Use the links below to skip to the section you wish to view:
- Significant Weaknesses in EDD's Claims Processing and Workload Management Leave It at Risk of a Continuing Backlog of Claims
- Because EDD Responded to the Claim Surge by Suspending Certain Eligibility Requirements, Many Californians Are at Risk of Needing to Repay Benefits
- EDD Took Uninformed and Inadequate Steps to Resolve Its Call Center Deficiencies
- Despite Multiple Warnings, EDD Failed to Prepare for an Economic Downturn
Significant Weaknesses in EDD's Claims Processing and Workload Management Leave It at Risk of a Continuing Backlog of Claims
- In December 2020, EDD reported that it had about 685,700 claims in its backlog. However, EDD has presented unclear and inconsistent information about the backlog leading to the belief that all claims in its backlog were waiting for payment. In fact, fewer than 20,000 of those claims had waited for payment longer than 21 days because of EDD's failure to resolve an issue with the claim.
- Although EDD has made improvements since the pandemic began to increase the number of claims it can process without manual intervention, it cannot rely in the long term on some of these adjustments because they are dependent upon its suspending critical requirements. As a result, EDD remains at risk of its backlog of claims continuing or increasing.
- EDD has failed to adequately plan for additional possible increases in UI claims when making staffing decisions. This failure to prepare leaves EDD vulnerable to future workload disruptions from spikes in claims caused by additional pandemic-related shutdowns or even predictable seasonal changes in employment levels.
EDD's Backlog Dashboards Misrepresent the Number of Claims With Delayed Payments
At the end of September 2020, EDD began reporting the numbers of backlogged initial and continued claims on dashboards on its website, using an approach recommended by the strike team. EDD represented this backlog as the number of claimants awaiting payment because EDD had yet to act on their claim. As of December 15, 2020, EDD reported that a total of about 685,700 initial and continued claims were remaining in its backlog. However, contrary to EDD's characterization of the backlog, this number does not represent the actual number of claims needing action so that claimants can receive payments. Instead, the count in the backlog also includes claims needing actions unrelated to issuing payments. This disconnect may cause confusion for the public and policymakers and also creates a false picture of the work EDD has done and needs to do. We asked EDD to modify its calculation to isolate the number of claims in the backlog that were waiting on payment due to incomplete work on EDD's part. That modified calculation showed that of the 685,700 claims EDD reported, fewer than 20,000 had incomplete work that EDD needed to perform so that the claim could be paid. Table 2 shows the key differences between EDD's reported numbers—which uses the approach that the strike team recommended and that EDD adopted—and the revised calculation.
EDD's Public Dashboards Are Significantly Overstating the Number of Backlogged Claims That Remain Unpaid Because of EDD Inaction
|Backlog as reported on EDD's Public Dashboards||Items that do not Depend on EDD Taking Action to Issue Payment||Backlog of Unpaid Claims That EDD Must Take Action to Resolve|
|Pending EDD Action||27,300||– 14,500||= 12,800|
|Claimant Must Submit Certification||345,700||– 345,700||= 0|
|Pending EDD Action||312,700||– 308,000||= 4,700|
Source: EDD's public backlog dashboards as of December 15, 2020, the strike team report, and analysis EDD performed using instructions we provided.
The strike team's approach to calculating the backlog included claims EDD had already paid, in order to better capture EDD's total pandemic workload. A key strike team observation was that EDD measured the timeliness of the first payments it made to claimants but that it did not adequately measure how frequently unresolved issues halted subsequent payments. In addition, to have fully addressed all pending work on a claim, EDD sometimes must make other determinations, such as examining whether the claimant received an overpayment. As a result of its observations, the strike team recommended a new definition of the backlog and provided EDD with a specific methodology for how to calculate the backlog.
The strike team's backlog methodology also included claims where payment has been delayed but where resolving the delay requires action by the claimant. Specifically, it includes a category that counts claims where the claimant has not submitted his or her eligibility certification. Although EDD does not pay a claim without this certification, these are not claims for which unresolved work on EDD's part is delaying payment. Because there are a large number of these claims, including them inflates the total backlog of claims for which EDD must take action. On December 15, 2020, there were about 346,000 claims for which claimants had not submitted certifications.
For those claims that require the claimant to submit their eligibility certification, EDD has taken action during the pandemic to help claimants who may not understand this requirement. Specifically, EDD extended the amount of time claimants have to submit their eligibility certifications from 21 to 120 days from the date EDD issues or mails the certification form to them.In January 2021, EDD reduced this extended amount of time from 120 days to 30 days. There was a corresponding drop in the number of claims represented in its public dashboard as EDD no longer considered over 250,000 claims to be in its backlog. EDD has also been reaching out to these claimants through mailed correspondence and emails to remind them to submit these certifications. Unless the claimant takes action, these claims will expire 52 weeks after the beginning date of the claim, at which point the claimant would have to file a new claim. Although it has taken some actions to remind claimants that they need to submit their certifications to be paid benefits, some claimants may have struggled to obtain from EDD effective assistance with their claim. We describe in greater detail later in this report the challenges that claimants have faced trying to reach EDD for assistance through its call center, as well as how EDD has at times been unable to help virtually any of the claimants that contact its call center and has not answered all web correspondence that claimants submit. Further, other actions that EDD took, such as its action in early January 2021 to require over 1 million claimants to undergo further identity validation, may exacerbate the difficulties that claimants face when contacting EDD for assistance.
According to the UI support division chief, EDD knew before publishing the first calculation of backlogged claims in late September 2020 that the calculation included claims that were not waiting for payment. According to her, the strike team's recommendation was to include these claims because the work EDD still had to perform could ultimately affect the total amount the claimant would receive. For example, in cases where a pending issue is related to possible overpayment to a claimant, EDD must determine whether it paid the claimant the proper amount and, if not, it must issue an overpayment notice to recoup the improperly paid benefits.
Although the strike team's calculation is useful to EDD for understanding its key workloads and managing staff to address that work, EDD's public characterization of the backlog has been different and, at times, inconsistent. EDD has spoken publicly about the backlog several times, with the context of its statements coming amidst concerns about late benefit payments. In that context, EDD spoke during a legislative hearing and in press releases about the backlog of work it has to process without clearly indicating that the backlog figures on the dashboards represented more than just the number of individuals awaiting payments. Notably, in a press release from mid-September 2020, shortly before it released the new backlog calculation, EDD stated that its backlog included over one million claimants who had "stopped receiving payments." We found that EDD missed chances to be more explicit about the composition of the backlog, including at a legislative hearing in October 2020 at which the former director of EDD did not clarify that the backlog included paid claims despite several questions about the backlog.
EDD must have an accurate understanding of its entire workload so it can provide clear information about that workload to the public and policy makers, and so it can determine its priorities. As we describe in more detail later in this report, EDD faces significant challenges in the coming months as it begins to address a substantial volume of deferred work while still fulfilling its duty to issue timely payment to claimants. Not all of that additional work will be captured by the strike team's method of calculating EDD's backlog. EDD will be able to manage those challenges more successfully if it has a comprehensive understanding of its workload. Such an understanding will allow it to better balance its efforts so that it is making informed decisions between expediting payments, combating fraud, and performing other deferred work. In the absence of a comprehensive workload measurement, EDD is at significant risk of placing its staff's attention on tasks that are less urgent than others it has not accounted for.
Therefore, the backlog calculation that the strike team recommended—or some enhanced version of it—will likely prove helpful to both EDD and external stakeholders for understanding the workload challenges EDD faces even though it does not provide an accurate measurement of unpaid claims. If EDD clearly indicates the difference between its total pending workload versus unpaid claims, this will add clarity to the discussions about EDD's performance, workload challenges, and the urgency with which the State or EDD must take action to address related conditions.
