Interbank Rate Uncertainty and Bank Lending

Layer 1: Overview

This paper asks whether uncertainty in the interbank market — distinct from general macroeconomic uncertainty — raises the cost of bank credit to firms, and whether bank-specific characteristics buffer or amplify this transmission. The question matters because interbank market disruptions were a central feature of both the 2007–2009 global financial crisis and the 2010–2012 European sovereign debt crisis, yet the empirical channel linking interbank stress to retail lending conditions had not been quantified at the individual-bank level.

The authors construct a novel measure of interbank rate uncertainty defined as the volume-weighted cross-sectional standard deviation of interest rates on overnight unsecured interbank loans in the euro area. This measure is extracted from individual transaction data in TARGET2, the main European payment system, using a Furfine-type algorithm that identifies interbank trades by matching outflow and inflow transactions between pairs of banks. Because it is based on overnight unsecured loans — not term loans — the measure is largely immune to uncertainty about the future path of monetary policy rates; it captures instead counterparty risk and precautionary liquidity hoarding in the interbank network.

The empirical strategy is a fixed-effects panel regression of bank-level lending rates on new loans to non-financial corporations against interbank rate uncertainty, interactions of that uncertainty with three bank-level variables (CDS spreads, ECB refinancing credit as a share of assets, and capital ratio), and a full set of controls including deposit rates, sovereign security holdings, interbank market borrowing, the three-month EONIA-OIS rate, and country unemployment rates. The panel covers monthly data for 323 individual banks across 18 euro area countries from June 2007 to February 2018, representing 80% of euro area Monetary Financial Institution assets.

Main quantitative findings: Heightened interbank rate uncertainty is robustly associated with higher lending rates on corporate loans. For the median bank in the sample, the average in-sample contribution of interbank rate uncertainty to lending rate spreads is approximately 35 basis points. The effect peaks sharply during crisis episodes: the contribution reaches around 90 basis points in Q4 2008 (following the Lehman Brothers collapse) and a historical maximum of around 120 basis points in Q4 2011 (during the acute phase of the European sovereign crisis). By end-2017, the contribution had declined to approximately 20 basis points.

The interaction terms reveal substantial heterogeneity. Banks with higher credit risk (higher CDS spreads, at the 90th percentile) tightened lending rates by approximately 70 basis points more than median peers in response to the 2011 uncertainty spike, while banks at the 10th percentile of CDS spreads responded similarly to the median. For capital: banks at the 10th percentile of the capital distribution tightened by about 25 basis points more, and banks at the 90th percentile tightened by about 20 basis points less, than their peers in response to the same episode. Banks with greater recourse to ECB funding (90th percentile of ECB credit) tightened lending rates by around 35 basis points less than their peers when uncertainty rose in 2011.

Crucially, these results are robust to controlling for the VIX (which itself enters significantly and positively) and for Euribor uncertainty (option-implied uncertainty about the three-month Euribor one year ahead, which is insignificant). The interbank rate uncertainty coefficients retain their sign, magnitude, and significance after including both alternative uncertainty measures, confirming that the measure captures interbank-specific stress — counterparty risk and liquidity hoarding — rather than general macroeconomic uncertainty or expected monetary policy volatility.

Policy implications: The findings support the bank-lending channel and suggest that macro-prudential policy (stronger capital buffers) and monetary policy operating through liquidity provision (ECB refinancing operations) both attenuate the transmission of interbank stress to corporate lending rates. ECB liquidity measures — fixed-rate full allotment, 3-year VLTROs, TLTROs — are visibly associated with declines in interbank rate uncertainty in the time-series plot.

Layer 2: Deep Dive

What is the identification strategy and what are the main threats to it?

The paper uses bank-level fixed-effects panel regressions. Bank fixed effects absorb time-invariant bank characteristics. The key identifying variation is the time-series movement in interbank rate uncertainty (a common aggregate shock) interacted with pre-determined or lagged bank-level characteristics. Because interbank rate uncertainty is constructed from overnight interbank transaction data — not from the bank lending rates themselves — it is not mechanically linked to the dependent variable. The main threats acknowledged or addressed are: (1) interbank rate uncertainty might simply proxy for general macroeconomic or financial uncertainty; the authors address this by including VIX and Euribor uncertainty as controls and showing the interbank uncertainty terms are unaffected; (2) non-linear effects of financial distress (not just uncertainty) could drive results; the authors include a quadratic term in bank CDS spreads, which is not significant, supporting the uncertainty interpretation; (3) the interactions with bank CDS spreads could reflect time-varying selection into risky lending rather than a pass-through mechanism; this concern is partially addressed by including controls for sovereign exposures, deposit rates, and interbank borrowing, though full identification of the causal mechanism is not claimed.

What are the main mechanisms proposed and how are they distinguished empirically?

