Friday, August 7, 2020

COBOL AND COVID

My pitching philosophy is simple: keep the ball away from the bat. ~ Satchel Paige 

Infrastructure. You’ve heard this word so many times it no longer perks up your or anyone’s attention. When I think of infrastructure, I imagine roadways, airports and bridges. Merriam-Webster’s rather non-descript definition of infrastructure is: “the system of public works of a country, state or region.”

Infrastructure isn’t exciting or stimulating. That’s basically why there are far more politicians who love to promise they’re going to amp up infrastructure spending than actually put dollars to work improving it. Infrastructure’s considerable value stems from enabling many appreciated economic and social activities to happen. Like Little Red Riding Hood driving her EV on her town’s freshly-paved streets to visit gramma’s house (and avoid the big bad wolf), or using her town’s public water system to clean her fine, red hoodie in her water-efficient washer.

Last fall, the Congressional Budget Office estimated that combined federal, state and local spending in 2017 on infrastructure was $441 billion, roughly 2.3% of GDP. This estimate represents the lowest level in more than 60 years. Infrastructure spending peaked at about 3% in the late 1950s, when the Interstate highway system was being constructed.

The Interstate System is perhaps the best-known Federal infrastructure program. Its 47,000 miles cost about $129 billion, making it the nation’s largest-ever public works program. Every state has an interstate highway. The first segment of an Interstate highway was begun in 1956 in the Show-Me state of Missouri. The state with the most Interstate miles is Texas (3,501.2mi.); the one with least is Delaware (40.6mi.).[1]  

Roadways are an example of traditional infrastructure that’s easily seen and used by the public. The Coronavirus pandemic has exposed a series of failings in our public infrastructure that is more hidden but nonetheless has created calamities for lots of us. I’ll talk here about two types of fading public works.

First, Covid testing. The CDC’s initial requirement of only using state and local government testing laboratories to diagnose Covid-19 in individuals produced tragic consequences. Despite #45’s completely spurious claims, neither the CDC or these public labs have accomplished their clearly-understood objective of providing sufficient, reliable and timely testing.

Six months into the coronavirus’ incessant attack, the US has yet to adequately deliver enough dependable testing in a timely fashion. In August the US ranks 12th of the 91 nations listed in terms of total per capita coronavirus testing – 173.8 tests/1000 people. Luxembourg is first with a testing rate that’s four (4)-times higher than the US. The media regularly reports people having to wait so long for their results that the findings are medically meaningless. Like all too many others, one San Francisco person finally received his test results 16 days after taking it.

Another veiled, unseen form of vital public infrastructure that’s stumbled in its efforts to surmount substantial viral challenges is the government’s digital infrastructure, its computer systems and personnel. These computer systems enable the panoply of federal/state/local government program payments to be received by intended recipients – like Social Security, SNAP/food stamps, Medicaid/Medicare, tax refunds and, of recent note, unemployment insurance benefits (UI). The pandemic has strained federal, state and local governments’ assistance programs beyond anything imaginable over the last 80 years.

Our public digital infrastructure has floundered because many public agencies haven’t prioritized or been allowed to update their systems for decades. They’re typically legacy mainframe systems, with a very capital “L”.

As of July 25, over 54.1 million American workers, representing more than one in three workers in our labor force, have filed for UI. That’s more than 37 times higher than normally expected, pre-virus. The national unemployment rate peaked in April at 14.7%, the highest since the Great Depression. Our real (inflation-adjusted) GDP fell 9.5% during the last quarter. Over 100,000 small businesses have likely closed for good. Well-known retail businesses have also declared bankruptcy like Lord & Taylor, Sur La Table, Brooks Brothers, Hertz, JCPenney and Chuck E. Cheese.

In response to this medical, economic, social and wholly-human catastrophe the Congress passed its first Covid pandemic support legislation on March 27, the $2.2 trillion CARES Act. Among countries, the US is not known for its generosity to the unemployed. However, the CARES Act provided significant fiscal support for workers, firms and many others harmed by the coronavirus, including a $600 per week supplement to the unemployed. This multi-faceted support totaled 13.2% of the nation’s GDP, placing the US program third largest in the OECD, a group of advanced nations, behind Japan and Canada. In this age of “do-nothing” government, the CARES Act embodies an impressive, timely accomplishment.

