County Clerk s Office Election Division Voting System

A Performance Audit of the County Clerk s Office Election Division Voting System Report Number 2015-05 August 2015 Office of the Salt Lake County Auditor

Executive Summary The Salt Lake County Council (Council) requested an audit and analysis of options to replace aging electronic voting machines used at polling places. The Council also desired a baseline analysis of current functionality for conducting elections in Salt Lake County (County), trends in voter preferences, and charging of fees to contracted municipalities for which the Election Division conducts elections. The Election Division has Three Options Available in Updating the County s Voting System Administrators in the Election Division of the Salt Lake County Clerk s Office (Clerk s Office) place high confidence in the reliability, accuracy, and durability of the current voting system. In this system, voters cast ballots on electronic touch screen tablets at polling places throughout the County. Election Division administrators feel that the current electronic voting machines will be viable at least through the 2020 election. Beyond that time, machines may need to be replaced due to wear and tear and a lack of availability of replacement parts. Reduced machine usage through implementation of a vote-by-mail model could further extend their useful life. The three options available to the Election Division are: Option 1: Option 2: Option 3: Replace the current electronic voting machine inventory with new machines and continue with the existing polling place configuration. Transition to a Countywide vote-by-mail system while still maintaining a few localized polling places known as voting centers. Implement an internet based electronic voting system. Although, based on our analysis, we have determined that this option is unlikely and currently not a viable option for the County. Replacing the County s Current Voting Machines Estimated costs to upgrade and replace the voting machines under the current Salt Lake County voting system configuration (Option 1) are as follows: $13.5 million to replace 3,000 electronic tablets used at polling places. $90,000 for software licensing fees, a one-time cost. Software license costs vary widely by vendor. $50,000 in annual maintenance agreement fees, which again, vary. $20,000 for training employees and poll workers on the new machines, at $2,000 per training day, considering 10 days of training. Office of the Salt Lake County Auditor i

Countywide Vote-by-Mail Voting by mail offers another option to the County. Dramatic increases in voters opting for a permanent vote-by-mail status demonstrate the growing awareness and popularity of this methodology. In 2005, about 3,000 voters opted for permanent vote-by-mail status. By the next year, this number increased 8 times to 25,000 voters. Today, over 133,000 out of the total 441,000 registered voters are permanent vote-by-mail participants. In 2015, all but two municipalities within the County opted to implement a vote-by-mail system for their municipal elections. The Clerk s Office Election Division will use these municipal elections as a test for the viability, operability, and reliability of a primarily vote-by-mail system on a Countywide basis. Research that we conducted indicated that a Countywide vote-by-mail system would reduce County expenses and would likely increase voter participation as well. Our analysis showed that transitioning to a Countywide vote-by-mail system, while still retaining approximately 30 voting centers throughout the County would affect yearly expenditures in the following ways: Reduce Election Division costs by approximately $538,000, due mainly from the need for fewer poll workers, but also from reduced numbers of other election workers and some reduced printing and hauling costs associated with transporting the electronic voting machines to over 300 polling locations. Costs reductions would fluctuate depending on the election year with lesser reductions likely occurring in oddnumbered municipal election years when fewer resources are required. Increase other costs, such as printing additional vote-by-mail ballots, postage, and temporary workers to scan and process mailed-in ballots by approximately $218,000. Create a net annual savings of approximately $320,000 for the County. Our analysis determined this would be a conservative estimate of the annual cost savings for the County. A one-time cost of $2,105,000 would occur from purchase of an inbound scanner, three additional electronic paper vote scanners, and 300 new voting machines for the 30 vote centers. Election Division Administrators stated that perhaps 30 vote centers would still be needed under a Countywide vote-bymail model. For each of these vote centers we conservatively estimated deployment of 10 voting machines. However, the County Clerk feels confident that the current inventory of voting machines would be adequate for the vote centers in the near future. Vote centers would still give voters the option of voting the traditional way at a physical polling place. Internet Voting or e-voting We examined electronic voting via the internet as a third, though unlikely option for the County. Due to public concerns over the security and reliability of internet voting results, it is unlikely that it would gain widespread support of the registered voting public.

Internet voting would reduce costs by eliminating setup and testing of electronic voting machines and the processing of paper ballots or memory card ballots. In addition, it provides a highly accessible and convenient method for voters to cast their votes. Additional costs, though, would arise from the need for information services personnel to program software and help ensure the highest level of voting system security. Currently, we are unaware of any fully-functioning internet voting systems within the United States, although some jurisdictions have experimented with the concept. Due to the lack of adequate County resources to validate voter identities, and general public mistrust of the security and reliability of internet voting, we have concluded that it is not a viable option for the County at this time. Voter identification and reliability of data transmission present fundamental obstacles to voting via the internet. Moreover, State law currently does not allow for internet voting. Baseline Data Provides a Basis for Determining the Best Options for Voting System Configuration With a population of approximately 1.1 million residents, Salt Lake County has 441,000 active registered voters and used 317 polling locations throughout the County during the 2014 general election. Trends in the number of active registered voters and polling locations, has remained relatively unchanged since 2004. To the contrary, the number of permanent vote-by-mail voters has steadily increased from only 3,000 in 2005, to over 133,000 in 2014. This trend can be attributed in large part to the efforts of the Election Division and the County Clerk along with political parties promoting the concept of voting by mail throughout the County as a convenient alternative to visiting a polling location on Election Day. The Election Division has a total of 3,146 voting machines. Administrators reported that all machines were used in the 2008 presidential election, though typically fewer machines are actually used. The Election Division deployed 2,736 voting machines in the 2012 presidential election and 1,951 in the 2014 general election. In odd-numbered municipal election years, even fewer voting machines are deployed. For example 1,680 machines were used in the 2011 municipal elections, and 783 in the 2013 municipal elections. The Election Division s 2014 total budget was approximately $4 million, half of which was allocated to direct costs of the primary and general elections. Some Election Division Costs Were Not Allocated to Contracted Entities We also examined the allocation of costs to elections, particularly as these related to billing of other municipalities and jurisdictions for which the Election Division contracts with to conduct local municipal elections. In our analysis, we found that some Election Division costs were not uniformly allocated, such as County government building and warehouse rent costs, and County overhead costs, which include support costs from other County entities. Annual County overhead costs charged to the Election Division total approximately $500,000, a portion of which, in our opinion, could be systematically allocated to municipalities for which the Election Division conducts elections. Office of the Salt Lake County Auditor iii

