Testimony of Scott Jensen, Research Improving People's Lives, on Unemployment Insurance after the Pandemic

Summary

Testimony of Scott Jensen, CEO of Research Improving People's Lives, before the House Committee on Oversight and Reform on March 31, 2022 at a hearing titled Genuine Modernization of Unemployment Insurance after the Pandemic. Jensen states that before joining RIPL he led Rhode Island's Department of Labor and Training through the unemployment insurance crisis caused by COVID-19. He states that Rhode Island paid benefits faster without being victimized by fraud any worse than other states. He cites the Brookings Institution as saying Rhode Island was the first to take Pandemic Unemployment Assistance claim applications and tied for the second fastest in paying them.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Cite this document

“Testimony of Scott Jensen, Research Improving People's Lives, on Unemployment Insurance after the Pandemic.” Original at docs.house.gov. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/hhrg-117-go24-wstate-jensens-20220331-fbaa62883e48-a0c97c49a3/, SHA-256 fbaa62883e487afc37fe9b1abeb93ce4da1ee9f8f4cdbafc4df1269bf04b82ce.

Full text

TESTIMONY OF SCOTT JENSEN, CEO OF RESEARCH IMPROVING PEOPLE’S
LIVES BEFORE THE COMMITTEE ON OVERSIGHT AND REFORM, U.S. HOUSE
OF REPRESENTATIVES HEARING: “GENUINE MODERNIZATION OF
UNEMPLOYMENT INSURANCE AFTER THE PANDEMIC” MARCH 31, 2022

My name is Scott Jensen, and I am currently the CEO of Research Improving People’s Lives
(RIPL), which is dedicated to the prudent application of technology and science to solve public
problems. Before coming to RIPL, I led Rhode Island’s Department of Labor and Training,
including through the unemployment insurance (UI) crisis precipitated by COVID-19. This
testimony concerns lessons from that difficult stretch.
While none of my team were satisfied with the service that UI customers received during the
pandemic, the approach we took meant that Rhode Island paid benefits faster without being
victimized by fraud any worse than our counterparts in other states. According to the Brookings
Institution, Rhode Island was the first to take Pandemic Unemployment Assistance (PUA) claim
applications and tied for the second fastest in paying them 1 (Texas paid fastest). Further, an
analysis by a Census Bureau official of how UI programs were handling the rush of claims early
in the pandemic put Rhode Island far out-front.2 In short, scrappy Rhode Island played the
terrible hand dealt to UI programs everywhere as well as it could be played.
It is important to assert the efficacy of Rhode Island’s performance, without denying short
comings, because, with benefit of hindsight, I am convinced that we made it through the COVID
odyssey because we squarely faced the central challenge that lies at the heart of “UI
modernization” and made genuine progress against it.
Before discussing this central modernization challenge directly, allow me a brief digression
about UI fraud and paying benefits. It is tempting to see a stark choice in UI administration, a
fundamental “either/or” – either pay claims quickly or protect against fraud. Such an
understanding is categorically mistaken. By law, and because it is UI’s duty to stabilize our
nation’s labor market, UI is, and must be, a “both/and”. We must both expeditiously pay benefits
and protect UI trust funds from the criminals who want to steal from them.
During COVID, Rhode Island did not have the option of straying from the “both/and” path.
Because it paid benefits quickly, fraud surfaced in Rhode Island quickly as well. On the ground,
for us this meant that early in the pandemic the state police, the FBI, and the U.S. Attorney’s
Office were proactive and insistent that the department combat fraud aggressively. However,
only three months into the crisis, the Rhode Island ACLU sued the department, complaining in
federal court that our anti-fraud measures violated the due process protections afforded UI

1
Ryan Nuss, Jana Parsons and Jay Shambough. “Incomes have Crashed. How much has Unemployment Insurance
Helped?” The Hamilton Project; May 13, 2020.
https://www.hamiltonproject.org/blog/incomes_have_crashed_how_much_has_unemployment_insurance_helped
2
Kevin Rinz. “Understanding Unemployment Insurance Claims and Other Labor Market Data During the COVID-
19 Pandemic”; May 25, 2020; p. 21. https://kevinrinz.github.io/covid19_labordata.pdf?bcs-agent-
scanner=baa9c163-c9e3-db41-bc1d-50b22e927ff8

