Bank Spreads & Financial Structure

The analysis of banks' interest rate setting behavior begins with the Monti-Klein model, which posits that banks maximize profits by setting prices in both loan and deposit markets. This model assumes banks possess some pricing power but cannot influence interbank or bond market interest rates. The market interest rate is considered the funding cost for loans and the opportunity cost for deposits. The spread between deposit and market rates represents the opportunity cost for depositors and the bank's profitability on deposits. However, the Monti-Klein model examines loan and deposit markets separately. The study then introduces the Ho and Saunders dealership model, which views banks as intermediaries setting interest rates to balance asymmetric loan demand and deposit supply. This model utilizes insights from the financial literature on broker bid-ask spreads. The combination of these models forms the theoretical basis for the study, offering a framework to understand the multitude of factors affecting bank interest rate setting behavior and the subsequent pass-through effect.

A crucial determinant of the spread between lending rates and market interest rates is credit risk—the possibility of loan defaults. While the study focuses on aggregate credit risk at the country level, reflecting the business cycle, it acknowledges that individual loans may carry idiosyncratic risks. Changes in credit risk are expected to positively influence lending spreads, unless credit rationing occurs. The study explicitly addresses the limitations of focusing solely on interest rate effects, given data constraints, and notes that banks use multiple tools to manage credit risk (collateral, covenants, etc.). Interest rate risk, due to the asynchronous arrival of loan demands and deposit supplies, compels banks to temporarily invest in the money market, creating reinvestment risk should market rates fall. These risks significantly influence banks’ pricing decisions and consequently, their responsiveness to market rate changes.

Market structure and competition are explored through different lenses. The “structure-conduct-performance” hypothesis suggests higher market concentration leads to less favorable consumer pricing due to potential collusion. The relative-market-power hypothesis argues that only firms with substantial market shares and differentiated products can exert market power and earn profits. In contrast, the efficient-structure hypothesis posits that concentration may increase sector efficiency. Contestable markets theory suggests that even concentrated markets can exhibit competitive behavior if barriers to entry are low. The study cites Corvoisier and Gropp (2006) to illustrate how internet access enhances contestability. Furthermore, the influence of bank capital is analyzed. High capital ratios may lead banks to seek higher interest rate spreads to offset the higher cost of capital; however, higher capital can also signal creditworthiness, potentially lowering deposit funding costs and increasing deposit spreads.

The availability of substitute banking products in financial markets influences bank spreads. The Ho and Saunders model highlights that loan demand and deposit supply elasticity affect bank spreads, implying that competition from non-bank financial intermediaries influences pricing behavior. For instance, easier access to direct debt financing for non-financial corporations puts downward pressure on banks' lending spreads. Similarly, household access to direct investment or money market funds affects deposit rates. The study extensively discusses financial innovation, categorizing it into increased prominence of non-bank financial intermediaries (Thornton, 1994) and the development of new financial instruments, services, and markets (González-Paramo, 2006; Tufano, 2003). This innovation influences bank spreads through increased competition, improved risk management, and reduced agency costs and information asymmetries.

Liquidity risk—the risk of insufficient cash or borrowing capacity to meet demands—is another critical factor. This risk forces banks to borrow at potentially higher costs (Angbazo, 1997). The proportion of funds in cash or cash equivalents impacts the liquidity premium in bank spreads. Incorporating liquidity risk (Prisman, Slovin, and Sushka, 1986) into the Monti-Klein model shows that the bank's cost of resources increases due to the expected cost of liquidity shortages (Freixas and Rochet, 1997). Management efficiency also influences interest spreads. Efficient management leads to optimal asset and liability composition, affecting interest margins (Angbazo, 1997). The study uses distance-to-default, a measure of bank quality, to account for management efficiency and bank stability (Gropp et al., 2006), given data limitations on quarterly balance sheet data.

The study employs a panel econometric approach to analyze the dynamic adjustment of bank spreads—the difference between bank interest rates and corresponding market rates—for various loan and deposit categories. This approach allows for the estimation of the bank interest rate pass-through from changes in monetary policy, considering both immediate and long-term adjustments. The researchers aim to deepen the understanding of the pass-through process by controlling for cyclical developments and differences in financial structures across Euro area countries. The methodology allows the speed of adjustment to vary across different financial structures and levels of financial innovation across the countries in the study. This contrasts with other studies in the pass-through literature, which typically focus on systematic measurement of pass-through extent and then relate it to bank price-setting determinants. This study instead directly estimates the response of bank rates to policy rate changes. The choice of econometric model is not specified, however, the document does refer to the potential use of Vector Error-Correction Models (VECM) which are often used to capture dynamic relationships and distinguish between short and long run impacts

The analysis utilizes two datasets: the National Reference Interest Rates (NRIR) sample and the Monetary Interest Rates (MIR) sample. While the NRIR sample provides a longer time period, the MIR sample offers higher quality and greater detail, based on harmonized definitions and methods across the Euro area. Spread calculation involves a three-step process: selection of a comparable market interest rate (more straightforward with the MIR sample's detailed maturity breakdown); aggregation of MIR spreads to match NRIR product categories (using outstanding amounts for deposits, and new business volumes for loans, with some exceptions); and adjustment of NRIR spreads to match any level differences with corresponding MIR spreads. The choice of market interest rate for each sample is carefully explained, with differences noted between the two samples, primarily due to the differences in the granularity and types of data available for each sample. The data challenges associated with measuring credit risk are highlighted, and the chosen metric—the slope of the yield curve (difference between 5-year government bond yield and 3-month interbank deposit rate)—is justified. This selection is made based on capturing banks’ assessments of borrowers' future repayment abilities.

