Applying behavioural theories and Tobin taxes to crypto assets
Today’s crypto-asset markets are irrational due to broad scale speculations of social media driven noise traders. The efforts of some sophisticated investors using fundamental analysis don’t pay off yet, given the high correlations of digital assets. The phenomenon of persistent mispricing for long periods of time is nothing new. It occurred in other asset classes and has been reduced through different measurements. The purpose of this article is to extend a similar line of inquiry towards crypto-asset markets, to discuss the key behavioural drivers behind the fundamental deviations in pricing and propose a means of algorithmic taxation as part of a solution.
Anomalies and behavioural theories
I turned to my course on Behavioural and Institutional Finance in the Booth School of Business where one of my core areas of learning was to understand the anomalies brought about in capital markets on account of the biases of asset managers and retail investors. These included anomalies pertaining to persistence of asset mispricing, value stock returns and the lack of uncovered interest rate parity, among others — but the one that really captured my attention was the effects of momentum and excessive volume trading which could potentially move the price of an asset far from its fundamental value. In the next sections I am going to shed some light on behavioural anomalies (what they are, consequences and core drivers) before discussing Tobin Taxes as one potential solution to the volatility problem.
Behavioural anomalies — categories and risks
I’d like to diverge here and provide a little background into what I refer to as behavioural anomalies.
By definition, anomalies are conditions persisting in the financial markets that potentially allow a deviation in fundamental asset pricing to persist for a time span– and these could be exploited, or could turn against an arbitrageur, thereby creating value excesses or minima in such portfolios.
One of the most well-known anomalies is the value anomaly. It means that a deviation is created through investments into value assets which have certain low KPIs, such as cashflow to price, book-to-market or earnings-to-price ratios and therefore could provide high alphas in certain financial cycles. Another well-known phenomenon is the momentum anomaly. It means that a deviation is created through investments into small-cap growth assets which have been consistently providing good returns over the span of the recent past by run ups following a wave of positive market sentiment — in crypto often triggered by (social) media noise and rumours. Other anomalies harm investors — such as noise trader risk, leading to asset mispricing in different markets or bubbles in correlated asset classes. In crypto the noise trader risks manifests itself through regulatory announcements or statements of high profile leaders from the conventional finance industries.
The key drivers of behavioural anomalies
Efficient market hypotheses (Fama-French, etc.) try to explain these effects through the reactions of rational market participants and how they create probabilistic aberrations from expected market behaviour, which over time (and across markets) has the tendency to revert to the mean. Seen from a behavioural perspective however, we can broadly classify all the key drivers across these scenarios into two broad categories: volatility risk (capture/avoidance) and information asymmetry. In short, there are risk levels that change dynamically thanks to the degree of specialisation in certain asset classes and the perception of the same (given investors’ margins of liquidity). In addition, the kind of understanding and the timing of information about the underlying asset classes drives a great degree of decision-making — and both of these have an underlying effect on asset prices and volatility.
The key point to be taken away is that these phenomena introduce mutually-reinforcing liquidity spirals among related groups of investors and asset classes which have the potential to create immense distortions.
Reinforcing liquidity spirals
Seen in this context, the momentum anomaly is driven by under-reaction to current earnings reports or in the case of crypto assets: news pertaining to long-term economic design capabilities (refer to Willy Woo’s and Chris Burniske’s NVT ratio) which creates an increased demand and valuation for assets until the constraints of margins/liquidity are hit. This fuels short-term market liquidity, high money-flow indices and trading volumes leading to the reinforcing phenomenon of market and trading liquidity (after Brunnermeir, Pederson 2009 — aka BNP). On the opposite side, there could be equilibria leading to reduced investor positions (on account of liquidity crunches in other financial markets), thereby causing a spike in the margins and losses on valuation. This is best explained by the following diagram:
The most interesting point BNP’s paper speaks about is the negative skewness of exchange rate movements with the interest differential for investment and trading currencies.
