CBDC Interoperability: Simulating Futures

In my previous post, I explored the utility of network analysis for mapping interoperability among retail CBDC projects at a global scale. Both across and in aggregating different design choices, namely the architecture, infrastructure, access, and interlinkages of different CBDC projects, I showed that even with early data there appear to be nascent clusters of interoperability. In theory, this portends a future in which those clusters can more easily move money across their respective domestic financial actors – especially retail banks – and wherein cross-community currency exchange among private banks may be more difficult. These initial results around the community-level interoperability of retail CBDC projects may be disheartening at first glance, and imply the need for often-difficult cross-border coordination in soft and binding technical standards for CBDCs which would embed these initiatives in broader geopolitical interactions.

However, these results offer some hope. Namely, given the sheer volume of countries whose projects remain uncommitted to design features among some or all of these choice vectors, there is ample room for this interoperability issue to grow more or less severe as more central banks announce decisions. As a result, while cross-border coordination is likely an inevitable necessity for CBDC use at global scale, those talks may be made easier with careful consideration of these community-level concerns among countries whose projects remain largely uncommitted to any design feature. Here, we can think not only about smaller countries who may choose sides among the larger existing projects, like China’s. We may also consider the looming decisions by countries like the United States, which will inevitably shape the network structure and, by extension, interoperability issues at scale when those decisions are announced.

This suggests a need not only for mapping interoperability among existing design decisions, as we explored in the last post, but also implies value in simulating outcomes across different possible choices among undecided countries. This post explores precisely that juncture, and uses the United States as a salient case of an undecided country whose choices might radically reshape the interoperability network among retail CBDCs. The post first explores simulated outcomes among two extreme vectors of CBDC design choices by the United States, including the most or least familiar design of a fiat currency, and describes the major economic and political implications of either choice set. The post then introduces a novel tool for simulating more specific future interoperability outcomes, with user-input on any of the major design choices available across all four relevant design vectors described in this and the previous post.

i. Simulating Interoperability at Two Extremes

As we explored in the last post, there are more and less radical design choices available to central banks in their retail CBDC projects. By radical, I am not making normative claims about what a CBDC should look like, but rather describe the degree to which a CBDC looks different from fiat money that we see today. As a refresher, we’ll review the more and less familiar design choices along each design vector. In architecture, the status quo choice would be indirect access, which resembles the two-tiered banking system we have today. Conversely, the most radical design choice here would be direct architectures allowing retail accounts held with the central bank. In infrastructure, centralized ledgers resemble today’s status quo whereas decentralized ledgers represent a more radical transformation of fiat currency ledger management. For access, an account framework resembles today’s status quo of currency ownership logic as compared to the more radical design option of token frameworks, which more closely resemble private and decentralized digital currencies like Bitcoin, for example. In each of these cases, we can consider hybrid design choices to be ‘middle-ground’ options between status quo and radical designs. Each of these is summarized below for reference in table 1. Here again, we will leave aside the issue of interlinkages, which more meaningfully shape whether interoperability is a concern at global scale given its determination around the use of retail CBDCs across borders.

Table 1: Retail Digital Currency Design Choices by Alignment with Status Quo

Design FeatureStatus QuoModerately DifferentRadically Different
ArchitectureIndirect ClaimsHybrid Claims ArchitectureDirect Claims
InfrastructureConventional LedgerHybrid Ledger InfrastructureDecentralized Ledger
AccessAccount AccessHybrid Access FrameworkToken Access

This table allows us to imagine two extremes of US CBDC design choices along the status quo – radical continuum. On one hand, the US might issue a CBDC which looks a lot like fiat currency today, with indirect architecture, conventional ledger infrastructure, and account access logic. On the other hand, the US might pursue fully radical designs with direct claims on the central bank, a decentralized ledger infrastructure, and a token-based access framework. In line with the prior post, we can simulate the network of inter-CBDC interoperability under either of these conditions and observe what happens to the community structure of the entire network. In plot 1 below, these two extremes are simulated in network plots with the status quo simulation on the left and the radical simulation on the right. As with the prior post, communities are identified with a 5-step walk-trap community detection algorithm, though we can relax this later on.

Plot 1: Interoperability Map Under Status Quo and Radical US CBDC Design Scenarios

US is square node, all other countries are circles. Community coloring automatically assigned.

