Issues of electoral fraud, and voter registration and turnout targets Submission 3 to Yash Ghai Commission Professor Wadan Narsey

Issues of electoral fraud, and voter registration and turnout targets Submission 3 to Yash Ghai Commission Professor Wadan Narsey This Submission addresses two areas that the Ghai Commission could usefully examine and comment on. First is the continued Regime allegations that there was electoral fraud in the 2006 elections. Second, some advice based on international experience, on what might be appropriate and costeffective targets for voter registration and voter turnouts in a rurally dispersed electorate such as Fiji. The Regime selectively quotes the EU Report, pointing to disenfranchisement of certain voter groups (not stated who exactly), flawed registration processes, lack of integrity in the electoral roll, old traditional wooden ballot boxes being used with some political parties claiming that "the boxes had sufficient gaps beneath the lids to allow ballot papers to be inserted after the boxes were sealed", no recount of some close votes, a 101 percent. voter turnout in one constituency; and electoral officials favouring the SDL. These are the serious allegations of possible electoral fraud that the Ghai Commission must examine objectively using the facts, and either accept or reject these allegations once and for all. There have been other allegations which even the Yash Ghai Commission would know to be merely inefficiencies which are undesirable but little to do with possible electoral fraud: such as, inappropriate allocation of polling stations and ballot boxes; high levels of invalid votes (bad electoral system); Electoral Commission lacking funding, lack of institutional knowledge due to the downsizing of the Office of the Supervisor of Elections; the main voter roll not ready on time for public scrutiny which resulted in about 20,000 corrections; and the strange Regime reference to "only 12% of polling stations were being headed by women". This submission tries to assist the Ghai Commission with an analysis of the "big picture" allegation of electoral fraud by the Fijian SDL (presumably against the Indo-Fijian FLP) using the 2007 Census data produced by the Fiji Bureau of Statistics. On the issue of appropriate registration and voter turnout targets, a key statistical point that the Ghai Commission might wish to consider is that having another 10% or even 5% of potential voters registering to vote, or voting, would have been extremely unlikely to have made any difference whatsoever, to the elections outcome- either in the past, or in the future. [A qualification: Let me state at the outset that I hold no brief for the SDL, FLP, or any other political party, although I have been accurately described in the media as a former NFP Parliamentarian (which I was between 1996 and 1999). Following the 1999 elections, however, I ceased my political affiliation, although I still have friends from many of the political parties during my three years in Parliament. I was able to assist the Electoral Office during the 2006 elections, as well as all the political parties that attended my voter education workshops throughout Fiji in 2005 and early 2006. I now focus on the substance of this Submission. 1

What could be indicators of electoral fraud? I suggest to the Yash Ghai Commission that if there is any substance at all to allegations of widespread electoral fraud by the Fijian SDL against Indo-Fijian voters and parties, then (a) the numbers of Fijian voters registered as a proportion of the actual population aged 21 and over, would tend to be systematically higher than the similar proportions for Indo-Fijians, both in individual constituencies and in aggregate; and (b) the numbers of Fijians voting as a proportion of those registered to vote, would tend to be higher than the similar proportion for Indo-Fijians, both in individual constituencies and in aggregate. The facts in Annex 1, Annex 2 and Annex 3, to this submission suggest completely the opposite. The 2007 Census data The elections were held in 2006 and the Census was unfortunately postponed to 2007. [This was much to the unhappiness of the Fiji Bureau of Statistics demographers and the few of us who understand how critical it was to not break the hundred year old cycle of the ten year gap between censuses. For political purposes, it would have been far more sensible and cost effective to have the census first, so that the electoral boundaries could be more easily established, given the requirements of the 1997 Constitution. In the end, the costs were wastefully duplicated.] Regardless of that, anyone can go to the Fiji Bureau of Statistics website and download all the 2007 Fiji Census data, by single years. [My considered opinion is that, contrary to blog allegations, there has been no political interference with any FBS data for the last six years, despite the recently retired Government Statistician being the older brother of Commodore Bainimarama]. Add up the 2007 numbers of potential voters (aged 21 and over) for Fijians, Indo-Fijians and Others in 2007. To estimate the numbers of potential voters for 2006, reduce the Fijian number by 1.9% (that is the annual growth rate of Fijian voters). And reduce the Indo-Fijian number by a much smaller 0.1%, the growth rate of Indo-Fijian voters around 2007 (but note that the growth rate of Indo-Fijian voters has been negative for the last five years- expect fewer Indo-Fijian voters at the next election in 2014). You will get the following interesting table for 2006: Table 1 Fijians Indo-Fijians Registered voters in 2006 256014 204470 Estimated Number Of Voting age in 2006 261876 205723 Percentage registered 98% 99% The last row indicates that 98% of eligible Fijian voters were actually registered to vote in 2006. 2