Other Counties We Surveyed Primarily Used a Paper Ballot System For comparison purposes, we surveyed the election departments of 18 other counties nationwide. Surprisingly, we found that 13 of these other counties used paper ballots instead of electronic tablets when voting. Our survey also showed a trend among many western States, particularly in the northwest, that have transitioned exclusively to a permanent vote-by-mail system. Replacing the County s Current Electronic Voting Machines Will be the Most Costly Option The Election Division of the Clerk s Office has three options in replacing its voting machines: 1) replace the current voting machines with new machines, 2) re-configure the elections system to one of primarily vote-by-mail with a minimal number of vote centers located throughout the County, or 3) implement internet voting. Replacing all voting machines, Option 1 above, will cost $11.4 million more up front than reconfiguring the system to vote-by-mail with vote centers, Option 2 above. Ongoing annual savings from converting to the vote-by-mail option we have determined would be approximately $320,000.

A Performance Audit of The County Clerk s Office Election Division Voting System Report Number 2015-05 August 2015 Scott Tingley, CIA, CGAP Salt Lake County Auditor Cherylann Johnson, MBA, CIA, CFE, CRMA Chief Deputy Auditor Audit Manager: Larry Decker, CPA, CIA Audit Staff: James Fire, MBA/Acc Todd Livingston, JD Leonardo Flores

This page intentionally left blank.

Contents Executive Summary... i Introduction... 1 Audit Scope and Methodology... 2 Section One: Baseline Costs and Metrics... 3 Finding 1: Election Division costs increased significantly after the acquisition of electronic voting machines....3 Finding 2: The Election Division allocated a smaller percentage of its total expenses to actual election costs in odd years than in even years....4 Finding 3: Under the current elections system, as many as 2,000 poll workers and 150 temporary workers could be required for each election....7 Finding 4: The number of polling places and the number of registered voters have remained relatively the same since 2004.... 11 Finding 5: The maintenance costs of the Clerk s electronic voting machines are reasonable, but not tracked as accurately as they could be.... 16 Section Two: Options to Replace the County s Electronic Voting Machine... 19 Finding 6: The cost to replace the County s current electronic voting machine inventory with similar machines would be approximately $13.5 million.... 19 Finding 7: Implementing a Countywide vote-by-mail system would result in election cost savings for the County.... 25 Finding 8: Internet voting is challenging to implement and maintain.... 28 Section Three: Allocating Election Costs and Billing for Election Services... 33 Finding 9: Some Election Division overhead costs were not billed to municipalities for election services.... 33 Finding 10: Less than 33% of personnel costs were allocated to fees charged to municipalities and other jurisdictions for election services provided by the County.... 34 Finding 11: Postage costs were not documented or managed for efficiency.... 35 Finding 12: Rent and most utilities costs were not charged to contracted entities for election services provided by the County.... 36 Section Four: Nationwide County Elections Survey... 39 Finding 13: Only 6 out of 19 counties we surveyed used an electronic tablet system for voting.... 39

Finding 14: Election costs increased as the number of polling places increased throughout the Country.... 40 Finding 15: Survey results indicate that election trends are leaning towards vote-by-mail as a preferred model for voting in elections.... 42 Conclusion... 44 Appendix A: Nationwide County Elections Survey Results... 47 Appendix B: Survey Details Table... 49

List of Tables Table 1: Percentage of Allocated Election Costs to Total Division Costs... 4 Table 2: Election Division Expenses Net of Revenues 2009-2014... 6 Table 3: Annual Costs to Maintain Voting Machines from 2008 to 2014... 16 Table 4: Vendor Electronic Tablet Costs... 24 Table 5: Voting Machine Replacement Costs... 24 Table 6: Prospective Analysis of Vote-by-Mail Cost Savings Using 2014 Election Data... 27 Table 7: Annual Election Divisions County Overhead... 34 Table 8: Election Division Allocation of Permanent Staff Expenses... 35 Table 9: Electronic Voting Machine v. Paper Ballot System... 39 Table 10: Polling Places, Registered Voters, and Overall Election Costs... 41 Table 11: Estimated Cost per Voter in 2014... 43 Table 12: Percentage of Registered Voters Compared to Population... 43 Table 13: Voter Participation Rates... 43

