What happened at SVB

Here’s my very simplified version of what happened to bring down SVB: and specifically, why no-one seemed to have seen it coming. Informed very much by Matt Levine’s excellent writing on the topic. Thoughts are purely my own, not representing any organisation.

At core, we need to look at a simplified model of what banks like SVB do, and especially, what then happens when interest rates change. Essentially, banks:

  1. Take in deposits, from individuals and businesses. Deposits are generally low-interest, and as interest rates rise, only a little of that is passed on to deposit accounts.
  2. Put all that money somewhere! Broadly, there are two options:
    • A. Loan-like instruments (e.g., home loans, business loans). These are often floating-rate, i.e., their interest rates follow market rates, but they are also very “illiquid” (hard to sell or otherwise turn into cash). If a bank makes a home loan for a specific house, it can’t easily get that money back immediately.
    • B. Bond-like instruments, like corporate debt. These are typically fixed interest rate, but they are liquid (easy to sell).

Now, what happens when interest rates go up? Deposits and bonds rates don’t really change much, but loan interest rates rise. This is an “endowment effect” that leads banks, all else being stable, to make more money when interests rise: their “Net Interest Income” (NII) rises as interest rates rise.

Great! Next question, what happens if, for some reason, a lot of depositors want their money back at once? The bank would eventually run out of cash reserves, and need to sell some bonds (as the loans are hard to sell). But here’s a problem: bonds have a fixed interest rate, but their market value decreases when interest rates rise, because new investors would rather buy new bonds offering a higher rate, than your old low-rate bonds. When a bank holds a bond to maturity, that’s not a problem — they get back the full face value of the bond. But, if a bond needs to be sold early, and the interest rates have risen, the seller will take a loss. At worst case, a bank being forced to sell lots of bonds could make a huge loss, which overwhelms its capital reserves, and leaves it insolvent.

Normally, this is irrelevant, as this only happens if a bank has to sell bonds early, i.e., has a massive outflow of deposits, a bank run. There are many mechanisms to prevent this:

  • deep relationships between the bank and it’s customers;
  • a wide variety of depositors, many of whom don’t really follow the finer points of financial news and so are fairly “sticky”;
  • deposit insurance;
  • capital buffers, regulatory supervision, risk modelling, etc etc.;
  • and hedges. Let’s talk about these.

Clearly, it would conceptually be useful for banks to be able to deploy cash in instruments that have both floating interest rates (and so do not lose market value when interest rates rise), and also highly liquid. You could imagine two ways to do that:

  1. Make loans more liquid, by, let’s say, packaging groups of similar loans into standardised instruments (call them “CDOs”), splitting them into tranches by risk, getting ratings agencies to rate them, and then create a liquid market for them. There’s a problem with though: it removes the risk from the loan originators, leading to perverse incentives that lead to bad quality loans, and you get the 2008 financial crisis. So, let’s not do this.
  2. Make bonds that don’t lose market value when interest rates rise. This can, broadly, be done by banks through hedging on interest rates. Then, when interest rates rise, the bonds lose market value, but the hedges make money to roughly counteract that effect, and vice-verse. This is a great idea, in general!

So why did SVB not have hedges in place? It seems that they were worried about what happens when interest rates fall: if hedges make money when rates rise, they obviously lose money when rates fall. Combined with the negative endowment effect on loans, this can make falling rates pretty bad for bank profitability. So, it seems that SVB dismantled much of their hedging in 2022, to take profits and to avoid losses if/when rates fell again. And this would have been fine, as long as we didn’t get both a rise in interest rates and a lot of depositors wanting their money back. Of course, that’s then what happened, and clearly the bank’s risk scenario testing was insufficient.

So let’s put this together into what led to SVB’s collapse:

  1. An (unrealised, theoretical) mark-to-market loss on bond holdings, due to:
    • lots of bonds relative to loans, at SVB, due to their client base of startups being relatively cash-rich and loan-light
    • insufficient hedging, due to concerns of the impact of hedges on profitability if rates were to fall.
  2. An unprecedented drop in deposits, due to:
    • a depositor base suddenly becoming less cash-rich, due to the sudden slowdown in VC funding to startups
    • a depositor base unusually prone to runs, because most of it was in deposits that exceeded the deposit insurance maximums, and came from depositors that were NOT diverse, as most startups (and especially their VC shareholders) were on the same Whatsapp groups
    • modern banking apps making it way easier to move cash out of a bank — no more queueing on the steps of the bank
    • some communication accidents and mistakes that flagged the theoretical massive losses on the bank’s bond holdings at market price.
  3. An inability to find extra liquidity to cover the gap:
    • SVB tried to raise further equity, but this failed and just contributed to the communication of the point just above, i.e., accelerated the deposit flight
    • emergency funding from the Fed, backed by bond holdings, would have had to have been done at market prices for bonds, thereby realising the theoretical mark-to-market losses, and leading to insolvency. Catch-22!

