Appeals to exclusionary complexity reduce public trust in expertise and mainstream politics.
, .
Complex factors.
On 28 January 2024 Kenan Malik had a piece in The Guardian about immigration and housing.
"What of the claim, made by critics such as Goodwin, that mass immigration drives up house prices and squeezes locals out of private renting?" he asks. "Housing demand is influenced by many complex factors and there is no 'direct relationship between population change (as a result of migration) and housing demand'" he reassures us, pulling from a Parliamentary Office of Science Technology note from 2017 which in turn references work by the ONS on International migration and the changing nature of housing in England.
It annoyed me. I know the literature on migration and housing quite well. I don't think it shows what he implies it does. I disliked the appeal to complexity especially when combined with that disagreement.
Lots of people shared his piece. Great piece, they said. I kept quiet.
Mass immigration and affordable housing.
On 14 February 2024, in an unusual Valentine's day gesture, Matthew Goodwin published a blog post promoting through provocation an upcoming event he is headlining in central London. It ended with his view that "Britain can either have mass, uncontrolled and unsustainable immigration. Or it can have affordable and available housing. It can't have both".
Goodwin annoys a lot of people, I certainly disliked his decision not to use an Oxford comma after uncontrolled. But often he annoys people for good reasons too.
I don't think uncontrolled is correct, since we have greatly increased our control of migration since leaving the EU. I don't think unsustainable adds anything to his point. He is not disciplined in how he examines pressures on Britain's housing stock, public services, or culture. Is high net immigration the problem? Or foreign-born people (many of whom are British)? Or is it some other "don't share our culture" issue? It feels like the problem is often whichever measure makes the point, even if the others don't.
But I reckon that his blog post and its final statement, within the context of his blog post, reasonable benefit of the doubt, and within discussions about housing in Britain, was fine. I'd have just said mass immigration. But I think it's fine, especially since Goodwin had talked about how supply of housing was also important to housing costs, linked to respectable think tank reports attempting to estimate the quantity of building it would require to house net immigration at current levels without causing housing costs to rise substantially, and I think reasonably assessed that this was unlikely.
Outlandish claims?
Ben Ansell, a Professor of Comparative Democratic Institutions and whose recent Reith Lectures on the future of our democracy I enjoyed, was less keen on it. In a popular thread on Twitter and without naming names or linking to pieces, he echoes Malik's piece. He reminds us that lots of factors determine the price of housing, sets out why he wasn't convinced that migration mattered that much to house prices, and implored people, with the assistance of lots of nice graphs, to look at the data a bit more before making "outlandish claims".
The most important of Ben's graphs can be reproduced using the Population by country of birth and nationality dataset from the ONS combined with the House price statistics for small areas in England and Wales dataset.
We see that the change in the foreign born population of a local authority over the decade to 2020 is barely correlated with the increase in house prices in that local authority over that period. The p-value is 0.002 if you care for those type of measures but the R² value of 0.03 means that the variation in the change in foreign born population barely explains the variation in the change in house prices.
Lots of people shared that thread. Great thread, they said. It appeals to their and my desire to not blame migrants for domestic problems or to give intellectual backing to those whose anti-migration rhetoric is driven by racism. So I kept quiet for a while. But then I couldn't.
Foreign born of international migrants?
It annoyed me that the change in foreign born population data was being used. I know that we have international migration estimates. These are closer to what I think Goodwin was claiming, and they're what we'd expect to matter. And I knew from past work that they showed a stronger, if still weak, correlation.
So I got the latest Local area migration indicators, UK dataset from the ONS, recreated a graph I'd made years ago, and sure enough, the R² value is up to 0.11. Still low, but bigger.
But there's more.
Direct and indirect links.
Like I said earlier, I know the data on immigration and house prices pretty well. A good place to start is Immigration and House Prices in the UK by Filipa Sà which came out between 2011 and 2015. A good follow-up paper for those who need more maths in their life is Decomposing the Impact of Immigration on House Prices by Rosa Sanchis-Guarner which came out between 2017 and 2023.
Both make with lots more maths the point I'll make in a single chart below,
Net domestic migration out of a local authority is highly correlated with net international migration into a local authority. Or in simpler language, international migrants move in to a place while domestic residents move out. You'd need to do more investigating, but a decent hypothesis to test might be that "immigration drives up house prices and squeezes locals out". What Goodwin wrote.
This pattern is what we'd expect if the UK had a housing market with quite low new supply and very little gap between supply and demand. Indeed the high correlation is one reason among many that we are confident that housing market analysts who believe that there is a large and growing surplus of homes in the UK are wrong.
Because there are so few vacant homes ready to absorb new demand, and such a slow and limited supply response to new demand, new arrivals into a housing area must substantially be balanced by existing residents leaving. This means that extra housing demand caused by international migration in a place quickly cascades outward. This cascade leads quickly to housing demand caused by international immigration into a single local authority being reflected across large areas. That reduces any relationship at the local authority level between international immigration levels and house prices.
Indirect relationships matter too.
I think it's pretty well known in the literature that the type of analysis that Ben did won't tell you much about the impact of international immigration on house prices. We know that there is 'no direct relationship between population change (as a result of migration) [in a local area] and housing demand' as the ONS wrote and POST repeated. But in a complex system, indirect relationships matter too. We know that there is quite a strong relationship in most cases.
Once you know that international and domestic migration are linked, the extra house price increases that extra demand caused by international immigration lead to can be modelled quite well. Papers such as Sanchis-Guarner include some good models, but there are lots more. This blog post is already going to be far too long, so I won't work through any here.
Instead I'll just claim that these models show that given reasonable expectations about UK house building, Britain cannot have net immigration at current levels and provide good quality and affordable housing to most of its population. What Goodwin wrote wasn't outlandish.
