Anyone who has read this blog consistently knows that I've written on and off about an important developing body of work showing how algorithms -- closely linked to Google's PageRank algorithm -- could be used to greatly increase transparency surrounding issues of systemic risk. The systemic risk linked to any one bank or even to any single financial transaction could be made apparent to everyone; call it "radical transparency." Coupled with other mechanisms, such transparency could provide a route to making the system safer on is own, through normal economic self organization. The idea, in essence, is to use computation to give everyone in the market the information they need to make better choices. I've written about this
. I've just put a kind of
of all this up at Medium.com. But I'd also like to just quote some of the nice discussion from the most recent paper of Stefan Thurner and Sebastian Poledna. I think they describe the idea with beautiful clarity; so here's a whole big section. After describing various ideas currently under consideration for dealing with Too Big to Fail and systemic risk more generally, they note that..
No matter how well intended these developments might be, they miss the central point about the nature of SR, and might not be suitable to improve stability of the financial system in a sustainable way. SR is tightly related to the network structure of financial assets and liabilities in a financial system. Management of SR is essentially a matter of re-structuring financial networks such that the probability of cascading failure is reduced, or ideally eliminated.
Credit risk is the risk that a borrower will default on a given debt by failing to make the full pre-specified re-payments. It is usually seen as a risk that emerges between two counterparties once they engage in a financial transaction. The lender is the sole bearer of credit risk, and figures the likelihood of failed repayments into a risk premium. Lenders usually charge higher interest rates to borrowers that are more likely to default (risk-based pricing). Credit risk is relatively well understood, and can be mitigated through a number methods and techniques [18]. The Basle accords provide an extensive framework dealing foremost with the mitigation of credit risk [19–21].
When two counterparties are part of a financial system, for example as nodes in a financial network, the situation changes, and their transaction may affect the financial system as a whole. The lender no more is the sole bearer of credit risk, nor does credit risk depend on the financial conditions of the borrower alone. The impact of a default of the borrower is no longer limited to the lender, but it may affect the other creditors of the lender (who also lend to the same borrower) as well as their creditors, and so on. Similarly, the lender is not only vulnerable to a default of the borrower but also to defaults from all debtors of that borrower as well as their debtors, etc. In financial networks credit risk loses the local character between two counterparties, and becomes systemic.
SR is the risk that the financial system as a whole or a large fraction of it can no longer perform its function as a credit provider and collapses. SR is a result of the network nature of financial transactions and liabilities in the financial system. It unfolds as secondary cascades of credit defaults, triggered by credit defaults between individual counterparties. These cascades can potentially wipe out the financial system by a de-leveraging cascade [22–29]. It is obvious that lenders have a strong incentive to mitigate credit risk. In the case of SR the situation is less clear, since the loss-bearers will in general not be directly involved in those transactions that trigger systemic damage. It is not obvious which players in the financial system have a true interest to mitigate SR. Management of SR is foremost in the public interest.
It is important to note that SR spreads by lending. If a systemically risky node lends to a systemically non-risky one, the later inherits SR from the risky node, since if the non-risky borrower should (for whatever reason) not repay the loan, the risky node would trigger systemic damage. In this sense SR spreads from the risky through lending.
SR is predominantly a network property of liability networks. Different financial network topologies will have different probabilities for systemic collapse, given the link density and the financial conditions of nodes being the same. The management of SR becomes a technical problem of managing the network topology of financial networks. The goal is to do this in a way that does neither reduce the credit provision capacity, nor the transaction volume of the financial system. Data on the topology of credit networks is available to many central banks. Several studies on historical data show typical scale-free connectivity patterns in liability networks [30–35], including overnight markets [36], and financial flows [37]. As a network property, SR can be quantified by using networkmetrics [38, 39]. In particular a relative risk measure (DebtRank) can be assigned to all nodes in a financial network that specifies the fraction of SR they contribute to the system (institution- or node-specific SR) [39]. As shown later, it is natural to extend the notion of node-specific SR to individual liabilities between two counterparties (liability-specific SR), and to individual transactions (transaction-specific SR).
The central idea of this paper is to introduce an incentive structure in form of a transaction tax that dynamically structures liability networks such that SR is minimized. Since every transaction in a financial network has an impact on the overall SR of a system, we suggest a transaction tax on all transactions between any two market participants that increase the SR of the entire system. The size of the tax is proportional to the SR contribution of the particular transaction. Market participants looking for credit will try to avoid this tax by looking for credit opportunities that do not increase SR and are thus tax free. As a consequence the network arranges toward a topology that, in combination with the financial conditions of individual institutions, will lead to a defacto elimination of SR, meaning that cascading failures can no longer occur. In the spirit of risk-based pricing as it is used for credit risk, here we propose a systemic risk premium. It was shown in [39] that SR can be drastically reduced by reducing borrowing from systemically risky nodes. This is achieved by distributing SR evenly over the network and by preventing the emergence of systemically super-risky nodes. The mechanism works in a self-organized way: risky nodes reduce their SR because they are blocked from lending, non-risky nodes become more systemically risky through their lending. A SR premium encourages borrowers to borrow from safer lenders (since the borrower pays the tax). Further, lenders have an incentive to become systemically safe so that no (or only little) SRT is added to their loan offers, and they can offer competitive rates. Since mitigation of SR is foremost in the public interest we propose to charge a systemic risk tax as a margin on every financial transaction that increases global SR.
Of course it will take lots of further thought to bring this into a practical form. But big ideas always start out small.