So my guess is, we’ll see regulatory changes and/or focus on requiring banks to model the impact of interest rate changes, not only on profitability and cash flow, but also on a bank’s ability to liquidate assets at short notice, without taking prohibitive market price losses.

Easily summarise PDFs (with AI)

Next chapter in learning and deploying AI tools: a tool to summarise arbitrary-length PDFs. Very useful for dry, long reports, legal judgements, etc. You could use it on a book, but it will be rather … dry?

Code at https://github.com/langabi/RecursiveSummarizer. It’s a simple (and not particularly elegant) Python script, but also a MacOS finder shortcut (right click on any PDF to summarise it, like a super-power on my computer!) Needs minor technical skill, and a (free) OpenAI.com API key.

The main point of this is to explore how to work around one of the key constraints of current Large Language Models like GPT-3: the limited size of the prompt. GPT-3 current version can, at time of writing, handle around 4000 tokens, or around 3000 words in the combined input and output. So, if you have a document bigger than this, chunking and recursive summaries are one way to go. It produces a decent result, at the risk of the summary losing a little of the overarching theme of a document.

More generally, a key technique to avoid AI models from, well, making things up, is to enrich the prompt with information relevant for an accurate answer to the prompt question. The limited prompt size is a constraint here too — you might find yourself using AI techniques like embedding to find exactly and only the needed content.

But certainly, I expect that there will be a lot of innovation in the next few months in how to give AIs working memory (long and short term), through things like architectural changes (bigger prompt windows), embedding-driven long-term memory, and maybe using the bots themselves to create a summary of a current conversation, that is repeated at the start of each new prompt window. And no doubt much more!

AI: The “spreadsheet” for copywriting

We just reduced new product copywriting time from 40 minutes to 15 minutes, with our first internal AI-powered tool. Here’s how I’m thinking about AI as the “spreadsheet for copywriting”.

Technically, the implementation relies on:

  • Enriching the “prompt” by replacing product- and order-specific references with all the information we have available, before passing the (much bigger) prompt to GPT-3, so that the AI is working on complete, current product and order information. Currently we’re just matching product/order IDs, but a more complete implementation will also used named entity extraction to find products, and then traditional search or embedding-based search to match and retrieve the matching products
  • Adding some boilerplate to the prompt about tone of voice, and instructing it to respond “I don’t know” when it is not confident in being able to answer the prompt.

Without this, we found (as have many others) that the AI makes up very plausible but completely fictional details.

For now, we’ve released the tool in an internal “chat bot”-style interface, but are looking at a Google Sheets function, to allow generating text for each of a list of products in response to a standard prompt — very useful for newsletter creation. Regardless, human review and editing is critical for all our use cases.

So far, we’re seeing fastest adoption by our copywriters, especially for listing new products. Faithful to Nature’s team, for example, spends hours reviewing each and every ingredient with our suppliers, before listing a product. This isn’t changing, but the process to convert that information into a standard, compelling product description, has dropped from 40 minutes to 15 minutes per product.

So where to from here? (I’m asking, and so, to be honest, are our copywriters).

I keep returning to a podcast by Planet Money on the impact of the spreadsheet on finance and accounting. Tl;dr:

  • spreadsheets used to be physical paper, calculated with hand calculators, laboriously
  • digital spreadsheets (Lotus 1-2-3, Excel) completely changed this, and away went all that human work
  • but, since then, finance employment has grown substantially
  • because spreadsheets changed from a reporting tool, to a scenario planning tool: what would happen if we opened another store? Changed our margin? So finance itself become far more strategic.