List of Figures Figure 1: Total Clerk Election Division Costs by Year... 3 Figure 2: Total 2014 eral Election Costs by Category... 5 Figure 3: Election Division Revenues from 2004 through 2014... 6 Figure 4: Number of Poll Workers per Election 2011 through 2014... 8 Figure 5: Total Poll Worker Stipends Paid by ElectionYear... 9 Figure 6: Number of Temporary Workers per Election Year... 10 Figure 7: Amount Paid to Temporary Workers per Election Year... 10 Figure 8: Number of Polling Places per eral Election 2004 through 2014 (Even Years)... 12 Figure 9: Number of Registered Voters per eral Election 2004-2014 (Even Years)... 13 Figure 10: Voter Turnout per Election Year 2009 through 2014... 14 Figure 11: Estimated Cost per Registered Voter 2009 through 2014... 15 Figure 12: Estimated Cost per Voter that Actually Voted 2009 through 2014... 15 Figure 13: TSX brand voting machines used at Salt Lake County.... 20 Figure 14: Number of Voting Machines Deployed by Election Year... 21 Figure 15: Average Number of Machines per Poling Location by Election Year... 22

Introduction The Salt Lake County Clerk s Office Election Division conducts all general elections for public office in Salt Lake County. The Division also contracts to conduct elections for most municipalities and local jurisdictions within the County, for example, Salt Lake City and Murray City. The Election Division also administers candidate filing for political office in even-numbered years. In odd-numbered years, city recorders administer this function for municipal elections. Any individual desiring to run for public office pays a candidate filing fee and completes the necessary paperwork. The Division registers voters for the entire County, generates and prints ballots, establishes polling places, recruits and trains poll workers and other temporary election workers, and deploys the County s electronic voting machines to polling places that the Division has selected. Furthermore, they tabulate and report election results for all Countywide and local municipal elections as well. By State Statute the County Clerk is the official voter registration agent. Utah Code, 20A-2-304, states: Each County Clerk shall: register to vote each applicant for registration who meets the requirements for registration. The Division has 16 full-time employees. In the weeks leading up to an election, additional temporary employees are hired to help train poll workers, assist with early voting, test voting machines, answer phones, enter data, and deploy voting machines. The Division rents a warehouse to store over 3,000 voting machines in its inventory. These touch screen voting machines allow voters to select candidates electronically without the use of a paper ballot. Voters insert a plastic electronic card, received from an election judge at the polling place, into the voting machine that brings the ballot into view. Votes are recorded electronically on a memory card and also physically on a paper roll printout. Paper rolls are attached to the machines. Votes are recorded on the paper roll printout for voters to view and compare their selections to those on the summary screen of the voting machine. Also, the paper roll serves an auditing function and to verify votes in any disputed elections. On election night, after the polls close, the memory cards are delivered to the Election Division at the County Government Center for tallying and reporting of Election Day results. Salt Lake County acquired most of its voting machines in 2006. Prior to that time, the County used a punch card system. The Federal Help America Vote Act (HAVA) of 2002 provided funding to states to replace paper punch card voting with systems that met requirements within the act, including alerting voters to any over votes, or voting for more candidates than allowed in any particular race, and allowing voters to make any changes. The State of Utah accepted HAVA funds with the goal of implementing a statewide electronic voting system. The County s current Diebold electronic voting machines were purchased by the State, and originally 2,844 machines were allocated to Salt Lake County to replace 5,000 punch card voting machines. Office of the Salt Lake County Auditor 1

Electronic voting systems prevented over voting and prompted the voter to correct any under voting for offices where the voter did not select a candidate. Salt Lake County acquired its current electronic voting machines through the State of Utah, which in turn used Federal funding to purchase them. Later, the County purchased a few additional machines when the State allotment was determined to be insufficient. Voting by mail has gained additional attention, popularity, and use among Salt Lake County voters. Any registered voter in Salt Lake County has the option of voting by mail, after completing an application and submitting it to the Clerk s Office. The Election Division mails paper ballots to vote-by-mail voters. Ballots mailed back to the Election Division are verified for an authentic voter signature and tabulated using high speed scanners. Other paper ballots, including provisional ballots, are also processed through these scanners. Audit Scope and Methodology Our audit scope comprised the manner and cost of conducting elections in Salt Lake County, options for replacing current voting equipment, and the allocation of costs to elections and billing to other municipalities for contracted election services. We examined overall costs from 2004 through 2014, with a more detailed examination of elections costs from 2009 through 2014. Our objective was to determine the best approach to replace the County s electronic voting equipment when considering costs and the risks involved with various alternatives. Other underlying objectives were to determine: 1. Baseline costs and other relevant data related to voting equipment and personnel in conducting elections. 2. Trends in various operational aspects, such as the number of poll workers and polling places, and the number of registered voters. 3. Accuracy and reasonableness of cost allocations to primary and general elections, especially as they related to billing of contracted jurisdictions. 4. Costs to replace the current system or to transition to a vote-by-mail, or other system. 5. Costs and trends in voting methodologies and systems in counties nationwide. To achieve the audit objectives, we interviewed employees in the Election Division, requested documentation of costs from them, and examined relevant costs as found in the County s general ledger. We generated a survey that we disseminated to selected counties nationwide, analyzed the data we gathered from the survey, and made comparisons. We contacted voting machine vendors to determine the types of voting systems available and the costs of those systems. We formulated a schedule of costs that would be added or deleted under a vote-by-mail system. We examined revenues and expenses, and compared the Division s cost allocations to cost data found in the County s financial systems. 2 A Performance Audit of the County Clerk s Election Division Voting System

