Full letter to the Guardian on Universal Basic Income

I recently had a letter published in the Guardian, here. I hacked it down considerably though to get under the word limit though. Below is a slightly more extended version. (I will spare you the original, even longer version!)


Dr Tom Holden

School of Economics, University of Surrey, Guildford, GU2 7XH



Dear Editor,

Monday’s Guardian contained a discussion of a new paper from Compass on Universal Basic Income (UBI). All of that paper’s calculations are predicated on changes to taxes and benefits having zero effect on labour supply decisions. Aside from being wildly implausible, such an analysis removes the key benefit of a UBI, that, if properly constructed, it increases labour supply, particularly amongst low income households.

The existing system creates a “benefits trap” where the removal/tapering of benefits means that the effective marginal tax rate can be astronomical on low incomes. While Universal Credit aims to smooth this, the tapering still leads to high effective marginal tax rates for some low earners. Indeed, IFS research suggests that around 700,000 still face effective marginal rates over 70% even under Universal Credit. By making benefits payments unconditional on income or employment status, a UBI solves this problem, bringing people back into the labour force.

UBI also enables simplifications to the tax system which reduce its administrative cost, enhance work incentives, and increase its fairness. The complications of the existing tax and benefit system both make it hard to administer, and mean that individuals cannot readily calculate how a change in their situation will change their tax liability. The Mirrlees review makes assorted wise suggestions for simplifying the system, but the presence of a UBI would enable us to go further, with a switch to a flat income tax that would also replace National Insurance. The system would still be progressive overall, thanks to the UBI payment being far more significant for people at the bottom of the income scale.

UBI would also enable us to increase VAT in a progressive way. In particular, a rise in the UBI level funded by a rise in VAT would leave those at the bottom of the distribution unambiguously better off. While reporting often focusses on wealth or income inequality, these are undesirable only to the extent to which they lead to consumption inequality, which VAT tackles directly.

In the Compass paper, much is made of the fact that their system leaves most groups with higher income. They do this by preserving virtually all of the complexity of the existing tax and benefit system, while adding additional taxes on the rich. Even aside from complexity considerations, this is fundamentally the wrong approach. Firstly, under a well-constructed UBI, average unemployment durations will be shorter due to increased incentives to work, and the net income of those on the minimum wage will be higher. Both facts mean self-insurance through savings will be more feasible, hence it is appropriate that the unemployed have lower income under a UBI than at present. (To support this, by default UBI payments should be paid into a government provided current account paying the market interest rate.) Secondly, the introduction of a UBI gives an opportunity to wipe clean the old inconsistencies of the tax system, which should not be wasted. Finally, there is no good reason why the cost of introducing a UBI should fall only on those with the very highest incomes. UBI is about protecting the very poorest in our society. It is only fair that these costs should be borne by those with middle incomes as well.

In the past, UBI has been advocated by figures across the political spectrum. It is vital that the progressive centre reclaim this idea before it becomes permanently tied to out-dated views from the far left.

Yours sincerely,

Dr Tom Holden

Lecturer in Economics, University of Surrey

Attaining proportional representation without giving up the constituency link

Traditional forms of proportional representation (PR) lead to one or more of the following undesirable features:

  • large multi-member constituencies (undesirable as it leads to weaker ties to any particular MP, and because it inevitably results in very disparate regions being lumped in together),
  • MPs without a constituency, or with a “regional” constituency that overlaps standard ones (undesirable as such MPs have different workloads, and weakened responsibilities to represent those who elected them),
  • party lists which make it impossible to vote against particular individuals without voting against their party (undesirable as it should be the voters, not political parties, who have the final say as to which individuals get elected).

Given these problems, one might be tempted to abandon PR entirely. In this post, I will give a mechanism without any of these problems, which nonetheless comes close to proportionality.

In my proposed mechanism, constituencies would remain their current size, and there would be no reason to reshape their boundaries that is not already present. On election day, voters would be presented with two ballot papers. The first paper would ask something along these lines:

“Which of the candidates below would you be prepared to have as your MP? Please place an X next to the name of any candidate who you would be prepared to have representing you in parliament. You may place as many Xs as you wish, indicating your support for no, one or many candidates.”