But that was lower than the 99% of Indo-Fijian voters who registered. There is little possibility of hordes of non-existent Fijian voters being registered twice by the SDL or any Fijian political party. And what percentage of those registered voters actually voted? The last row of Table 2 tells you that 87% of registered Fijian voters actually voted, compared to a higher 89% of registered Indo-Fijians who voted. Table 2 Fijians Indo-Fijians Registered voters in 2006 256014 204470 Actually voting 222660 182476 Percentage voting: 87 89 Nationally, a higher proportion of potential Indo-Fijian voters were registered than Fijians. AND a higher proportion of registered Indo-Fijian voters actually voted, than Fijians. Whatever happened in that one Cakaudrove East constituency, certainly was not replicated throughout the constituencies in aggregate (see Annex 3) nor in individual constituencies (see Annex 2). Annex 2 shows that the Cakaudrove East result (of more voters than registered) was just one constituency out of 46, and only in 2006. There was no such result in either 1999 or 2001, when Fijian parties were also in control of the election processes. i.e. 1 anomaly out of 138 communal constituencies (and I show below that even that was trivial). The Yash Ghai Commission should insist on hard evidence from anyone who keeps alleging that there was widespread electoral fraud in the 2006 elections. If it wants to satisfy itself, the Yash Ghai Commission can commission similar analysis at the division and the province level. Just request and please pay for assistance from the last few remaining demographers at the Fiji Bureau of Statistics (before they emigrate to better paying jobs at regional organisations and elsewhere). To get the Ghai Commission started, I present Annexes 1, 2 and 3 at the bottom of the paper, of some analysis I did three years ago, to see if there was any evidence to support the allegations of electoral fraud in any of the constituencies. Annex B suggests that these allegations of electoral fraud are not substantiated by these numbers. Annex 3 shows that for all three elections (1999, 2010, and 2006) a higher proportion of Indo- Fijians registered, actually voted than did Fijians. What of the anomaly in Cakaudrove East, where there was indeed a 101% voter turn-out. 3

What of the Cakaudrove East anomaly? Of course, you cannot have 1% more voters than the number supposedly registered. But was this clear evidence of electoral fraud by the SDL, perhaps with their hand-picked electoral officials secretly stuffing the ballot boxes with extra votes for SDL, through gaps below the lids of wooden boxes? If you examine this anomaly closely, you find that not only was Cakaudrove East a small rural constituency (with only 7639 voters), but the "extra" 1% voters amounted to a mere 52 votes (that is right, fifty two). Right alongside was another Fijian constituency, Cakaudrove West, where a much larger 1987 registered voters did not vote. I would not be surprised to find that some voters registered in the Cakaudrove West mistakenly voted in Cakaudrove East. The Ghai Commission should also note that the SDL won Cakaudrove East with 6120 votes, and a massive margin of 5353 votes over all the others combined. 52 extra votes was a drop in that big bucket. I doubt if any one from the SDL would have bothered to cheat in that constituency, even if some political parties alleged that "the boxes had sufficient gaps beneath the lids to allow ballot papers to be inserted after the boxes were sealed"). The Ghai Commission should note that the FLP had one year in the Interim Government, and the Military Regime has had more than five years, to find any evidence of electoral fraud. They have not found any. Continued repetition of allegations of electoral without an iota of evidence should be rejected by the Ghai Commission, and seen for what they are: a refusal by political parties to abide by the "rules of the game" when the game goes against them, and other agenda. The other flimsy excuses Extremely strange are the Regime allegations that "only 12% of polling stations were being headed by women", as if that amounts to electoral fraud. The women members of the Ghai Commission would know that most female civil servants (and civil servants are usually the polling officers) will not want to be working at odd hours in polling stations, with their families worried about their safety, or probably more likely male family members clamouring at home: "who is going to cook the dinner?". Such a complaint is indeed strange coming from an all-powerful dictatorial Regime which has appointed only 1 female Minister in an otherwise all male Government, especially when one Superman is allegedly looking after 7 ministries of his own, and probably another 6 as well for the Boss. (Goodness me. The Ghai Commission could even recommend that the Fiji Cabinet can do with just 2 Ministers- one Superman, and one SuperWoman- to have gender balance! But paid one salary each, of course.) 4

