# Governments don't create growth

_Inflation falls, the ECB looks worse, and chip stocks lose their minds_

Neil Woodford · 3 July 2026 · 5 min read

![PM-in-waiting Andy Burnham](https://cdn.sanity.io/images/v3acfbvo/production/5fd75eeeb8d9d82b8e939baaefda7bee9854acb3-2696x1798.jpg?w=1600&fit=max&auto=format)

---

Like last week, this has been a strangely quiet week for geopolitics and markets _(editor’s note: it was not quiet at Noise Cancelling HQ as you can hopefully tell!)_. 

There have been a few stories that grabbed my attention, but they are all relatively niche and didn’t really warrant that much attention from the wider financial community. 

One was a sort of policy speech from the UK’s PM-in-waiting, another was the US government’s relaxation of rules it had imposed on access to Anthropic’s Fable model, another was the lack of any oil or financial market ripple following last weekend’s exchange of fire between Iran (on a ship) and the US’s military response. Both sides have apparently agreed not to repeat the conflagration and appear to still be in dialogue mediated by Pakistan and Qatar. 

The final interesting development this week was the extraordinary volatility in the share prices of global semiconductor stocks, particularly in Micron, Samsung and SK Hynix, which have become the darlings of the AI industrial revolution this year.

_[Watch: Watch the first episode of Season 2 of Noise Cancelling — Three Central Banks Made the Same Mistake in One Week](https://www.noisecancelling.co/the-show)_

## Burnham's devolution debut

Andy Burnham, who will become the UK’s new prime Minister in a few weeks, [gave a policy speech this week](https://www.youtube.com/watch?v=SHOcYcIqzsA) that, given what is about to happen, attracted considerable attention, at least from the media. 

_[Embedded media](https://www.youtube.com/watch?v=SHOcYcIqzsA)_

It outlined his vision for more devolution in the UK, which he said would result in more growth and a levelling up of the regions with the south east, and particularly London, which for years has had better productivity and higher per capita incomes. For someone who has apparently recruited some highly regarded economic advisors, this was for me a pretty disappointing debut for the soon-to-be PM. 

Aside from the fact that the electorate has virtually no interest in devolution, the history of the last twenty-five years of increased devolved powers for Scotland, Northern Ireland and Wales shows that there has been no closing of the gap in GDP per capita since 1998. In fact, measured against average UK GDP per capita, Scotland, Wales and Northern Ireland were pretty much the same in 2023 as they were in 1998: 93% for Scotland, 83% for Northern Ireland and 74% for Wales. _(Neil in the margin: GDP per capita is output divided by population — a rough proxy for average living standards. Holding steady relative to the UK average since 1998 means these nations neither caught up nor fell behind, despite the extra powers.)_

![Relative GDP per head has barely moved since devolution](https://r4at4qm6kmohrtvq.public.blob.vercel-storage.com/charts/c8ca7ab2-d939-4fdd-8ca1-ae135776a6f6-cd47202849ad-light.png)

Once again, it seems that politicians are wilfully ignorant of the facts (or just ignorant). I am also struggling to see how local government that appears incapable of managing its current workload (planning and potholes, for example) could be equipped within a reasonable time frame to take on more revenue-raising and public-spending responsibilities. Whilst it is true that, compared with other developed economies, the UK has a very centralised public sector, I am really struggling to see how more of what has objectively not worked over the last twenty-five years will deliver the miracle of higher regional growth as outlined in Burnham’s speech. 

