Learning a Little from Ponzi Schemes

diff.substack.com · by Byrne Hobart

In this issue:

  • Learning a Little from Ponzi Schemes

  • Coinbase

  • Tiger and Turnover

  • The Finance Dialectic

  • The Hostile Amazon Bundle

  • Cybersecurity Updates

Bernie Madoff, ponzi schemer extraordinaire, died yesterday at the age of 82. At the moment his scheme collapsed in December 2008, Madoff's customers thought they had assets of $65bn; they actually had around $13m left. There are still mysteries surrounding the scheme and its collapse: Madoff claimed that he didn't start his ponzi scheme until the 1990s, but as early as 1962 he borrowed money from his father-in-law to make his brokerage customers whole after a market decline. Most Madoff clients got steady returns of 1% a month; some got triple-digit returns back-to-back, or convenient tax losses (one of those clients died of a heart attack soon after Madoff's business imploded; his widow returned $7.2bn in profits). So it’s still unclear how long the scheme went on, and who was in on which parts.

Before Madoff's fund was famous for collapsing, there were questions about how he earned the profits he reported, and the absence of fees on his funds. Separately from the fund, Madoff had a sizable market-making business, which at one point was handling 5% of the NYSE's volume. This gave rise to two mutually-incompatible conspiracy theories:

  1. One possibility was that Madoff was running a profitable investment strategy as a loss-leader, and then monetizing it by overtrading with his market-maker operation. In that model, Madoff was still adding value, just skimming an unknown amount of it for himself.

  2. Another version was that Madoff's fund was being retroactively assigned profitable trades from the market-making operation. Since market-making involves frequent trades with a hit rate close to 50/50, you can construct a subset of profitable trades. As long as trades can be assigned to a different account after the fact, a high-volume trading operation has the right raw material to create a high-return one.

Ponzi schemes often get attributed to "greed," but Madoff's track record doesn’t line up with that theory. In fact, when a fund of funds associated with Renaissance Technologies passed on investing in Madoff, part of their argument was that he wasn't greedy enough: Madoff was letting other entities charge fees just for funneling money to his fund, instead of charging for the fund itself. Even if he had some way of monetizing the fund, he was clearly under-monetizing it. And the external feeder funds that raised money for him were charging plenty; Fairfield Sentry took 20% of profits and was able to raise $4bn, indicating Madoff's pricing power. And even if they were good at sales, it's hard for them to have been $100m worth of good-at-sales.

Of course, we know in retrospect why Madoff was willing to give up so much of his potential returns: he needed the cash, and was paying feeder funds to keep the scheme topped up. This may also explain why he wasn't charging high fees: to do so, he'd either have to report lower returns to his investors, or would have to report higher gross returns, which would make the scheme less plausible. Since one of the limiting factors for Madoff was scale—the strategies he purported to use would have run out of opportunities at the level of assets he managed—it was optimal for Madoff to minimize the amount of pre-fee excess return he had to report in order to produce a given post-fee return to his backers.

Running a ponzi scheme has the serious downside that the expected value over a long enough period is -100%, but in the short term there are some perks. For example, ponzi schemes have a comparative advantage at risk management; since the numbers are fake, they can choose whatever level of volatility they want. This is a mark of how sophisticated Madoff was. Typical ponzi scheme busts promise high returns; this recent one offered 35%. Madoff promised returns in the low teens, and reported very consistent month-to-month numbers. In a way, this makes Madoff a very modern, sophisticated operation: he was pitching an investment with great risk-adjusted numbers even though the absolute returns were more modest. This is a recipe for keeping a ponzi scheme alive for as long as possible: at a given Sharpe ratio, withdrawals will be a lot higher for the fund offering 30% than for the fund claiming 12%.

So, in some ways Madoff ran the ideal ponzi scheme: he minimized cash needs, he incentivized people to raise money for him (while keeping his pitch to them vague enough that there wasn't really a story to keep straight), he paired it with a legitimate business that gave him semi-plausible excuses for the money he claimed to be making, and provided a source of emergency liquidity towards the end.

But there's still the lingering question: why? Madoff made less money on his scheme than at least one of his investors, and if we assume that feeder funds believed one of the theories that his returns were real-but-illicit, then most of the Madoff scheme money was made by people other than Madoff. To the extent that he had an exit strategy, the plan seems to have been dying at home in his sleep rather than in prison.

The best available evidence is that Madoff's scheme was not a cynical attempt to maximize his returns, but instead a cynical attempt to tell people what they wanted to hear. When those people were everyday investors, charities, and capital allocators, what they wanted to hear was that it was possible to get equity-like returns with bond-like volatility. When those people were sharper-elbowed operators, what they wanted to hear was that Madoff was willing to fake whatever returns they needed.

The Madoff failure happened just a few months after Lehman Weekend, around what turned out to be the peak of the financial crisis. He wasn't the only one; Allen Stanford's collapsed a few months later, with less money left over. A few months earlier, another scheme, Petters Group, also fell apart. When markets crashed in early 2020, I assumed we'd see another ponzi scheme unravel. And yet, there weren't any. The drop might have been too short and sharp, or perhaps ponzi schemers have learned to require longer lock-ups. Or, quite possibly, Madoff was a sufficiently well-executed ponzi scheme that, once everything he did became a red flag, it was impossible to run a remotely similar operation. Just as it's hard to copy successful people, both because they're unusual and because they've already done whatever it is that you're copying, it's hard to copy record-setting ponzi schemers; Madoff was a talented crook, who invented or perfected numerous tricks that are now obsolete thanks to him.

