CBDC and Bank Disintermediation | BIS Working Paper 1280

CBDC and Banking Stability: The Dual Disintermediation Challenge

Central bank digital currencies represent one of the most consequential innovations in monetary policy since the abandonment of the gold standard. As central banks worldwide accelerate their CBDC research programs, a critical question has emerged: how will retail CBDC reshape the commercial banking system? The BIS Working Paper No. 1280, authored by Rhys Bidder, Timothy Jackson, and Matthias Rottner, provides the most comprehensive analysis to date of this dual-edged challenge.

The paper introduces a powerful framework distinguishing between two forms of CBDC-driven disintermediation. Slow disintermediation occurs when households gradually shift deposits from commercial banks to CBDC during normal economic conditions. This process can actually reduce banking sector fragility by shrinking overleveraged balance sheets. Fast disintermediation, by contrast, occurs during financial stress when CBDC provides a frictionless escape route from troubled banks, potentially triggering or amplifying systemic bank runs.

This distinction is critical because most existing research examines only one dimension. Policymakers designing CBDC frameworks need to understand both effects simultaneously. The BIS research combines an empirical household survey with a sophisticated macroeconomic model to demonstrate that without careful design — particularly holding limits and remuneration mechanics — CBDC could increase financial instability even as it modernizes the payment system. For readers exploring how financial research is transforming policy decisions, this paper sets a new benchmark.

BIS Working Paper 1280: Research Methodology and Household Survey Design

The empirical foundation of BIS Working Paper 1280 rests on the Bundesbank Survey on Consumer Expectations, conducted in April 2023. This representative survey captured responses from approximately 5,700 German households, making it one of the largest CBDC preference studies ever conducted in a major economy.

The survey embedded five carefully designed CBDC-related questions within the broader consumer expectations framework. Respondents were asked to hypothetically allocate a €1,000 monthly income across multiple asset classes — bank deposits, cash, government bonds, equities, and a hypothetical digital euro (dEUR) — both with and without CBDC availability. This allocation exercise provided granular portfolio-level data on substitution patterns.

A crucial innovation in the survey design was the inclusion of a banking stress scenario. Respondents were given a hypothetical situation where they held €5,000 in bank deposits and faced generalized banking stress. They were then asked how much they would withdraw and to which assets they would reallocate. Half the respondents received additional information emphasizing that the digital euro, unlike bank deposits, is a direct liability of the central bank — creating a natural experiment on the role of perceived safety in run dynamics.

The theoretical component employs a medium-scale New Keynesian DSGE (Dynamic Stochastic General Equilibrium) model, globally solved to capture nonlinear crisis dynamics. The model features an endogenous systemic bank run mechanism driven by banks’ risk-shifting incentives and state-dependent leverage constraints. This is not a standard linearized model — the global solution method is essential for capturing the threshold effects that distinguish normal times from crisis episodes.

Digital Euro Adoption in Normal Times: What 5,700 German Households Reveal

The survey results reveal substantial latent demand for a digital euro even without interest payments. A striking 43.2% of respondents indicated they would hold a positive amount of an unremunerated digital euro in normal economic conditions. This baseline adoption figure challenges assumptions that CBDC without remuneration would see negligible uptake.

Among all respondents (including those choosing zero CBDC), the average portfolio allocation to a digital euro was approximately 10% of the hypothetical €1,000 monthly income. However, among “keen” adopters — those reporting positive CBDC holdings — the average allocation rises to roughly 21% of their portfolio. This concentration effect means that policy design must account for a substantial minority holding meaningful CBDC balances, not just token amounts.

The CBDC-to-cash ratio among respondents was approximately 66%, indicating that digital euro holdings would reach roughly two-thirds the size of existing cash holdings. This figure has important implications for monetary aggregates and the ECB’s digital euro project. If these survey preferences translate to real behavior, a digital euro would become a significant component of household liquid assets within a few years of launch.

The introduction of a digital euro triggers meaningful portfolio rebalancing. Deposits decline by approximately 14 percentage points, and cash declines by about 4 percentage points on average. This confirms that CBDC predominantly substitutes for bank deposits rather than cash — a finding with direct implications for commercial bank funding costs and the monetary transmission mechanism. Banks that rely heavily on retail deposit funding face the greatest exposure to this slow disintermediation channel.

CBDC Remuneration Sensitivity and Portfolio Reallocation Effects

One of the most policy-relevant findings concerns the sensitivity of CBDC demand to remuneration. When the digital euro offers the same interest rate as respondents’ current bank accounts, adoption jumps from 43.2% to 54.3% — an 11 percentage point increase. This sensitivity is asymmetric and economically significant.

A 50 basis point decrease in CBDC remuneration relative to bank deposits reduces adoption by approximately 23 percentage points. Conversely, a 50 basis point increase boosts adoption by approximately 14 percentage points. This asymmetry — where the downside response exceeds the upside response — suggests loss aversion plays a role in CBDC demand, consistent with behavioral finance literature on household financial decision-making.

