I thank the Chairman and members for inviting me and giving me an opportunity to make this presentation. I covered two areas in my written submission. One was constituency boundary revision and the other related to the political implications of change in the electoral formula. At the committee's request, I will address only the first of these two areas.
I propose to make three points at some length. The first relates to problems with the current mechanism for constituency boundary revision. I will look at what I describe as the normal mechanism under proportional representation as practised in continental Europe. I will then look at the practical implications of a switch to the European mechanism in this country. I have outlined these arguments at much greater length in a set of articles in Administration in 2007 and 2008.
The first point relates to problems in the current system. To varying degrees, it seems that this system arguably alienates the public. It creates unfamiliar and unstable boundaries that breach those of recognised territorial units. Deputies already made this point earlier. There is some evidence that this causes significant electoral alienation. Colleagues, like Adrian Kavanagh in NUI, Maynooth, have made this point.
From the perspective of politicians, a certain element of irritation must creep in. The unpredictability of the future shape of constituencies, for example, is something that cannot be pleasing to those contesting elections. It arguably wastes the time of officials, it is an inefficient use of resources and it is expensive. I will try to justify this point in a moment.
The number of constituencies created since 1922 is enormous. The number is 348 and by my computation, we only needed 35 over that period of time. We could have started off with 25 constituencies in 1922. A few amendments could have been made since then resulting in ten additional constituencies.
These constituency boundaries are largely unfamiliar to voters and even to Deputies. The use of micro-level units, such as electoral divisions and even townlands and in urban areas imaginary geometric lines, must bewilder many people. Traditional boundaries are commonly breached. The case of Leitrim is the most recently celebrated one. These breaches are arguably often unnecessary. By way of illustration of this, I mention the case of Leitrim. Why not group it permanently with Sligo and have a constituency of Leitrim-Sligo or Sligo-Leitrim? Similar arguments could be made in respect of other counties.
We get extraordinarily long descriptions of constituency boundaries. The annexes to electoral Acts typically extend to something in the region of 3,000 words. Again, by my computation, 350 words would be sufficient if larger territorial units were used.
There has been only one extraordinarily stable constituency created in 1920 by the Government of Ireland Act, which was then King's County-Queen's County and is now Laois-Offaly. This is the only constituency that had unaltered boundaries over a very long period of time. However, it has fallen victim to the Electoral (Amendment) Act 2009 which has made a small change.
If one looks at the new definition of this constituency, it consists of the County of Laois and the County of Offaly, except for ten named electoral divisions which are being transferred to Tipperary North. If we look at these ten electoral divisions, even Deputies from that part of the country would have great difficulty recognising their borders. There is very good reason for this. They were last used for administrative purposes in the local elections of June 1914 — for election to the board of guardians of Roscrea Poor Law Union and to Roscrea No. 2 Rural District Council. These bodies are not recognised now. They are still used by the CSO for purposes of statistical reporting but they are otherwise unfamiliar to the public, and I believe, to Deputies.
How are boundaries drawn elsewhere in countries that use proportional representation? I emphasise the comparison I am making is not with the British system or the American system in which the committee has seen the number of constituencies where different issues arise, but with the continental European system. The typical approach is that territorial boundaries are fixed, they correspond with administrative units — for example, with the Swiss cantons and with the provinces in Belgium, except that one province, the central province of Brabant, is divided — and these administrative units remain stable over the decades. The target size of the chamber is defined and using a simple mathematical formula, seats are allocated periodically between these constituencies on the basis of population after each census. Over time, some boundary changes may be made to avoid excessively large or excessively small constituencies, but these changes are very infrequent and, typically, the new constituencies are instantly recognisable.
This is applied in all countries of continental Europe using proportional representation. There are some special cases such as Germany with its two-tiered system, Italy where there is no prior allocation of seats to the regions which have a function partly as constituencies, or the Netherlands where there is a single 150 member constituency covering the entire country. Obviously, in the case of single-member districts the system cannot work and boundaries must be continually changed to ensure that the principle of electoral equality is maintained.
My third general point has to do with the possible application of this system to Ireland. We immediately encounter a difficulty here, that is, that the single transferable vote system of proportional representation imposes a practical upper limit on constituency size. We cannot have 75 member or 150 member constituencies under this system because it is unrealistic to invite voters to rank the large number of candidates that would then result.