The allegation that the composition of the polling staff did not reflect the balance of Fiji s ethnic communities may have some substance- but I suspect simply reflecting whoever volunteered for these tasks and perhaps some insensitivity of the SDL government to this issue- hardly any evidence per se of attempted electoral fraud by them. If the Regime is to be consistent about the issue of ethnic balance in electoral officers, the Ghai Commission might record in their Report that if the Regime continues to use the Fiji Military to conduct the bulk of the voter registration exercise, that will also reflect the 99% indigenous Fijian balance in the military while Fiji's ethnic balance would require 33% of these officers to be Indo- Fijian. As an economist, I would not object to the Military to be used to conduct the voter registration exercise, for three reasons: first, they are likely to be fair (what can they be unfair about?); second, it is a productive use of tax-payers funds currently spent on the military salaries; and third, running around the rural areas where even four-wheel drives cannot reach, might make them physically fitter (but I suspect that this would apply only to the senior officers, some of whose current media sideways profiles suggest that they could do with this exercise). Why do most Indo-Fijians still believe the allegations of SDL electoral fraud? I have little doubt that if the Ghai Commission were to ask a large number of Indo-Fijians if they believed that there was electoral fraud by the SDL in 2006 or 2001, I suspect the majority would say "yes". Most Indo-Fijians believed the FLP's allegations of electoral fraud in 2001 and 2006, and these allegations have not been retracted to this day. In Australia and NZ, there are also powerful media propaganda machines which keep peddling these views internationally, despite the lack of any hard evidence, and indeed despite any evidence to the contrary. It is important for the Ghai Commission to understand the reasons for this continued but misplaced belief. The harsh reality is that the Indo-Fijian community have not forgotten the 1987 and 2000 coups which removed their political leaders from control of government, and all the associated random and targeted violence against them. Those wounds have not healed and the few racist political statements since the 2000 coup have not helped either. Such violence has never been targeted against the general indigenous Fijians population even after the 2006 coup, although many have suffered violence at the hands of the military. It should be noted that while the elected Fijian leaders may have been deposed by the 2006 coup, they have been replaced by another set of Fijian leaders, albeit from the military. Two Indo-Fijian swallows in the Bainimarama Cabinet do not make for an Indian summer, however prominent one may be in the media. 5

It is to be hoped that the current rapprochement between all the political parties such as SDL, FLP, NFP, and UPP will result in genuine reconciliation between all the parties, most of whom have by now made the mistake of supporting one military coup or another. [NAP, SVT, PANU, BLV, MV etc. may surface one of these days- in one form or another, once they understand the likely advantages to themselves, should a proportional electoral system come into being for the 2014 elections.] What are sensible targets for voter registration and voter turnout? There is currently a frenzy of spending of tax-payers funds, on electronic methods of voter registration, with the objectives of improving the proportions of voter registration, and voter turnout. These are theoretically good objectives in themselves. But the Yash Ghai Commission should note that Fiji's registration rate and voting rates are already incredibly high by international standards. Have a look at the international comparisons here: http://en.wikipedia.org/wiki/voter_turnout#international_differences This sensible article points out that voter turnout depends on "trust in government, degree of partisanship among the population, interest in politics, and belief in the efficacy of voting. For Fiji, Annex 3 shows that the voter turnout rate declined for ALL ethnic communities between 1999 and 2001, not just for Indo-Fijians: why bother voting when the resulting government is going to be removed at gun-point? But even in 2001, the Fiji voter turnout rates were among the highest in the world. Note also, that the voter turnout in 2006 returned to the much higher levels of around 89% of 2006, indicating that the vast majority of voters were once more engaging with the electoral process. I submit that the Ghai Commission should note the following four aspects of voter turnout rates and voting effectiveness in the Fiji case. First, the proportions of invalid votes in future will almost certainly be drastically reduced by the likely changes in the electoral system and the simplification of the ballot papers. Secondly, for many voters who live far from the polling stations, especially rural indigenous Fijians, the logistics and costs of getting to the polling stations far outweighs any benefits of voting for their party of choice. (Regardless of other benefits that the two former Fiji parliamentarians on the Ghai Commission will remember, with mixed feelings no doubt, such as the free transport of voters, food, grog, and jovial company that usually awaits voters at polling stations, often merrily enjoyed without necessarily giving the bribing political aspirants, their vote, in the secrecy of the polling booth). The third point is that some 5% of potential voters in Fiji are currently aged 70 years and over, and this proportion is going to rise rapidly in the future given our demographic trends. A large fraction of this elderly group may have no wish to vote, or would find it physically onerous to 6