I add quickly that we expect net immigration levels to reduce in the coming years. 700,000 was probably an outlier. And that there are some hopes of a new UK government increasing house building. And of course there are other things we can do to make housing more affordable such as taxing second homes and holiday homes more and updating housing taxes to encourage people to downsize their homes.
But right now, with net international migration into the UK of around 700,000 people while we complete 230,000 new homes, we are going to be pushing up house prices significantly versus the counterfactual (which given recent increases in interest rates should be falling quickly).
The complexity is the problem.
At this point I hope that you think I've made a reasonable argument. And you might then be thinking that this is a great example of complex consideration of a problem leading us through a competition of ideas to a place closer to the truth. Sadly I don't think that's true.
Ben has studied immigration and housing market interactions a great deal. I have studied immigration and housing market interactions a great deal. Many others have studied immigration and housing market interactions a great deal. And we all come to very different conclusions. In fact, I don't think we shed much more light on this issue with access to all of our studies and our complex models than an interested member of the public.
For at least the past decade in England there's been expert disagreement about housing at a far more fundamental level than the complexity above describes. Everyone sensible eventually agrees after half an hour of shouting at each other that the price of housing is set by supply and demand, as with almost all other market goods. But the components of both supply and demand, the relative importance of each component, and the relative political and fiscal costs of changing each component is heavily disputed.
Ian Mulheirn is probably the leading advocate of the "we've got enough homes" line of thought in England. His blog post "the current policy focus on boosting housing supply does not offer a solution to the housing crisis" is a good summary of his long-held and well-argued position. He is very influential, especially since his position provides intellectual backing to the English public's widely held desire to avoid building more homes near them.
I think he heavily underweights the importance of building new homes in setting the price of housing. I've written a very long blog post with the specific details of why I disagree with his interpretation of the housing data we have. Others have written more about why they think that household formation is constrained more than he does by the undersupply of homes and how this makes it look like there isn't an undersupply of homes when in fact there is.
Fuck nuance.
I suspect those people are right too. And indeed some of Ian's replies to my work are also probably right. The OECD's empty homes comparisons across countries do use importantly different definitions of empty home count for example. I haven't even got onto some work by The Bank of England that probably supports Ian's views more than mine. I could write another two thousand words about that I suppose.
As in the very long blog post I linked to above, I get exhausted at this point. We're trying to understand the comparative importance of dozens of interacting factors on the price of housing without any reasonable chance of doing that well. We are luxuriating in the complexity and nuance of the system instead of doing the hard work of accepting that we shouldn't. As Kieran Healey famously wrote, Fuck Nuance .
Imprecise measurements in complex systems.
In England we try to maintain, through planning restrictions, a national vacancy rate of around 2%. That is an astonishingly tight market. And we do this despite us knowing that our best estimates of the population of our major cities is regularly off by as much as 7%. Our estimates of the UK-resident EU population eligible for settled status were off by as much as a third, millions of people, with the error continuing to grow.
What we measure about housing supply and housing demand in England is, unavoidably, measured with high uncertainty. But there are also things such as the size and condition of housing that we barely measure at all. We know surprisingly little about what rents people pay, where, and when. And unlike most countries, the size of homes is not a primary feature of listings and data collection, so we lack good proxies for the quality and quantity of housing that people are paying for.
Complexity, self-delusion, and bias.
My experience in this field and many others is that adding complexity usually moves us further away from useful debates and good policy. And whenever a sufficiently complex system is constructed that can be used to prove or disprove a point, the defeated party simply demands that further complexity is added to the model, and more nuance permitted in its interpretation.
At the core of the problem is the property of complex systems that they are flexible. This is especially true when there is a large uncertainty in how well we can measure each component of the system. And the more complex a system the more flexibility there is. This flexibility lets highly-educated people with a lot of time to devote to narrow investigations mould the projection of that system onto policy in a way that aligns with their own political preferences.
Most worryingly I have observed that at very high levels of complexity these people start to believe that they are able to build their ever more complex systems without biasing them in favour of their own views. These people would benefit from friendly challenge from those with other views, but the complexity of their models and the systems that select the people given the time and education to understand such models is such that they exclude most people who could provide that criticism in ways they would accept.
Less complexity and better outcomes.
In summary, I think Kenan Malik and Ben Ansell in appealing to complexity have made worse arguments than Matthew Goodwin's simpler ones on the interaction of migration and housing.
But my more important point is that the appeals to complexity themselves are damaging to public policy debate. The public are aware of the left-leaning bias of academics especially, and of experts more generally, and they are right to be sceptical of appeals to complexity that they see as exclusionary. They see frequent appeals for new national specialist institutes to ensure long-term thinking at arms-length from government as a way for those who benefit from exclusionary complexity to create more of it. They see democratic innovations such as citizens assemblies as a way for biased experts to guide the public towards their interpretations of those ever more complex systems. And they see that these ever more complex systems simply don't deliver good results. We spend £300m and create 300,000 pages of complex documentation so that we can fail to build a road tunnel under the Thames. We are unable to build Nuclear power stations or new railways such as HS2 at anywhere near the efficiency we could in the past.
Appeals to "drain the swamp" and "defeat the blob" and to "get back to common sense" policy are all rational responses to the over-complexification of policy.
So that's why I wrote this tweet. It wasn't very friendly, but I hope I've explained myself.
And I'm not just moaning, I've written before about how, without sacrificing necessary expertise, we could fix some of the creeping complexity that paralyses Western politics and pushes people to vote for populists. I won't repeat those arguments here, this blog post is already far too long, far too complex, and far too nuanced. I'm sorry.