Final Report - Analysis of the Financial Impact of Electronic Voting in Portugal

In order to find out information to weigh up whether to adopt an in-person e-Voting system in polling stations, The Knowledge Society Agency (UMIC) commissioned Deloitte to carry out a study in 2006, the report from which, entitled “Final Report – Financial Impact Analysis of E-Voting in Portugal”, was concluded in March 2007.

This study has helped clarify the following aspects, as summarised below:

• Cost of the traditional voting model in Portugal;
• International benchmarking of e-Voting adoption;
• Estimated costs for an e-Voting system in polling stations on Election Day;
• Legal aspects and the need for legal amendments.
  

Cost of the traditional voting model in Portugal

The Deloitte study holds that the costs of elections in Portugal differ according to the type of election and have the orders of magnitude presented in the following table:

Cost of the traditional voting model in Portugal in different types of elections

Acesso alternativo:
pressione tabela abaixo para a ampliar

Source: Deloitte.

It can be seen that the highest portions of the cost of elections are remuneration of polling station staff and the cost of “airplay” in the media. These are followed by the cost of purchasing paper and printing ballot papers and transporting them, which make up between 3% and 4% of the total.

Thus the study concluded that:

“All things being equal, it is not foreseen that adopting the e-Voting system could significantly influence the cost categories that have the greatest impact on the traditional voting model."

International benchmarking of e-Voting adoption

The Deloitte study presented a summary table comparing e-Voting aspects in political elections in 9 countries. This was then used as a basis for the following table with updated information:

International comparison of the adoption of e-Voting in 9 countries

Alternative access:
click on table below to enlarge

Source: Deloitte.

The main conclusions from the international benchmarking process with the countries examined were as follows:

1. The “vote anywhere in the country" option that Portugal is interested in, i.e. in person at polling stations, was only allowed in the Netherlands in 2004 and 2006, but e-Voting was cancelled by this country in May 2008 due to the possibility of fraud.
   
   
2. Of the 9 countries in the study, e-Voting has only fully been rolled-out in India, and only has a significant incidence (of 50%) in one other country – Belgium. In both cases “voting anywhere in the country” is not possible.
   
   
3. Voting online is only available to all voters in Estonia, where it is only used by 3.4% of voters. For voters living abroad, it was made available in a test phase in France and the Netherlands, although the number of voters who voted online was low.
   
   
4. Information on e-Voting costs in the countries examined in the study is scarce and irregular. As regards e-Voting in polling stations, the cost per voter in Belgium is estimated at three times that of paper voting. In experiments conducted in the United Kingdom, the costs per voter using electronic devices varied from 100 to 600 times more than voting on paper. In Ireland, where the number of voters is roughly 2.7 times lower than the number in Portugal, 41.2 million Euros were spent on voting machines and publicity without e-Voting being mandatory due to security-related doubts.
   
   
5. It is possible to vote electronically at specific polling stations in Germany, France and the United Kingdom, but use is experimental and the number of these polling stations is very low. Furthermore, the voting machines used in Germany and France are the same as those used in the Netherlands, where security faults were detected, leading to them being withdrawn from use in May 2008.
   
   
6. There have been setbacks in the adoption of e-Voting in Ireland, Norway and, more recently, but with a greater impact, considering that the system had been widely used and commonplace for many years, the Netherlands.

A broader international overview can be gained at Experiments in and Adoption of e-Voting in Political Elections Worldwide.

Estimated costs for an e-Voting system in polling stations on Election Day

The estimated costs calculated in the Deloitte study are based on a series of assumptions regarding the voting model and cost prediction, plus the answers received from two potential suppliers of the three consulted.

• Making the equipment profitable over 10 years, the anticipated material obsolescence timeframe.
• Assumption of 2 national elections of each type over the 10 years, therefore not including regional elections.
• Repayment of investment costs through assigning to the total number of voters registered.
• Assumption that savings will be made on paper, based on the idea that no paper trail audit will be required, in contrast to what is being adopted in several other countries due to detected security risks and a loss of transparency.

The costs presented by the two suppliers vary excessively, with one putting forward costs that are between double and two and a half times more than those presented by the other. This fact causes considerable uncertainty with regard to the foreseeable costs assessed in the study. Suppliers are likely to underestimate expenses in their estimates for this type of study, perhaps because they have to provide a voluntary free evaluation which has to be carried out without drawing too heavily on resources, meaning that it is natural that not all the aspects involved are considered. Therefore the estimated costs alluded to here come from the response from the supplier that gave higher costs. Moreover, this supplier was the only one that had participated in specific projects developing e-Voting projects in other countries, which also provides more certainty when it comes to the estimated costs given.

The total estimated costs for elections with the introduction of e-Voting in all polling stations on touchscreen machines taken from the information given by the supplier with the higher costs would be about 37 million Euros per election for national-scale elections (Presidential, Parliamentary and European elections) and 4 Euros per registered voter. This equates to a cost almost four times that mentioned above for traditional elections, where the ball-park figures are 10 million Euros and 1.10 Euros respectively. Over a total 10-year period, this corresponds to an estimated 346 million Euros for the different types of elections (including local and referenda but excluding regional elections) with the chosen e-Voting solution, compared with 82 million Euros with the traditional voting system

The study also assessed by how much the estimated cost of elections with e-Voting could be lowered if paper were eliminated altogether, i.e. if there were no paper trail audit. The conclusion was that by taking paper out of the equation, 6% of the cost of traditional elections could be saved; in other words about 0.6 million Euros per election, 7 Euro cents per registered voter in each election, and 4.9 million Euros over the 10-year timeframe examined. However, as mentioned above, this saving would make no difference to the order of magnitude of the increased cost caused by e-Voting against traditional voting. The costs would still remain four times as much.