EDD's Inefficient Processes Were Unable to Handle the Claim Surge, Resulting in Late Payments
As we discuss in the Introduction, EDD did not make timely payments to a significant percentage of claimants during the claim surge. According to data EDD reports to the Department of Labor, its rate of first-payment timeliness declined significantly from April to September 2020 compared to the same months in 2019. In a regular year unaffected by the pandemic, the Department of Labor measures California's first-payment timeliness by assessing the percentage of claims paid within 14 days of the end of the first week a claimant is eligible for benefits. For claims submitted in April through September 2019, EDD reported that it paid about 75 percent within that 14-day period. In contrast, for the same period in 2020, EDD reported that it made only 61 percent of payments within 14 days. Although these delayed payments occurred in part because of the unprecedented number of claims EDD received, its inefficient claims processing also played a significant role. State law requires EDD to periodically review policies and practices in the UI program and identify those that result in delayed eligibility determinations or benefits payments, those that increase its workload, and those that provide little or no value in identifying fraud or abuse. However, instead of continually improving its policies and practices, EDD has allowed inefficient manual processes to remain. In the three months preceding the claim surge, EDD's automatic initial claim processing rate was at about 30 percent. Since the claim surge began, these inefficient processes have delayed benefits for claimants who require them for essential needs, such as food or shelter.
Most notably, nearly half of the claims EDD processed in the first six months of the claim surge required additional intervention to complete filing after claimants submitted them online. In total, about 4.7 million of the 9.9 million claims EDD processed during this period—about 48 percent—were not filed automatically in UI Online, EDD's online UI application service.As noted in the Introduction, EDD also accepts claims filed on paper or by phone. These claims are not included in the information we present about the number or percentage of claims filed online that were manually processed versus automatically filed because they are, by definition, manually processed claims. Many of the remaining claims required staff involvement to verify claimant identities or resolve issues related to employment information, as Figure 3 shows. For example, when a claimant submits an application with a first or last name that does not precisely match the existing name in EDD's benefits system, staff need to manually review the claim to resolve the mismatch. EDD's workload reports indicate that activities like manual identity verification require significant time for its staff to complete, when compared to other manual work performed by EDD staff for UI claims.
EDD also struggled to efficiently process work related to continued claims during the claim surge. As of September 2020, continued claims represented the majority of those claims for which EDD had pending work to perform that it had not resolved within 21 days. As we explain in the Introduction, after the approval of their initial claims, claimants must certify every two weeks that they continue to meet eligibility requirements. EDD's processing of these continued claims has frequently required significant staff attention. For example, sometimes claimants submit eligibility certifications that contain employment or wage information that does not match the information on their original filings. Staff must investigate these unmatched certifications, and EDD does not pay such claims until it resolves the issues in question.
Most Issues That Prevented Automatic Claim Filing Related to Prior Employment or to Identity Verification
Source: EDD data and internal reports for the month of August 2020.
Note: Data in the figure are from August 2020. We also compared these data to the months of March and June 2020 and found the percentages to be comparable.
According to the strike team's report from September 2020, EDD's failure to promptly process both initial and continued claims was in part the result of its staffing decisions. Most notably, the strike team reported that EDD assigned its most experienced claims processers to help train newly hired claim-processing staff. This responsibility left these experienced individuals too little time to focus on actually resolving claims. The strike team developed a staffing and workload projection tool (workload tool) that isolated 16 critical areas of work for EDD to focus on to eliminate its work backlog by January 2021. The strike team noted that this tool would help address the most time-intensive areas of work that required additional staff to increase productivity.
EDD Recently Increased the Percentage of Claims It Processes Without Staff Intervention, But It May Struggle to Maintain This Progress
EDD asserted that it more quickly paid claims that were delayed by implementing several measures, namely, by redirecting its experienced staff to claims processing, using the workload tool from the strike team, and adopting the automation measures we discuss in this section. Figure 4 summarizes those automation measures. One of those measures was implementing a new method of identity verification as a part of its online claim filing process. EDD chose a tool known as ID.me, which is an identity authentication platform that new claimants must now use to verify their identities before they can file a claim online. EDD implemented ID.me after the strike team recommended that it adopt automated identity verification to reduce its manual workload. Before ID.me, EDD performed manual identity verification. According to EDD's IT branch deputy director, the department had been researching and meeting with vendors in July 2020 to discuss automating identity verification. When EDD implemented ID.me in October 2020—nearly seven months after the claim surge began—it helped to increase the automated processing of new initial claims from about 57 percent to slightly more than 90 percent. In other words, EDD can now automatically process more than 90 percent of new claims filed online without intervention from staff.
However, ID.me is not wholly successful at automating identity verification, and there is a continued need to manually process some claims. If claimants cannot verify their identity through ID.me, EDD requires them to file their claims either by paper or phone instead of online. Data from the first eight weeks of EDD's use of ID.me show that among the estimated number of legitimate claimants who attempted to validate their identities, about 20 percent—just under 144,000—were unsuccessful in validating their identity.ID.me estimates the number of unsuccessful attempts to validate identities that were suspected of fraud. Those attempts are not included in the 144,000. Because EDD has not implemented a way for these claimants to file their claims online, they are forced to use the more laborious process of filing their claim by phone or paper, which also creates more work for EDD staff. Moreover, EDD asserted that it cannot use ID.me on existing initial claims in its backlog that require additional identity verification; instead, EDD staff must complete identity verification using traditional manual methods.
EDD Significantly Changed Its Operations to Issue Payments More Quickly
Source: EDD claim filing reports and interviews with EDD staff.
EDD has also improved its automatic processing of initial claims by implementing an emergency claims processing tool (emergency processing tool) in March 2020. The emergency processing tool is an IT solution that allows claims to progress through EDD's systems under more relaxed claim filing rules than would normally apply to claims. If a claim may require manual processing, the emergency processing tool reviews it and attempts to resolve any issues so that the claim can advance toward payment without requiring intervention from staff. For the most complex of claims, this tool may not be effective and staff may still need to intervene. Since its implementation and through September 2020, the emergency processing tool successfully filed 20 percent of all processed claims. Although EDD instituted this tool at the outset of the claim surge, it has modified it since to increase the number of claims that the tool routes for automatic filing.
EDD was not able to project whether it will sustain the rate of more than 90 percent of initial claims filed automatically once the pandemic conditions subside. ID.me is likely to continue to support EDD's efficiency even during regular operations; however, it is less clear that EDD will continue to sustain the level of automation resulting from using the emergency processing tool. Although EDD temporarily relaxed certain rules so it could issue more timely payments, reapplying these rules once EDD's workload decreases is likely important. For example, one rule relates to determining whether regular UI claimants had been paid at least the minimum required amount of wages in the qualifying period before the date they filed their claim to be eligible for benefits. Relaxing this rule helped to reduce EDD's manual workload because Pandemic Unemployment Assistance (PUA) claimants are not required to have earned a minimum amount of wages to receive a minimum weekly benefit amount. Therefore, when claimants filed for UI benefits without indicating wages but had noted that COVID-19 had affected their ability to be employed, this relaxed rule allowed EDD to automatically file their claims as claims for PUA benefits without needing to manually process the claim. However, when the PUA program expires, EDD will likely find it cannot continue to automatically file claims that have this particular issue.
Depending on how much functionality EDD can preserve in this tool, its automation rate may decrease considerably. When we asked EDD about whether it would be able to continue using the relaxed rules integral to the emergency processing tool to sustain its level of automation, the UI support division chief indicated that EDD will eventually analyze the tool to determine which relaxed rules will be important to reinstate and which can now be incorporated into the UI Online application process. The earliest EDD expected it could perform this analysis was spring 2021.
Because of a recommendation from the strike team, EDD now has a significant opportunity to increase its automation and efficiencies as it emerges from its current backlog. In November 2016, EDD initiated a multiyear Benefit Systems Modernization (BSM) project to modernize its UI, State Disability Insurance, and Paid Family Leave benefit systems by implementing a single, integrated benefit system. EDD had planned to implement the UI portion of the BSM solution last, with an expected completion date of March 2024. However, at the recommendation of the strike team, EDD paused the implementation of BSM. The strike team recommended that EDD restart the project with an incremental approach of modernizing the benefits systems based on areas of critical need. To make sure that it retains as much of the improvement in its automated claim filing rate as possible, EDD should identify opportunities for incremental system modifications that it can implement in the near term to improve its claim processing, such as implementing online applications for claimants who are currently required to file on paper or by phone.