As we can see in the plots, the community structures look fairly different contingent on US choices; the US is portrayed as a square node, while the rest of the countries are circles. The status quo plot largely retains the three-community structure we previously saw in the empirical plots of actual design choices, whereas the radical choice plot breaks one major community in two, with the US in the smaller community. Notably, we can consider the implications of these two community outcomes for the US along economic and political dimensions. Economically, the community structure of each plot divides US trade partners in different manners. As shown in table 2 below, the status quo world includes a higher volume of total US trade (in log values) within the community than outside of the community, as compared to the radical scenario. The ratio of out-community trade flows to in-community trade flows is approximately three times larger in the radical scenario than in the status quo scenario.

Table 2: Trade Implications of Status Quo and Radical US CBDC Design Scenarios

Trade MeasureStatus Quo ScenarioRadical Scenario
Total Trade: In Community77.132.0
Total Trade: Out of Community141.4186.5
Ratio of Total Trade: Out / In1.835.82

Politically, we can consider the network position of the US within each of these plots as indicative of its leverage in CBDC negotiations. Namely, the centrality and brokerage positions of the US imply different structural positions from which the US could bring other countries to the table in establishing soft and binding standards for CBDC interoperability. Table 3 below offers network positions for the US in each of these plots, with two measures of centrality and two measures of brokerage. Degree centrality weighs the proportion of connections the US has to the total number possible in the network, and eigenvector centrality normalizes this value by the popularity of the US’ connections. Conversely, betweenness measures the proportion of total shortest paths in the network on which the US sits, while constraint indicates the proportion of US connections which are not redundant (e.g. already connected amongst themselves). As we can see in table 3, while both brokerage measures are largely identical across scenarios, the centrality measures are notably higher in the status quo world. Importantly, the movement in eigenvector centrality from degree centrality is positive in the status quo world, implying more popular connections in the US ego-network, as compared to the downward shift in the radical design scenario, which suggests a more isolating normalized centrality position for the US.

Table 3: Network Position Implications of Status Quo and Radical US CBDC Design Scenarios

Network MeasureStatus Quo ScenarioRadical Scenario
Degree Centrality0.650.40
Eigenvector Centrality0.890.34

ii. A New Tool for Interoperability Simulations

Notably, both of these tables imply a more privileged position for the US under a fully status quo design scenario as compared to a fully radical design scenario. However, this comparison is made between two extremes on a multi-dimensional scale. This section concludes the post with a new tool for simulating more granular outcomes along specific, unbundled US CBDC design choices.

The tool is available at this link (https://timarple.shinyapps.io/CBDC_Explorer/), and includes four pages that change dynamically with user input. Inputs, on the left-hand side, allow the user to select one of several design features across all four vectors of US retail CBDC design choices. The user can also decide whether hybrid choices are universally interoperable (e.g. a hybrid architecture is interoperable with both direct and indirect architectures), or not. If the user selects this option, they can also decide whether this interoperability is compound, giving a weight of 1 to hybrid to non-hybrid choices and a weight of 2 to all exact choice matches. If this is not selected, all edges are weighed as ‘1’. Finally, the user can also specify a different number of steps for the walk-trap community detection algorithm in the slider at the bottom of the left-hand side, ranging from 2 to 10 to allow for more or less restrictive community detection walks contingent on these design choices.

The simulation tool has four pages. The first page shows the same network images produced from the simulations here, with walk-trap community detection and shape identifiers for the US versus other countries, based on user input. The second page also shows a network image, on this page with a dynamic layout that allows the user to click and drag nodes for inspection of their names and neighbors. The third tab includes replicas of tables 2 and 3 in this post, showing the trade and network position implications for the US based on the design features the user selects in the tool. The fourth tab finally lays out a table of all network members (countries which have atleast one CBDC design choice), with columns to show whether the US design choice is interoperable across any of the four vectors, along with a count of interoperable features and indicator of whether a country is in the same community as the US within the network.

This tool is hopefully a first step to more rigorous simulation efforts in empirically informing US CBDC design choices, and can serve as a template for simulating other countries’ choices as well. Moving forward, it will be critical to update this tool with new data as it rolls out, and to expand the simulation framework to include selection options for other undecided countries or other undecided features among projects for more granular inspection of choice implications at scale.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s