However, in the past in this State we have had relatively large constituencies. For example, in the Senate election of 1925 there were 19 vacancies, 76 candidates and 67 counts. Unfortunately, the count went on over three weeks. It discredited this system, perhaps unfairly, at the time. However, there have been other instances of very large constituencies: Galway, from 1923 to 1935, and several other constituencies up to 1947. In the Northern Ireland Assembly, from time to time there are relatively large constituency sizes.
In showing how this system might be applied in Ireland, I assumed a minimum of constituencies with three members and a maximum size of nine to show what this would look like. Essentially, the counties would become constituencies with certain exceptions. Some small counties would be merged involving familiar enough combinations such as Leitrim-Sligo, Cavan-Monaghan and Carlow-Kilkenny. Cork and Dublin would have to be divided. Under a list system, it would not be necessary to divide them but were we to impose an upper limit of nine Deputies per constituency, these counties would have to be divided. The 1923 Act came close to having this kind of pattern. It stuck closely to the existing county boundaries. Then one would need a decision on the overall size of the Dáil, whether it is to have 166 Deputies or some other number. Then seats would be allocated in a simple mathematical process. Typically, in other countries an official the equivalent to the director of the CSO would make this allocation.
I worked out carefully the result of this and the committee will find it in an appendix to my written submission where there are four tables where I look at what the outcome would have been had this system been in operation since the 1920s. With stable boundaries certain counties where population was dropping would see a fall in their number of Deputies. For example, Mayo would drop from nine to five over this long period — it would now be a five-seater having started off as a nine-seater.
Constituencies near Dublin would see an increase. Kildare would have risen from three to seven seats, and both Meath and Wicklow would have risen in seat numbers. Some would have remained stable and there would have been very rare changes in constituency boundaries. Dublin would have required four changes, in 1936, 1961, 1979 and 1986, because of the rapid process of population expansion there. Roscommon would have fallen eventually below the threshold for being a three-member constituency and I suggest in my simulation that this would be linked to Sligo-Leitrim following the 1971 census.
An assessment of this system shows it would be relatively stable, predictable and efficient. It would result in some large constituencies, and a nine-member, an eight-member or even a seven-member constituency would cause certain political cultural problems because since 1947 we have been accustomed to the notion of five-seat constituencies as the maximum size.
However, it would ensure greater compatibility with the Constitution. It is easy to forget that the Constitution stipulates, not only the single-transferable vote but proportional representation. With an average constituency size of four members where the electoral threshold is 20%, one could argue that a party with 15% of the vote which is spread nationwide would end up without any seats and that this is not proportional representation. It might result in slightly larger deviations from the present Deputy-population ratio, possibly resulting in conflict with the Constitution. This would depend on how the Constitution is to be interpreted.
I summarise in a slide in my presentation the kind of spread that would have existed under the system I suggest here over time — a range or deviation of 17%, which is quite high, in 1991 to 29% in 1971. This refers to the spread between the lowest and the highest Deputy-population ratio. The current system has ranged from 8% in 1961 — the lowest ever — to 28% in 1947. Allocation would be much quicker than under the current system. It might require a constitutional amendment; this is unclear.
In my presentation, I summarised the effects of this system on the Deputy-population ratio. The red bars refer to the system that I have simulated in my submission to the committee and one will note that the deviation from the average can sometimes be quite high. However, one may compare this with the existing electoral Acts, which are shaded in green, and the presentation shows the 1923 Electoral Act based on the 1911 census; the 1935 Electoral Act based on the 1926 census, and so on, where the deviations were often quite high too.
I remind the committee that I have double-checked the underlined portion of this text because the implications of this seem to be devastating, and I will read it carefully. Of eight constituency commissions since 1979, not one saw its recommendations translated into legally defined constituencies which were used at a general election before the next census intervened. Admittedly, in 1980 and in 1998, the results of the new census were not available, but a population census had taken place. What I mean by this is that if we look at the first electoral commission, reported in 1980, the results of which were based on the 1979 census and enacted in legislation in 1980, the first election under its terms took place in June 1981, but a new census had already taken place, in April 1981. The two elections of 1982 both took place under 1979 census data, not under 1981 census data. This pattern has been repeated ever since.