Three mechanisms are proposed. First, counterparty risk: when interbank rate uncertainty rises, banks perceive uncertainty about what rate they will face if they need to borrow from the interbank network; banks with higher own credit risk (higher CDS spreads) face a compounded problem because they are likely to borrow at worse rates within that dispersed distribution, and they pass these higher funding costs onto corporate borrowers. Second, precautionary liquidity hoarding: uncertainty about interbank rates induces banks to hold more precautionary liquidity rather than lend, and this tightening is reflected in higher loan rates. Third, capital buffers: well-capitalized banks are more insulated from funding shocks and less likely to engage in risky lending, so they raise rates by less. Fourth, central bank liquidity substitution: access to ECB refinancing operations provides an alternative funding source that shields banks from interbank market stress. The paper distinguishes the counterparty risk/interbank-specific mechanism from general macro uncertainty by showing the VIX adds explanatory power independently but does not subsume the interbank uncertainty effect, and that Euribor uncertainty (which includes policy rate expectations) is not significant.

What heterogeneity across banks and time is documented?

Time heterogeneity: the uncertainty contribution averages 35 basis points across the sample, peaks at ~90 bps in Q4 2008 and ~120 bps in Q4 2011, recedes to ~20 bps by end-2017. The trajectories closely mirror the evolution of the interbank rate uncertainty measure itself, which spikes around Lehman (Sep 2008), subsides in 2009, rises again from mid-2010, peaks in late 2011, and then declines following ECB VLTRO announcements. Cross-bank heterogeneity by CDS spread: the 90th-percentile CDS bank tightened ~70 bps more than the median in 2011; the 10th-percentile CDS bank responded similarly to the median. Cross-bank heterogeneity by capital ratio: 10th-percentile capital banks tightened ~25 bps more, 90th-percentile capital banks tightened ~20 bps less than peers in 2011; this differential is relatively persistent over time. Cross-bank heterogeneity by ECB credit access: 90th-percentile ECB credit banks tightened ~35 bps less than peers in 2011, and this relief also persisted.

What robustness checks are run?

Four main robustness exercises are conducted. First, the baseline is estimated with and without the full set of bank-level controls; the signs and significance of interbank uncertainty terms are stable across all four specifications in Table 2. Second, alternative uncertainty measures (VIX and Euribor uncertainty) are added separately and jointly in Table 3; the interbank uncertainty terms remain significant and similar in magnitude. Third, nonlinear interaction terms are explored in Table 4 by adding quadratic interactions of uncertainty with CDS spreads (decomposed by above/below-median CDS) and capital ratio (decomposed by above/below-median capital); the quadratic CDS interaction terms are not significant, confirming the baseline’s linear specification for CDS; the quadratic capital interaction is significant for above-median capital banks, indicating that the marginal buffering effect of capital declines at high capital levels, but the linear term remains strongly significant. Fourth, the inclusion of a quadratic own term in bank CDS spreads (to rule out non-linear distress effects being mis-attributed to the interbank uncertainty interaction) is part of the baseline specification itself.

How does this paper relate to and differ from prior work?

The paper sits at the intersection of three literatures. In the banking network/fragility literature (Acemoglu et al. 2015; Allen and Gale 2000; Gai et al. 2011), existing work reconstructs interbank networks from loan data to study systemic risk; this paper instead uses a single scalar summary of network stress — the cross-sectional dispersion of interbank rates — that is empirically tractable and quantitatively links interbank conditions to corporate lending rates. In the uncertainty literature (Bloom 2009, 2014; Baker et al. 2013; Jurado et al. 2015), most measures are economy-wide (VIX, policy uncertainty indices, macro forecast dispersion); this paper offers an uncertainty measure that is explicitly financial-sector and interbank-specific, orthogonal to the VIX and Euribor uncertainty after conditioning. In the credit channel literature under uncertainty (Buch et al. 2015; Bordo et al. 2016; Valencia 2017), prior work examines how aggregate uncertainty measures affect bank lending; the present paper’s novelty is (a) the bank-level interbank-specific uncertainty measure constructed from transaction data rather than market prices, and (b) the interaction with bank balance-sheet heterogeneity at the individual-bank level for a large cross-country euro area panel. The paper also connects to work on interbank market disruptions during crises (Afonso et al. 2011 for the U.S.; Frutos et al. 2016 for the euro area) and to the bank-sovereign loop literature (Altavilla et al. 2017; Acharya et al. 2014).

What are the policy implications and their scope conditions?

Three policy lessons follow from the estimates. First, macro-prudential and micro-prudential policy that raises bank capital standards can reduce the sensitivity of corporate lending rates to interbank stress: the capital interaction term is negative and significant, and the marginal protective effect is largest at below-median capital levels. This implies capital requirements have diminishing returns as a buffer against interbank uncertainty at high capital levels (the quadratic robustness check). Second, monetary policy operating through liquidity provision — the paper points to fixed-rate full allotment operations, 3-year VLTROs, and TLTROs as concrete examples — reduces interbank rate uncertainty directly (as shown in the time series) and also shields individual banks from its effects via the ECB credit interaction term. Third, monitoring interbank rate dispersion provides a parsimonious, real-time indicator of the stress being transmitted to broader financing conditions. Scope conditions: the paper covers the euro area only, with its specific institutional architecture (ECB as LOLR, common monetary policy, country-level sovereign risk variation). Results hold over a sample dominated by two severe crisis episodes; generalizability to more tranquil periods or other banking systems is not directly tested. The paper does not examine quantities (loan volumes), only prices (lending rates), so the total credit contraction effect during crises is not fully captured.