Section One Baseline Costs and Metrics Finding 1: Election Division costs increased significantly after the acquisition of electronic voting machines. A significant increase in Election Division costs occurred following acquisition of electronic voting machines in 2006. Increased costs for equipment and training, regular and temporary personnel, and warehouse rent to store the voting machines factored into this increase. Costs that were $2,073,662 in 2005, increased to $4,979,582 in 2006, an increase of 140%. Most recently, Election Division costs totaled $4,167,938 in the 2014 election. Figure 1 shows trends in overall Election Division expenditures from 2004 through 2014. Figure 1: Total Clerk Election Division Costs by Year $7,000,000.00 Total Clerk Election Division Costs by Year $6,000,000.00 $5,000,000.00 $4,000,000.00 $3,000,000.00 $2,000,000.00 $1,000,000.00 $- 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total Election Division costs fluctuated depending on the type of election. The cost spike in 2006 resulted from added training and implementation costs of the new electronic voting machines that replaced the punch card voting system. The Help America Vote Act (HAVA) provided funding to states to replace their punch card systems with systems that complied with requirements of the act. High costs shown in 2008 related not only to high voter turnout for the Presidential election that year but also to a third election, the Western Presidential Primary, for which the Election Division received a Office of the Salt Lake County Auditor 3

rebate. After 2008, costs declined due to what management deemed as efficiencies gained through better poll worker training. With increased proficiency came a reduced need for additional personnel and outside training, which factored into reducing the overall costs of elections for the County. Costs fluctuated significantly from year to year depending on the type of election conducted. Actual voter turnout drives election costs. A higher voter turnout leads to increased actual costs for the Election Division and increased actual costs that are billed to the cities and other entities for conducting elections. Costs trended lower in odd-numbered years when municipal elections occurred. Evennumbered year elections are Countywide elections, and consistently have higher voter turnout than odd-numbered years. Presidential elections created not only the highest voter turnout rates but also the largest Election Division expenditures. The impact of Presidential elections can be seen in 2008 and 2012 when costs were $6,445,040 and $5,591,663, respectively. Finding 2: The Election Division allocated a smaller percentage of its total expenses to actual election costs in odd years than in even years. The Election Division allocates a portion of its total costs to each primary and general election. In this manner, they arrive at a total cost for each election. All Division costs are not allocated; only those costs that administrators determine are directly related to the election. These cost allocations provide the basis for billing other jurisdictions for contracted elections that the Division conducts on their behalf. There are statutory duties such as maintaining voter registration, voter lists, providing the voter registers for the elections to the cities and local districts that are solely the statutory responsibility of the County Clerk in both odd and even years. The Election Division cannot charge the costs of those statutory duties to the cities and local districts. Full-time personnel, temporary employees, poll workers, supplies, printing, and postage costs are among the types of costs allocated. Though other entities may be billed in even-numbered years, billing typically occurs in odd-numbered years when municipalities and other jurisdictions hold elections. Table 1 shows costs assigned to each primary and general election from 2009 through 2014, and the percentage they bear to total Division costs. We gathered elections cost data from the County general ledger (G/L), either from the old AFIN system or from PeopleSoft. Table 1: Percentage of Allocated Election Costs to Total Division Costs Percentage of Allocated Election Costs to Total Division Costs Year Allocated to Allocated to Total Allocated Total Election % Primary Election eral Election Election Costs Division Costs Allocated 2009 $682,151 $1,079,560 $1,761,711 $3,854,286 46% 2010 $985,573 $1,388,937 $2,374,510 $4,015,510 59% 2011 $489,359 $943,085 $1,432,444 $3,319,154 43% 2012 $1,230,950 $1,656,863 $2,887,813 $5,591,663 52% 2013 $671,514 $816,542 $1,488,056 $3,620,784 41% 2014 $847,073 $1,314,720 $2,161,793 $4,167,938 52% 4 A Performance Audit of the County Clerk s Election Division Voting System

A greater percentage of costs are allocated to election costs in even years, when Countywide and presidential elections occur. As shown in Table 1, about 40% to 50% of all Division costs were allocated to elections. To allocate costs, fiscal personnel used invoices as their source or they computed certain amounts. For example, full-time staff costs were allocated based on a computation of wage rates and benefits multiplied by the number of weeks, typically 6 to 10, that Election Division management felt were devoted exclusively to each election. Table 1 also shows a greater percentage allocated in even years than in odd years. We did not identify specific reasons for the percentage disparity, but our analysis showed fewer temporary and poll workers hired in odd-numbered years. Lower costs associated with fewer workers would reduce the percentage allocation given that other costs remained static from year to year. Municipalities and other jurisdictions are billed in odd years. Also, Election Division administrators allocated a portion of permanent full-time employees salaries to the costs of each election based on the number of weeks determined to be devoted to that election. erally, 10 weeks worth of salary was allocated, but in 2013, only 6 weeks was allocated. In other years, though, this number remained consistent at 10 weeks. Election Division administrators allocate not only full-time personnel costs but also poll worker costs, costs for printing, public notices, rent, and other categories. Figure 2 shows percentages by category in allocated costs of $1,314,720 for the 2014 general election. This chart approximates the types and percentages of costs assigned in other election years. Figure 2: Total 2014 eral Election Costs by Category Total 2014 eral Election Costs by Category Rent Poll Places 1% Other 3% Maintenance 3% Postage 3% Communications 0% Utilities & Phone 1% Printing 16% Public Notices 11% Rent 4% Full-Time Staff 19% Poll Workers 23% Overhead 6% Poll workers expenditures are the greatest percentage of cost by category. Temporary Workers 9% Supplies 1% Office of the Salt Lake County Auditor 5