travel long distances to vote. That would leave a mere 5% of potential voters who do not vote for whatever reason- cost, illness on the day, or even very legitimate personal inclination such as total mistrust and dislike of all political parties and politicians. But the fourth and probably the most important point to consider is a statistical one, related to the ultimate objective of all elections, which is to identify accurately and fairly "who the people want to govern the nation" in their, and the public interest. How big a voter turnout do you really need? Every good statistician and Bureau of Statistics knows that if a proper random sample is taken of the entire country of voters, then a mere 5% (I repeat, a mere five percent.) would tell you quite accurately which party is likely to be the winner (don't take my word for it, go and ask a good statistician at the FBS or USP). This great statistical result is what household surveys by bureaus of statistics, good "opinion polls" or "exit polls" rely on, in the developed world. Nobody questions that a "sample" or "voter turnout rate" as large as 48% (which is apparently the voter turnout rate in US) or 58% (in the world's largest democracy, India) or 75% (in UK, the origins of the Westminister system) would give you statistically reliable results, accepted by wining and losing parties alike. The Ghai Commission should consider that increases of voter registration beyond 90% or voter turnout beyond 90% is extremely unlikely to change the result of any election: why would the last 10% of potential voters be any different in political views than the first 90% who have already voted?) All accountable and resource-scarce countries in the world understand that once you have reached the 85% mark (as Fiji already has) then the "marginal costs" of increasing both the registration rate and the voter turnout rate will result in negligible marginal benefits in identifying winning parties, while imposing great cost to tax-payers- as we may end up doing currently. The Ghai Commission should guard against costly and un-necessarily high targets for voter registration or voter turnout, especially when there are many more urgent needs for the use of taxpayers' funds, such as in poverty alleviation, health, education or rural development. All political parties would similarly gain, if they mutually agreed to not provide all the incredibly costly "bribes" that voters have come to expect from aspiring candidates, often discouraging poor candidates from standing. While this is something that cannot be enforced (even though there is absolutely no evidence that in Fiji such electoral "bribes" actually work), the Yash Ghai Commission might wish to make a recommendation on this issue, and the political parties might wish to come to some agreement on this (to reduce their own expenditures). Let the voters vote, based on their commitment. Conclusion I urge the Yash Ghai Commission to ensure that they do not repeat or give any credibility to any allegations of alleged electoral fraud in either 2001 or 2006, without definitive and objective evidence. 7

It is accepted that the Regime's new arrangements for electronic electoral registration, individual voter cards may be improvements on the past systems and should be welcomed by all the political parties- provided they are not too costly and they not suffer from glitches (have a look at the FBS disastrous belated attempt to use electronic scanners for the 2007 Census forms). However, I submit to the Yash Ghai Commission that they keep in mind that such minor improvements in the logistics of the electoral processes are extremely unlikely to make any great difference to the eventual election outcomes, or confer any significant benefits to the tax-payers and the nation. Far more useful for the country's improvement of electoral processes would be a genuine dialogue, rapprochement and the building of goodwill, between the political parties and the Military Regime, with independent NGOs as facilitating intermediaries. 8