How is the interbank rate uncertainty measure constructed and what does it capture?

The measure is the volume-weighted standard deviation of interest rates on overnight unsecured loans between euro area banks in a given month. It is constructed by applying a Furfine-type algorithm to individual payment data from TARGET2. The algorithm identifies interbank loans by matching outflows from one bank to an inflow the next day from the same counterparty of a nearly identical amount (principal plus a plausible interest rate), thereby recovering the implied rate on each overnight loan without direct observation of loan contracts. The monthly cross-sectional dispersion across all such identified transactions is the uncertainty proxy. Because the loans are overnight, the rate is insensitive to expectations about the future path of monetary policy (which would require a term premium for uncertainty about future rates). The measure instead reflects counterparty risk — if banks are uncertain about the creditworthiness of potential counterparties, they will lend to some at much higher rates than to others, widening the cross-sectional dispersion — and precautionary liquidity hoarding, which similarly creates a tiering of rates across banks of different perceived creditworthiness. The authors explicitly contrast it with Euribor uncertainty (a term measure incorporating policy expectations) to sharpen this interpretation.

What are the data sources and sample characteristics?

Four proprietary or confidential datasets are combined. Bank-level balance-sheet variables (main assets, bank capital, interbank liquidity) come from the ECB’s Individual Balance Sheet Items (IBSI) database. Bank lending rates on new loans to non-financial corporations and deposit rates come from the Individual MFI Interest Rate (IMIR) database. Banks’ recourse to ECB refinancing operations (both standard and non-standard, including LTROs and TLTROs) is provided as confidential ECB supervisory data. Bank CDS spreads are from Thomson Reuters Datastream. The interbank transaction data are from TARGET2. The sample is 323 individual banks across 18 euro area countries, observed monthly from June 2007 to February 2018, representing 80% of the assets held by euro area Monetary Financial Institutions. The panel is unbalanced: the full specifications with all interaction terms use approximately 12,850 observations, compared to 27,418 for the simpler specifications, reflecting data availability for CDS spreads and ECB credit data.

Are there limitations or caveats noted in the paper?

The authors focus exclusively on loan prices (lending rates), not loan quantities; the full effect of interbank uncertainty on credit availability (extensive margin) is not estimated. The Furfine algorithm, while standard, may misclassify some transactions or miss some interbank loans, introducing measurement error in the uncertainty measure. The regression imposes linearity of the uncertainty effect in bank-level moderating variables (with the exception of the capital quadratic robustness check); more flexible functional forms are only partially explored. The findings are specific to the euro area institutional context; the ECB’s role as a direct liquidity provider to banks is a key moderating factor that may not generalize to banking systems without a comparable LOLR. The sample is dominated by two unusual crisis periods; the average 35 bps effect masks that the contribution is modest (around 20 bps) in the tranquil post-2014 period, so the uncertainty channel may be primarily a crisis-period phenomenon.

Key Concepts

Interbank rate uncertainty: As defined by the authors: the volume-weighted cross-sectional standard deviation of interest rates on overnight unsecured loans between euro area banks in a given month, extracted from TARGET2 transaction data via a Furfine-type algorithm. Distinct from uncertainty about future policy rates; interpreted as reflecting counterparty risk and precautionary liquidity hoarding in the interbank network.

Furfine algorithm: A procedure for identifying interbank loans from payment system data by matching outflow and next-day inflow transactions of similar size between two banks, and inferring the implied interest rate from the difference between the two transaction amounts. Used here to extract overnight interbank loan rates from TARGET2.

Bank-lending channel: Used in the paper’s sense to describe the mechanism by which interbank funding conditions (specifically, uncertainty about the rate at which a bank can borrow overnight from peers) translate into higher lending rates charged to non-financial corporate borrowers, with the transmission depending on the bank’s own credit risk, capital position, and access to central bank funding.

Euribor uncertainty: An alternative uncertainty measure constructed as the interquartile range of the option-implied probability density function of the three-month Euribor one year ahead. Unlike the interbank rate uncertainty measure, it captures uncertainty about future interbank rates (including monetary policy expectations) rather than current cross-sectional dispersion in overnight rates. It is used as a control to identify the component of interbank rate uncertainty orthogonal to policy rate uncertainty.

ECB credit (over main assets): Banks’ total recourse to ECB standard and non-standard refinancing operations (LTROs, TLTROs, etc.) as a share of total assets, used as the measure of central bank funding access. Higher values are associated with a dampened sensitivity of lending rates to interbank rate uncertainty.

Lending rate spread: The difference between the lending rate charged by a bank on new loans to non-financial corporations and the three-month overnight index swap (OIS) rate, used as the dependent variable in robustness comparisons and for visual depiction of cross-sectional dispersion over time.

Capital ratio: Bank capital divided by main assets (total assets), used as the measure of balance-sheet soundness. Higher capital ratios are associated with attenuated sensitivity of lending rates to interbank rate uncertainty, consistent with well-capitalized banks being more insulated from funding shocks.