August 05, 2021
August 05, 2021
Co-Chairs Hackett and Fraizer and members of the Unemployment Compensation Modernization and Improvement Council, thank you for the opportunity to testify at today’s hearing. I am Zach Schiller, research director of Policy Matters Ohio, a nonprofit, nonpartisan research institute with a mission of creating a more vibrant, equitable, inclusive and sustainable Ohio. As you know, Policy Matters has previously made recommendations to the council, in testimony last February and in a memo to members last month.
Ohioans have benefited from the regular meetings and hearings that the council has held. You have created a window into the unemployment compensation (UC) system that we hope you will keep open. The process of modernizing and improving the UC system will not end with your report, and we are heartened that the council will continue to meet in future.
That said, you did not make the full recommendations in your draft report available in time for the public to provide an informed reaction. This means that public input – including that of UC claimants – is inadequate. This is unfortunate, given the time and resources that have been put into this effort. Even if it is not reflected in your official report, we recommend that you seek such public input and make changes accordingly. Our testimony today responds to your July 29 draft report and we hope you will follow up on these issues whether or not the report does so. We add a few observations based on copies of the executive summary and full report made available late yesterday. We see a number of positive recommendations but have not been able to do a comprehensive review of the report.
The agency has significantly reduced attempted fraud. The council should include in its recommendations and future communications to the General Assembly that care must be taken to ensure that anti-fraud efforts do not lock out legitimate claimants. According to the data provided by the Ohio Department of Job & Family Services (ODJFS) last week, there remained as of July 26 some 501,133 Pandemic Unemployment Assistance (PUA) claims that were flagged for fraud and still pending, as well as 53,252 fraud-flagged traditional unemployment compensation claims. These include hundreds of thousands of PUA claims that have been pending since last year, and thousands of traditional UC claims pending since early this year. Based on the share of claims flagged for fraud that have been approved – 15.1% of traditional UC claims and 43.6% of PUA claims – it seems clear that many thousands of claims have been held up for many months that are likely to receive ultimate approval. Even if the proportion of these claims eventually approved is considerably lower than the shares to date, this means thousands of Ohioans are without benefits who should be getting them. As the council has heard from unemployed Ohioans who have been unable to access their benefits, this is resulting in incalculable human suffering.
The council needs to address this issue head-on. This includes a review of who has been flagged incorrectly and why, as well as whether this impacts Ohioans of different ages, races or other demographic groups differently (as we have recommended previously, demographic data should also be reported on benefit denials). It means a review of the assumptions used to create these flags and the claimant personas described by Jeff Ficke of the public-private partnership at least week’s meeting of the council. Making the claims experience “seamless and easy” for Kevin, the 28-year-old young professional with a young family who was recently laid off, as described in the presentation, may be fine. But are steps being taken to be sure that the creation of such personas is even-handed and treats all applicants fairly?
The issue of bias in artificial intelligence systems is getting increasing attention; businesses are raising start-up money to offer tools monitoring AI services and identifying potentially biased behavior. In June, the Software Alliance, a trade group representing some of the biggest software companies, offered a detailed framework for fighting bias in AI, “including the recognition that some automated technologies require regular oversight from humans,” according to the New York Times. In its framework, the association says: “While AI can be a force for good, there is a growing recognition that it can also perpetuate (or even exacerbate) existing social biases in ways that may systematically disadvantage members of historically marginalized communities. As AI is integrated into business processes that can have enormous impacts on people’s lives, there is a critical need to ensure that organizations are designing and deploying these systems in ways that account for the potential risks of unintended bias.”
These principles should be used as a starting point for the use of AI. In addition, it should be clear upfront that the systems used in the private sector financial industry to prevent fraud may not always be appropriate, or may need significant adjustment, for a public benefit system. This is especially so if the agency pursues more self-servicing and leveraging of two-step verification, as the council recommends. Critically, unemployed Ohioans and claimant representatives must be involved in testing and designing the new information technology system Sagitec is building for UC benefits to ensure that it is easy to use, accessible and does not screen applicants in a discriminatory fashion.
Therefore, we applaud the recommendation to “involve claimants in the development of future communications, system changes, etc., to ensure issues individuals may face are learned prior (to) changes going public.” We are glad to see as well that the council wants to “maintain while improving traditional forms of communication for individuals who may not be able to use more self-servicing functionality.” And we support your recommendations to ensure all Ohioans can understand the documents, while improving access to alternative forms of communications. These are all positive steps. We hope to see real implementation of them—and that the council will monitor that.
The draft report made available at your July 29 meeting omitted some key issues. These included:
n A significant discussion of nonfraud overpayments and the agency’s implementation of a waiver policy. As noted previously, Ohio needs to adopt an “equity and good conscience” rule so that like in most other states, recipients of nonfraud overpayments who experience financial hardship do not have to repay them. It also remains a question why only 11% of the 688,000 claimants sent notices about the chance for waivers have responded. For ODJFS to send out a contradictory notice telling claimants to ignore language in the same notice that they only have 48 hours to respond is hard to imagine. Further outreach efforts are necessary.
n What is being done to speed up the appeals process, and will the agency meet the judge’s ruling that PUA appeals be addressed within 21 days? Director Damschroder said on July 29 that ODJFS now was training staff and moving staff to appeals. However, data in the agency’s July 29 report show the number of state staff devoted to adjudications and appeals grouped together has declined in the last few months and now is well below where it was at the end of May.
n When will federal timeliness standards for paying claims be met?
n Deloitte, the vendor that set up the PUA system, has not appeared before the committee. Mr. Ficke has provided modest information on the review of vendors. However, there has been nothing approaching a full-scale, public look at vendor performance. This is essential in holding contractors accountable and determining their role going forward; engaging with private industry, as the council is recommending, must depend first on a full review.
Finally, as we have previously recommended, the council should propose General Revenue Fund spending to support a minimum level of operations so that that the agency can provide the access to benefits it should and is not badly short-handed when the next recession comes. This is necessary if the agency is to have “a prepared and adequate workforce that is able and ready to assist Ohioans,” as the council recommends, and “to set departmental standards for call and claim support to ensure positive claim experience.”
We have not had time to do a full assessment of the report and recommendations. However, there are clearly several useful recommendations – and some significant gaps. We hope the council will follow up on both, to monitor implementation and to take up those issues that have not been fully addressed. Thank you very much for the opportunity to testify. I would be happy to answer any questions.
 Schiller, Zach, “Testimony before the Unemployment Compensation Modernization and Improvement Council,” Policy Matters Ohio, Feb. 25, 2021, at https://www.policymattersohio.org/research-policy/shared-prosperity-thriving-ohioans/basic-needs-unemployment-insurance/unemployment-insurance/testimony-before-the-unemployment-compensation-modernization-improvement-council
 Schiller, Zach, Memo to Rep. Fraizer and Sen. Hackett on the draft UC Modernization and Improvement Council report, July 5, 2021
 Metz, Cade, “Using A.I. to Find Bias in A.I,” The New York Times,” June 30, 2021, updated July 1, 2021, at https://www.nytimes.com/2021/06/30/technology/artificial-intelligence-bias.html
 “Confronting Bias: BSA”s Framework to Build Trust in AI,” https://www.bsa.org/reports/confronting-bias-bsas-framework-to-build-trust-in-ai
1 of 22