This basically refers to how carry trade positions are built up with a long left-tail up until the point of funding constraints. However, any small withdrawals/losses are followed by a sudden, discontinuous drop in market liquidity and asset price. There could be multiple points of fragile equilibria — and a small outflow/fund withdrawal could be a sharp trigger (given the high skewness):
- A crunch in margins/borrowing capability of investors in money/other financial markets, or investor runs on financial and asset management institutions
- Asset sales for capital outflows
- Price declines across the asset class, and therefore, in the balance sheets of investors
- A tightening of specialized investors to buy these assets at these prices, and a more general flight of capital to safer assets
- Corresponding volatility and risk spikes on these accounts, leading to increased margin requirements
We see a similar discussion in Caballero and Krishnamurthy 2008, which speaks of how Knightian effects (idiosyncratic uncertainty) kicking off this capital flight phenomenon across unrelated high-margin/volatility asset classes (this could explain the not-insignificant correlation between asset classes price drops).
This can be seen especially with regard to the build-up of the VIX spread (representing margin buildups on account of liquidity constraints) and the drop in the money flow index corresponding to the early months of 2018. This holds true even for the years in 2014 representing an increased sell-off on account of these constraints.
BNP’s model also speaks about how market liquidity declines with the increase in fundamental volatility (and the perception of the same in financial backers) — how volatile assets could be exposed to shocks that would subject them to highly discontinuous, illiquid equilibria.
Metrics driving activity
There are certain factors which cause a build-up of assets in investor portfolios — and as suggested by the afore-mentioned paragraphs, these are:
- Co-movement of funding liquidity (as indicated by VIX spreads, or treasury bond yields) with market liquidity (or price changes from baseline)
- Indications of high money-flow indices and higher transaction volumes
- Co-movement of market liquidity among asset-classes (could mean marginal investing by speculators)
- Increased bid-ask spreads cross assets
- Interest rate differentials in investment currencies (i.e., assets having the potential to be perceived as interest-earning) compared to funding currencies (assets used as a SoV); based on BNP’s 2008 model, these differentials are inversely correlated with the underlying nature of the currency and the crash risk skewness. The corresponding parallel in the crypto world would be to utilize the relative rate of return driven by the nature of the asset (utility value vs. share of profits on future cashflows) — and this would work when there is a movement of capital from funding currencies (which offer higher stability and lower economic rents) to investment currencies (securities, reputational rewards among others until a point where there are some margin constraints hit, following which there would be a much steeper drop in exchange rates
- Ratio of active addresses vs. number of transactions
- Cost of achieving consensus (as reflective in the rate of adding new mining architecture/increased number of on-chain transactions)
Fighting against volatility drivers with Tobin taxes
Given the slow build-up of carry positions and the high crash risk profiles built up on account of the fundamental uncertainty around crypto-assets and their co-movements we need to think about how to offset this phenomenon. Are there any adaptive measures that could be taken to reduce this extreme skewness seen in assets, and reduce the commonality of fragility?
I had the opportunity to question Prof. Rob Vishny on the needs of curtailing excessive volatility driven by noise trader activity — and one of the discussions led to Tobin taxation schemes. Tobin taxes, as proposed in the 70’s, were a means of imposing a tax on currency/exchange transactions in order to cushion exchange rate fluctuations. This definition was however broadened to propose taxes on general financial transactions to serve as a means to capture other trades which weren’t reflective in direct foreign exchange trades. This is done to prevent speculators from indulging in short-term capital flights, which could potentially cause developing ecosystems to charge much higher interest rates, thereby falling in the margin constraint trap as outlined above.
