The Drill

France: Closer To a Coin Flip Than the Spread Suggests

In short: Le Pen’s odds are closer to 50-50 than the market fears, and even if she does win, she will not have the level of impact that many investors fear. France might be a medium-term buy here.

The market is pricing in a regime change

The French 10-year yield hit 4.99% on 1 October, the highest since 2002. The spread to Bunds peaked at 152bp the next day, the widest since 2011, and has since come back to around 130bp. On the monthly data, France now trades level with Italy and well above Spain.

Equities tell the same story. Since Macron dissolved the Assembly in June 2024, the CAC 40 is down 2%, while the Stoxx 600 is up 22%, the FTSE MIB 47% and the IBEX 70%. Japanese investors have sold French bonds week after week, and as Andreas wrote on Friday, positioning in France is about as sour as it gets.

The Runoff Polls Overstate Her Odds

On the headline numbers, Le Pen is a clear favorite. She polls around 33% in the first round and leads Philippe 57-43 and Mélenchon 69-31 in the runoff polls.

I would discount those runoff numbers. In the last two elections, she ended well below her final polls: 39.3% in the polls against 33.9% on the day in 2017, and 45.4% against 41.5% in 2022. The late deciders went against her both times. If the average miss of those two elections repeats, her current 55.7% against Philippe becomes something like 51-49.

So the question is who Le Pen meets in the second round, and that is wide open: Philippe polls 14-15% and Mélenchon 14-16%. Against Mélenchon, she wins even with a polling miss of that size. Against Philippe, I would call it 50-50. Polymarket has her at around 42%, and Macquarie’s Thierry Wizman puts an RN presidency at “near 50%”.

My take: those are the right numbers. The spread is priced as if the outcome were already known.

The Budget Does Not Add Up, and the Bond Market Rallied Anyway

On Monday Le Pen presented the RN’s counter-budget. The main pieces are a constitutional golden rule put to a referendum, €140bn of savings by 2032, net of at least €30bn of tax cuts, a primary balance within 18 months and a deficit below 3% by 2030. In the same speech, she promised the retirement age back at 62, “even 60 in some cases”.

Run the numbers and most of it does not hold. Her rule keeps the deficit at least half a point below the level that stabilises debt. On our numbers, that lowers the debt ratio by about half a point a year, and from the 112% she targets for 2032, France would get back to 60% around 2136. A primary balance in 18 months means finding roughly €95bn by mid-2028, at a time when Lecornu is struggling to get €54bn through parliament. Pensions are supposed to save €15- 20bn in the same plan that lowers the retirement age, and almost €100bn of the savings is not itemised. Allianz’s Ludovic Subran made the same point: the savings only work if the RN drops the retirement promise.

OAT yields still fell 14bp on the day, the biggest one-day drop since May. Bond investors care more about which way she is heading than about whether the sums work.

If Le Pen Wins, She Will Most Likely Not Have a Majority

The Élysée is only half of it. Domestic and economic policy runs through a prime minister who has to survive the National Assembly, and it takes 289 of 577 votes to bring a government down. The RN and its allies won 142 seats in 2024, and the RN and Ciotti’s UDR groups count around 140 deputies today.

A new president can dissolve straight away, and Le Pen would have every reason to. The latest snap-election projection puts the RN and allies at 285 seats, with a range of 231-340 and roughly a 45% chance of a majority. Note that in 2024 the final projections gave the RN 157-250 seats, and it won 142. To be fair to the other side, every new president from 2002 to 2017 won a majority in the legislative election that followed. Macron in 2022 was the first to fall short.

My take: Le Pen in the Élysée with an RN majority behind her is roughly a one-in-five outcome. If she wins, my base case is a president without a majority.

Bardella Would Do the Governing

If Le Pen wins, Jordan Bardella becomes prime minister. That was the arrangement when they launched the campaign together in July. Without a majority, his government would be in the same position as Barnier, Bayrou and Lecornu: budgets forced through with Article 49.3, and survival dependent on Les Républicains, who will want fiscal restraint and business-friendly policies in return.

Bardella is comfortable there. In the 2024 campaign, he dropped the pledge to leave NATO’s integrated command and put the pension reversal on hold pending an “audit” of the public finances. This spring he went to MEDEF, and when Le Pen backed a windfall tax on TotalEnergies in April, he said the same day that it was not a priority in a country with a 43.6% tax burden. On Europe he talks about reforming the EU from within. Frexit and leaving the euro left the programme years ago.