Along the lines of standard approval voting, the number of Xs for each candidate would be summed up, in order to produce a percentage approval for each local candidate. The candidate with the highest approval would be selected if one of the following was true:

  • At most one candidate obtained over 50% approval.
  • Two or more candidates obtained over 50% approval, but the one with the highest approval was not affiliated to a national political party.
  • Two or more candidates obtained over 50% approval, but all or these candidates except perhaps the one with the highest approval were not affiliated to a national political party.
  • The vote was for a by-election.

However, if none of these situations obtain, so multiple candidates from national political parties obtained over 50% approval, then each of them would be considered a possibility, and the one selected would be determined by trying to match the distribution of MPs to voters overall preferences, as determined by the second ballot paper.

This second ballot paper would ask something along these lines:

“Which political party would you like to form a government? Please place an X next to the name of the single political party you would like to lead the government. You should place precisely one X on this ballot paper.”

These ballot papers would be summed up nationally, not at constituency level, in order to give a picture of what a representative parliament ought to look like. Then, in the constituencies in which multiple candidates obtained 50% approval, candidates would be selected in order to bring the distribution in parliament as close as possible to the results of the second paper. The technical details of this are given below, but as a practical way of understanding how this would work, suppose that in the eventual distribution of MPs, party A was over-represented and party B was under-represented, according to the proportionality criterion. Then in any constituency in which MPs from both party A and party B obtained over 50% approval, the candidate from party B would be selected, even if they had lower approval that the candidate from party A. While selecting candidates with slightly lower approval may not be ideal, any candidate with over 50% approval is supported by a majority of voters, and so may legitimately represent the views of the constituency’s voters.

A natural concern with this system is that the result might not be fully proportional, as too few constituencies would have multiple candidates with over 50% approval. However, the system is sure to be an improvement over the current one at least, and sacrificing full proportionality seems a small price to play to preserve the constituency link. Furthermore, given the make up of the UK political landscape, in fact many constituencies might end up with multiple candidates with over 50% approval. For example, many Labour and Conservative voters would also support a Liberal Democrat MP, so we can expect support for the Liberal Democrats on the first ballot to be high. However, given that they would have relatively low support on the second ballot, they would still not be over-represented in parliament. More generally, this system gives voters greater freedom to vote for minority candidates on both ballots. Indeed, a good approximation to a voter’s optimal voting strategy on the first ballot is to vote for whichever of the expected winner and the expected runner-up they prefer most, and then to also vote for every candidate they prefer to these. This means that parties like the Greens or UKIP may be able to get 50% approval in sufficiently left/right wing constituencies, despite not having a track record there. Again though, it is worth stressing that such parties will not obtain significant numbers of MPs unless they can also convince the broader electorate on the second ballot.

I conclude by giving the full details of the procedure for assigning seats and ensuring proportionality. First, we make an initial allocation by assigning each constituency to the candidate who obtained the highest approval, breaking any exact ties by the result of the second ballot. Next, we form a list of all of the candidates from national political parties who obtained over 50% approval in their constituency, but are not selected to win in it, and we give them a score equal to their approval minus the approval of the candidate currently allocated to win in their constituency. We then sort this list, and step through it starting with the highest score. We continue stepping through until we find a candidate from a political party that is under-represented nationally, where the winner of that constituency is from a political party that is less under-represented (comparing percentage errors, not numbers of seats). If after changing the winner of the seat to the currently considered candidate, it would still be true that that candidate’s political party is more under-represented than that of the previously allocated winner, then we make the swap, otherwise we continue searching through the list. Once we have made a swap, the new winner is removed from the list, and the old winner is added to it, and we recommence our search from the start of the list. Eventually, we will complete a scan of the entire list without making a single swap, at which point the algorithm terminates, and the current allocation of winners to seats is accepted.

To me, this seems like a good compromise between the conflicting aims of proportionality, simplicity, the representation of preferences and the preservation of constituency links. Are there any obvious flaws that I have missed?

Jean Tirole wins the 2014 Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel

Tirole has made many huge contributions to the field of Industrial Organisation that we’re studying this semester. I know I’m a bit late on this as I’ve been away, but here’s a nice summary of why Tirole deserved to win the prize.

We’ll be seeing some of his work later in the Industrial Organisation course, and this perhaps gives you an additional reason to consult his excellent textbook.

For more information, read this non-technical introduction to his work, or this more comprehensive one.

Welcome to my site

Thanks for visiting this site.

The site is currently under development, so may be in a state of flux for the next few weeks.

I am unsure as yet how often I will be posting in this blog, but I will try to at least share interesting econ links here as I stumble across them.