When I took the introductory fine arts course in college (which actually was a tough course), one of the ideas that stuck with me was devolution. The instructors used that word in a very particular way, to apply to the changes that took place in late Roman art. Drill technology improved to the point where drills could be used to help create sculpture: they were faster and cheaper to use than chisels and hammers. But the results were cruder. It was easy to identify a late Roman bust: the curls and eyes were much coarser than in earlier Roman or Greek work.
Now it’s probable that drills, by making it cheaper to produce sculpture, allowed for more sculpture to be made. But I never heard that mentioned in that long-ago course, so I’m not sure. And I didn’t hear of a two tier market, with the more labor intensive, more finely crafted work being produced for a more discerning clientele, and the drill work being more of a mass product. The impression I had was that the new drill technology became a new normal, replacing the older methods.
It’s become fashionable to discuss the creeping decay in advanced economies, particularly the US, both in term of third worldification and end of empire. The more apocalyptic turn to theories of collapse from writers like Jared Diamond and Jacques Tainter. But I think they miss one aspect that may prove to be important, that of how the pursuit of efficiency doesn’t always produce net gains, as economic theory might tell us. The measure of productivity, more stuff per unit input, misses how service/product quality can deteriorate. Some of this is deliberate: I have readers in comments regularly lament how old durable goods and tools were more reliable and lasted longer than contemporary versions. But there are other aspects of the downside of the willy-nilly pursuit of efficiency that have become so routine we accept these indignities and often don’t recognize them (unlike other ones that remain annoying years after the change, such as the widespread implementation of call routing and prompts in place of humans answering phones).
Let’s look at banking. The public has become desensitized to the fact that banks do something that is still rather amazing: they handle a ginormous volume of transactions, and they give you a report, every month, of what happened to you. And they were able to do this in the bad old days, when computers were mainframes, a lot of banking work was done on adding machines (when I started on Wall Street in the 1980s, I did spreadsheets on green ledger paper with a calculator, copying information from hard copies of SEC filings), and you got your paper checks back with your bank statement. You really need to sit back and think on what a computational and logistical challenge it was. And this worked, transaction processing was highly reliable.
The first wave of technology implementation was led by people who were sufficiently concerned about maintaining old service standards that my impression is that the service change was a net gain. Citibank’s computerization of its check processing was a singular success and set John Reed on the path to becoming Citi’s CEO. But in the last year, I’ve experienced and/or heard of things that would have been seen as serious failings of transaction processing, and the number is so large that I no longer think this is just random, I think it’s a sign of a bad nexus of more complex technology installations (remember, new tech in banking for the most part is bolted onto existing infrastructure) combined with ongoing cost cutting and underinvestment (which is a long-lived managerial fad; remember this is taking place in an industry which loves to cavil about profit problems but whose survival risks come from risky wagers, not retail banking margins). And we will put aside the horrorshow of mortgage servicing, we’ll stick with bread and butter processing.
I used to do a lot of international wire transfers when I lived in Australia in the early 2000s. I never never had a problem. By contrast, I’ve had a 50% fail rate with them in the last two years, despite having no errors in the information provided. And in one of the fails this year was to someone who’d been sent a wire successfully from the same account at my bank to the same recipient at another bank. And to add insult to injury, despite being required to provide all sorts of info about each of us, including phone numbers, no one tried calling either of us even though the wire was hung up for five days. And I was charged by the recipient bank for their inability to find their own customer.
I also a check to a cat sitter mistakenly get in the envelope to my health insurer. Not only did the health insurer deposit the check, my bank honored it. By contrast, the bank I used in the 1990s would call me at least once every six weeks about checks presented to them for payment. My signature wouldn’t look enough like the one they had on file and they wanted to make sure I really had signed it.
Oh, and I had some ambiguous branch charges on my business account. I had a guess as to what they were but I need real explanation for my taxes, not guesses. It was like pulling teeth to get documentation from the bank. I find this particularly alarming given another innovation: that you can now deposit a check using your smart phone. Scan the front, endorse and scan the back, and voila! All done. I had American Express as a client in the 1990s and 2000s, and Amex spend a small fortune keeping abreast of forgery technology, frequently updating the sort of paper and printing it used to beat counterfeiters. Scanning an image rather than even taking an actual check so a somewhat trained human can at least give it a once over? This seems to be an invitation to fraud, and the banks finding a way to cover those losses with other customer fees.
When I mentioned my wire gripe to a friend, a much worse story came to light. A friend of theirs needed to overpay a Citibank business card at year end (I am mentioning the name of the guilty banks because the unhappy customer wants the perps embarrassed; I have also double checked the details of this account) because some payments he had authorized weren’t going to post (shaggy dog story details will be omitted), and he wanted them to take place fro tax purposes in 2012. Overpaying would solve that problem (as in what mattered was the money getting into Citi’s hands by year end). But Citi won’t let you do that on line; the only way is by sending a check or wie transfer.