Further reading: The SEC's report on all their failed opportunities to stop Madoff makes for grim reading. The Wizard of Lies is a good narrative.



Coinbase successfully completed its direct listing: the stock was indicated to price at $250, opened at $394, rose briefly, and closed at $345. Direct listings seem to be a purer form of price discovery than IPOs. Instead of a structured mutual leaking of purchase and sale intentions—the company quotes an offering size and a price range, potential buyers indicate interest, both numbers shift until they converge, and meanwhile both sides have some combination of legal restrictions on quickly realizing profits (i.e. lockups) or social restrictions (i.e. IPO flippers don't get great allocations). Direct listings have elements of this, but are much closer to a pure market. This makes them volatile (from peak to trough, Coinbase shares lost 27.8% of their value before recovering a bit). But it's also simpler. Since Coinbase is minting money right now, it doesn't need to raise much from its public offering. It may, however, want to provide liquidity to employees and investors, many of whom have been through more crypto cycles than the average Coinbase buyer yesterday.

I previously wrote up Coinbase here, and before that I noted that Coinbase competes with other stocks that represent bets on Bitcoin, most notably, at that time, GBTC. Microstrategy, another equity proxy for Bitcoin, closed down 13% yesterday. And, for historical interest, here's Brian Armstrong's request for a cofounder, with feedback ("Bad idea" and the classic "Trivially easy to implement.")

(Disclosure: still long Bitcoin.)

Tiger and Turnover

Everett Randle of Founders Fund has a good piece on the math of Tiger Global's growth-stage PE practice. The core argument is that capital velocity is a substitute for percentage returns as long as outside investors are satisfied. In other words, if a fund can deploy $x per year with a 25% hurdle rate, or $2x with a 20% hurdle rate, the latter is better. This is the unspoken argument for why many funds grow their assets under management, and it's easy to come up with stylized arguments that show that it's unfavorable to outside investors. Suppose a fund can generate a fixed dollar amount of alpha each year, and the optimal strategy for the fund is to expand assets under management until fees exactly equal alpha. But in this case, it enables a genuinely new model: a lower hurdle rate allows investments at higher valuations, which is one way to get into competitive deals. And raising more money means that each investment is a smaller share of total assets under management, so single-company risk is mitigated.

The strongest case for this strategy is that there's a cohort of investors who try to add non-financial value, but don't, especially. Those investors can't switch their narrative to "we just write big checks" without adverse signalling, so they're stuck competing in a business that they don't have a competitive advantage in. Tiger—and anyone else who practices the lower-hurdle/higher-velocity approach—has a pricing advantage relative to these companies. But what's especially interesting about this is that the high-velocity/low-touch model is a direct complement to the low-velocity/high-touch approach. Right now, the main effect that Tiger seems to have is that it's made late-stage funding far more competitive, but the long-term effect may be to further widen the gap between the top-tier funds and everyone else. The best companies will have less dollar-dilutive financing, and less board-control dilution per dollar, all of which will make early-stage checks from top investors go further.

The Finance Dialectic

A study, summarized here, argues that activist campaigns are more likely to succeed when a stock has high short interest. There are two good stories you can tell here:

  1. Some companies that get the attention of short sellers have poor management, and activist investors are a solution to that.

  2. High short interest is a sign that a company's plausible valuations are bimodal, or at least have a wide distribution. A company probably won't get unusually high short interest if the median short seller thinks it's overvalued by five or ten percent, but it will get high short interest if there's a strong case that the price could drop by half, or that it's a zero. If there's generally high uncertainty, it's more likely that a short seller and an activist can be convinced that there's a wide gap between perception and reality, even if they disagree on the direction.

This is a subset of one of the functions of markets. As I've argued before, they aren't just for price discovery, but for distribution discovery. And if the wide distribution of potential prices is caused by uncertainty of judgment rather than unknowable reality, then the time value of money motivates people with positions in the stock to resolve the gap in beliefs as soon as possible.

The Hostile Amazon Bundle

The WSJ has a piece on Amazon's negotiations across different product categories ($): giving a device company better placement on Amazon retail in exchange for data, or blocking counterfeits only if a company paid for advertising. Amazon's economics to its customers are a testament to the power of bundling, and a charitable interpretation of their behavior here is that they're bundling for merchants, too. If everything Amazon offers has a different supply and demand curve, then the more areas Amazon and sellers can negotiate over, the more likely they are to come to an agreement. That's the theory, but in practice the economics look more like this: Amazon finds where their merchants' demand is most inelastic, and then finds every conceivable way to charge them the maximum possible.

(Disclosure: I'm long Amazon.)

Cybersecurity Updates

  • Vice has a profile of a ransomware gang, including an interview with the organizer. One interesting note: "Unlike other ransomware groups such as Netwalker, REvil, and CONTI, Cl0p doesn't run an affiliate program, meaning they don't share their malware with other cybercriminals to get a share of their proceedings." An affiliate program makes sense if a group is exploiting one vulnerability that they expect to get patched over time, since it parallelizes the hacking process. Cl0p's refusal to do this implies that they think they have a steady supply of future hacking options.

  • Microsoft's latest round of patches includes nineteen critical vulnerabilities (up from ten last month), and five 0-days.

  • The Biden administration is planning to improve power grid security. I've occasionally wondered if the first sign of an invasion of Taiwan will be when the lights go out in the US; it's a hard risk to assess, but an important one to worry about, because the bigger the outage, the more costly it is to restart.

diff.substack.com · by Byrne Hobart