EDD's Failure to Plan for Another Rise in UI Claims Leaves It Vulnerable to a Continuing Backlog
In the fall of 2020, EDD did not plan for another increase in UI claims in the winter months, which left it susceptible to a rise in backlogged claims. EDD has used the workload tool that the strike team provided to help understand and manage its backlog of claims and address the most critical areas of work as its highest priority. The workload tool has guided EDD in deciding to increase or to decrease the staff it assigns to claims processing. However, because the workload tool uses data EDD enters to estimate its upcoming workload, the tool's usefulness is directly related to the quality of the information EDD provides, and EDD has not used available data about upcoming claims or modeled possible scenarios it could face if there is another round of economic shutdowns. In the absence of such modeling, EDD's overly optimistic projections of its workload may lead it to shift staff and other resources away from claims processing, leaving it shorthanded again if claim volumes spike.
A key input to the workload tool is the number of claims that EDD expects to receive in coming weeks: a higher number of expected claims results in a higher projection of the level of work EDD will need to address. Nonetheless, at the time we began our audit in October 2020, the management team overseeing the tool was updating EDD's projections each week by using the number of new claims from the previous week as the expected number of claims for each week in the upcoming three months. Effectively, this decision resulted in EDD's projected workload for a multiple-month period being entirely reliant on the claim rate of just one week.
Such a narrow input can have a dramatic effect on EDD's understanding of its upcoming work. For example, during the last three weeks of October 2020, the number of new claims submitted dropped each week. When EDD updated its workload projection each week, it entered the decreasing numbers of new claims, causing its projected workload for the coming months to fall. In November 2020, the EDD management team altered its approach slightly and began using an average of the new claims from prior weeks going back to mid-October as its new expected number of claims, and used that approach through the end of our fieldwork. However, this is still a retrospective as opposed to a prospective approach. A prospective approach would involve using a reasonable increase or decrease factor based on expected unemployment rates and historical data.
EDD had information it could have used to implement a more accurate approach. For example, UI claims historically trend upward at the end of the year and into January. From 2015 through 2019, claims rose an average of 25 percent between October and January of the following year. EDD explained that claims generally increase during this period because of winter weather's effect on seasonal industries, such as agriculture, and because of January layoffs in the retail industry following the holidays. However, EDD did not take these historic trends into account in forecasting its workload.
Perhaps more importantly, EDD failed to take into account indicators that the pandemic might worsen and result in significant economic repercussions. EDD was aware that in November 2020, COVID-19 cases rose in California. On December 3, 2020, the Governor issued regional stay-at-home orders to be enacted if the capacity of regional hospitals' intensive care units fell below 15 percent. These events increased the likelihood of a rise in UI claims as businesses shut down or reduced their operations. In fact, a forecast that EDD's Program Estimates Group (estimate group) released in October 2020 predicted a 21 percent increase in new claims from October through January 2021. However, EDD's management team decided not to use this forecast for workload planning. In addition, EDD did not take into consideration the possibility that the federal government would approve additional UI benefits or make more changes to existing UI programs that would increase EDD's workload. Recent action by the federal government in late December 2020 brought about such changes and now EDD faces additional work that it had not planned to perform.
EDD's process improvement manager (improvement manager)—who is part of the management team that uses the workload tool—agreed that modeling multiple scenarios, including a large increase in claims, would be beneficial. However, she indicated that the management team has not had the time or opportunity to do so. According to the improvement manager, EDD's management team decided not to use the estimate group's forecasts for new claims from November 2020 through January 2021 to project EDD's workload because, among other reasons, the group's forecast had overestimated the number of claims EDD would receive in October 2020. The improvement manager acknowledged that as of the beginning of November, the team had not held extensive discussions about how an increase in claims resulting from either new federal legislation or another economic shutdown would affect EDD's projection of work. Under the current economic climate, EDD's failure to include these factors in its modeling is a surprising omission and a significant misstep. If EDD continues to project its work based only on recent claim submission rates, it risks being unable to quickly address spikes in its workload and issue timely payments to Californians in need of assistance.
To ensure that EDD's claims processing is as effective and efficient as possible, the Legislature should require EDD to convene a working group to assess the lessons learned from the claim surge and identify the processes that EDD can still improve. That working group should do to the following:
- Include representatives from EDD's UI branch, IT branch, and executive management. It should also include representatives from the strike team.
- Issue a report on the lessons learned from the claim surge by no later than January 2022. The report should identify any improvements that the working group recommends that EDD make and include a review of EDD's implementation of the strike team's recommendations.
To provide a more transparent picture of claims in its backlog, by March 2021 EDD should revise its public dashboards to clearly indicate the number of claims that have waited longer than 21 days for payment because EDD has not yet resolved pending work on the claim.
To ensure that its identity verification processes are as robust as possible, EDD should determine by June 2021 the reasons why claimants cannot successfully complete their identity verification through ID.me and work with its vendor to resolve these problems. EDD should thereafter regularly monitor the rate of successful identity verifications to ensure that it consistently minimizes unnecessary staff intervention.
To retain as much automation in initial claims processing as possible, by June 2021 EDD should determine the automation modifications achieved through its emergency processing tool that it can retain and by September 2021 it should make those a permanent feature of its UI Online application.
To ensure that it does not delay needed improvements to its IT systems, EDD should, by June 2021, identify the elements of the BSM that can assist it in making timely payments and that it can implement incrementally. It should then prioritize implementing the elements most likely to benefit Californians.
To ensure its ability to respond in a timely fashion to fluctuations in its workload, EDD should immediately begin modeling workload projections that account for possible scenarios that would cause a spike in UI claims. EDD should plan its staffing around the likelihood of those scenarios, including having a contingency plan for less likely scenarios that would have a significant impact on its workload.
Because EDD Responded to the Claim Surge by Suspending Certain Eligibility Requirements, Many Californians Are at Risk of Needing to Repay Benefits
- In March 2020, EDD halted most of its work related to determining whether UI claimants were eligible for benefits. Although that step likely resulted in more timely delivery of payments to individuals in need, it conflicted with core UI program tenets. EDD is now faced with a workload of about 12.7 million deferred eligibility issues that affect up to 2.4 million claimants, and also related efforts to recover benefits it paid to any of those claimants it may deem ineligible.
- For eight weeks in the spring of 2020, EDD suspended the requirement that claimants certify their eligibility to continue receiving benefits after their initial claims were paid. This decision created another large pending workload for EDD and left nearly 1.7 million Californians at risk of needing to repay benefits.
To Mitigate the Number of Delayed Payments, EDD Stopped Determining Whether All Claimants Were Eligible to Receive Benefits
As claims began to surge in March 2020, EDD halted most of its work determining whether claimants were eligible for UI benefits. This action curbed the size of its claims backlog significantly and resulted in more timely payments to Californians. However, it also compromised the integrity of the program and may hinder the ability of the department to conduct day-to-day operations in the future. As we discuss in the Introduction, claimants must meet various eligibility requirements to qualify for UI benefits. In response to the pandemic, the Department of Labor issued guidance about the flexibility states had to interpret and understand key eligibility requirements. Specifically, it advised states about how to determine whether an individual was able and available for work in light of the pandemic. However, the Department of Labor maintained that states must still apply the able and available criteria. If a claimant might be ineligible, EDD staff must conduct additional work—such as an interview with the claimant—to make an eligibility determination. Indications of potential ineligibility include a claimant reporting that he or she voluntarily quit a job or was discharged for work-related misconduct. However, as Figure 5 shows, EDD suspended the work necessary to make most eligibility determinations in order to better manage its workload.
EDD Delayed Eligibility Decisions to Keep Its Claim Backlog From Growing
Source: Review of March 20, 2020, memo from agency secretary, data from EDD, and federal guidance.