This is how I’m thinking about these AI copywriting tools: by replacing the “hand-held calculator” of having to write every single word by hand, we will achieve two things:

  • Unlock opportunity for copy, that was previously marginal. Ideally we’d have a page on our site for every possible combination of interests and questions for every small segment of customers, but that’s a lot of work. If the work halves, a lot more become possible — and indeed, we’re already working on newly-opened opportunities.
  • Make copywriting a strategic “what if” function. Currently, the marketing flow is linear: come up with a campaign concept, choose a tone of voice, identify copy requirements, write the copy, send to customer. This can become iterative: if 90% usable copy is generated instantly, we can experiment with different concepts based on how they turn out, before finalising the brief.

Clearly, elevating copywriting from  execution to what-if strategy is very applicable elsewhere, for example, in the legal industry.

If you’ve deployed AI tools yet, what have your experiences been?

Weekend idea

Weekend idea: a thought-provoking book I’m reading [1] made the good point (with data to support) that the more choices we evaluate from a long list, the better our final conclusion, but the LESS HAPPY we are with the final choice. Very relevant for picking a meal from a long menu/UberEats, or a holiday from Airbnb/bookings, or the best book from Amazon, or lots more.

So here’s a suggestion: for your next choice when the answer doesn’t REALLY matter, and there are lots of choices, restrict yourself to finding e.g. FIVE plausible options, and then choose the best from those. And then be happy with the result!

[1] “Barking Up the Wrong Tree: The Surprising Science Behind Why Everything You Know About Success Is (Mostly) Wrong”, by Eric Barker

Another COVID wave?

Maybe not

Time for a bold take. There won’t be a fourth COVID wave!

We’re exiting the third wave in South Africa, and (depending on how you count) globally too. Most people seem to be expecting another wave in Dec/Jan. I think there’s at least a 50% chance it won’t happen. (Yeah, that’s only 50%, but higher than many people are guessing).

Why? Because waves are driven by variants. Another wave would need a variant that is substantially more infectious than Delta, and/or evades immunity, and/or there is a massive decline in vaccine effectiveness. Some of those happened with Beta, and then Delta, but there’s no sign yet of a similar variant (early days I know). Historically, though, respiratory pandemics (eg 1918 flu) have 1-2 serious variants and that’s it. Plus, the hurdle for a successful breakout variant grows with every vaccination performed.

So it’sm far from certain but let’s put a stake in the ground 🙂

PS. Get vaccinated. COVID can kill outside waves, and we need to keep that hurdle climbing else a fourth wave is much more likely.

More Predictions

It’s time for some more predictions! Last time I did that (My predictions for the next 10, 20, 30 years) I was, if anything, too conservative Although, the section of “Black Swans” at the bottom was particularly accurate… (and has NOT been edited since it was written).

Photo by Drew Beamer on Unsplash

So here goes! Where do I think common opinion is wrong, especially in South Africa?

Electric vehicles and renewables

  1. By 2030, 40%+ of new cars sold in South Africa will be pure electric, with East Africa (eg Kenya) a bit behind, and West Africa a bit further behind. This seems inevitable when looking at the promises from major Western car companies, and the pace of innovation and falling prices from Chinese car companies.
  2. Due to 1., by 2030 the demand for petrol in South Africa will be falling 3%+ per year; and electricity demand will be growing 1% per year due to electric cars (though it may be falling for other reasons).
  3. Despite decarbonisation of electricity, and the growth of electric cars, the price of electricity in major global markets will NOT rise significantly from today’s levels, and may even fall, due to rapid rollout of solar and other renewables. South Africa is a special case depending on Eskom’s finances.
  4. By 2030, there will be large businesses built on taking advantage of near-free electricity during sunny hours (e.g., bulk hydrogen production), in several global markets.
  5. There will never again be a major (>300MW) coal power station built in South Africa.
  6. Kusile coal power-station will stop operating (or at least have been converted off coal) well before 30 years (2050), despite a design lifetime of 50+ years (around 2070). Which means an even bigger disaster for its return on capital.

Consumer trends over next 10-15 years

  1. The distinction between FMCG company / “brand” (designs and coordinates the manufacture of products, high margins, high marketing spend %, little direct consumer interaction) and retailer (sells products from brands, low margin, high volume,  low marketing spend %) will continue to blur in both directions, to an effective spectrum; plus there will be new logistics business models beyond traditional retailers, that aggregate deliveries from multiple other players (i.e., Instacart model evolved further).
  2. Traditional monolithic brands will fragment in favour of increasing numbers of niche brands with more authenticity and story. New “meta-brands” will appear, in the form of structured ranges of endorsements by influencers.
  3.  By 2030, 20%+ of “meat-like” products sold in upper-end grocery stores will be plant-based (i.e., non-animal).
  4.  By 2035, we will routinely take individualised medical probiotics in order to tune our gut biota, as treatment for a wide variety of complaints.