1
applicants.3 If we were going to make it through the pandemic, we had to steer a middle course
between the Scylla of timely payment, and the Charybdis of fraud.
Because Rhode Island was forced to both combat fraud and pay efficiently, we gained what I
believe is an important insight into what I referred to above as the central challenge of UI
modernization: that UI needs to use the data it generates to reimagine its fundamental approach
to program administration, and it must do this reimagining in light of the possibilities afforded by
contemporary technology.
Please allow me to elaborate upon this compact observation.
First, I want to be clear about what I do NOT mean. The old technology used by most state UI
programs did not melt down wholesale during the pandemic. If, when thinking about what went
wrong in UI programs during COVID, the image of a sputtering old truck, on a very hot day,
climbing a steep hill with a heavy load comes to your mind, then you have conjured the wrong
picture. Further, these old systems were not “hacked” by fraudsters who exploited 1980s cyber
technologies. This, or anything like it, did not happen in Rhode Island.
Were the failure merely due to old tech, then the fix would be simply new technology with more
horsepower, or silver-bullet identity verification systems. Contemporary technology is a
necessary, but not sufficient, condition for addressing UI’s “both/and” mission. The proof of this
assertion lies in the fact that no state UI program performed well during COVID, and this in spite
of the fact that some states already had new systems and cutting-edge “gate keeper” software.
If old tech is not, what is the heart of the UI modernization problem?
For some context, we have all been approaching the administration of UI in the same way since
the program was created in the 1930s. Roughly, UI programs proceed according to the following
three steps:
1) Information about the person who has lost their job is collected (e.g., who they are, why
they were separated and how much they made, which allows the amount of a possible
benefit to be determined).
2) A UI staff person assesses the information presented, may gather some more information,
and then makes a non-monetary judgment about whether benefits are warranted; and
then, if payment is warranted, also makes a monetary judgment about how much the
benefit will be (these judgments are subject to elaborate appeals).
3) If warranted, a benefit payment is issued.
During the pandemic in Rhode Island, the most precious asset we had was the time of expert UI
staff in the call center, because these professionals drive all the work in step (2) described above.
Under normal circumstances, UI staff have time to both take calls in the call center and do the
“back-office work” of making determinations about claims. In this structure, common to all the
UI programs with which I am familiar, the same people who take general calls to help citizens

3
RIACLU press release, May 27, 2020. https://riaclu.org/en/news/aclu-sues-over-frozen-unemployment-insurance-
payments

2
are also the ones who determine what claims will be paid benefits. This means we are counting
on staff simultaneously to execute the roles of, as the proverbial saying goes, “chief cook
(determine claims) and bottle washer (provide general assistance to the public).” Such a strategy
is rarely a wise idea.
The weakness inherent in UI’s fundamental approach becomes an acute problem during
pandemics and recessions because it does not scale. When claim loads double or triple, UI
programs must double or triple the staff in the call center and also train them, which takes a long
time. And worse, more unemployed people means more calls to the call center; and, because
people do not just call once, it means an unmanageably steep increase in call volume with which
no operation can keep up. For instance, on a particularly hectic day during the height of the
pandemic only 10% of the people calling Rhode Island’s UI call center called only one time
during the course of that day.4 90% called numerous times, some even armed with robo-dialers.
Our COVID UI strategy in Rhode Island was straightforward: do everything we could to free up
trained UI staff to make the determinations that only they could make. We worked on hundreds
of tasks to this end, but I will broadly characterize how we approached the main tasks of
preventing fraud and paying claims expeditiously to illustrate why freeing, and then using, data
must be at the center of reimagining UI administration.
Fraud was all about the “Who are they?” question. In Rhode Island we knew that the ACLU’s
lawsuit would not allow us to erect burdensome electronic solutions to identity verification,
because we could not respect due process rights if we stopped too many people. But there is a
practical dimension to why erecting impossible barriers is a bad idea as well, which I have
touched on. Desperate people call the call center repeatedly. Justifiably, they clog the lines and
tie up UI staff. This reality caused us to turn toward the cyber version of “probable cause”. When
a claimant interacts with the UI system, she or he leaves a rich set of electronic “fingerprints” in
the “metadata” – information including what IP address a computer uses, how long a claimant
lingers on our website, how many clicks an applicant makes, and so on. Naively, our plan was to
distinguish wholesome from suspicious actors, and then subject the suspicious to more intense
identity verification. Our theory was to winnow down who might be an identity fraudster so that
we could devote the resources necessary to find the bad guys while respecting the due process
rights of everyone.
Expeditious payment was about where we could find efficiency in step (2) enumerated above.
Our plan, which we will see also turned out to be naïve, was to distinguish “clean claims”, the
positive determination of which was easy, from “complex claims” that required scrutiny by an
expert. If we could establish this distinction, we could pay the former in days and set up
assembly lines of talented UI agents to do the real thinking on the more nuanced issues. (Please
note we chose not to deny claims without expert, human review.)
The reason I described our approach to both these challenges as naïve, and what kept Rhode
Island’s arguably-best-in-the-nation pandemic performance from being good by any objective