He sent a check overnight to the address on the website for overnight payment after verifying with a rep that that was the address to use (they also showed a PO Box, and he was going to use overnight mail, but was told to use a different address instead). He is doing end of year transactions on his bank account and sees the check has not cleared, which does not surprise him (the address was out of state) but figured he’d look at his account at Citi. No indication that the check got there. Not sure if this was unusual, he decides to call to see when the payment will appear. The rep tells him it should have been posted, and takes his routing and account number to see if any such payment is pending. No dice.
The now concerned businessman asks how this could have happened. The rep asks where he sent the check. He reads the overnight payment address from the Citicard website. The rep says they don’t accept payments at that address. The customer asks what will happen to his check (if Citi will eventually process it, all will be OK because he can prove they had it by year end). He says they will either process it or return it.
The fact that the rep has no idea what will happen is a big problem. The poor sap isn’t sure about sending another check to the right overnight address (there was still time to do that) so he decides to go to his bank, TD Bank, which is open late, to send a wire and stop the check that is floating around.
TD asks for wire instructions, which he has. But TD wants a recipient bank branch address, and Citi (on the phone with super harried customer) says this is unnecessary, they’ve never heard of such a thing. TD Bank rep says the wire can’t be processed without an address. The Citi rep gives an address and makes clear this address is pretty arbitrary. The poor customer is worried as to whether the wire will be credited properly. He can see another mess coming.
Next day, customer gets an e-mail from his accountant, and the implication is he could still send a check, he has one last chance today. Customer does not like sending a check but is even more worried about the wire, and so goes back to the branch to see if the wire can be stopped (it could not go out until Monday and this is Saturday).
Much consternation ensues. The people in the branch look in all the electronic registers where the wire should be. Not in the wire log. Not in the wire queue for pending branch approval. Not in the queue for pending regional approval. A wire initiated after Friday evening should be in the wire log and locatable in the branch system. They call the back office, and the back office looks in places the branch can’t access and see no evidence of the wire either.
The branch staff tells the customer he is perversely lucky, it looks like somehow that the wire he didn’t want to go isn’t going anyhow. But just in case it somehow got to the wire room, which they think is impossible, they send a stern message to the wire room not to send the wire and also make sure the people who open up the branch on Monday will call the wire room to make double sure the wire isn’t sent somehow.
The poor guy goes to Fedex to send a check overnight.
Monday morning, he gets a call from the branch. The wire went through. Oh, and by now, he can see the first check he sent was processed by Citi and was posted to the account.
So get this: the poor sap had tons of stress, wasted tons of time, paid two stopped check fees to his bank, two overnight fees, and one wire fee. Citi at least agreed to waive its bounced check fee for the check sent to the incorrect address on its website. And this reveals a moderate competence fail on the part of Cit and a major one by TD Bank, which gets top customer service ratings from JD Powers.
It may seem as if I’m making a big deal over a series of transactions, but you need to understand the importance of payment systems to get the significance. Bank employees should understand how their systems work, since systems mow are much of what banking is about. And banks have long been designed to give high priority to achieving accuracy in payments processing. The kinds of failings I’ve seen in bank payment processing is like seeing the power dim in an electrical system when it’s not at peak load. These are signs that corners are being cut, and not just in one place.
And, to come full circle, this trend to devolution may be a driver of collapse. At this point, all I have is anecdote and intuition. But one of the things I found frustrating about Tainter is that he posited that increasing social complexity led to increasing energy demands, and those led to collapse. He also pointed out that sometime the elites in a society succeeded in pulling it out of a dive path, and other times they didn’t. Yet he refused to give social factors any explanatory power in his theory (like, say, rising levels of corruption among the ruling classes).
The reason I’m not sure increasing energy demands are an adequate explanation is that, before modern times, population level were pretty static, as were standards of living. So you would not see the two main drivers of our resource demands as having anywhere near as significant an impact.
But any complex system has a tradeoff between efficiency and robustness. This is something we discussed at length in ECONNED, how economic theory bizarrely assumes stability by assuming that economies have a propensity to equilibrium. That in turn gives them a bias in policy prescriptions to favor more efficiency and not even think about stability. Yes as Richard Bookstaber stressed in his book Demon of Our Own Design, highly efficient systems are prone to catastrophic failures, since disruptive processes propagate through the system too quickly for anyone to stop them. Taleb also argue that robust systems, like biological systems, are inefficient by virtue of having extra capacity (two kidneys, for instance).
Since the early 1990s, some seemingly minor changes in prevailing idea of corporate governance have had disproportionate bad effects. Share-price-linked CEO pay has led to underinvestment and short-termism, which produces more short term profits at the expense of stability and reliability. And if you do that across an entire economy, it isn’t hard to see that this over-optimization has led to much more fragility. We now have extended supply chains and the media has identified potential catastrophic points of failure, such as China having a stranglehold on rare earths. And even absent disaster scenarios, we also have an ongoing, hidden tax of things not working as promised but it being too costly in terms of time and money to get that rectified (I’ll spare you examples outside banking and health insurance, but overpromising and underperforming appears to be another new normal).
While extreme outcomes make for better stories, it may be that the US (and perhaps other advanced economies) go through a period of whimpers rather than bangs, of a grinding fall in competence that consumers accept because it’s just too hard to demand better. I don’t relish the prospect but I’m also at a loss to suggest remedies.