  1. By 2035, it will be functionally impossible for “legitimate” companies and individuals to use tax havens and financial engineering to pay near-zero taxes on profits or income.
  2. There are fortunes that will still be made in simplifying the payment of paper (or PDF) invoices, using machine learning text recognition to automatically load payment requests via bank apps/APIs. This will happen far faster than we can persuade people to stop using paper-based invoices for billing.

The July 2021 Riots

A thought to ponder

My enduring image of the 2021 looting will not be empty shelves or burning warehouses, but a short video of a boy, maybe 10 years old, carrying a MrPrice bag. He’d been stopped by a passer-by who went through his bag. It had a t-shirt or two, a pair of tekkie shoes, one 5-pack of underwear, and one or two other things, neatly folded away.

It didn’t have toys. Or flatscreen TVs. Or even food.

It had a few clothes, and a single pack of new underwear.

The bag was closed again, and he went on his way, in the grimy and thin clothes he was wearing, into a Johannesburg winter night that was at about freezing point.

There are many things to be said about attempted coups, and factional battles, and senseless vandalism, and the desperation of yet another COVID lockdown, and a terribly slow police response, and the community response that showed the best and sometimes the worst of us, and about how the ultimate victims of the looting are almost entirely going to be the people who can least afford it. And it’s all true and important, but also … not really anything new. It’s just where we are, the unsurprising result of the last few years, and also the last few decades, and also the last few centuries, as we take steps forwards and backwards.

But when the sun comes up in the morning, we need to get up, and put down our social media feeds, and get back to work. Cleaning up, building, teaching, healing, starting businesses, spending tax money responsibly, serving customers, processing paperwork, even sending emails. Because doing those things better is the only way I can think of that we can gradually build a society that is good enough for us all.

Not (just) so that looting doesn’t happen, but so that a 10-year-old child doesn’t need an outbreak of looting to get a new set of underwear.

Why does the start up industry beat corporates?

We seem to have, today, an unparalleled explosion in young, new companies, pioneering new products or ways of doing business, and thereby disrupting seemingly invincible pillars of our economy through explosive growth — commonly called startups. How is this possible?

Photo by Ian Schneider on Unsplash

Startups face a seemingly impossible challenge: they seek to build successful businesses from nothing. To do so, they need products that are so much better than alternatives that customers choose to use the new products, despite the lack of any brand recognition. These products need to be built on a shoe-string budget (at least initially), and quickly, by a team of founders that are working with limited resources, limited structures and few established commercial relationships. How can this ever work? Why don’t bigger companies, with access to all the same new technologies, lots of resources and skilled staff, a brand, and sales and marketing teams, win every time?

The answer often comes down to two things: startups have a completely crazy idea that actually works, and/or they are unreasonably good at something.

Continue reading “Why does the start up industry beat corporates?”

Thoughts: Lockdown

End of month one of lockdown: some thoughts as someone lucky enough to still be getting a salary.

Following the president’s example, I’ve donated a third of my April salary. Everyone’s choices of where to donate will differ, here were mine:

  • First priority was Silvertree staff (and ex-staff, in companies that have had to shut down). This month, between UIF and crowdfunding, everyone received a salary
  • I didn’t donate to Solidarity Fund this month, as (right now at least) we seem to have enough PPE, which seems to be their main focus
  • Next, I tried to donate or buy vouchers at restaurants+coffee shops I usually support. Not yet easy! Hopefully, initiatives like https://www.saveyourlocal.co.za/ will help
  • Lastly, I focused on NGOs that I know and that have established infrastructure to reach people that are hungry. This month, that included Thembisa Feeding Scheme, Streetlight Schools (https://www.streetlightschools.org/donate) and Gift of the Givers (https://giftofthegivers.org/make-a-difference/)
  • Economically, our problem is that the velocity of cash has dramatically slowed, as people can’t spend due to lockdown. This is causing a demand-side slump. So, your duty if you have cash: spend as you would have previously, and if you can’t, donate!

[Let me know your great ideas for causes to donate to!]