Annex Tables Annex 1 Voters Listed Numbers Voting No Constituency Type 1999 2001 2006 1999 2001 2006 1 Bua Fijian Fijian Comm. 6357 6972 6749 5966 6050 6245 2 Kadavu Fijian Fijian Comm. 5845 6540 6089 5371 5328 5476 3 Lau Fijian Fijian Comm. 6807 7536 6612 6343 6197 5943 4 Lomaiviti Fijian Fijian Comm. 8131 8743 7650 7265 7009 6906 5 Macuata Fijian Fijian Comm. 9377 9964 9823 8545 8076 8956 6 Nadroga/Navosa Fijian Fijian Comm. 16051 17415 19044 14718 13672 16704 7 Naitasiri Fijian Fijian Comm. 11449 12488 12067 10511 10214 10874 8 Namosi Fijian Fijian Comm. 2856 3053 3340 2658 2531 3066 9 Ra Fijian Fijian Comm. 9570 10589 10880 8831 8586 9590 10 Rewa Fijian Fijian Comm. 6289 6832 7341 5798 5636 6675 11 Serua Fijian Fijian Comm. 3903 4065 4473 3630 3423 4112 12 Ba East Fijian Fijian Comm. 10019 11115 11836 9201 8955 10215 13 Ba West Fijian Fijian Comm. 12435 13141 15348 11076 10077 12650 14 Tailevu North Fijian Fijian Comm. 8946 9534 9682 8407 7838 8687 15 Tailevu South Fijian Fijian Comm. 8738 9635 10303 7938 7934 9389 16 Cakaudrove East Fijian Fijian Comm. 8054 8808 7587 7120 6923 7639 17 Cakaudrove West Fijian Fijian Comm. 9062 9855 11609 8426 8328 9622 18 North East Fijian Fijian Com.Urban 13234 14477 17155 10785 10618 14560 19 North West Fijian Fijian Com.Urban 15307 16306 18864 12965 11531 15550 20 South West Fijian Fijian Com.Urban 12070 13215 15093 10174 9728 12518 21 Suva City Fijian Fijian Com.Urban 11653 12663 12707 9914 9337 10435 22 Tamavua/Laucala Fijan Fijian Com.Urban 12573 13701 16068 10801 10139 13491 23 Nasinu Fijian Fijian Comm. 11538 12417 15694 9857 8980 13357 24 Suva City General General 3772 4107 3523 3231 2956 2896 25 North Eastern General General 4556 4894 4701 3860 3694 4042 26 Western/Central General General 5701 5942 5593 4890 4328 4657 27 Vitilevu East/Maritime Indian Ind.Comm. 7760 8230 7256 7324 7006 6621 28 Tavua Indian Ind.Comm. 8477 9197 8536 8070 7873 7912 29 Ba East Indian Ind.Comm. 10049 10487 8203 9394 8912 7532 30 Ba West Indian Ind.Comm. 10188 11240 11538 9450 9149 10155 31 Lautoka Rural Indian Ind.Comm. 9667 10253 11200 9104 8304 9841 32 Lautoka City Indian Ind.Comm. 11849 12356 12308 10806 9285 10634 33 Vuda Indian Ind.Comm. 11286 11584 10526 10413 9316 9239 34 Nadi Urban Indian Ind.Comm. 12336 13019 13081 11437 10088 11453 35 Nadi Rural Indian Ind.Comm. 9678 10160 11467 9079 8629 10394 36 Nadroga Indian Ind.Comm. 11179 11833 11240 10552 9879 10350 37 Vitilevu South/Kadavu Indian Ind.Comm. 7839 8290 8407 7222 6623 7586 38 Suva City Indian Ind.Comm. 13280 14435 12568 11837 10055 10618 39 Vanualevu West Indian Ind.Comm. 8839 9186 7754 8200 7612 7193 40 Laucala Indian Ind.Comm. 14453 15343 18610 13171 11374 15983 41 Nasinu Indian Ind.Comm. 12090 13075 14789 11218 10393 13327 42 Tailevu/Rewa Indian Ind.Comm. 10875 11519 11641 10257 9108 10525 43 Labasa Indian Ind.Comm. 9668 9996 10248 8793 8148 8986 44 Labasa Rural Indian Ind.Comm. 9775 10113 7416 8806 8568 7012 45 Macuata East/Cakaudrove Indian Ind.Comm. 8332 8721 7682 7641 7203 7115 46 Rotuma Rotuman Comm. 5232 5567 5373 4682 4255 4737 9