Looking at the metrics outlined in the previous section, we see imposing a Tobin-type tax would be most useful if it could prevent the reversal of money-flow indices, especially in times of endogenous(lower liquidity, higher spreads, closing/shutting of exchanges) and exogenous liquidity shocks(based on oracle data reflecting higher VIX, capital outflows, higher margins in money markets, downgrade of credit risks, capital gain tax withdrawals). The tax itself would be dedicated towards restoring some of the liquidity in such situations, through:
- Something akin to what the seignorage-style stablecoins/central banks do by acting as liquidity buffers and buyers of investment currencies
- Providing higher economic incentives to miners and consumers to perform more on the network
- Provide something like a call-option to arbitrageurs wishing to exit now at a premium price
Issues and implementation
The issue with Tobin taxes, as they may be categorized today, are the following:
- No clear sign on reduction on volatility (some studies point out to reduced liquidity and increased price deviations)
- Differentiating between strategies used to hedge against currency movements and those intended to be speculative strategies
- Less impact in deeply-established markets with high degree of liquidity already existing in them
- No clear practical implementation possible at the counterparty exchanges
Prof. Vishny was of the opinion that imposing a Tobin tax would potentially lead to making the carry trade less attractive for arbitrageurs, and increase the interest rate disparities, thanks to the home bias and the nature of sleepy/slow-moving capital.
In other words, this would mean that there would be a fall in exchange liquidity and more capital flow/remain with the asset classes that are most trusted by investors, even at the expense of lower returns (which might lead to short-term deviation of asset prices). The other outcome would be seeing more trades performed in other instruments/exchanges which are not tied to the place where the tax is imposed. Note however that this also reflects some of the empirical work performed on Tobin tax imposition on markets with a high mix of seasoned asset managers and institutional investors.
There are also obvious risks in reduced trading volumes and how they make prop up easy-to-form asset bubbles once there is more institutional capital flowing in — and the question of how this taxation would be imposed at the exchanges, or in off-chain transactions — but from a theoretical behavioural finance perspective, this would aid in reducing liquidity spirals through reduced economic constraints.
The major underlying point on all of what Prof. Vishny was speaking about boils down to the issue in controlling capital flows across varied asset classes and across different markets. Simply put, you could choose to off-shore our transactions to another nation’s exchanges and financial markets — or avoid being penalized for transacting in certain nature of asset classes by building on other, more opaque instruments. There is no easy way to track capital flows even in today’s globally-integrated financial markets.
My take on the same is quite different when it comes to the crypto asset market — and this has to do with the very nature of these assets.
Given the relative degree of immaturity across asset classes, the easier association of addresses involved in transactions , the limited number of exchanges and the highly community driven nature of stakeholders, it would probably be easier to impose algorithmic governance through the imposition of a tax on the exchanges themselves or even going a step further: as part of the token and exchange infrastructure.
Decentralized exchanges and relayers (like Radar Relay, 0x) could potentially be game-changers in driving an implementation of this. We could have Bancor-like systems monitoring the interchange of non-fungible tokens among multiple users and also impose a general financial tax on off-chain transactions by the on-chain verifier contract (based again on internal and external factors).
This would all be reflective of the pronounced degree of negative skewness in speculative investments made in crypto assets (esp. the ones with higher market caps) — and this is something that could be reduced by imposing a Tobin tax based on market-capitalization, or on the money-flow index of the asset that is bought into. Note that this would be something that would have to be treated as a function of the market trade volume and the price of the underlying asset at the exchanges/swaps where these trades take place and also as a factor of any exogenous changes reflected in the other financial markets. This would be triggered into the broader ecosystem of crypto-asset data marketplaces, and the increased degree of transparency in accounting for investments/expenses made across all spectra of users.
Papers referenced:
- https://www.princeton.edu/~markus/research/papers/liquidity.pdf
- http://www.nber.org/papers/w14473.pdf
- https://econpapers.repec.org/article/eeejfinec/v_3a88_3ay_3a2008_3ai_3a2_3ap_3a299-322.htm
- https://ms.mcmaster.ca/~grasselli/DeLongShleiferSummersWaldmann90.pdf
The ideas and opinions expressed in this article are my own views and by no means are to be taken as representative of Consensys Ventures.
Many thanks to Alexander Lange and Prof. Vishny for the extensive time taken to review and make suggestions on the same, and to Elfi Sixt for her comments. Thanks to Prof. Stephen McKeon for taking the time to respond on the same note.
Get in touch
If you share or disagree with our convictions, build a piece of cutting edge technology (or know someone who does) or just feel like connecting, please don’t hesitate to do so :)