Allocated costs are billed to municipalities and other jurisdictions for contracted election services. Revenues received from these other entities are shown in Figure 3. Revenue spikes occur in oddnumbered years when other municipalities conduct their elections and these revenues are collected. Figure 3: Election Division Revenues from 2004 through 2014 $1,600,000.00 Election Division Revenues from 2004 through 2014 $1,400,000.00 $1,200,000.00 $1,000,000.00 $800,000.00 $600,000.00 $400,000.00 $200,000.00 $0.00 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Most of the Election Division revenues are received in odd-numbered years when municipalities are billed. Revenues received from municipalities significantly offset the impact of election expenditures. Table 2 nets revenues against total Election Division expenses by year. Table 2: Election Division Expenses Net of Revenues 2009-2014 Election Division Expenses Net of Revenues 2009-2014 Year Expenses Revenues Net Expenses 2009 $3,854,286 $1,269,567 $2,584.719 2010 $4,015,510 $110,315 $3,905,195 2011 $3,319,154 $1,091,079 $2,228,076 2012 $5,591,663 $94,843 $5,496,820 2013 $3,620,784 $1,291,141 $2,329,643 2014 $4,167,938 $104,000 $4,063,938 Revenues are higher in odd years when municipalities are billed for their elections. 6 A Performance Audit of the County Clerk s Election Division Voting System

Finding 3: Under the current elections system, as many as 2,000 poll workers and 150 temporary workers could be required for each election. When combined, employing both poll workers and temporary workers represent the largest expenditures by category for an election. Elections require additional workers to operate polling places, test and set up voting machines, sign-in voters, and aid in ballot tabulation. To carry out these functions, the Election Division hires poll workers, temporary workers, election night workers, and rovers. Poll workers are broken down into functional categories of poll manager, touch screen technician, provisional clerk, table judge, and alternate judge. Pay rates per day for each of these categories are as follows: Poll Worker Category Pay Rate per Day Polling Location Manager $275 Touch Screen Machine Technician $220 Provisional Clerk $185 Table Judge $175 Alternate Judge $100 ($230 if deployed) About 40 rovers, who travel to various polling places and provide help as needed, are paid for the average 21 hours they work. Rovers pay for their own vehicle, gas, and cell phone expenses. Approximately 13 election night workers designated either as lead or assistant workers are paid $100 and $60 respectively. Poll worker staffing levels fluctuated widely depending on the election size and expected voter turnout. Figure 4 shows the number of poll workers in each primary and general election from 2011 through 2014. During this period, poll worker numbers peaked at 2,005 in 2012, a Presidential election year. Office of the Salt Lake County Auditor 7

Figure 4: Number of Poll Workers per Election 2011 through 2014 Number of Poll Workers per Election 2011 through 2014 2,500 2,000 2,005 1,500 1,298 1,299 1,000 500 364 860 417 454 982 0 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 Pri 2014 The highest number of poll workers was needed in even-numbered years during Countywide elections. Poll worker stipends correlate to the number of poll workers used. Figure 5 shows total stipends paid to polls workers from 2011 through 2014. Stipends paid in 2011 ($88,920 in the primary election, and $97,755 in the general election) were the lowest of any year. The two highest elections for stipend payments were the 2012 general election, a Presidential election year, at $362,730, and the 2014 general election at $284,578. 8 A Performance Audit of the County Clerk s Election Division Voting System

Figure 5: Total Poll Worker Stipends Paid by Election Year Total Poll Worker Stipends Paid by Election Year $400,000.00 $362,730 $350,000.00 $300,000.00 $284,578 $250,000.00 $200,000.00 $150,000.00 $100,000.00 $50,000.00 $78,420 $97,755 $- 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 Pri 2014 The highest amount of poll worker stipends paid occurred in the 2012 Presidential election year. In addition to the workers just mentioned, temporary workers are hired to perform tasks over a period of several weeks in preparing for the election. Their duties include training poll workers, registering voters, data input, and voting machine testing and deployment. As with poll workers, temporary worker numbers also reflect election size and expected voter turnout. Fewer temporary workers than polll workers are used in an election, but since they work over several weeks, their total pay approximates the total paid to poll workers. Figure 6 shows total temporary worker numbers in each primary and general election from 2009 through 2014. Figure 7 shows totals paid to temporary workers in thesee same elections. Office of the Salt Lake County Auditor 9

