Brexit and Hate Crime

After the UK voted to leave the EU it was widely reported that there was an increase in hate crime. I was initially fairly sceptical of this and thought that a large proportion of this increase was likely to be explained by individuals being more likely to report crimes committed against them as hate crimes. For example if individuals believed that the Brexit campaign involved significant levels of xenophobic rhetoric they might be more likely to interpret crimes committed against them as being racially or religiously motivated. This effect would likely be compounded by reporting programmes like True Vision. In addition, the hate crime increase appeared to affect LGBT people despite the Brexit campaign barely touching on LGBT issues – leaving no clear causal link between Brexit and the increase in hate crimes committed against LGBT individuals.

The chart below shows quarter on quarter changes in hate crime across England and Wales. There appears to be some degree of seasonality. As a result even though there was an increase in hate crime in the quarter leading up to and including the period immediately after the vote (Q2 – April-June), this appears unremarkable compared to the similar quarter on quarter changes in previous years. What does stand out is the period after that (Q3 – July-September), where the increase in hate crime was larger than usual.

 The task of disentangling whether the increase was real or driven by reporting is a non-trivial one – at first glance both might equally explain the change.  If we assume that some small but fixed proportion of Brexit supporters committed these crimes then we would expect to see regional variation in the increase in hate crime related to the proportion of people voting for Brexit. If the change was driven by reporting we would probably expect to see a more uniform change across the country (although might still see some effect, if people in predominantly Leave supporting areas were more likely to be worried about hate crime and flag crimes within this category or if police forces tended to prompt individuals in these regions more about potential hate crime links).

The police crime and outcomes data are reported on the basis of Police Authorities, which can then be used as a basis to aggregate the Brexit referendum results data. All crimes which are listed as “religiously or racially aggravated” are tagged as hate crimes. The City of London is excluded from this analysis as the profile of the small number of residents who live there likely differs substantially from that of the individuals who work or visit the square mile. Running a regression of the Q3 proportional increase in hate crime on the proportion voting Brexit produces the following results:

HATECRIME16Q3 = -0.279 + 1.061*LEAVE
se:                              (0.387)   (0.696)

The relationship is statistically insignificant (p=0.14) although this isn’t entirely surprising given the low number of observations able to be used for this analysis (n=42). Furthermore the proportion voting Brexit only appears to explain around 6% of variation in hate crime – which suggests that this is not an overwhelming driver. Although this doesn’t meet conventional levels of statistical significance these results do at least suggest that there is more likely to be a correlation than not.

There are however a few major cautions in believing even this limited story. First if you look at the quarter leading up to Brexit when most of the supposedly “toxic” campaign rhetoric emerged there is no correlation between the increase in hate crime and the regions that voted leave:

HATECRIME16Q2 = 0.391 – 0.235*LEAVE
se:                              (0.370)  (0.665)

Secondly if you look at the quarter-on-quarter increase in hate crime in Q3 there is a much more substantial correlation with voting for Brexit for 2015 than 2016, when the campaign actually happened. This result is statistically significant (p=0.014) and explains a greater proportion of variation than the subsequent year (R^2 = 0.142 as compared with R^2 = 0.055). As a result it may be the case that there are other reasons for variation in hate crime that happen to correlate with voting Leave but are unrelated to the Brexit vote itself:

HATECRIME15Q3 = -0.632 + 1.277*LEAVE
se:                               (0.276)   (0.497)

This is clearly insufficient to give a definitive answer on what caused the increase in hate crime. However it suggests that at the very least the story is a lot more complicated than the one that we have been left to infer – that Brexit supporters whipped up by xenophobic rhetoric went on to commit these crimes. There may of course be reasons why those most likely to commit hate crimes were more evenly distributed than Leave voters – although these don’t seem immediately obvious.

The Uterus Tax

Today the Independent published an article arguing that the UK ought to remove the 5% VAT paid on sanitary items given a broad range of non-essential goods and services are currently zero-rated or tax exempt, including exotic meats. While I enjoy a good kangaroo steak as much as the next guy, it seems pretty obvious that there is something wrong with the UK’s tax system. Sanitary items are not the only goods that seem less essential than “edible sugar flowers” and “alcoholic jellies” to be taxed at a higher rate – also included are insulation and children’s car seats within the same 5% VAT band. Removing one apparent absurdity will not be enough to change a system of inconsistency. Instead the UK needs more radical reform of its tax system.

Obviously there are arguments to be made that essentials ought to be taxed more in an optimal taxation system. If a government uses the Ramsey Rule to minimise the deadweight loss of taxation then goods ought to be taxed in proportion to the inverse of their elasticities. This provides a compelling argument for why increases in excise taxes on tobacco might be desirable past the point where the external costs of smoking are internalised in the consumption decision. I will however ignore this discussion as it seems fairly clear there are political constraints that would render such a tax system untenable. Given the current plethora of exemptions for essentials like bread and milk it is difficult to imagine any government implementing such a hypothetical optimum. A more realistic choice is eliminating the complexity and maintaining a single rated VAT for all goods and services.

Using OECD data, in 2003 the UK had a c-efficiency ratio of just 46.4. This compares to 96.4 in world leading New Zealand. The c-efficiency ratio is calculated by determining VAT revenue as a percentage of the hypothetical tax revenue if all consumption had been taxed at the non-exempt and standard tax rate. In general countries with higher c-efficiency ratios will tend to have lower average headline tax rates, as a result of having a broader tax base. The ratio suggests that the UK is losing over half of the potential revenue of consumption taxation (assuming no changes in consumption) to a combination of non-compliance and more importantly exemptions. The EU places some constraints on the ability to move towards a broad low rate structure. Imagine an economy currently exempting half of all consumption tax and with a headline rate of 20%. They might decide to lower the tax rate to 10% but remove all discrepancies. This is impossible because of an EU directive that sets the minimum headline VAT rate at 15%. Incidentally EU rules are also the reason why the petition to zero-rate sanitary items will fail. It was intended that the lowest rate of VAT would be 5% across the EU VAT area. The UK, Ireland and others have grandfather clauses that allow the continuation of existing zero ratings but not the addition of new items.

Despite these constraints the UK ought to move to replicating best practice in other countries and creating a low rate and low exemption VAT system insofar as this is possible. This clearly increases consistency, avoids regulatory capture and means the tax system is easy to understand and without relative price distortions. Without this clarity we instead find farcical cases like Nix v Hedden where SCOTUS was left to clarify whether a tomato was a fruit and Lansell House v Commissioner of Taxation (H/T: Kiwiblog) where the Australian High Court decided whether a Mini Ciabatte product was a bread or a cracker.

The primary judge considered he was able to form his own view as to the significance of similarities or differences between biscuits, crackers and bread, having regard to the evidence before him. Various indicia were considered in comparing Mini Ciabatte with crackers and that, with the exception of water and yeast, the ingredients of Mini Ciabatte fell within the ranges of the ingredients for crackers. The primary judge attached little significance to the fact that water and yeast were outside the range of those ingredients in crackers, on the basis of the appellants’ evidence that there may be exceptions with crackers that fall outside those ranges and that the percentages of ingredients of the Mini Ciabatte were ‘rough estimates’.