EDD enacted these changes at the very beginning of the claim surge. Specifically, on March 20, 2020, the agency secretary directed EDD to temporarily pay all claims without determining whether claimants met key eligibility criteria: being able to, and available for, work. The agency secretary made this directive after receiving a recommendation from EDD to do so. This directive remained in effect at the beginning of December 2020. However, the agency secretary's directive required EDD to maintain its identity verification practices as well as to continue to allow employers to contest unemployment claims.Because employers fund unemployment insurance through taxes, they are stakeholders in the process. An employer could contest a claim for various reasons, including because it fired the employee for work-related misconduct. In response, EDD stopped making some of the specific eligibility determinations the agency secretary had identified, but it also suspended its review of many additional eligibility issues that it would usually examine. These issues included, for example, determining whether a claimant who reported voluntarily quitting a job or refusing suitable work had good cause to do so. Effectively, EDD stopped making most required eligibility determinations. According to EDD's general counsel, the Labor and Workforce Development Agency did not learn of this additional action on EDD's part until several months later.
As we noted earlier, EDD struggled to pay claims on time during the claim surge in spite of the fact that it made this decision. Had EDD performed additional work required for eligibility determinations before issuing payments, the backlog would have been even greater, and it almost certainly would have further slowed access to critical benefits for many Californians.
However, the Department of Labor determined that this decision likely compromised the integrity of California's UI program. In September 2020, the Department of Labor learned about EDD's decision to suspend many eligibility determinations. On December 4, 2020, the Department of Labor notified EDD that it believed those actions conflicted with a core tenet of the UI program, namely not paying benefits to ineligible claimants. It noted that without conducting eligibility determinations, EDD could not be certain that individuals are eligible for benefits and instructed EDD to immediately resume all eligibility determinations. It also directed EDD to begin examining all the suspended determinations that had accumulated. As of December 3, 2020, EDD's claims processing system had flagged about 12.7 million potential eligibility issues on claims since March 2020, affecting up to 2.4 million claimants.
Although EDD's actions likely allowed it to pay benefits faster, EDD now faces an impending workload for which it has no clear plan to address and that could have significant consequences for claimants. Resuming all eligibility determinations will slow how quickly EDD can process claims and pay benefits. Processing the 12.7 million suspended determination issues on prior claims will also present significant challenges. EDD estimates it takes 30 minutes on average to resolve a determination. Even if it only had to resolve half of the suspended determination issues, it would still take EDD over 3 million hours to do so. According to the UI support division chief, this calculation overestimates the number of determinations needing significant work to address. However, EDD's own analysis of the work that must be done is still preliminary. As of mid‑December 2020, the UI support division was in the process of drafting a plan for resuming all eligibility determinations and addressing deferred determinations. The plan was due to be finalized in January 2021. Ensuring that benefits are paid in a timely fashion while simultaneously reviewing the unprecedented number of claims for which it suspended determinations represents a workload never before seen by the department, which is already struggling, and threatens its ability to effectively conduct ongoing operations.
When it conducts these eligibility determinations, EDD will likely find that some of these claimants were in fact not eligible for the benefits they received. For example, EDD told us that in 2019, it disqualified about 164,000 claims because it found during its eligibility review that the claimants had voluntarily quit or been dismissed because of misconduct. Although some ineligible claimants are actively attempting to defraud the UI program, others may be genuinely confused about the eligibility requirements. These claimants now face the possibility that they may have to repay some or all of the benefits they received in good faith—and many will have already spent these benefits.
If a claimant has to repay benefits, EDD offers installment payment options. For nonfraud repayments, claimants typically have 48 months to repay the amount they owe. Further, EDD generally has the ability to waive repayment when cases meet certain criteria, such as when there is no fraud involved. However, it is unclear the extent to which it can do so for the current claims because it has not yet analyzed them. State law outlines a two-year period during which EDD can issue an overpayment notice requiring the repayment of benefits in cases that do not involve fraud.
However, in the case of fraudulent claims, the likelihood that EDD could recoup the payments may be very low. For example, EDD informed us that in September 2020, it flagged about 250,000 claims as having been filed using suspicious addresses. In the event these claims represent fraudulent activity, EDD may find it difficult to identify the bad actors who filed the claims and to pursue recovery of the benefits it paid out. That difficulty will come primarily because these claims were likely filed using another person's identity and address at which the perpetrators likely do not live.
As it moves forward and implements the Department of Labor's directive, EDD must employ a strategic approach to ensure—to the extent possible—that it provides eligible Californians with benefits to which they are entitled in a timely fashion. A key factor in planning the approach EDD takes will be assessing the risk level of the various eligibility issues. For example, EDD may identify that some eligibility issues have historically been more common reasons to disqualify an individual or are more indicative of fraud. Further, it may find that because the federal government temporarily broadened certain eligibility requirements, it is unlikely to disqualify individuals for certain reasons it may have used to do in the past. Once it has developed a risk assessment model, EDD would need to use it to process the 12.7 million deferred determinations to ensure that it is addressing those that represent the highest risk of ineligibility or fraud first.
If EDD cannot complete the risk assessment modeling on its own, it should consider partnering with another state agency or hiring contracted help to expedite the work. The improvement manager noted that although outside help could take some of the burden off EDD's limited staff resources, EDD staff would still need to provide assistance in obtaining and understanding data and EDD processes. In her opinion, it would be a better return on investment to hire additional staff to perform data analysis, as these staff could serve as a continuing resource for EDD.
Because EDD Told Claimants Not to Certify Their Continued Eligibility, It Faces Another Significant Workload It Must Manage
A second decision by the agency secretary to suspend required work has created another large, pending workload for EDD and has left close to 1.7 million Californians at risk of needing to repay benefits that they received in the first half of 2020. UI claimants must certify every other week that they remain eligible for benefits, and EDD must validate those certifications by confirming the information the claimants report in order for the claimants to continue receiving payments. These certifications are California's way of ensuring compliance with federal program integrity requirements. However, on April 23, 2020, the agency secretary directed EDD to suspend the certification requirement for claimants for eight weeks, from mid‑March through early May 2020. The agency secretary stated that she was issuing the directive in light of a persistent inability of many claimants to access EDD's website to submit certifications because of a historically high volume of claims and the pressure they had placed on the website.
Following this direction, EDD automatically paid claimants without requiring them to submit certifications for their continued claims (autocertifications). Autocertifications are different from the eligibility issues we discuss in the previous section, wherein claimants submitted information about their claims but EDD deferred eligibility determinations. In the case of autocertifications, EDD did not require the necessary certifications from any continued claimants during the eight-week period. According to information from EDD, it paid nearly 1.7 million claimants more than $5.5 billion in benefits over this period.
A few days after the agency secretary's directive, EDD's director recommended to the agency secretary that California stop its autocertification process earlier than originally planned. The director noted that autocertifications did not reduce traffic on EDD's website, created a larger manual workload for staff when some claimants chose to mail in certifications, and did not meet standards in federal law and regulations. In response, the agency secretary reaffirmed her direction to continue autocertifying claimant eligibility. Correspondence we reviewed shows that the agency secretary coordinated with the California Office of Digital Innovation (ODI) when making this decision. The communication shows that ODI staff agreed with the EDD director that autocertifications had not reduced traffic on the EDD website and separately noted that the autocertifications were not having the desired effect because claimants chose to certify anyway. However, ODI also noted in its assessment that EDD's systems had experienced problems because of certifications and that certification volumes were likely to grow in the immediate future, which left EDD vulnerable to additional system problems in the future. Finally, ODI noted that ending the autocertification process early when the time period for autocertification was almost completed could lower claimants' level of trust with EDD and could make it harder for EDD to effectively communicate with claimants in the future. Separately, the correspondence we reviewed also show that the agency secretary shared ODI's concerns about system stability and that EDD would lose credibility because of the conflicting messages to the public. In May 2020, the Department of Labor issued a letter to all states reminding them of the requirement to continue collecting biweekly certifications and that even the temporary suspension of claimant certification does not meet standards in federal law.
