not just bonds.
The Counter-Thesis: Why the Hike May Never Come
The strongest case against the model's call is also the simplest: the MPC has a symmetric 2% target, and its central projection already shows inflation returning to 1.9% by 2029 without requiring a near-term hike above the current 3.75%. Governor Bailey has repeatedly emphasized that monetary policy cannot influence energy prices, and that the job is to ensure any increase is temporary. If the conflict in the Middle East de-escalates, energy falls back, and the July inflation uptick proves to be exactly the mechanical base effect the Bank described, then the three July dissenters become a shrinking minority rather than a growing majority.
There is also the growth constraint. The Bank projects GDP growth of just 1.1% in 2026 and 2027, with excess supply of 1.1% of potential output - an economy operating below capacity. Tightening into a below-potential economy with unemployment rising toward 5.3% is a policy error the MPC will work hard to avoid. As forecasters in the September survey put it: they continue to expect the MPC to hold this year before cutting in 2027, judging market pricing for Bank Rate too high.
The model, trained on inflation dynamics, may be overweighting the price signal and underweighting the growth constraint. Machine-learning systems excel at pattern recognition in the data they are given; they are less reliable at incorporating the political and distributional constraints that shape real committee decisions. A 6-3 vote to hold is not a committee on the verge of hiking - it is a committee split, and the swing voters have so far chosen patience.
The falsifying signal for the "no hike" view is specific and observable: if core CPI and services inflation both print at or above 0.3% month-on-month for two consecutive releases into early 2027, accompanied by regular wage growth above 4.5%, the cyclical-fade thesis is wrong and the February-hike scenario moves from tail risk to base case. Conversely, the falsifying signal for the model is equally clear: if CPI falls back below 2.5% by the first quarter of 2027 while wage growth decelerates toward 3%, the model has priced a regime shift that never arrived.
What to Watch and the Scenarios Ahead
The near-term catalyst is the September 17 MPC meeting. A hold is the base case, but the accompanying Monetary Policy Report and the Governor's press conference will be scrutinized for whether the inflation-risk language hardens. The October budget is the second catalyst - fiscal loosening would add demand-side pressure to the supply-side energy shock and materially raise the probability of a February move.
Three scenarios frame the path:
- Base case - hold through 2026, data-dependent into 2027. Energy stabilizes, services inflation grinds lower, and the MPC holds at 3.75% at the September, November and December meetings. The February decision then becomes a genuine toss-up, driven by the autumn wage and inflation prints.
- Upside case for rates - the February hike arrives. A renewed energy spike or stubborn services inflation pushes the MPC toward a 25-basis-point increase to 4% at the February 4 meeting. Gilts sell off, sterling strengthens, and the hiking-cycle narrative takes hold. This is the AI model's call.
- Downside case for rates - cuts return in 2027. The energy shock fades faster than expected, growth disappoints, and unemployment rises faster than projected. The MPC pivots to cutting toward 3.5% in 2027, as some forecasters already expect. In this scenario, the model's signal was a false positive born of overfitting to the inflation spike.
For investors, the asymmetry is clear. UK rate-sensitive assets - housebuilders, banks with large mortgage books, highly leveraged companies - are pricing a hold. A February hike would reprice them lower. Conversely, a confirmed path to cuts would support duration assets and the most indebted borrowers. The model is not a trading signal, but it is a reminder that the market's consensus "hold" is far from unanimous - inside the MPC, in money markets, and now in machine-learning forecasts, a meaningful constituency is preparing for the opposite.
The central bank's own numbers contain the irony: the Monetary Policy Report already embeds Bank Rate at 4.2% by 2027, yet the Governor insists policy is appropriately set today. The AI model has simply noticed that a central bank whose own forecast requires higher rates cannot indefinitely pretend it does not. Whether February is the month that pretense ends depends less on the model than on the next wage deal signed in Britain's boardrooms - and on whether the Middle East conflict that started this inflation pulse decides to fade, or flare.
Explore more exclusive insights at nextfin.ai.