Figure 6: Number of Temporary Workers per Election Year Number of Temporary Workers per Election Year 180 164 160 140 120 100 121 107 80 70 60 40 46 41 53 42 53 33 39 47 20 0 2009 Pri 2009 2010 Pri 2010 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 Pri 2014 Temporary workers perform election tasks over a number of weeks leading up to Election Day. Figure 7: Amount Paid to Temporary Workers per Election Year Amount Paid to Temporary Workers per Election Year $350,000 $312,041 $300,000 $250,000 $200,000 $185,352 $150,000 $135,516 $100,000 $50,000 $60,749 $40,628 $95,745 $- 2009 Pri 2009 2010 Pri 2010 2011 Pri 2011 2012 Pri 2012 2013 Pri The highest temporary worker expenditures occurred in even-numbered general election years.. 2013 2014 Pri 2014 Temporary worker expenses dropped significantly in 2013 and 2014, due to efficiencies achieved in the Election Division. These efficiencies included acquisition of new high-speed scanners that reduced the number of employees required to feed and process paper ballots into ballot scanning machines. Also, in 2014, fewer temporary workers were needed to process voter registration forms because of the State s 10 A Performance Audit of the County Clerk s Election Division Voting System

online registration tool. More people registered online and saved staff time from keying in the information from paper registration forms. Finding 4: The number of polling places and the number of registered voters have remained relatively the same since 2004. Though the vote-by-mail option has gained popularity, Salt Lake County continues to follow an election model primarily configured around hundreds of neighborhood polling places. Since 2006, voters have cast their ballots on an electronic touch screen tablet. Despite the increase in permanent vote-by-mail voters, conducting elections exclusively by mail was not authorized in law until passage of H.B. 172 in 2012. Election Division management select the number of polling places based on the type of election. A general election in an even-numbered year, especially a Presidential election, results in significantly more polling places. Conversely, perennially low voter turnout for municipal elections in odd-numbered years requires fewer polling places. Past trends in voter turnout also influence this number. In 2015, the Election Division will test the viability of a Countywide vote-bymail option. In 2015, the Election Division will significantly reduce the number of polling places due to municipalities and other jurisdictions opting to conduct vote-by-mail elections, while still retaining a limited number of vote centers throughout the County. Fourteen of 16 cities within the County opted for completely voteby-mail elections. Taylorsville City and West Valley City, the two exceptions, decided to stay with the standard polling place model. The additional data and experience gained from these municipal elections will give Division administrators a basis for deciding whether to propose to the County Council that they adopt the voteby-mail method for Countywide elections. In odd-numbered municipal election years, the number of polling places followed an erratic pattern driven by the number of entities conducting elections and the types of elections. For example, the Election Division set up 280 polling places for the 2011 general election, but only 143 for the 2013 general election. In 2013, two cities opted to use a vote-by-mail model, thus eliminating the need for most polling places in these two jurisdictions. As opposed to odd-numbered years, polling places in even-numbered year elections followed a more consistent pattern, as show in Figure 8, which details polling place numbers in elections from 2004 through 2014. Office of the Salt Lake County Auditor 11

Figure 8: Number of Polling Places per eral Election 2004 through 2014 (Even Years) Number of Polling Places per eral Election 2004 through 2014 (Even Years) 400 350 300 250 200 150 100 50-388 370 2004 2006 336 328 322 317 2008 2010 2012 2014 The number of polling places has decreased only slightly since 2004. Since 2004, general election polling place numbers declined by 18%, which is only a slight decrease compared to the 433% increase in registered voters who have become permanent vote-by-mail voters. Despite a growing population, the number of registered voters within the County has not changed significantly over the last ten years. The number of registered voters has hovered in the 440,000 to 500,000 range since 2004. The only exception has been during the 2008 election, when the number of registered voters reached 521,039. Despite an increasing Salt Lake County population, registered voters since 2004 have largely remained unchanged. In fact, the Election Division recorded 480,807 active registered voters in 2004 compared to 440,524 in 2014, an 8% decrease. Election Division Management explained that their statistical data was influenced by the move in 2010 to the Statewide VISTA database that allowed them to more readily delete inactive and duplicate voters from the database. Therefore, lower numbers of registered voters in recent years were due, at least in part, to more efficient data management. 12 A Performance Audit of the County Clerk s Election Division Voting System

Figure 9: Number of Registered Voters per eral Election 2004 through 2014 (Even Years) Number of Registered Voters per Election 2004 through 2014 (Even Years) 540,000 521,039 520,000 500,000 480,000 460,000 440,000 420,000 480,807 491,108 470,301 444,196 440,524 400,000 2004 2006 2008 2010 2012 2014 The number of active registered voters has decreased only slightly in recent years. Voter turnout varies by year and is affected by the types of elected offices and election items appearing on the ballot. As expected, a larger voter turnout reflects voter interest in the election, with Presidential election years creating the largest voter turnouts. Figure 10 shows voter turnout in each primary and general election from 2009 throughh 2014. Office of the Salt Lake County Auditor 13

