Home » Election Analysis » Is Increase in Postal Votes driven by Ghosts?


Maps: Districts are shaded by the number of postal votes cast in the Presidential elections from 1982 to 2015.

Suspicious Rise in Postal Votes

The numbers of postal votes increased 10-fold between the Presidential Elections in 1982 and 2015 (to 1.1 million) while population only increased by a third (to 20.5 million). Consider also the following:

  • While postal votes rose by 50% in the sixteen years prior to 1999, it increased by 300% in the next sixteen years.
  • Less than 20% of the government employees used the postal voter facility before 1999. Since then this ratio has doubled and in four districts has exceeded 60%.
  • The increase of personnel in the armed forces by about 200-240,000 explains a fraction of the increase of postal votes there is still a significant unexplained rise.
  • Even after the numbers in the armed forces stabilized after 2009, the postal votes continued to increase.

While some of the ten-fold rise can be explained, the rise is highly suspicious in some districts. We looked into postal voting to assess the ways in which voter fraud could be taking place. What comes out is that the fraud through postal voting probably does not exceed 3 lakhs.

Rise of Ghost Voters

While there were 15 million voters on Sri Lanka’s electoral list, the population that are of voting age was only 14.2 million. Some of the excess is masked due to under-registration in districts in the Northern and Western provinces. Taken together with those who are not registered because they are displaced, overseas or did not care or for administrative reasons, there is evidence that there are 1-2 million “ghost voters” in the 2014 electoral roll (See Dr. Laksiri Fernando (Colombo Telegraph, Jan 15, 2015) and ZL Mohamed (Colombo Telegraph, June 6, 2015) ).

After getting these 1-2 million ghost voters on the electoral rolls, those directing the ghost voting, have to deliver the ballots to the district counting centres. Postal voting takes place with more lax monitoring than regular polling and it could be one of the mechanisms by which such fraudulent votes are cast.

Postal Voting

Postal voting refers to early voting and is offered to government employees (and even a few others) who are engaged in “essential duties” at the time of elections. It seems that the government employee decides which he or she is engaged in essential service.   The voter has to apply for postal voting. The voter registration information is “posted” to the employee at their place of work. “Postal voting” is actually done at their workplaces or sometimes in others government offices. The postal voting takes place about one to two week prior to the regular polls. Sometimes, the employees have been offered an opportunity to present themselves at the Kachcheri at a second date if they missed the vote at the office. If the employee does not receive the voter registration card, they revert to polling at the regular polling place.

There have in the past been around 1200 postal voting sites. In some of the polling stations, neither independent monitors nor party representative have attended the postal voting, or followed the ballot boxes being transferred   to the counting centres.

Postal Voting in the 2015 Presidential Elections

The Deputy Commission of Elections (e.g. Sunday Times, 26th July 2015) noted that there were 60,000 applicants who were either attempting to cast duplicate votes in the elections or not filling up the forms properly of the 629,000 applicants for postal voting. The Elections commission was able to establish the existence of the large numbers who were trying to vote multiple times due to its database of national identity card (NIC). This points to either ineptitude or fraud among a minority of government officials.
The registration for postal voting has increased from 525,000 in the Presidential election in January to 567,000 in the coming Parliamentary elections even though the same voting list from 2014 is being used. It may be that this increase is due to the voter registration by the GCE A/L examination personnel. While no one could be added to the voter list from the Presidential polls since mid-2014, some government employees could have retired or died and the numbers in the armed forces did decline. This rise by 8% is possible but suspicious.

Checks of Postal Voter Fraud

The Elections Commission and the Government Agent and their staff have to ensure that avenue for multiple voting through postal votes and regular polls are closed. Even if this were the case, some postal voters could be deployed to impersonate others at the regular polls. Some of the checks against impersonation – such as the inking of the finger and the difficulty in being physically present at multiple voting places or multiple districts – do not constrain these 6 lakhs of postal voters. The NIC checks do not foreclose multiple voting by the same person through document fraud.
Last month, the Elections Commissioner requested the assistance of all groups and citizens to help identify voter fraud.

Deterrence for Fraud through Postal Vote

The Department of Elections has rejected 60,000 postal vote applications for the 2015 parliamentary elections. Still apart from rejecting the voter registration, no punitive steps are being taken. Indeed, there is a loophole for the electors to be registered twice and still not lose the vote.
Section 99 (4) of the Constitution: “Each elector whose name appears in the register of electors shall be entitled to only one vote notwithstanding that his name appears in the electoral register in more than one electoral district.”
Indeed, a person who is registered twice can even be elected to parliament. This clause can only be there to cover up errors in the election registration process but this is an invitation to fraud. Note that the draft of the 20th Amendment intended to perpetuate that loophole.

Rise in Postal Votes by District

At a district scale, the rise is significant in Kurunegala and Gampaha in 1994; dramatic in Kurunegala, Anurhadapura and Polonnaruwa between 2005 and 2010; postal voting in Kegalle, Kandy and Badulla also rose out of proportion to the national rate. In 2010 and 2015, the rise continued in all these districts and extended to Kalutara, Ratnapura, Galle and Matara. By looking at the spatial spread sequentially, there was one centre of rising postal voters was Kurunegala picking up in an election with notorious fraud in 1999. In subsequent election years, the spurt in postal voting spilled over to the neighbouring districts.



Maps: For each Presidential Election from 1999 to 2015, the total number of postal votes in each district is shown as a percentage of the number of government employees in that district. The numbers of government employees were available for census years and we interpolated the data for election years.