But nothing lasts forever including this Act, by design. Alas, the Dems and Repubs have wasted time pointing fingers of blame rather than agreeing what the scope and form of a successor Covid support program should be.

The House Dems easily passed their $3 trillion follow-on package, called the HEROS Act, on May 15. Meanwhile the Senate Repubs are tardily drafting their follow-on package, entitled the HEALS Act, that could provide $1 trillion of benefits. [Don’t you just love how every piece of legislation needs an acronymically-suitable title.] The winner of the HEROS v. HEALS political prizefight has yet to be determined.

No matter how the new legislation finally gets squeezed through the Congress’ political meatgrinder, it will likely be very challenging to implement by state agencies that administer unemployment benefits. Why? Because the vast majority of state unemployment insurance agencies use ancient computer systems.

The age of state benefits agencies’ computer systems ranged between 22 and 42 years old, according to a survey that was performed over 10 years ago. These mostly mainframe hardware systems are digital dinosaurs that use obsolete programming languages like COBOL. The first COBOL program ran in August 1960, that’s 60 years ago folks, on an RCA 501 mainframe computer.[2] Because of their nearly-geologic age, state unemployment insurance agencies’ computers are costly to run, fragile, inflexible and error-prone. Their operation requires continued attention by knowledgeable personnel. It’s no surprise that very few computer jockeys know (or want to know) anything about such archaic systems – most have retired long ago. Hiring knowledgeable staff for such computer systems is a nightmare. Doc Brown’s DeLorean might be a useful staffing tool.

During the 2007-09 recession, when these agencies were last obligated to produce rapidly rising numbers of benefits payments, their legacy computer systems regularly failed. No real system changes were needed, only larger than usual volumes. Many shut down for days after the systems attempted to handle the elevated claim levels.

This Spring, the CARES Act implementation required substantial amounts of angst-inducing time and effort by state UI systems’ operators to redesign their systems to add the uniform $600/week federally-funded supplementary payment to qualified recipients. The federal Treasury Department had similar, time-consuming technical difficulties in providing millions of citizens with their $1,200 payment checks. Many of these challenges centered on the agencies’ obdurate computer systems.

Notwithstanding their possible merit, if the provisions of the Senate’s proposed HEALS Act go into effect, such challenges would multiply. States would need to provide supplementary UI money that would equal 70% of an employee’s lost weekly wages (but not exceed $500/week), when added to existing state benefits. This represents a major change from the CARES Act $600/week supplementary payment.

As such, this HEALS Act stipulation would give many states’ UI systems yet another coronary. To make this seemingly straightforward 70% calculation would require rapid introduction of new revisions to legacy software together with likely linkages to additional databases. What could go wrong when changing lumbering, inflexible, archaic hardware/software systems? How long would it take? Don’t ask and don’t hold your breath.

Perhaps these recognized obstacles are the reason the Senate’s HEALS Act surprisingly provides $2 billion to help states “upgrade” their ancient UI systems. Surprising because the vast majority of Repubs are loath to provide any funding that assists state and local governments.

I would be amazed if the final House-Senate compromise legislation includes any payments that depend on such semi-sophisticated computations. The regrettable state of public digital infrastructure will limit Congressional negotiators to arguing about levels of dollars per week, not percentages of lost wages. It’s another case of public law and action being effectively constrained by woebegone infrastructure.

In this confrontation between COBOL and Covid, keeping it simple – following Occam’s razor – is the only expedient maxim. 

August 11, 2020 postscript. Another COBOL & Covid interaction. The CA state health director abruptly resigned on August 10 amid CA’s rising case/death rate and a big snafu with Covid-related health data. Why the problems? From the story: The governor on Monday vowed to quickly overhaul what he described as the state’s outdated information technology systems, which he blamed for not just the testing data snafu, but also for a staggering backlog of unemployment claims .

 



[1] Until the District of Columbia convinces Congress it should become a state, it’s just a district with 12.3mi. of Interstate. You shouldn’t hold your breath for DC’s latest statehood effort to be successful.

[2] At that time, my father served as the senior RCA executive in charge of its computer systems business. 



No comments:

Post a Comment