The option of including just an electronic electoral register was considered. This would bring improvements to electoral roll management and election logistics. As well as not making adopting e-Voting necessary, it would enable people to “vote anywhere” in national elections where you could consider the country as one electoral ward for vote counting purposes (i.e. in all elections except local and regional ones, yet also opening these up to the possibility of considering "vote anywhere" schemes, even though they would be more complex). The cost of introducing an electronic electoral register alone has been estimated at 53 million Euros, which corresponds to the cost of traditional elections with the electronic book increasing from 82 to 135 million Euros (up 65%) over the 10 years under consideration. This equates to an increase from 1.1 Euros to 1.7 Euros per registered voter in each election.

To sum up, the following table provides a cost comparison summary of elections over the 10 years under consideration, based on holding two elections of each type, not including regional elections and taking into account traditional voting arrangements with and without electronic electoral registers:

Cost comparison of traditional elections and different types of electronic voting over a 10-year period

Alternative access:
Click on table below to enlarge

Source: Deloitte.

*E-voting with a paper trail audit would not engender paper savings and would present considerable added costs in relation to the indicated points. These have not been estimated, but are linked to the production system and processing of the paper trail.

The study also examines the possibility of rationalising the distribution of polling stations, given that more than half (2,351) of all parishes (4,260) have fewer than one thousand voters (and there are even parishes with fewer than 40). All parishes currently have polling stations. It is estimated that the number of polling stations could be cut by 30%, although this estimate is excessively high as it is based on the assumption that no voter would remain further than 50 Km from a polling station, which seems too far.

The study concludes that, among other points:

“(…) bearing in mind the cost of the traditional model, any of the solutions effectively represent very significant cost increases.

(…) the introduction of e-Voting can primarily be justified by introducing the concept of mobility, and could be rolled out partially for national elections with the implementation of the electronic electoral register management system. This would therefore significantly reduce implementation costs."

Legal aspects and the need for legal amendments

The Deloitte study identifies a vast array of legislation related to electoral procedures, namely regarding Electoral Roll Legislation, Referenda Election Laws, Electronic Voter Authentication and Instituting e-Voting. Moreover, it raises several issues related to the legal amendments necessary, focussing on the following::

“(…) the Council of Europe, through the "European Commission for Democracy through Law" (Venice Commission ) (...) started the process of creating a code of good conduct for elections (…) which culminated in the presentation (…) of a report (…). In this context, we would also highlight the Recommendation (2004) 11 adopted by the Committee of Ministers of the Council of Europe on 30 September 2004 on legal, operational and technical standards for e-Voting.

1. Revision of the electoral roll system, making registration automatic and linking it to the civil identification database.

(…) In this context, the intention is to go even further and replace electoral registers in their current format with the electoral roll database, which would then act as a central electronic electoral register.

2. Consolidation of an in-person e-Voting system

3. Combining the voter card and the Citizen’s Card

(…) the technology solution selected will indubitably have to meet security standards that guarantee utmost respect of the fundamental values that are intended to be safeguarded, especially when the information on the Citizen's Card regards personal data.

4. Presentation of a proposal to modernise and consolidate the electoral process in a single piece of legislation, from the electoral procedure to all elections and referenda.

(…) if the e-Voting system is effectively implemented, polling stations should be equipped with qualified personnel who can deal with the system and provide voters with the necessary clarification and assistance so voters can duly exercise their right to vote. A recommendation of this kind may impact not only on the current wording of electoral laws/referenda themselves, but also on Law 22/99 of 21 April, which regulates the funding of electoral officers and remuneration of polling station and polling place members in elections and referenda.

As regards electoral roll records, and as mentioned before, checks will need to be made on the convenience of using such and possibly adapting their format to render them compatible with the desired e-Voting system for implementation, depending on the benchmark scenario, i.e. the existence of a centralised database laid out according to the objectives mentioned above in paragraph 1. Moreover, combining the voter card and the Citizen's Card will have an impact on the current wording of the election/referendum legislation, as explained above in paragraph 3.

As regards the procedure foreseen for “early voting", in cases where it is allowed, it is recommended that the procedure be compatible as required with the e-Voting procedure, if one is actually put in place.

Of further note as regards the concept of the “location for carrying out the act of suffrage" is that this should be tweaked to include the issue of mobility as a way of exercising the right to vote at a polling station other than the one for the location where the voter in question is actually registered.

(…) The voting procedure, from start to finish, including the method and means used and, subsequently as regards the count, verification and allocation of votes, shall be adapted to take in the new electronic voting system, should such be implemented fully or on a mixed basis, i.e. alongside the traditional voting method.

In this regard, the very idea of the ballot box and paper will need to be modified, as will the “method used for each voter to vote” from presenting themselves at the polling station, to the identification procedure using the Citizen's Card and subsequent authentication using the central voter database, the act of voting itself and, lastly, verification.

Lastly, it is worth pointing out that the types of electoral transgressions will also have to include predictions of e-Voting misdemeanours.

(…) the utmost attention needs to be paid to personal data protection legislation.

It is also recommended that the e-Voting system chosen for implementation be covered legally by its own legislation as regards the technical aspects that characterise it.”