To check whether the rise in particular districts is driven by rise in government employees, we can check the ratio of the postal votes to the number of government employees. The second panel of maps (below) shows that Kurunegala had the highest ratio of postal votes / government employees in 1999. However, in 2010, Polonnaruwa had the highest ratio followed by Trincomalee, Anurhadhapura and Matale. In 2015, intriguingly, the ratio for Polonannaruwa dropped and Trincomalee, Anurhadhapura, Matale, and Kurunegala had the highest tier of ratios.
Some of this rise in could be due to an expansion in the armed forces. These personnel are likely to register for postal voting at higher rates (up to 60-70%) compared with other government employees.
At a national level, the rise in armed forces by approximately 200,000 is outstripped by the rise in postal voters. One cannot explain the rise in postal voting after 2009, as the armed personnel numbers stabilized or declined slightly.

Voting Patterns by District during the 2015 Presidential Polls

If one looks at the postal voting patterns for the incumbent and challenger in the Presidential Polls (see graph), there is a split between the candidates. The voter turnout approached an upper limit. There was the skew towards higher turnout in districts which favoured the incumbent. Such a pattern is usually a signature of voter fraud. But the range in postal voter turnout is small unlike in the case of the general population (See ZL Mohamed, Colombo Telegraph, May 2015). Thus while there is suspicious patterns in the voting here too but one can rule out a wholesale dump of fraudulent ballots for the incumbent.


Graph: For each of the 22 districts of the Eleciton department, the above graph shows the gain for Rajapakse over Sirisena versus the turnout. The y axis shows the difference in percentage of vote obtained by the Incumbent (Rajapakse) and Challenger (Sirisena). The x-axis shows the turnout as a function of registered postal voters.


Colombo is home to the largest number of government employees from 1982 to 2015. Yet, by 2015 there are eight districts that have more postal votes than it. Some observations regards Kurunegala, Trincomalee, Jaffna, Gampaha, Colombo are provided below. The districts of Anurhadhapura, Kandy, Matale, and Matara are also suspect.
• Kurunegala: Has had both the highest number of postal voters and the highest number of ghost voters since 1999 – a year which has been widely reported to have had outrageous voter fraud which was not punished. If indeed fraud through postal voter did it take place, it could explain up to 40% of the ghost voters.
• Trincomalee: has the highest ratio of postal votes to government employees. Is it correct to surmise that the recruitment to the armed forces from this district shall be limited given its demographics? If so what accounts for its large number of postal voters? Of relevance is the discovery of 15,000 fraudulent ballot papers in Kantalai 10 days prior to the 2015 parliamentary elections and the arrest of a deputy minister for walking in unauthorized into a postal voting booth in Trincomalee.
• Jaffna: Why did postal voting in Jaffna rise up dramatically in 2015? Is this simply the removal of bottlenecks to voting – if so how does one explain the 60% ration of postal votes/government employees ?
o Can we surmise there is little hiring to the armored forces from Jaffna residents?
o Are there estimates of persons who re-settled or have migrated to Jaffna?
o Could this be explained by relocation of residence of officials from the Wanni to Jaffna after 2009? Of relevance is that the voter lists from 2008 were used in the 2010 elections.
• Gampaha: has low voter registration than the age-eligible population but its share of postal voters has been rising dramatically in relation to Colombo. The rise is quite large after 1999 and has continued in recent years particularly after 2010. This was a district tended by Basil Rajapakse for the 2015 elections.
• Colombo: has had the highest number of government employees, but its contribution to the postal votes ranks 9th of 25 districts. This may be partially explained if the recruitment to the armed forces is extremely low in Colombo.
• Polonnaruwa: The district that had the highest ratio of postal votes to government employees in 2010 dropped to the second tier in 2015. Could it be that the voter fraud operation in the district in 2015 was compromised as one leader went over to the opposition ?

The rise in postal voting in Kalutara, Ratnapura, Moneragala, Nuwara Eliya and Hambantota districts while significant was relatively smaller than the rise in ghost voters – so fraud through postal voting shall be less important in those districts.

What have we learned?

In attempting to unravel the 10 fold rise in postal voting, we have been limited by
• the availability of data at electorate or polling division level
• the lack of district based estimates for armed forces domicile
• lack of information on changes in practices and policies
• Insights from different government establishments.
Still the data shows that
• Postal voting shows excessive rise in districts with higher ghost voters – but not all districts that have highest ghost voters are in the high tier of postal voting.
• The rise in postal voting can only account for a modest fraction of ghost voting . At most, there could be only about 2-3 lakhs of votes cast on behalf of ghosts – that was estimated to range from 10-20 lakhs
• The unusually large postal voting shows a spatial pattern in its evolution. It starts in Kurunegala in 1999 and spreads to neighbouring districts in a creeping pattern from election to election. There is a second locus of more modestly rising postal votes in the South.
• The rise in ghost voters corresponds with the rapid increase in postal voters in Kurunegala, Anurhadapura, Matale, Polonnaruwa.
• The exceptions to this are in Trincomalee, Jaffna and Gampaha. These districts may be also having high voter fraud.

These insights bear refinement and we hope you can contribute to this effort.
This work has been carried out by a volunteer effort by a collective of Sri Lankans (Sri Lankan Election Analytics) – mechanisms for contributing and participating in this work is described at www.slelect.net with updates via @slelect or by email to mail@slelect.net  

Leave a Reply

Your email address will not be published. Required fields are marked *


  (To Type in English, deselect the checkbox. Read more here)
Lingual Support by India Fascinates