Price stabilization mechanisms

While stablecoins are a relatively new entrant into the financial arena, the core concepts and methods of price stabilization underlying them are familiar to traditional economics. To explain how and why Cogito’s price stabilization methods work, it is valuable to briefly review some of the economic background.

Supply & demand model

At the most basic level, all methods for stabilizing the exchange rates of currencies rely on a simple supply and demand based economic model. In essence, this sort of model states that a change in the market price (i.e. the exchange rate relative to a given reference currency, commodity, or basket) is due to changes in relative supply and/or demand. Given this assumption, the system can hold the stability around a target price if it counteracts this change, e.g. by buying or selling quantities of currency to impact market properties and behavior. This basic logic applies whether one is looking at fiat currencies, tokens pegged to fiat currencies, or to synthetic indices as with Cogito.

Leveraging AI

Artificial intelligence algorithms can be helpful here to the extent they can predict changes in supply and/or demand before they happen. In this case, a stabilization system can carry out financial transactions aimed at palliating supply or demand changes when they are in their initial stages, preventing them from becoming too extreme. Modern machine learning and reinforcement learning algorithms can be very powerful here; in this context, Cogito will leverage AI crypto-finance work done in the last 2 years by SingularityNet and SingularityDAO, using deep neural networks, multi-agent systems, and other methodologies.

Price stabilization theories

These basic economic concepts can be expanded upon in various ways, arriving at a spectrum of theoretical analyses of price stabilization, each of which has its pluses and minuses. Makiko Mita et al discussed and compared several theories on price stabilization mechanisms for decentralized payment systems, including Quantity Theory of Money, Tobin Tax and Speculative Risk. These offer a relevant background for understanding the design choices behind Cogito.

Quantity Theory of Money (QTM)

Quantity Theory of Money (QTM) is an early economic theory that is concerned with controlling the money supply and stabilizing the price of the currency. In a conventional fixed-exchange-rate regime, a central bank stands ready to use its foreign reserves to exchange for domestic currency if there are persistent deviations from the peg. When the domestic currency’s value trades below the peg, the central bank reduces the supply of domestic currency by selling foreign reserves. The stability mechanism is thus supply-driven in the case of a central bank-managed peg. QTM helps us understand how money moves and how to adjust the money supply.

QTM was later updated by Milton Friedman who argues that a central bank policy should aim at keeping the growth of the money supply at a rate commensurate with the growth in productivity. According to Friedman, "The stock of money [should be] increased at a fixed rate year-in and year-out without any variation in the rate of increase to meet cyclical needs" (Friedman, 1960).

Historical data analysis shows that QTM explains practical monetary behavior only in certain contexts and within certain bounds. Statistically, there has not been a strong correlation between changes in money supply and inflation/deflation, either among traditional currencies (Wang, 2017) or cryptocurrencies (Withiam, 2020). Partly, this has been because other factors such as financial innovation, market sentiment and political dynamics have proved dominant. However, it has also likely been partly because central bankers and their crypto analogues have tended to avoid extremely unwise money supply adjustments due to the warnings posed by QTM theory.

Tobin tax theory

The Tobin tax theory is based on the concept of a tax on international financial transactions designed to control exchange rate volatility. Makiko Mita points out that the Tobin Tax mechanism tells us how to stabilize stablecoin prices by including transaction fees. When users buy or sell a stablecoin and cause the price to deviate from other currencies (eg. USD), a high transaction fee is applied to disincentivize trading. On the other hand, when users buy or sell a stablecoin and cause the price to match other currencies, a low transaction fee is applied and users are incentivized to continue.

Transaction fees can also be utilized to help disincentivize destructive behaviors by bad actors. As Makiko Mita pointed out, as long as a stablecoin is designed to peg to something, it is exposed to the risk of speculative attacks aimed at moving the value away from the peg so as to profit from the necessary efforts to move it back. To mitigate speculative attacks, Spahn proposed a high transaction tax for speculative trading showing hallmarks of a speculative attack. However, he also pointed out that it is “virtually impossible to distinguish between normal liquidity trading and speculative “noise” trading.” Spahn expanded the Tobin tax to “a two-tier rate structure consisting of a low-rate financial transactions tax, plus an exchange surcharge at prohibitive rates as a piggyback. The latter would be dormant in times of normal financial activities and be activated only in the case of speculative attacks”. There is a clear potential role for machine learning here, in accurately identifying patterns of behavior characteristic of the early stages of a speculative attack.

Bank runs

Bank runs are another risk that those managing stablecoins must deal with, in a manner somewhat comparable to traditional fiat banks. Diamond and Dybvig formulate bank runs in terms of game theory, pointing out that a bank run can be a Nash equilibrium because when one depositor thinks that other depositors will withdraw their deposits even when they do not need to, the withdrawal makes that one depositor’s utility increase. To avoid this game theoretic dynamic and prevent bank runs, they fall back to the tried and true mechanism of deposit insurance. Obstfeld adopts a game theory approach to fixed exchange rate currencies that depend on a reserve fund, coming to similar conclusions.

Theories in practice

We observe the adoption and variants of these economic theories in the design of different stablecoins. With collateralized stablecoins, for instance, there is generally no equivalent of a central bank actively participating in the market to stabilize the peg. Instead, most stablecoin systems generate price stabilization through demand-driven flows to arbitrage differences between the intended peg and the rate in the secondary market. As soon as the price of the stablecoin rises above parity, there is an incentive to deposit fiat currency to mint coins and sell them in the secondary market. The stability mechanism is thus supply-driven in the case of a central bank-managed peg. This is the elemental design of stablecoins such as USDC and USDT.

With algorithmic stablecoins, we see Ampleforth is theoretically based on the Quantity Theory of Money, whereas Fei adopts the transaction fee theory in its design. Although we have not seen any stablecoin projects following Friedman Theory, one could interpret the downfall of Luna as providing a piece of evidence regarding what happens when Friedman’s ideas on money supply are egregiously ignored. Terra/Luna grew to be the largest algorithmic stablecoin, in part due to a practice of over-printing UST stablecoins substantially beyond what any macroeconomic fundamentals would suggest; this, among many other reasons, led to the collapse of the giant.

Cogito's approach

Cogito’s price stabilization model is based on these general economic principles as applied to DeFi ecosystems. Cogito’s underlying quantitative-finance algorithms provide a solution to a core question: How can the protocol manage a digital asset that keeps its price stable over the long run and aligns its intrinsic value to the progress of humanity? In what follows, we will explore how this solution works – first introducing the conceptual principles underlying Cogito, then addressing Cogito’s detailed architecture, stability mechanisms, and price curve.