Similar to the decision to suspend eligibility determinations, the agency secretary's decision to suspend the certification requirement likely resulted in more timely payments to UI claimants. However, that decision again threatens EDD's ability to effectively administer the UI program at a time when residents of California still depend on timely assistance. These pending retroactive certifications represent a significant workload that EDD will need to process, in addition to the upcoming work in addressing the millions of eligibility determinations that it has postponed since March 2020. During July and August 2020, EDD notified the nearly 1.7 million affected claimants that they would need to retroactively submit all certifications by November 21, 2020, for the weeks they received benefits. To avoid placing an additional burden on its unemployment benefits information system that typically receives the online certifications, EDD contracted with a vendor to implement a new system for accepting these retroactive certifications. As of November 23, 2020, about 67 percent of these claimants—or 1.1 million—had submitted retroactive certifications, representing the equivalent of almost 4 million weeks of benefits that EDD will need to process. As it continues to collect retroactive certifications, EDD will need to validate that claimants were eligible for payments and that they were paid the appropriate amount of benefits. For those claimants who do not submit their retroactive certifications, EDD must investigate the claims and potentially initiate the process to recoup the overpaid amounts.
These retroactive certifications add to EDD's already sizeable volume of work. In October 2020, EDD told us that it did not know how quickly the Department of Labor expected it to complete processing these millions of retroactive certifications. The UI support division chief indicated that EDD would begin to process the retroactive certifications when it finished processing the backlog of claims, which it expected to complete in January 2021. However, as EDD finishes its work on backlogged claims, it faces not just these retroactive certifications but also the continuing economic repercussions from the COVID-19 pandemic and the Department of Labor's recent directive to resume all eligibility determinations that we discuss in the previous section. In other words, EDD is at risk of having to manage another influx of claims while also processing the retroactive certifications and its deferred eligibility work. Given the previous challenges that EDD has had in managing higher than normal levels of work, it is not clear that EDD will be able to effectively and efficiently manage all three of these workloads.
Further, if EDD determines that a claimant was ineligible or that it paid claimants more than they were eligible to receive, it will need to begin a process to recoup the identified overpayments. Some claimants may have experienced a change in eligibility or a return to partial employment after filing their claims that would have affected the benefit amount they received had EDD collected and reviewed their certifications as usual. Figure 6 shows an example of such a situation in which even a claimant who followed all of EDD's instructions could be asked to repay benefit payments. In each of the three calendar years preceding 2020, the percentage of the total benefits that EDD paid to claimants for which it later issued overpayment notices was about 3 percent. EDD's UI support division chief does not believe that this historic ratio will necessarily apply to the $5.5 billion it paid in benefits during the autocertification period because the eligibility requirements have been more lax for some claim types during the claim surge. Nonetheless, as a guide, we used the historic overpayment rate of 3 percent to estimate that of the $5.5 billion EDD paid to claimants, it will likely need to issue repayment notices to claimants that total more than $160 million. As we discuss in the previous section, EDD has the authority to forgive repayment on some, but not all, claims, and for nonfraud cases, EDD allows claimants up to 48 months to repay benefits. Nonetheless, for those claimants who justifiably believed they received the correct benefit amount or who made unintentional errors in their claim files due to their unfamiliarity with the claims process, these overpayment notices could represent a serious financial strain.
Some Claimants May Owe Repayment of Benefits Because EDD Did Not Collect Eligibility Certifications
Source: A hypothetical example based on state and federal law and regulations, EDD policies and procedures.
Finally, the claim surge and corresponding delayed payment on claims has generated significant levels of interest from the public and the Legislature. EDD's claims processing has been the subject of many news reports and legislative hearings, and members of the Legislature report fielding numerous calls from constituents about their claims. It is almost certain that a similar level of interest will exist for information about how many Californians may be subject to overpayment notices and how far EDD has progressed in processing that workload. Consequently, the Legislature would add to the public transparency of EDD's operations by requiring it to report on these and other related metrics when it resolves both the retroactive certifications we discuss in this section and the deferred eligibility decisions we discuss earlier.
To ensure transparency in EDD's operations and provide information to policymakers, the Legislature should require EDD to report on its website at least once every six months the amount of benefit payments for which it must assess potential overpayments, the amount for which it has issued overpayment notices, the amount it has waived overpayment on, and the amount repaid related to those notices. The reports should encompass benefit payments EDD made from March 2020 until the time when it resumes all eligibility determinations. EDD should be required to publish these reports until the repayment period for all the notices has elapsed.
To continue providing timely payment of benefits to Californians in need while also effectively responding to the Department of Labor's directive regarding immediately resuming all eligibility determinations and resolving all suspended determinations, EDD should do the following:
- Perform a risk assessment of its deferred workloads, including deferred eligibility determinations and retroactive certifications. EDD's assessment should take into account the relative likelihood that it issued payments to ineligible claimants by considering historic overpayment trends as well as the new or altered eligibility requirements the federal government adopted in response to the pandemic. If necessary, EDD should either partner with another state agency or contract for assistance in performing the analysis in support of this assessment.
- Develop a workload plan that prioritizes its deferred workloads based on the risk assessment and determine the staffing and IT resources needed to accomplish the work within expected time frames.
- Hire and train staff as necessary in order to carry out the workload plan.
- Using the workload plan, EDD should process the deferred work in alignment with the following: the need to pay timely benefits to new or continued claimants, federal expectations about the urgency of the deferred work, and any deadlines by which EDD may no longer be allowed to recoup inappropriately paid benefits.
EDD Took Uninformed and Inadequate Steps to Resolve Its Call Center Deficiencies
- Even before the claim surge, EDD struggled to answer a high rate of calls. Then, at the start of the surge, EDD answered less than 1 percent of calls and failed to answer hundreds of thousands of requests for assistance that claimants submitted online.
- Although EDD added thousands of staff members in response to the claim surge, it failed to adequately address the significant weaknesses in its call center's performance, in part, because of its lengthy training program and in part, because it has not collected critical information about why claimants call for help.
- EDD has not implemented best practices for managing its call center that would help it operate more effectively and improve its customer service.
The Claim Surge Worsened the Already Poor Performance of EDD's Call Center
EDD provides several different avenues through which UI claimants can request assistance, including its UI Online website, email, online chat, and call center. As of January 2020, the call center had about 1,270 EDD agents located in field offices across the State. These agents provide customer service by answering questions and providing assistance to Californians who need help with UI claims, and they also spend time performing off-phone work processing UI claims. When contacting the call center, claimants call a toll-free number and navigate a series of prerecorded messages—known as Interactive Voice Response—that routes callers based on the options they select.
Even before the claim surge, EDD struggled to meet a critical benchmark for its call center's performance. Specifically, in 2014, EDD made a commitment to the Legislature to answer 50,000 claimant calls per week as a result of receiving increased funding to hire staff. EDD refers to this commitment when it makes staffing decisions and it pointed us to other documents related to that commitment as a way to measure the success of its call center. However, from January to mid‑March 2020, EDD answered only about 42,000 calls per week on average. Although it answered at least 50,000 calls in four of the 11 weeks in the period, it failed to meet this benchmark in the other seven weeks, and it was often significantly below 50,000. Worse yet, the calls EDD answered represented less than 25 percent of the 184,000 calls requesting to speak to an agent it received on average each week. In fact, EDD's phone system blocked 17 percent of call attempts because of its technical capacity limitations.
When we asked EDD about the performance of its call center during this period, it cited two factors—a low unemployment rate that resulted in less funding for staff and the number of holidays early in the year—as contributing to its struggle to answer 50,000 calls each week. However, EDD often failed to meet the 50,000-call requirement even during several weeks without holidays. Further, the 2014 commitment that EDD made to the Legislature did not have an allowance for answering 50,000 calls some weeks while failing to do so in other weeks. Regarding the high percentage of blocked calls, EDD asserted that the majority of these calls occurred during particularly busy times of day, such as the first few minutes that the call center was active on Monday mornings. However, call data also show that once total calls increased at the start of the claim surge, a significant portion of calls were blocked during all call center hours.