Figure 10: Voter Turnout per Election Year 2009 through 2014 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% Voter Turnout per Election Year 2009 through 2014 Red Line - Vote-by- Mail Turnout Blue Line -Polling Place Turnout 0% 09 Pri 09 10 Pri 10 11 Pri 11 12 Pri 12 13 Pri 13 14 Pri 14 Voter turnout at polling places has only outpaced vote-by-mail turnout in the 2012 general election, a Presidential election year. Vote-by-mail percentages generally ranked higher than those for polling place voting. The only exception occurred in the 2012 Presidential election, when voter turnout at polling places was 87% compared to 78% of vote-by-mail voters. erally, the data we reviewed showed higher participation rates among vote-by-mail voters, compared to those registered voters who choose to vote in person at a polling location. Figure 10 shows higher participation among vote-by-mail voters, a factor favoring a vote-by-mail elections option. The cost per voter indicates efficiency of resources used. We analyzed this factor based on the amount allocated to each election by Election Division administrators. We examined this data two different ways. First, we compared the cost per active registered voter, and second, the cost per voter who actually voted. Figure 11 shows this first comparison, and Figure 12 shows the later, respectively. 14 A Performance Audit of the County Clerk s Election Division Voting System

Figure 11: Estimated Cost per Registered Voter 2009 through 2014 Estimated Cost per Registered Voter 2009 through 2014 $4.00 $3.50 $3.00 $2.50 $2.00 $2.78 $2.08 $2.08 $2.95 $2.84 $2.51 $2.70 $3.73 $1.78 $1.64 $1.95 $2.98 $1.50 $1.00 $0.50 $- 2009 Pri 2009 2010 Pri 2010 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 Pri 2014 The estimated cost per registered voter has been consistent from 2009 through 2014. Figure 12: Estimated Cost per Voter that Actually Voted 2009 through 2014 Estimated Cost per Voter that Actually Voted 2009 through 2014 $25.00 $20.00 $22.14 $21.89 $20.00 $15.00 $10.00 $10.67 $10.49 $13.37 $14.38 $13.50 $11.06 $5.00 $5.44 $4.28 $5.73 $- 2009 Pri 2009 2010 Pri 2010 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 Pri 2014 The cost per voter that actually voted is much higher in primary elections, due to low voter turnout. Office of the Salt Lake County Auditor 15

Primary elections are costly due to low voter turnout. Figure 12 shows the relatively high cost per actual voter in primary elections since 2009. Compared to the general elections in the same year, primary elections are not cost effective under the current elections system. In these cases, costs per voter always exceeded general election costs per voter. Low primary election voter turnout compared to extensive equipment, personnel, and polling place requirements drives this inefficiency. Finding 5: The maintenance costs of the Clerk s electronic voting machines are reasonable, but not tracked as accurately as they could be. We analyzed the maintenance costs of the Election Division s electronic voting machines to better understand if the costs are controlled at a reasonable level, when compared to the benefits derived. When not controlled, there is a greater risk of costs increasing to a level that is unreasonable to the benefits derived. We focused on four major categories of costs: replacement parts purchased, repair and maintenance labor costs, warehouse rent, and utilities. Information was gathered through expense reports from the County s financial system, reviewing invoices for purchases, and interviewing appropriate personnel. Table 3 below describes the costs calculated from 2008 to 2014 in each major category. Table 3: Annual Costs to Maintain Voting Machines from 2008 to 2014 Annual Costs to Maintain Voting Machines from 2008 to 2014 Cost Category 2008 2009 2010 2011 2012 2013 2014 Parts 983 1,445 1,364 1,145 125,834 11,463 - Labor 53,690 23,135 27,801 25,205 68,346 37,303 31,094 Warehouse Rent 110,891 86,400 86,400 86,400 87,129 90,217 91,272 Warehouse Utilities 26,252 28,367 27,448 25,520 29,736 29,705 30,718 Total Cost $191,816 $139,347 $143,014 $138,269 $311,046 $168,687 $153,085 Rent and utilities are the bulk of costs relating to voting machine maintenance. Each cost category, shown in the table above, indicates a consistent trend from year to year, with the exception of labor costs in 2008, parts costs in 2012, and labor costs in 2012. Higher labor costs in 2008 resulted from contracting out the testing of voting machines to a vendor. Since 2008, the level of labor costs lowered because testing of machines was performed by election personnel. The 2012 increase resulted from a large purchase of batteries for voting machines. We adjusted both the Warehouse Rent and Utility categories to 80% and 90%, respectively, of their total individual cost. The warehouse is primarily used to house and maintain the voting machines; however, the warehouse serves other purposes as well. Therefore, an adjusted cost was calculated to reflect the costs associated with the voting machines only. Invoices were reviewed to substantiate the Parts cost category. Labor costs included calculations of the repairing, testing, and inspection of voting machines. 16 A Performance Audit of the County Clerk s Election Division Voting System

Even though costs showed a consistent trend, no detailed records were kept for parts purchased and the repairs made with those parts. Additionally, records tracking the labor to repair the machines were absent. Implementing a tracking process will aid in controlling costs and will help management determine if costs have risen to a level that is unreasonable to the benefits derived from using machines. We concluded that costs to maintain the voting machines are controlled and therefore minimize the risk of increasing to an unreasonable level when compared to the benefits derived. We found that costs were controlled in different ways, mainly by: Removing parts from machines no longer in operation to repair machines in operation. Changing labor processes over time to improve efficiency and economy. Rotating the deployment of voting machines to minimize wear and tear. Performing annual detailed maintenance inspections on every machine in inventory to increase longevity of machines. Repairing and testing machines in-house to save on outside repair and maintenance costs. As the machines continue to age, it is expected that parts purchased for repair and labor to perform the repair will continue to increase relative to the frequency in using the machines. Frequency in using the machines may be alleviated as the Election Division changes the method of voting throughout the County. This could result in prolonging the life expectancy of the voting machines and in continuing to control costs associated with the machines. Recommendations We recommend that Election Division management: 1. Reexamine the percentage disparities in election cost allocations between oddnumbered and even-numbered year elections to ensure that all costs are equitably allocated. 2. Implement a process to track parts purchased, repairs made, and labor costs for maintaining the County s current voting machines. Office of the Salt Lake County Auditor 17