These remuneration sensitivity estimates provide central banks with a powerful calibration tool. By adjusting the CBDC interest rate relative to market rates, central banks can effectively control the pace and magnitude of deposit migration. The paper demonstrates that a remuneration formula tied to the policy rate — where CBDC pays the policy rate minus a fixed spread — creates an automatic stabilizer. During crises, when central banks typically cut rates, the attractiveness of CBDC automatically diminishes, reducing run incentives precisely when the banking system is most vulnerable.

For the European Central Bank and other institutions developing CBDC frameworks, these findings suggest that remuneration is not merely a technical parameter but a primary financial stability instrument. The choice between zero remuneration, fixed remuneration, and policy-rate-linked remuneration will significantly shape the CBDC’s impact on the broader financial system and monetary policy transmission.

Fast Disintermediation: How CBDC Amplifies Bank Run Dynamics

The most concerning findings emerge from the banking stress scenario. Without a digital euro available, respondents facing generalized banking stress reported they would withdraw more than 50% of their €5,000 deposits on average, primarily fleeing to physical cash. This baseline reveals the inherent fragility of fractional reserve banking even without CBDC.

When a digital euro is introduced into the stress scenario, the dynamics shift dramatically. Cash withdrawals decline to slightly above 40% of deposits, as some run-to-cash demand redirects to CBDC. However, the critical finding is that overall withdrawals from banks increase. CBDC does not merely substitute for cash as a run destination — it expands the total volume of bank runs. This is the fast disintermediation channel in action.

The mechanism driving this amplification is what the authors term storage-at-scale. Physical cash imposes increasing marginal costs as holdings grow — storage, security, insurance, and transaction costs all rise with the quantity of cash held. These frictions naturally limit the speed and scale of bank runs. CBDC eliminates these friction costs entirely. A household can move €50,000 from a bank deposit to CBDC instantaneously and without physical storage burden. This removes a natural brake on run dynamics.

Perhaps most revealing is the behavior of non-adopters under stress. Among respondents who said they would hold zero digital euro in normal times, approximately one-third (35% among those given information about central bank backing, 25% in the control group) indicated they would nevertheless withdraw to CBDC during banking stress. This demonstrates that CBDC’s crisis-time appeal extends far beyond its normal-time user base — crisis demand for CBDC is structurally different from steady-state demand.

The informational treatment experiment adds another layer. Respondents who received information emphasizing that the digital euro is a direct central bank liability showed significantly higher crisis-time CBDC withdrawal rates (58% versus 49%). This suggests that public communication about CBDC safety features could inadvertently amplify run dynamics by reinforcing the perception that CBDC is safer than bank deposits — even if deposit insurance provides equivalent protection up to guaranteed limits.

Trust in Central Banks and CBDC Demand Heterogeneity

Trust in the European Central Bank emerges as a powerful predictor of CBDC adoption, independent of demographic and financial characteristics. High trust in the ECB raises the probability of adopting a digital euro by approximately 10 percentage points. Conversely, low trust reduces adoption probability by about 22 percentage points — a striking asymmetry that mirrors the remuneration sensitivity pattern.

The survey also uncovers a fascinating historical dimension. Respondents who grew up in pre-1989 East Germany show approximately 6 percentage points lower CBDC adoption probability, controlling for other factors. This finding captures deep-seated attitudes toward government-issued financial instruments shaped by experiences with state surveillance and control under the German Democratic Republic. For CBDC design in other contexts — particularly in countries with histories of authoritarian monetary controls — this finding carries important cautionary implications.

Demographic heterogeneity extends beyond trust and history. Younger respondents, those with higher digital literacy, and those already using mobile banking show higher CBDC adoption rates. Income effects are more nuanced: middle-income households show the strongest CBDC preference, while very high-income households — who have access to sophisticated financial instruments — and very low-income households — who may rely more heavily on cash — show relatively lower interest.

These heterogeneity patterns have direct implications for financial inclusion policy. A one-size-fits-all CBDC design may inadvertently create a digital divide where technologically sophisticated, trust-rich populations benefit disproportionately while vulnerable populations face disruption to their preferred cash-based transaction patterns.

DSGE Model Framework: Endogenous Bank Runs and Storage-at-Scale

The theoretical core of BIS Working Paper 1280 is a medium-scale New Keynesian DSGE model with several innovations that distinguish it from standard macroeconomic frameworks. The model is solved globally rather than linearized around a steady state — a computationally demanding approach that is essential for capturing the nonlinear threshold dynamics of bank runs.

The banking sector in the model features endogenous systemic runs driven by two interacting mechanisms. First, banks engage in risk-shifting: they take excessive leverage because limited liability means they capture upside profits while depositors and deposit insurance bear downside losses. Second, leverage constraints are state-dependent — when asset prices fall, bank equity erodes, triggering margin calls and fire sales that further depress asset prices in a classic amplification spiral.