The claim surge exacerbated EDD's struggles at answering calls to the point at which the call center effectively stopped providing service to almost all callers. As Figure 7 shows, the number of callers trying to reach the call center spiked drastically from 120,000 per week in the middle of March 2020 to more than 1.7 million in a single week by the end of April 2020, while the number of calls EDD answered increased only marginally. During this period, individuals attempting to reach EDD's call center were almost universally unsuccessful in speaking to an agent—hundreds of thousands of callers were unable to speak with an agent each week, and EDD answered an average of only 0.5 percent of total calls per week. In fact, based on the number of unique callers and total calls at the end of April 2020, the average unique caller attempted to reach EDD at least 10 times. This suggests that if EDD had been able to more immediately answer questions from a greater number of unique callers, the total number of calls would have likely fallen. We provide additional details about the calls EDD received in Appendix A.
Although it took some action to address the large volume of calls, EDD failed to meaningfully improve its call answer rates or customer service. On April 15, 2020, the Governor signed an executive order requiring EDD's oversight agency to expand call center hours and ensure sufficient staffing levels to process claims. Following this order, EDD quickly implemented a preliminary, minimal version of a new phone system—known as a virtual contact center (VCC)—in late April 2020. The VCC allowed agents who were working remotely because of stay-at-home orders to answer claimant calls, but it lacked some functionality that EDD's previous phone system featured. Additionally, EDD expanded call center hours and began adding thousands of agents to answer calls and perform other tasks related to claim processing. We discuss the shortcomings of EDD's hiring efforts in the next section.
Hundreds of Thousands of Callers Did Not Reach EDD Agents When Calling During the Pandemic
Source: EDD call center data, staffing records, and interviews with EDD staff.
These two actions resulted in a gradual increase in the number of full-time equivalent agents answering calls, from fewer than 100 at the start of April 2020 to about 1,000 by the end of August 2020.Full-time equivalent (FTE) is a metric that measures the equivalent number of full-time employees based on hours worked rather than the total number of employees. Because some agents spend only part of the day on the phone, EDD's FTE total is lower than the total number of staff who answer calls. This increase in staff may have had some positive effect, as total calls decreased significantly during the month of August. However, the number of unemployed Californians and claims filed during August 2020 also dropped, which likely also contributed to the decrease in total calls. Although EDD agents answered about 161,000 total calls during the first week of August, these answered calls still represented only 16 percent of the roughly 1 million unique callers who contacted EDD during that week, highlighting the fact that the vast majority of callers faced continued difficulty in speaking with agents.
EDD's poor call center performance during the claim surge is of special concern because EDD also failed to answer hundreds of thousands of questions claimants submitted online during this period. EDD stated that the questions generally involved concerns claimants encountered as they filed claims. For example, claimants often sought to correct mistakes they had made in reporting wages because they were worried about penalties they might face for overpayments. In just two of the help categories that claimants could select—both of which pertained to claimants trying to alert EDD to mistakes on their claim documentation—over 400,000 unanswered questions accumulated from March through October 2020.
In November 2020, EDD decided to automatically resolve messages in these two categories without addressing them. After reviewing a small number of these messages, EDD determined that many were likely no longer pertinent because the claimants had likely received their UI payments since they sent their messages. The chief of the UI support division, which in part provides technical and administrative assistance to the rest of the UI branch, asserted that staff time could be spent more productively on other work related to paying claims rather than responding to these messages. However, EDD's failure to respond to such a large number of online messages from claimants—while simultaneously struggling to answer the vast majority of calls it received—highlights the degree to which it left hundreds of thousands of claimants without answers during uncertain times.
Although EDD Added Thousands of New Agents, They Were Often Unable to Assist Callers
EDD added thousands of agents in response to the claim surge. As we mention earlier, in January 2020, EDD had about 1,270 agents responsible for answering calls to its call center and processing UI claims. EDD refers to these agents as employment program representatives (EPR agents). According to EDD's UI southern operations division chief, EPR agents generally spend four hours a day answering phone calls. EDD indicates that the job duties of an EPR agent include answering calls and providing prompt, accurate, and courteous customer service, but EPR agents also have other job duties such as interviewing claimants to verify information on their applications. From January through October 2020, EDD hired more than 2,000 additional EPR agents.
However, the newly hired EPR agents could not be immediately available to answer calls because they needed to first complete a lengthy training program. EDD trains EPR agents in each aspect of the UI claim process—including how to file claims and how to determine whether claimants are eligible for benefits—so that they are well equipped to answer most questions from callers. EDD indicated that as a result, until mid-March 2020, the training program took from 10 to 13 months to complete. Although EDD made changes in response to the claim surge that eliminated some training time, fully training a new EPR agent still took nine months. In fact, EDD explained that training may take longer because managers sometimes divert new employees from training to help complete other priority assignments.
EDD's two UI operations division chiefs said that they sometimes assign EPR agents to answer calls partway through training, after the new agents have learned how to file claims. EDD estimates that reaching this stage of the training takes an average of 90 days. According to the UI program training manager, at that point EPR agents should be ready to answer questions about filing claims but are unable to answer questions about eligibility determinations. The EPR agents whom EDD hired in April and May 2020 are not expected to begin training in eligibility determinations until January 2021.
In part because of the considerable time EDD needs to train its 2,000 new EPR agents, EDD added another 2,000 employees as tier 1 agents. Before the onset of the claim surge, EDD did not employ any tier 1 agents. EDD told us it created this tier of agents with the expectation that it could train and deploy them more quickly than EPR agents. EDD acquired these tier 1 agents by hiring new staff, redirecting existing staff from other branches, and contracting with a vendor. EDD initially taught tier 1 agents only how to provide basic technical assistance, such as how claimants could register online or reset a password in contrast to the more robust training that it provides to EPR agents. In total, by October 2020 EDD had quadrupled its call center staff for the UI program to more than 5,600 people, as Table 3 shows.
According to EDD, tier 1 agents began answering calls on its new VCC phone system in late April 2020, just days after the Governor ordered the increase of call center staffing; however, EDD had no pre-existing plan for training tier 1 agents, which left it little time to develop a new training plan. Further, EDD had not adequately determined the skills those agents would need to effectively answer calls, hindering the training team's ability to develop appropriate training that aligned with the needs of callers.
EDD Has Quadrupled Its Call Center Staff in Response to COVID-19
|Call Center Staff||January 2020||October 2020*|
|Tier 1 Agents||Office technicians||0||850|
|Redirected EDD staff||0||710†|
|Tier 1 total||0||2,060|
Source: EDD staffing data.
* Staffing totals include new employees who are still in training.
† Total redirected EDD staff as of November 1, 2020.
‡ Total staff, regardless of number of hours worked. Some agents spend only part of the day on the phone.
Again according to EDD, it did not use specific data on the reasons people call EDD when designing this training program because it did not have these sorts of data. Thus, EDD taught the tier 1 agents only how to provide basic technical assistance, in large part because it already had available training materials on these topics and because its EPR agent training was too long and complex to feasibly teach in a short window of time. However, it quickly recognized that callers needed more than technical assistance. Within three weeks, EDD expanded its training to include additional skills, such as how to add certain wage or certification information, in hopes of assisting more claimants. Even though the expanded tier 1 agent training covered some of the same topics as the EPR agent training, it did not cover them in as much depth.
The fact that EDD did not incorporate into its training for tier 1 agents the specific reasons people call has had significant consequences for the callers who reach these agents because they have been unable to assist many callers. When tier 1 agents are unable to answer a caller's question, they can try to transfer the caller to an EPR agent. However, because EPR agents are not always available to accept the transfer, EDD has its tier 1 agents keep a list of unresolved calls that they cannot transfer (callback list). When we examined callback list data provided by EDD for late April through late October 2020, we found that on average, tier 1 agents added to the callback list 29 percent of the calls they answered. This suggests that tier 1 agents were often unable to provide the same degree of prompt customer service that EDD expects from its EPR agents.