Le Pen has pulled the programme back towards the state over the summer. She put the age-62 pledge back in August, and Bardella’s liberal economic adviser François Durvye left the campaign. I still think the balance tips after the election. Le Pen’s line is the one that wins votes; Bardella’s is the one a minority government can pass. The weaker the RN is in the Assembly, the more it needs LR, and the more the policy ends up looking like Bardella.

Even With a Majority, the French State Constrains Le Pen

Take the one-in-five case where Le Pen wins, and the RN gets its majority. The programme would still have to get past the Senate, the Constitutional Council, Brussels, the ECB and the bond market.

Les Républicains and the centrists still hold around 185 of 348 Senate seats after September, against 15 for the RN. The Senate cannot block ordinary laws, but every constitutional amendment needs it, and that includes both the golden rule and “national priority” for benefits. Le Pen’s way around it is an Article 11 referendum, the route de Gaulle took in 1962, and one most constitutional lawyers consider unconstitutional. The Constitutional Council has already taken the benefit restrictions out of the 2024 immigration law and ruled that a five-year residency test for benefits goes too far.

The EU budget and its own-resources framework are agreed unanimously, so the RN’s target of a €5bn net contribution is not something Paris can decide alone. Access to the ECB’s TPI, the backstop against a disorderly widening of spreads, is conditional on a set of fiscal and macroeconomic criteria, including compliance with the EU fiscal framework. Note that the half- point annual adjustment in Le Pen’s rule is the same benchmark the EU uses for countries in its deficit procedure. She knows where the line is.

The bond market is the hardest constraint of all. Mitterrand won the presidency in 1981, then an absolute majority, nationalised banks and industry and lowered the retirement age to 60. Within 22 months the franc had been devalued three times and lost more than a fifth of its value against the Deutsche Mark, and in March 1983 he turned to austerity. Le Pen has no franc to devalue, so the pressure would go straight into the OAT spread, and faster.

Orbán and PiS were able to rebuild their states because they had constitutional majorities, and the RN is nowhere near one. Meloni is the closer case: she campaigned against the Fornero pension law, then cut Italy’s deficit from 8.1% of GDP in 2022 to 3.1% in 2025, and the BTP-Bund spread fell from 234bp to around 80bp.

Trade: A Medium-Term Buy

My take: France is priced for a policy rupture its institutions would heavily constrain. I would build exposure over the coming months with April and May as the horizon, through the broad index (iShares MSCI France, EWQ, or the Amundi CAC 40 ETF) and the banks that carry the spread risk: BNP Paribas, Société Générale and Crédit Agricole. I would stay away from the motorway concessions, Vinci and Eiffage, as renationalisation is still in Le Pen’s programme.

On the upside, a spread retrace to the 80bp Italy trades at is worth about 4% on a 10-year OAT relative to Bunds, and closing half of the CAC’s 24-point gap to the Stoxx 600 is about 12% in relative terms. The dates to watch are the budget vote by 17 November, the race for second place between Philippe and Mélenchon, and the first round on 18 April.

Spain: The Election Is Priced; the Bank Tax Is Not

In short: a right-wing government is consensus and already in the price, so I would skip the index and own the one policy change nobody has priced: the end of the bank tax, via CaixaBank and Bankinter.

Sánchez has called a snap election for 29 November after Congress voted down his housing decrees. The PP polls at 33%, the Socialists at 26% and Vox at 18%, which gives PP and Vox about 206 seats against the 176 needed. In 2023, the right polled 177 seats and ended on 170, which kept Sánchez in office. This time the margin is 30 seats, and Vox looks set for its first role in government.

Most of this is in the price. Spain trades about 45bp over Bunds, the IBEX is up 12% this year and 70% since mid-2024, and neither moved on the announcement. Housing and nuclear names have already run. The banks are the exception. The 1-7% tax on net interest income and fees covers the 2024-2026 tax years and has to be renewed to continue. My base case is that a Sánchez government would extend it and a PP government would let it lapse. CaixaBank booked €611m of the tax in 2025, roughly a tenth of its €5.9bn profit, yet it is up 19% this year and Bankinter 12%, against 16% for European banks. I see no premium for the tax ending in those numbers.

My take: the IBEX has more to lose from a messy result than to gain from a clean one, and Spain has gone back to the polls before. Long CaixaBank and Bankinter, neutral on the index.

US midterms: A blue Wave Cannot Stop the AI Train

In short: Congress cannot halt the AI build-out before 2029, the backlash against data centers is bipartisan and local, and the market has already priced the political risk in utilities. Hold AI, buy any election-week dip, and use regulated utilities as the event trade.

