| 6 mins read
On 27 July 2019 Prime Minister Boris Johnson announced The Towns Fund, a £3.6 billion scheme designed to improve transport and communications infrastructure in English towns. This fund brings together several existing and new schemes, including the Town Deals scheme, in which selected towns are each invited to bid for up to £25 million in funding.
A statistical review shows that Conservative-held areas (and in particular marginal Conservative-held areas) were much more likely to be selected for the scheme, and that this association remains—even when controlling for the ranks that civil servants awarded towns on the basis of qualitative and quantitative criteria. The Towns Fund is clearly an example of ‘geographically-targeted particularistic spending,’ It represents an example of pork barrel politics, or the use of public money for electoral advantage
Civil servants restricted eligibility for the fund to towns with above-median levels of income deprivation, resulting in a longlist of 541 towns. The towns were ranked:
"based on a formula that combined scores against seven criteria chosen to reflect local need and growth potential: income deprivation, skills deprivation, productivity, EU Exit exposure, exposure to economic shocks, investment opportunity and alignment to wider government intervention."
These criteria do not all point in the same direction. For example. although there is a moderate positive correlation between income deprivation and skills deprivation (r = 0.54), both of these variables are negatively related to Brexit exposure, and many correlations between the criteria are small in absolute magnitude.
Had these weighted summary scores uniquely determined the scheme outcome, then most of the selected towns would have been from the North West (32) or from Yorkshire and the Humber (22). Instead, civil servants recommended that ministers choose a specified number of towns from each region, and produced a ranking based on towns’ performance within their region, forming the basis for three priority groupings on which decisions would be made. Ministers agreed to fund all ‘high priority’ towns, but exercised discretion when deciding which sixty further towns from the low and medium groupings to invite to apply.
Analysing the decisions
My analysis of the effects of political complexion upon invitation to bid for funding shows the following:The proportion of towns in Conservative-held seats which secured funding was much higher than the proportion for towns in all other seats (14 per cent compared to 9 per cent).
The data strongly suggests that marginal seats were targeted. The highest success rates are for seats where the Conservatives are slightly behind, but the rates for Conservative-held marginals are much higher than the rates for safe Conservative seats, which in turn are much higher than the rates for safe seats held by other parties.
When we split towns into medium and low priority groups, the success rate for Conservative towns in the medium priority group is very much higher than the rate for all other towns, and this pattern is repeated for the low priority group.
Testing the data
In order to test the possibility that Conservative towns might have had higher ranks, or scored higher on criteria which were prized by ministers but given shorter shrift by civil servants, I estimated a series of logistic regression models, predicting the odds of towns being selected. This showed that the effects of political variables are always significantly different from zero, whether or not we control for the other criteria formulated by civil servants. These effects are not just statistically significant, they are substantively significant. Compared to the reference category, and averaging across all the towns in the data, the effect of being in a Conservative ultra-marginal increases the chances of being selected by fully 45 percentage points (95 per cent confidence interval: 31 to 58 percentage points).
Bearing in mind the difficulty of establishing causality from observational data, a supplementary sensitivity analysis finds that in order to change our conclusions about the (positive) effects of being in a Conservative seat, the strength of the association between any ‘lurking variable’ and the outcome would have to be thirty times the strength of the association between rank and success. If ministers were not selecting on the basis of politics, they would have to be selecting on the basis of an incredibly potent factor which inexplicably did not occur to civil servants.
Pork barrel politics
In 1995, the Committee on Standards in Public Life set out a list of seven principles, known as the Nolan principles, intended to govern behaviour in public office. This list included a requirement to ‘act and take decisions impartially, fairly and on merit, using the best evidence and without discrimination or bias’. The evidence I have set out here suggests that two ministers (Robert Jenrick and Jake Berry, at all relevant times ministers in the Ministry of Housing, Communities and Local Government) took decisions that were biased in favour of Conservative marginals, decisions that the civil service accepted when given rationales for selection that now seem to be entirely adventitious. If we are not living in a post-Nolan world, it would be right for the Prime Minister to ask the Cabinet Office to investigate the matter formally.