4
This was September 14, 2020, and there were 54,189 calls made to Rhode Island’s UI call center that day.

3
standard, is the main problem faced by UI modernization. Our AS400 main frame would only
report out the data we needed distinguish clean from complex claims if our COBOL
programmers specifically wrote code from scratch to dig that data out of the system, which they
did not have time to do. Similarly, older web interfaces and analog call-center platforms are not
designed to make the kind of metadata our fraud strategy needed readily available. In other
words, the pandemic made clear to me that unemployment insurance was not being managed
with data – evaluated by data maybe, but not managed. This is like driving a car without a
speedometer or a fuel gauge. It can be done, but you are always going either too fast or too slow
and you keep running out of gas.
It goes without saying that we need to replace old UI technology if we want better outcomes. But
the larger lesson is that replacing technology is the beginning, not the end, of the work we have
ahead to avoid the next UI meltdown. We must also use the data that contemporary technology
makes available to get into a virtuous cycle of continuous improvement for how, fundamentally,
we approach the administration of unemployment insurance. Specifically, we need to constantly
remain one step ahead of fraudsters by keeping them off the virtual streets of our UI systems
through the use of metadata to expose who they are. For benefit payment, the goal is to make
getting to yes on obvious claims easy and keeping people informed about the status of their claim
applications, so that trained UI staff can prioritize the complicated work that only they can
perform.
All of us are becoming accustomed to better and better service delivered by very smart
applications of contemporary technology. I am always surprised that buying on Amazon gets
ever easier, and I have ordered from Domino’s just to watch the pie approach my front door on
their “Pizza Tracker”. The public sector can do such things too. But there is not an easy, off-the-
shelf fix that is going to do this hard government work for us no matter how much money we
spend. We have to change the way we approach, in UI’s case, benefit delivery by finding prudent
opportunities to deploy technological innovations like cloud computing and AI/ML to improve
service, and we need to free our administrative data (and use it) to discover where the
opportunities are to alter our current approach. And once we make changes, we need to keep
using data to find more opportunities to make further improvements … and so on. This is the
virtuous cycle of continuous improvement. The private sector is good at this, and government
needs to do the same thing but in accord with its mission.
I want Congress to know that even after all the UI mainframes are gone, it is going to take hard
work in fits and starts to realize genuine improvement. Anyone who says there is an easy, purely
technological fix to the problems we lived through in UI during the pandemic did not get close
enough to the challenge at hand to make a good assessment of it. That is the bad news. The good
news is that the manufacturing, retail, and even pizza delivery sectors have made remarkable
strides. Surely government can too. In fact, I know we can, because we saw it with our own eyes
in Rhode Island during the COVID 19 crisis, where necessity truly was the mother of invention.

File and source

File: hhrg-117-go24-wstate-jensens-20220331-fbaa62883e48.pdf
Size: 545,376 bytes
SHA-256: fbaa62883e487afc37fe9b1abeb93ce4da1ee9f8f4cdbafc4df1269bf04b82ce
Our copy: hhrg-117-go24-wstate-jensens-20220331-fbaa62883e48.pdf
Original: docs.house.gov