If I were advising the new PM, I would suggest his focus should be on issues that could unlock higher growth everywhere quickly, starting with deregulation and tax cuts, alongside getting rid of stamp duty as a decent place to start. I suppose the fundamental point here is that governments don’t create growth; they typically get in the way of it. Their job is to facilitate it and create the conditions in which it can happen, but as long as this government maintains its current completely insane energy policy, which has produced the highest industrial energy prices in the world, its claim to be doing all it can to deliver a growth agenda is totally disingenuous. _(Neil in the margin: Stamp Duty Land Tax is a transaction tax on property purchases. Economists widely dislike it because taxing moves gums up the market — discouraging people from downsizing, relocating for work, or freeing up family homes.)_

## Anthropic's Fable, unbanned and repriced

A little over two weeks ago, the US government applied export controls on Anthropic’s two latest models, Fable 5 and Mythos 5, over security concerns. This required the business to restrict foreign nationals’ access to the models, whether inside or outside the US. Because Anthropic didn’t know its users’ nationalities, the order resulted in the business suspending access to these models for all users. _(Neil in the margin: Both Fable 5 and Mythos 5 share the same underlying model, but the Fable version has stronger safeguards, making it safer for more general use. )_

_[Embedded media](https://www.anthropic.com/news/redeploying-fable-5)_

Either way, after a group of Amazon researchers [discovered a way to bypass safety controls](https://www.forbes.com/sites/zacharyfolk/2026/07/01/anthropics-fable-and-mythos-models-are-back-heres-what-the-government-was-concerned-about-in-the-first-place/), the software has been patched, and, following collaboration with the government, the models have been rereleased. 

Beyond the frustrations of suspending access to these high-performing models for just over two weeks, this development again highlights the politicised nature of AI model development, especially at the cutting edge. 

The economics has also evolved following this suspension of service. Following the rerelease, the cost of using Fable has increased significantly, and it may not become as widely used as was expected for the foreseeable future. Quite what is behind this repricing is unclear, but I suspect it may be a marketing experiment to test the economic boundaries of what users are prepared to pay for the highest-performing models. 

Meanwhile, [some of the best Chinese models](https://benchlm.ai/best/chinese-models), although inferior to these frontier examples, are free and open source.

## Inflation and rates

In a week when oil prices dipped further below pre-war levels, I was interested to see that Eurozone inflation fell further than expected in June to 2.8%, rather than the 3% expected by consensus, reflecting lower energy prices and the absence of the much-feared (by central bankers) second-round effects.  _(Neil in the margin: The central bankers' nightmare: an initial price shock (say, energy) feeding into higher wage demands and then into other prices, so inflation embeds itself rather than fading. Their absence here means the spike stayed a one-off.)_

This report once again highlights how wrong it was for the ECB to raise rates a few weeks ago and for members of the committee to add insult to injury by subsequently trying to justify the decision in a series of speeches and articles. I expect inflation to continue to fall pretty much everywhere, as the full extent of the oil price reduction in recent weeks has yet to work through to the headline numbers.  _(Neil in the margin: The European Central Bank sets interest rates for the euro area. Hiking rates into falling inflation was a policy error — tightening just as the pressure was already draining away.)_

This should mean that interest rates in the US and the UK will hold through the summer months until later in the Autumn, when the scope for rate cuts will return.

## Semiconductor volatility

Finally, the extraordinary volatility of global semiconductor stocks has continued this week and, if anything, has become even more extreme. 

For example, on Thursday this week, SK Hynix fell by over 14%, and today it rallied nearly 9%. Some volatility is to be expected given concerns about data centre overbuild in the US and Apple’s complaints about semiconductor price increases, but this would be exceptional for small caps, let alone businesses with market caps above $1trn.  _(Neil in the margin: Market cap is share price times shares outstanding — the total equity value. Double-digit daily swings are normal for tiny speculative stocks; for trillion-dollar giants they imply enormous sums sloshing about on thin conviction.)_

Quite what happens next is not at all clear. My view is that volatility will calm, and the relentless price increases seen in Q2, when the US semiconductor sector index rose 81%, are probably now behind us. _(Neil in the margin: An 81% gain in a single quarter for an entire sector index is extraordinary — the sort of move that usually marks either a genuine regime change or the frothy top of one.)_

## What to look out for next week

Next week looks set to be another relatively subdued one for macro data. There are more jobs reports in the US to add to this week’s data, which showed a pretty weak labour market in June (May’s data was also revised down), and in the UK there isn’t much to be concerned about other than some ISM survey data and Halifax house prices for June. 

The corporate calendar is also quiet ahead of the Q2 reports which will come thick and fast later in July.