The CBDC innovation enters the model through the storage-at-scale channel. In the model, cash has increasing marginal storage costs — holding small amounts is virtually free, but large cash hoards require expensive physical security. This cost structure naturally limits cash-based runs. CBDC, by contrast, offers constant marginal storage costs (effectively zero), meaning that the volume of a CBDC-facilitated run is limited only by depositor preferences, not by logistics.

The model reveals a key tension. In normal times, CBDC competition reduces the liquidity premium banks extract from depositors, shrinking bank balance sheets and reducing leverage. This slow disintermediation effect is welfare-improving — a smaller, less leveraged banking sector is less prone to systemic crises. However, in crisis states, the availability of frictionless CBDC storage increases both the probability of crossing the run threshold and the severity of runs that do occur, because more deposits can flee faster and further.

The net welfare effect depends critically on CBDC design parameters. Without holding limits or remuneration adjustments, the crisis-amplification effect can dominate the stability benefits of a smaller banking sector, resulting in net welfare losses. This finding provides theoretical rigor to the intuition that CBDC must be designed holistically, considering both its payment functionality and its store-of-value implications.

Optimal CBDC Holding Limits: The €1,500–€2,500 Sweet Spot

The paper’s most actionable policy finding concerns the calibration of CBDC holding limits. Through systematic variation of the per-person holding cap in the DSGE model, the authors identify an optimal range between €1,500 and €2,500. This range emerges from a trade-off that can be understood intuitively.

Below €1,500, holding limits are so restrictive that CBDC cannot serve its intended payment and partial store-of-value functions. Households are unable to hold sufficient CBDC balances for everyday transactions, reducing adoption and eliminating the benefits of slow disintermediation (the welfare-improving contraction of the banking sector). The payment modernization goals of CBDC are effectively sacrificed.

Above €2,500, holding limits become loose enough that CBDC begins to function as a large-scale safe haven during banking stress. The storage-at-scale problem re-emerges because households can shift substantial deposit balances to CBDC during runs, amplifying systemic instability. The fast disintermediation channel dominates, and the net welfare effect turns negative.

Within the €1,500–€2,500 range, the model identifies a balance where CBDC provides sufficient utility for normal payment and savings behavior while the cap prevents destabilizing runs of arbitrary size. It is important to note that these figures are calibrated to the Euro-area context using German household data. The IMF’s CBDC Virtual Handbook emphasizes that optimal parameters will vary across jurisdictions depending on financial structure, deposit insurance coverage, cash usage patterns, and payment system infrastructure.

The ECB’s current digital euro proposal includes a holding limit, though the precise calibration remains under discussion. The BIS findings suggest that the ECB’s reported consideration of limits in the €3,000 range may be slightly above the stability-maximizing threshold, a point that will likely generate significant academic and policy debate as the digital euro legislation progresses through the European Parliament.

Policy Implications for Central Banks and Financial Regulators

The BIS Working Paper 1280 delivers several actionable policy recommendations that should inform CBDC design globally. First and most fundamentally, policymakers should not assume that CBDC will be neutral for financial stability. Even a well-intentioned, unremunerated CBDC can increase systemic risk through the fast disintermediation channel. Design choices are not peripheral technical details — they are primary financial stability instruments.

Second, holding limits are effective and pragmatically implementable. Unlike complex remuneration schemes, per-person holding caps are transparent, easy to communicate, and technologically straightforward. The calibration should be evidence-based and jurisdiction-specific, using household survey data and macroeconomic modeling tailored to each economy’s financial structure.

Third, remuneration design tied to the policy rate creates an automatic stabilizer. When central banks cut interest rates during downturns, CBDC remuneration falls in lockstep, reducing the relative attractiveness of CBDC as a safe haven. This mechanism works in the same direction as conventional monetary policy, reinforcing rather than undermining the central bank’s stabilization toolkit.

Fourth, communication strategy matters critically. The survey’s informational treatment experiment demonstrates that emphasizing CBDC’s superior safety compared to bank deposits can amplify run dynamics. Central banks must balance transparency about CBDC’s nature as a central bank liability with careful framing that does not inadvertently undermine confidence in the deposit insurance system.

Fifth, complementary prudential measures remain essential. CBDC design should not be developed in isolation from deposit insurance reform, macroprudential policy, and lender-of-last-resort frameworks. The interaction between CBDC holding limits and deposit insurance coverage levels requires careful analysis — households effectively face two safety thresholds (CBDC limit and deposit insurance cap) that jointly determine their crisis behavior.

Finally, the paper highlights significant gaps requiring further research. Corporate deposits — often large, uninsured, and highly mobile — represent an additional fast disintermediation channel not captured in the household survey. Cross-border CBDC flows, wholesale CBDC implications, and interactions with stablecoins and decentralized finance all require separate analysis. The BIS framework provides a foundation, but the full picture of CBDC’s impact on financial stability remains an active and urgent research frontier.