Moreover, the actual number of callers whom tier 1 agents were unable to assist may be greater than our analysis of callback data suggests. Our ability to determine the number of callers EDD actually assisted is limited because EDD does not track whether it resolves callers' questions. As a result, it cannot demonstrate whether it addressed the concerns of callers who are not on the callback list. In September 2020, the strike team estimated that the maximum percentage of answered calls that tier 1 agents successfully resolved in the first week of August 2020 was just greater than 1 percent.
EDD has made some improvements to its tier 1 training since tier 1 agents began answering calls in late April 2020. Specifically, it has added new topics to make tier 1 agents more effective at answering calls. It has also implemented a training database to make available key information that can help tier 1 agents answer claimants' questions. In addition, in October 2020 the training manager began analyzing callback list data to identify the types of questions tier 1 agents need the most help answering. However, EDD undertook these improvements only recently, and tier 1 agents remain unable to complete many responsibilities that EPR agents can complete.
Despite Its Ongoing Struggles, EDD Has Not Adopted Best Practices to Effectively Manage Its Call Center
EDD has not implemented several best practices that would enable it to improve the performance of its call centers. Improving its ability to successfully respond to callers is of critical importance, as providing customer service to claimants is one of EDD's key responsibilities. We reviewed numerous reports and articles related to managing both commercial and government call centers to identify best practices for improving customer service that EDD should adopt. Figure 8 summarizes these practices. EDD either has not adopted these practices or lost the features that enable them when it implemented its new VCC phone system in April 2020. Consequently, EDD is not operating as effectively as it could be nor always resolving callers' questions.
EDD has yet to adopt a critical and foundational best practice that commercial and government call centers use: collecting and analyzing call data to understand the specific reasons why customers are calling. In 2011, we recommended that EDD track this kind of data, and, in 2017, it provided evidence that it had performed some of this type of analysis. Yet it appears it is no longer doing so with its new VCC phone system. Although EDD has data that tracks the specific reasons why people call, EDD's staff confirmed that it has not yet begun analyzing these data or using them to better manage its call centers. If EDD analyzed call data that show the specific questions driving the highest volume of calls—such as questions about filing a new claim or verifying identity—it could adjust call center operations to better address these concerns.
Knowing why people call for help would enable EDD to more efficiently train agents to answer the questions driving the highest call volume. As we previously discuss, it currently takes EDD nine months to train a newly hired EPR agent to answer all claim‑related questions. However, with a better understanding of why most people call for assistance, EDD could quickly train agents to answer common inquiries. Further, both EDD's former phone system and its new VCC phone system route callers to agents based on the issue with which they need assistance, a feature known as skills‑based routing. EDD could therefore quickly train agents in specific skills and then route callers to those agents as appropriate, allowing it to more efficiently train agents for answering calls. When we asked EDD's training director about the possibility of using call data and skills-based routing to make its training more nimble, she replied it would be both feasible and beneficial, and she would like to implement this approach in the future. Leveraging call data and skills-based routing to efficiently train staff would have particularly valuable benefits during periods of high call volume, such as recessions, because it would allow EDD to more quickly hire and train additional staff.
EDD Should Implement Best Practices to Improve Call Center Performance and Customer Service
Source: Review of call center best practices and documentation of EDD's call center operations.
Although faster training and use of new staff is a key benefit EDD could derive from analyzing call data, it is perhaps even more critical for EDD to use its call data to understand why claimants are turning to the call center in the first place. EDD has implemented self-service options—such as UI Online and a frequently asked questions (FAQ) webpage—that allow Californians to obtain assistance with minimal intervention from an EDD agent. Given that EDD will likely never have the number of trained agents needed to answer all of the calls it receives at peak volumes, it should take steps toward identifying how it can improve its self‑service options so that fewer individuals need to call for assistance. By analyzing call data to better understand the specific reasons why claimants use the call center, EDD could revise its self-service options to better address the questions and challenges claimants face. For example, EDD could pinpoint specific steps or issues in the claim process that result in claimant difficulties and then update UI Online or add content to its FAQ webpage to help claimants resolve these issues without agent intervention. In doing so, EDD would likely reduce the number of claimants requesting assistance through its call center, increasing the likelihood that callers promptly receive the assistance they need.
Further, EDD's UI branch is also not tracking one of the most critical performance indicators for call centers. The sources we reviewed consistently identified as essential the practice of tracking the number of callers whose questions were answered on their first call. This indicator is referred to as first-call resolution. Both commercial and government call centers track first-call resolution because it demonstrates whether call center customers are actually being helped, not just whether the calls are being answered. As we explain earlier, EDD made a commitment to the Legislature in 2014 to answer 50,000 calls a week; however, simply answering a call does not ensure that a claimant is helped, and EDD must track and prioritize first-call resolution to ensure that it is actually helping those claimants who connect with an agent. EDD stated it would be possible for the new VCC phone system to provide an after-call survey that asks callers whether their concerns were resolved; in fact, it has already implemented a similar survey in its disability insurance branch. However, despite its ability to measure first-call resolution and the importance of using this metric to gauge performance, EDD indicated that it would be difficult to track first-call resolution for UI customers and that it has no clear plans to begin tracking this metric.
By choosing not to measure first-call resolution, EDD is failing to collect critical data on how well its call center is performing, and it is missing opportunities to use these data to adjust operations to ensure that it adequately assists callers. According to EDD, the first-call resolution metric is not a straightforward metric to track because it is unclear whether EDD should measure the agent's or customer's opinion that the call was actually resolved. Additionally, EDD stated that the VCC phone system lacks the built-in tools needed to easily track this metric. However, given that EDD has already implemented similar functionality for another branch, we see no reason why EDD should not measure first-call resolution for callers to the UI branch.
Finally, when EDD quickly implemented the preliminary, minimal version of the new VCC phone system in April 2020, it lost valuable functionality featured in the old phone system for improving efficiency and the claimant call experience. For instance, when it transitioned to the new VCC system, EDD lost the ability to play a series of prerecorded messages advising claimants of their rights and responsibilities after they file their claim. Currently, EDD agents must spend the time needed to manually read this information to claimants. EDD implemented the prerecorded messages for its previous phone system because it estimated that by spending less time on calls with individuals, agents would be able to answer more than 6,500 additional calls each year. Thus, the current need to verbally read advisements to claimants effectively reduces the amount of time EDD agents have available to answer other calls. EDD estimated it would implement this functionality by mid-January 2021.
Another feature EDD abandoned when switching to the new VCC system was the ability for callers to choose to have the next available agent call them back or to schedule a specific time for callback from an EDD agent. This feature benefitted callers by allowing them to engage in other activities instead of waiting on hold until an agent became available. EDD stated it was looking into the possibility of adding this feature, though it did not offer a timeline for when it will determine whether and when to implement this functionality.
To ensure that it is able to take informed steps to provide better customer service through improved call center performance, EDD should implement a formal policy by no later than May 2021 that establishes a process for tracking and periodically analyzing the reasons why UI claimants call for assistance. By no later than October 2021, and every six months thereafter, EDD should analyze these data to improve its call center by doing the following:
- Identifying and resolving weaknesses or problems with the ways in which it provides assistance to UI claimants through self‑service and noncall-center options.
- Developing specialized training modules to quickly train its call‑center staff on the most commonly requested items with which callers want assistance.
To assess the effectiveness of its call center, by May 2021 EDD should implement a policy for tracking and monitoring its rate of first-call resolution. EDD should review first-call resolution data at least monthly to evaluate whether it is providing effective assistance to callers.
To maximize the number of calls that its staff are able to answer, as soon as possible EDD should add the prerecorded message functionality to its new phone system to advise claimants of their rights and responsibilities after they file their claim with an agent.
To provide a more convenient customer experience, as soon as possible EDD should implement those features of its new phone system that allow callers to request a callback from an agent instead of waiting on hold for assistance.
