Why we need greater exposure to data, and less exposure to Donald

German Gas. A year back a major row kicked off in Germany about cutting gas supplies from Russia (before the Russians themselves cut gas flows in August). Companies/unions warned cutting imports would mean an economic catastrophe, while some economists (including the always great Ben Moll) called for imports to end earlier to avoid funding Putin’s war effort, and argued the economy could adjust quicker than many claimed. The latter turned out to be right. In a new paper Moll and co-authors claim victory, showing that even if supplies had ended in April 2022, Germany would have still got through this winter with gas reserves of 25 per cent (and reject the argument that this was just good luck based on a mild winter). The Germany economy, like ours, has flatlined recently – but the feared apocalypse never arrived.

Generational gains. A recent labour market note from the Race Disparity Unit caught my eye with its focus on labour market outcomes varying by ethnicity among second-generation migrants residents. You won’t be surprised to know they tend to do better than the first generation – although the scale of progress for Pakistani/Bangladeshi women is huge, and second-generation Indians buck the trend with higher inactivity rates than their predecessors. But the variation among different ethnic minorities also stands out. Some second generations have better outcomes than white British counterparts (second generation Indian men are less likely to be inactive, and second-generation Bangladeshi men and black African women are less likely to be unemployed) but others worse (black Caribbean men and women). Divergent education outcomes explain some, but far from all, these differences. Britain is more diverse, and so are labour market outcomes for second generation ethnic minorities.

Prime platforms. There was a big row in the US media this week about whether Trump should be given a prime-time slot on CNN the day after he was found guilty of sexual assault. The platform/no platform row felt like 2015 and 2016 all over again. And you can tell why these rows happen, because coverage really does matter in elections – as research on Italian politics shows. The basic argument is that the surprise resignation of Pope Benedict ahead of the 2013 election saw TV coverage switch away from politics – and in particular from Berlusconi. The researchers show the 26 percentage point loss of TV coverage translated into a 2 percentage point loss of vote share (which was more than the margin by which he lost the election) because of his reliance on the medium compared to other parties’ internet focus. If nothing else the paper will remind you there were sex pest politicians who needed cancelling long before Trump.

Disappearing data. Data’s fairly important to the whole science thing. Sometimes that data’s public – like the surveys on household finances or the labour market we rely on – which is good because then other people can examine it to see if they agree with what a researcher says they’ve discovered. But lots of data isn’t publicly available like that, for example when it’s about a new experiment that’s been conducted. So, there’s pressure on academics (not least from the journals they publish in) to promise to make the data they’ve used available upon request. The problem? They promise to do so and then… don’t. A short article looking at psychology articles over the last five years finds that 42 per cent included promises to share data. Good. But it turns out only 17 per cent of authors actually did so when requested. Less good. Turns out we need to remind scientists, not just the kids, that sharing is caring.

Evaluating evidence. Everyone (at least theoretically) favours evidence-based policy making. Or used to at least. But do policy makers (think civil servants) using evaluations to design new policies and researchers doing those evaluations value the same things? No, finds an interesting paper that experimented on policy types and researchers at World and Inter-American Development Bank events. When asked to choose between policy evaluations to pay attention to, policy makers focused on the context for studies to decide how relevant they were (i.e. where they took place), while researchers behaved like good statisticians and focused on sample sizes/method quality. Obviously both matter – something reinforced by the most interesting finding that policymakers who behaved more like researchers (caring about method quality) and researchers who behaved more like policy makers (valuing the context) did best overall in predicting policy effectiveness.

Chart of the Week

The Bank of England made it 12 rate rises in a row yesterday. The impact on the economy will be significant (which is the whole point) but it’ll also be different compared to previous rate surges (you have to go back to the 1980s for one this fast). Think about mortgages – where fewer people will be affected due to changes in the housing market: we’ve got fewer mortgagors, with more older households owning outright, and fewer younger families owning at all. Changes in the mortgage market also mean the impact will take longer to be felt, as COTW shows. 20 years ago less than half of mortgage lending was via fixed-rate deals. But that changed post-financial crisis – now we’re talking 90+ per cent. And the deals are getting longer – five-year fixes recently overtook two-year ones as the most popular. The result of these changes is rather than lots of mortgagors getting jabbed every time the Bank raise rates, a smaller number of them get one huge blow when they come to re-mortgage. We’ll have lots on the scale and distribution of those blows in our latest Macro Policy Outlook, published tomorrow.