My predictions for the next 10, 20, 30 years

The latest in my once-every-two-years blog posts — oops. Over the New Year, I thought I’d make some predictions for the longer term. I’m looking forward to laughing at them in 2025, 2035 and 2045!

EDIT: some typos fixed

2025 (10 years time)

  • Physical signatures on paper will start becoming less common, replaced with electronic signatures and third-party document management systems. Over the next few years, security breaches or failures of some of these companies will lead to greater regulation of the industry. The result will eventually look similar to the credit rating agency or stock exchange industries of 2014 – several private companies running businesses in an industry heavily shapen by and working alongside regulatory agencies.
  • Hipster becomes accepted mainstream, as the desire for possession of mass-produced physical items is increasingly replaced with the quest for experience and “story” via artisanal and niche products. An increasing share of these products are virtual. Provision of these products and services will avoid massive unemployment, despite continuing decline in jobs in many of the careers that provided employment in 2014.
  • The global call-centre industry will finally peak (at a massive size), as new generations prefer to interact with computers and search for answers online. Content writing for helpdesks and forums will be the new outsourced growth industry, though it will not create as many jobs as the call centre industry.

2035 (20 years time)

  • As had happened to chess by 2014, computers will be unmistakably better than humans at “hard” AI problems from early 2000s, e.g., face recognition, speech recognition, “discovery” (reading and finding relevant content in huge troves of documents), medical diagnosis. However, AI will not be much closer to human-level consciousness, as we increasingly discover consciousness is not a single brain system, but rather an emergent property of many finely-balanced subsystems in our brains, built by our evolutionary past, that are very hard to abstract away from our brain structure. That is, computers won’t be “conscious” because we discover our “consciousness” is an increasingly slippery and less-generalisable concept than we had imagined.
  • More than 75% percent of seafood will be farmed rather than wild-caught. The exceptions will be either very high-end (and the target of growing environmentalist critiscism) or low-end. Farmed fish breeds will look and behave increasingly different to their wild ancestors.
  • The car industry will be in trouble as individual car ownership becomes less common. In advanced economies, shared self-driving cars summoned by smartphone are the default for many people. The only healthy parts of the industry are high-end luxury cars, low-end cars for emerging markets (though massive congestion is pushing public opinion away from car ownership here too), and self-driving electric cars designed for sharing.
  • Road congestion in advanced economy cities will be much reduced compared to 2014 (as happened to air pollution in these cities in a previous era, e.g., London after 1800s and LA after 1960s). This will be due to reduced private car use, but more so to self-driving cars and much better traffic management (traffic lights, automatic car re-routing).

2045 (30 years time)

  • CO2 emissions will be steadily falling, with global temperatures on track for a 3.5 degree rise. Agriculture will be steady, thanks to most of the world’s famers using genetically modified crops. Widespread but localised wars and revolutions will have happened, all with political proximate causes but with incidence strongly correlated with areas of greatest climate disruption. Large movements of people will also have occurred, leading to dramatic pro- and anti-immigrant upheavals, but these migrants will be largely described as economic- rather than climate-driven.
  • The dominant socio-economic issue will no longer be poverty and financial inequality as measued by Gini coefficient and similar, as this will be superseded by inequality in duration and quality of life. Improved medical technology will leave the top 1% with an expected lifetime almost double that of the bottom 50%, and much better quality of life in the meantime. The advantages of expensive biotech will threaten the assumption that all are born equal, as the offspring of the wealthiest gain developmental advantages, and society faces the danger of a biologically entrenched upper class.
  • The tertiary or “services” sector will employ nearly all workers, with industry following agriculture to become virtually irrelevant in formal employment. Production of physical goods will have followed energy use, to be largely uncorrelated with GDP, as non-physical goods become the bulk of GDP by value. Economists will split services into subsectors, such as traded services (finance, media and content) and non-traded services (hospitality, experiences, personal services).
  • First steps will be taken to in some countries to ban human drivers on certain roads (e.g., long distance highways), for safety reasons. These will be very controversial, pitting clear evidence of massive reductions in death toll due to self-driving cars, versus people’s right to drive themselves, and the rights of those who don’t yet own self-driving cars.

Black swans (that could make the above invalid)

  • Global pandemic of an easy to catch, slow to incubate but deadly virus. Might be caused by rogue biotech labs
  • War between super-powers

What do you think?