Despite Multiple Warnings, EDD Did Not Prepare for an Economic Downturn
- Before the claim surge, EDD did not adopt a comprehensive plan for how it would respond to economic downturns when its UI program is in higher demand. Having such a plan would have strengthened its poor response to the 2020 claim surge.
- EDD has for years been aware of many of the problems in its UI claims processing and customer assistance efforts that this report identifies. In fact, key problems related to its management of the UI program in 2020 were also present during the Great Recession of 2008 and 2009. Nonetheless, EDD did not take adequate steps to address these deficiencies.
EDD's Failure to Plan for a Recession Hindered Its Response to the Claim Surge
When the claim surge began in March 2020, EDD was far from prepared. As we note in the Introduction, the rise in claims was unprecedented in its size and speed, and we recognize that it is not realistic to expect that EDD would have flawlessly responded to such a challenge. Nonetheless, the key factors that limited how effectively EDD responded to the claim surge resulting from the pandemic—inefficient processes; a lack of readily available, qualified staff; and poor management of its call center—are the same factors that would degrade its ability to respond to a more regular occurrence like an economic recession. According to the National Bureau of Economic Research—an entity that tracks recessions—the United States has entered a recession approximately every five and a half years on average since January 1950. As the department that oversees California's UI program, EDD should be well aware that recessions regularly occur and that its operations may be stressed when handling the resulting increased workload. Consequently, we expected that EDD would have a plan for scaling up its UI program in response to a recession so that it could provide timely assistance to Californians. However, EDD had no such plan ready, and as Figure 9 shows, its failure to prepare left it poorly positioned to respond to the claim surge.
Although EDD has recognized that having a plan for an economic downturn is important, it only very recently took steps to create such a plan. EDD indicated that its UI branch began planning for a recession in 2019—almost a full 10 years after the end of the Great Recession of 2008 and 2009. When we asked EDD to explain its delay, it noted that it has implemented numerous improvements to its business processes since the Great Recession, such as training agents to both file claims and confirm that claimants are eligible for benefits instead of only one or the other. Although we acknowledge these efforts, they are not the same as adopting a comprehensive recession plan.
EDD Was Not Prepared for the Economic Shutdown
Source: Analysis of past audit reports, EDD's recession planning documents, claim data, and call data.
Among other key features, a fully developed recession plan would include the indicators that EDD's management would use to guide its decision making about staffing and process changes, descriptions of the available adjustments to business practices that EDD could make, and policies and procedures to facilitate these adjustments. These features are among those broadly recommended in a recession plan published by the State of Oregon Employment Department in 2017 that was featured at a 2019 conference of the National Association of State Workforce Agencies. Put simply, a recession plan would allow EDD to respond to economic downturns with a predetermined strategy that has considered the full effect on its operations rather than responding in the moment with untested actions.
EDD's delay in developing such a plan cost it valuable preparation time. EDD published an initial draft of a plan in January 2020 that articulates its overall vision for recession preparedness. But when the economic effects of the COVID-19 pandemic began, EDD was only in the beginning stages of developing specific policies, tools, and metrics that its staff would use during a recession to respond to the increased workload. EDD has since suspended its recession planning in order to respond to the claim surge.
EDD has long been aware of the kinds of problems in its operations that have hindered its response to the claim surge. For example, earlier in this report, we detail how UI claims became backlogged and went without payment in part because of EDD's inefficient claims processing practices. This situation closely resembles deficiencies we reported in March 2011.Employment Development Department: Its Unemployment Program Has Struggled to Effectively Serve California's Unemployed in the Face of Significant Workload and Fiscal Challenges, Report 2010-112. In that report on EDD's administration of the UI program, we noted that EDD had failed to meet federal timeliness standards for making payments for several years leading up to the Great Recession and that in 2010, its performance worsened to the point that it was making only 62 percent of its payments on time. We also noted that EDD needed to increase its efficiency to meet acceptable performance levels in the long term. EDD's payment timeliness has improved only marginally since 2011, which suggests that the actions EDD has taken in the interim have been insufficient.
To effectively address the issue of timely payments, EDD needs to continue identifying inefficiencies and streamlining its processes as we recommend earlier in this report. More efficient processes will assist EDD in making timely payments regardless of how high demand for UI benefits becomes. However, to be prepared for times when claims for UI benefits grow beyond its normal capacity, EDD should identify noncritical work or tasks that it can pause or stop until the economy has improved. Identifying these types of tasks would minimize the need for drastic steps like halting eligibility determinations, which, as we note earlier, increased the risk of payment being made to claimants who were ineligible, including those who deliberately filed fraudulent claims.
Similarly, EDD's lengthy staff training program—which has impeded its ability to quickly prepare agents to answer phone calls—is not a new problem. We reported in March 2011 that EDD took an average of three to nine months to train agents, during which time those employees were likely not as productive. EDD repeated this point in the draft of its recession plan. Reflecting on the Great Recession, EDD explained that its inability to quickly hire a qualified and skilled workforce made meeting its workload demands difficult. EDD also stated that several factors, including a lack of preparation and an insufficient number of trained staff, pressured it to lower its hiring expectations and implement a reactive hiring strategy during the Great Recession. Although EDD was aware of problems with its ability to quickly scale up its workforce, it did not take steps to address those problems for almost 10 years. Had EDD begun the recession planning process earlier, it likely would have been able to provide more timely assistance to more Californians during 2020.
In another example, we recommended in our March 2011 report that EDD analyze data to gain a better understanding of why people who call for assistance request to speak to an agent so that it could take steps to reduce the number of calls it receives. In 2017, EDD provided us with examples of various types of analyses that it performed in response to our recommendation. However, EDD has not incorporated the practice of analyzing the reasons for calls into its regular UI operations. Specifically, EDD was unable to provide us with any meaningful analysis related to the reasons why people called for assistance in March and April 2020. Had EDD continued to conduct the type of analysis we first recommended in 2011, it could have developed targeted trainings to use when scaling up its staff, which might have helped it avoid adding and training agents who often could not assist claimants. Such an analysis might also have allowed EDD to quickly identify ways to curb the increase in calls by providing the needed guidance to callers through other means, such as through its FAQ webpage.
To better serve Californians, EDD needs to make recession planning a priority. Given its history of inadequate preparation, it is reasonable for the Legislature to require EDD to develop a recession plan and to keep the plan up to date. Planning effectively for economic downturns includes two key analyses. The first is an assessment of the necessary changes to EDD's operations to make the department more flexible and adaptable to changes in demand. For example, to make sure it can quickly increase its pool of qualified staff, EDD could cross-train non-UI staff and use rotations to UI assignments and refresher training to keep those staff ready to assist in the event of a claim surge. Second, EDD needs to decide on the indicators that it will use to determine when and how to adjust its practices to respond to economic downturns. For example, EDD must monitor economic indicators that suggest it may face increased UI workloads in the near future. Any efforts that EDD can take to prepare are an investment in its own success and that of Californians affected by future economic downturns.
To ensure that EDD is better prepared to provide effective services and assistance to Californians during future economic downturns, the Legislature should amend state law to require EDD to develop a recession plan that takes into account the lessons learned from previous economic downturns, including the pandemic. At a minimum, the Legislature should require EDD's plan to include the following:
- The indicators EDD will monitor and use to project the likely upcoming workload that it will face.
- The steps EDD will take to address increases in its workload, such as cross-training non-UI staff, changing its staffing levels, prioritizing specific tasks, and adjusting the way it performs certain work.
- The altered policies or procedures that EDD will activate if a rise in UI claims becomes significant enough to warrant that step.
The Legislature should require EDD to develop the plan within 12 months of the effective date of the related change to state law. To address new developments in UI processes, programs, or other relevant conditions, the Legislature should require EDD to update its recession plan at least every three years thereafter.
We conducted this performance audit in accordance with generally accepted government auditing standards and under the authority vested in the California State Auditor by Government Code 8543 et seq. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on the audit objectives. We believe that the evidence obtained provides a reasonable basis for our findings and conclusions based on our audit objectives.
ELAINE M. HOWLE, CPA
California State Auditor
January 26, 2021