This page intentionally left blank. 18 A Performance Audit of the County Clerk s Election Division Voting System

Section Two Options to Replace the County s Electronic Voting Machines Finding 6: The cost to replace the County s current electronic voting machine inventory with similar machines would be approximately $13.5 million. The County s current electronic voting machine type is no longer manufactured. The most significant factor to consider in replacing the County s current electronic voting machines is that no vendor currently manufactures similar voting machines. All of the manufacturers that we contacted during our audit do manufacture and sell electronic voting tablets. However, in every case, their products lack an integrated vote recording and tabulating process, a distinguishing feature of Salt Lake County s current system. The Election Division currently has an inventory of 3,146 voting machines. The Diebold Company manufactured these under the brand name TSX, and sold them to the State of Utah in 2006. The State then distributed these machines, at no cost, to the counties, including Salt Lake County. Purchased through federally awarded funding of $10 million, the machines became a necessary purchase because of the Help America Vote Act (HAVA). Through HAVA, Federal funding was offered to states to eliminate punch card systems and replace them with systems that complied with HAVA requirements. At the time, the County was using a punch card system. Diebold sold its voting machine manufacturing and marketing operation and is no longer involved in the business. A successor company currently markets the TSX brand identical to Salt Lake County s machines, but no longer manufactures them. Considered now as legacy or obsolete equipment, the remaining inventory stock sells for $2,500 per unit. Office of the Salt Lake County Auditor 19

Figure 13: TSX brand voting machines used at Salt Lake County. The County s current TSX brand electronic voting machine features paperless recording and processing of votes. Election Division administrators highly favor this system for its ease in vote recording and vote tallying that otherwise eliminates time and effort in gathering, sorting, tabulating, and storing individual paper ballots. Salt Lake County s cultural environment of community camaraderie, trust, and confidence lends itself to this paperless, touch-screen system where votes record on and tabulation derives from the same electronic memory card. The public at large has voiced little to no opposition to it. In considering replacement costs, we examined voting machine usage, and found that only a portion of the machines were used in any election. Since 2009, the closest the Division came to using the entire inventory was in the 2012 Presidential election when 2,736 machines were deployed. Figure 14 shows machine deployment history from 2009 through 2014. 20 A Performance Audit of the County Clerk s Election Division Voting System

Figure 14: Number of Voting Machines Deployed by Election Year 3500 Number of Voting Machines Deployed by Election Year 3000 2500 2000 1500 1000 500 534 1,092 978 2,314 610 1,680 1,705 2,736 Machine Total # Deployed 1,650 719 783 1,951 0 2009 Pri 2009 2010 Pri 2010 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 Pri 2014 The most voting machines are deployed in even-numbered general election years. In addition to the memory card, the machines have an automated paper mechanism attached that also records votes cast. The mechanism records votes on a paper roll. Similar to a cash register tape, these paper rolls serve to verify votes in an auditing process, but are not used on election night to tally votes. The roll also allows voters to compare a paper record of their vote to the voting machine summary screen. The number of voting machines at each polling place per election from 2009 through 2014 is shown in Figure 15. The average number of machines during this period was 5.1. Office of the Salt Lake County Auditor 21

Figure 15: Average Number of Machines per Poling Location by Election Year Average Number of Machines per Polling Location by Election Year 7.1 5.0 6.0 5.2 6.1 5.5 5.5 5.3 6.1 3.6 3.5 3.0 2009 Pri 2009 2010 Pri 2010 2011 Pri 2011 2012 Pri 2012 2013 Pri 2013 2014 2014 Pri On average, five to six voting machines are deployed at each polling location during an election. All products currently marketed by companies that produce election machines use a paper ballot. Either a paper ballot is manually marked and then scanned, or a paper ballot is printed from votes made on a touch screen and then read by a scanner, either at the polling place or another central tabulation location. One of the largest election machine companies stated they will market a product by the end of 2015 similar to the County TSX electronic tablet, though this new product still produces a paper record for each ballot cast. While the new product features a memory card to integrate both the recording and tabulating of votes, similar to Salt Lake County s process, an encoded paper must be printed and inserted into the voting machine to bring the ballot into screen view. Each polling place would require a printer to print out the encoded paper. The machine also records votes on the piece paper, which falls into a bin after the voter completes the voting process. Vote tallying occurs from the memory card but the individual papers serve to allow for auditing election results. The Clerk s Election Division has a favorable view of the Denver City/County model. Recently, the city/county of Denver, Colorado adopted a new voting system that features a new generation of electronic voting tablets. The company marketing this product considers it a prototype and has not yet offered it for sale to the general public or even posted the product description and price on its website. The Election Division is aware of and has offered a favorable view of this system that Denver has adopted. 22 A Performance Audit of the County Clerk s Election Division Voting System