Annex 2 Percent. Voting Percent. Not Voting 1999 2001 2006 1999 2001 2006 1 Bua Fijian Fijian Comm. 94 87 93 6 13 7 2 Kadavu Fijian Fijian Comm. 92 81 90 8 19 10 3 Lau Fijian Fijian Comm. 93 82 90 7 18 10 4 Lomaiviti Fijian Fijian Comm. 89 80 90 11 20 10 5 Macuata Fijian Fijian Comm. 91 81 91 9 19 9 6 Nadroga/Navosa Fijian Fijian Comm. 92 79 88 8 21 12 7 Naitasiri Fijian Fijian Comm. 92 82 90 8 18 10 8 Namosi Fijian Fijian Comm. 93 83 92 7 17 8 9 Ra Fijian Fijian Comm. 92 81 88 8 19 12 10 Rewa Fijian Fijian Comm. 92 82 91 8 18 9 11 Serua Fijian Fijian Comm. 93 84 92 7 16 8 12 Ba East Fijian Fijian Comm. 92 81 86 8 19 14 13 Ba West Fijian Fijian Comm. 89 77 82 11 23 18 14 Tailevu North Fijian Fijian Comm. 94 82 90 6 18 10 15 Tailevu South Fijian Fijian Comm. 91 82 91 9 18 9 16 Cakaudrove East Fijian Fijian Comm. 88 79 101 12 21-1 17 Cakaudrove West Fijian Fijian Comm. 93 85 83 7 15 17 18 North East Fijian Fijian Com.Urban 81 73 85 19 27 15 19 North West Fijian Fijian Com.Urban 85 71 82 15 29 18 20 South West Fijian Fijian Com.Urban 84 74 83 16 26 17 21 Suva City Fijian Fijian Com.Urban 85 74 82 15 26 18 22 Tamavua/Laucala Fijan Fijian Com.Urban 86 74 84 14 26 16 23 Nasinu Fijian Fijian Comm. 85 72 85 15 28 15 24 Suva City General General 86 72 82 14 28 18 25 North Eastern General General 85 75 86 15 25 14 26 Western/Central General General 86 73 83 14 27 17 27 Vitilevu East/Maritime Indian Ind.Comm. 94 85 91 6 15 9 28 Tavua Indian Ind.Comm. 95 86 93 5 14 7 29 Ba East Indian Ind.Comm. 93 85 92 7 15 8 30 Ba West Indian Ind.Comm. 93 81 88 7 19 12 31 Lautoka Rural Indian Ind.Comm. 94 81 88 6 19 12 32 Lautoka City Indian Ind.Comm. 91 75 86 9 25 14 33 Vuda Indian Ind.Comm. 92 80 88 8 20 12 34 Nadi Urban Indian Ind.Comm. 93 77 88 7 23 12 35 Nadi Rural Indian Ind.Comm. 94 85 91 6 15 9 36 Nadroga Indian Ind.Comm. 94 83 92 6 17 8 37 Vitilevu South/Kadavu Indian Ind.Comm. 92 80 90 8 20 10 38 Suva City Indian Ind.Comm. 89 70 84 11 30 16 39 Vanualevu West Indian Ind.Comm. 93 83 93 7 17 7 40 Laucala Indian Ind.Comm. 91 74 86 9 26 14 41 Nasinu Indian Ind.Comm. 93 79 90 7 21 10 42 Tailevu/Rewa Indian Ind.Comm. 94 79 90 6 21 10 43 Labasa Indian Ind.Comm. 91 82 88 9 18 12 44 Labasa Rural Indian Ind.Comm. 90 85 95 10 15 5 45 Macuata East/Cakaudrove Indian Ind.Comm. 92 83 93 8 17 7 46 Rotuma Rotuman Comm. 89 76 88 11 24 12 10

Annex 3 Percentage voting in elections of Percentage Not Voting in elections of 1999 2001 2006 1999 2001 2006 Fijian Communal 89 78 87 11 22 13 Indian Communal 92 80 89 8 20 11 General Communal 85 73 84 15 27 16 Rotuman 89 76 88 11 24 12 Percentage Change 1999 to 01 2001 to 06 Fijian Communal -12 11 Indian Communal -13 11 General Communal -14 14 Rotuman -15 15 11