A Democratic sweep is the consensus.

As of 7 October, Polymarket prices a Democratic sweep at 65%, a Republican Senate with a Democratic House at 28% and a Republican sweep at 8%. That is 93% for the Democrats in the House and 65% in the Senate. As we wrote in the September 30 edition, the House looks very likely to flip, and the Senate is where the real uncertainty sits. The sell side has picked up on it: Michael Hartnett at BofA sees more than 10% downside for US equities in a sweep, and JPMorgan’s Dubravko Lakos-Bujas calls Republican control the most supportive backdrop for this year’s roughly $730bn of hyperscaler capex. The message we keep getting from readers is that you cannot own the AI build-out into a blue wave.

Congress Cannot Witch It Off

Trump signs or vetoes every bill until January 2029. The projections give the Democrats around 223 House seats and 51 Senate seats, a long way from the 290 and 67 needed to override a veto. The Sanders and Ocasio-Cortez moratorium bill is a messaging exercise.

What a sweep can do is make the build-out slower and more expensive. Expect hearings and subpoenas (Wolfe Research expects a Select Committee on AI next year), Senate holds on nominees to FERC and the NRC, which matter for grid connections and nuclear licensing, and spending riders once the new House takes over. It would also end any hope of federal preemption of state AI laws, which leaves 50 sets of state rules in place until 2029. That is the highest structural cost, and the build-out keeps running through it.

The Backlash Runs Through Both Parties

In Michigan, both Senate candidates support data center moratoriums, including the Republican Mike Rogers. Ohio has more than 125 local moratoriums in force and 14 data center measures on the November ballot, out of 28 nationwide. In Texas, the Republican governor paused projects in August. Gallup now has 39% of Americans saying AI does more harm than good, up from 31% a year ago, and 47% among under-30s.

My take: the Senate result matters less for AI than the headlines suggest, because the local vote against data centers wins under either party. It pushes the build-out towards permissive states, federal land, and on-site power. It does not change the size of the capex budget, which is set by AI revenue and the cost of capital.

The likely Democratic agenda is a toll on the industry. Senate Democrats blocked the ratepayer bill, which passed the House 417-3, only because they want it mandatory, and Senator Kelly’s AI Horizon Fund already has Microsoft’s backing. Data centers end up paying for their own grid upgrades and some extra tax, and they keep building.

The Market Has Already Priced the Risk — In Utilities

Since 1 August, Southern Co. is down 11%, Dominion 12% and the utilities sector 8%, while semis are up 18% and Bloom Energy 43%. Rising yields explain part of the utility move, but the ratepayer backlash is clearly in those prices, and the chip stocks are not pricing any of it. A mandatory ratepayer law would settle who pays for the grid, which reduces the rate-case risk the utilities are currently discounted for.

What Could Surprise On Election Night

The biggest upside surprise is a Republican hold of the Senate, which Polymarket puts at 36% and DDHQ at 44%. That would give the utilities relief and put preemption back on the table. A Democratic wave to 53 seats or more would need Texas, Iowa, and Florida to flip. It would bring “AI referendum” headlines and a sharp dip, but still no veto override. The downside case for Southern and AEP is a local revolt, where the data center ballot measures pass, and the Democrats take control of Georgia’s utility commission. Senate control may also stay open for weeks, since Alaska and Maine count ranked-choice ballots and a Georgia runoff would fall on 1 December, ten days before the government funding deadline.

The historical record helps. The S&P 500 has risen in the 12 months after all 19 midterms since 1950, by 15.5% on average on our numbers, and by 33%, 28% and 7% after the three elections where the opposition took both chambers.

Trade: Hold the Build-Out, Fade the Fear

My take: I would hold semis and memory through the vote and add on any election-week weakness, via SMH or SOXX, Micron and SanDisk. The regulated utilities with data center load (Southern, Dominion and AEP) are the event trade: a small position into the vote, more after. I would not chase Bloom Energy after a 239% run this year, and GE Vernova is the cleaner power name if the bipartisan permitting bill passes in the lame-duck session. The hyperscaler results in late October, with the first signals on 2027 capex, will matter more for these stocks than anything that happens on 3 November.

Mikkel Rosenvold

Mikkel is the core geopolitical analyst in Steno Research and host of ‘Macro Mondays’. Mikkel hosts geopolitical talks and provides weekly geopolitical analysis in the ‘The Drill’ article format. Whenever things go down in the realm of warfare, conflict and international affairs, Mikkel has the eye for how developments will impact the market and what the immediate and long- term outlook is.