The empirical model incorporates control variables to account for factors that influence bank spread dynamics. These variables include measures of credit risk (using the slope of the yield curve), interest rate risk (using the standard deviation of government bond yields for loans and money market rates for deposits), and bank soundness (using banks' distance-to-default). The use of the standard deviations of various interest rates acts as a proxy for controlling for the effect of the overall level of interest rates on bank spreads. The rationale behind each variable's inclusion is detailed; for example, the yield curve slope is used to reflect market expectations of economic outlook and potential credit risk changes. The selection of these control variables is driven by the desire to capture aspects of the banking sector's financial conditions that are likely to change over the business cycle. These variables are considered to be important for controlling for dynamic changes that may mask or confound the effects of the core variables of interest in the analysis.

The study's primary finding is the significant difference in the speed of adjustment of bank interest rates to market interest rate changes between loans and deposits. The pass-through is considerably faster for loans than for deposits. This difference is particularly pronounced for demand and savings deposits, where the pass-through is described as especially sluggish. The results indicate that, on average, lending spreads decrease by approximately 39 basis points following a 100 basis point increase in market rates within the same quarter, implying that lending rates increased by only 62 basis points. However, a lagged effect is observed, with lending spreads increasing by about 25 basis points in response to a 100 basis point increase in the previous quarter. This suggests an almost complete pass-through (86 basis points) after two quarters. In contrast, deposit spreads show a slower and less complete pass-through. A 100 basis point increase in market rates leads to a 59 basis point increase in deposit spreads in the same quarter but only a 17 basis point decrease in the following quarter, resulting in a total upwards adjustment of deposit rates of only 58% after two quarters. This difference is directly linked to the observed 'stickiness' of bank interest rates and has significant monetary policy implications.

The analysis further examines the effects of control variables on bank spreads. Changes in banks' distance-to-default (a measure of financial health) negatively affect bank spreads, suggesting that healthier banks price their products more in line with market rates. Conversely, increases in interest rate risk, as measured by the standard deviation of government bond yields (for loan spreads) and money market rates (for deposit spreads), lead to higher bank spreads, indicating that banks facing higher uncertainty operate with wider margins. The slope of the yield curve, reflecting market expectations and economic outlook, negatively affects loan spreads—suggesting lower spreads are demanded when economic prospects improve, likely due to lower credit risks. This highlights the influence of macroeconomic conditions and risk perceptions on bank pricing decisions, adding another layer of complexity to the pass-through process and explaining some of the variations around the core pass-through findings. The study briefly mentions considering the effect of overall interest rate level on spreads but ultimately excludes it due to its insignificance in their models.

The results show substantial variations in the pass-through across specific loan and deposit segments. While the pass-through for loans is generally complete after two quarters, the immediate impact varies considerably. For instance, short-term loans to non-financial corporations adjust by only around 50% after one quarter, while the adjustment for mortgages and long-term loans is significantly faster. The pass-through for deposit rates is even more heterogeneous. Overnight and savings deposits display extremely sluggish pass-through (25-30% after six months, less than 20% after one quarter), potentially indicating a non-competitive market or the existence of non-interest adjustments. In contrast, short-term and long-term time deposits show much quicker pass-through, reaching 70% and 100%, respectively, suggesting a greater responsiveness to market changes in these specific market segments. These segment-specific differences in speed and completeness of pass-through underscore the importance of disaggregating the analysis beyond broad loan and deposit categories. This also highlights the differing dynamics for various products and the potential importance of other factors that are not easily captured in the aggregated data analysis.

The study investigates the impact of competition within the banking sector on the speed and completeness of interest rate pass-through. The Panzar-Rosse H-statistic, a measure of market power, is used as a key indicator of competition. While the overall pass-through for loans and deposits does not show significant differences between high and low competition levels, a disaggregated analysis reveals that higher competition (lower market power) leads to a faster and more complete pass-through for most loan categories. The effect is primarily observed in the short run, with less competitive markets 'catching up' in subsequent quarters, resulting in minor differences in the long-run pass-through. This suggests that competition primarily influences the speed of adjustment rather than the ultimate level of pass-through. For deposit rates, the differences are less pronounced, although a statistically significant impact is detected for long-term time deposits. The study acknowledges the debate surrounding the use of concentration ratios to measure competition, citing alternative indicators like the Panzar-Rosse H-statistic and the Herfindahl index, and using a composite H-statistic derived from various recent studies (Carbo et al., 2005; Bikker, 2004; Claessens and Laeven, 2003) to account for the lack of a universally accepted measure.

The influence of competition from financial markets is analyzed using several proxies. Initially, the stock market capitalization to GDP ratio is used to represent the overall size and depth of financial markets. Results indicate that larger financial markets lead to quicker and more complete pass-through of bank lending rates, primarily affecting the speed of adjustment. A more detailed analysis disaggregates the data into specific product segments and employs additional metrics to capture non-bank competition. For lending to non-financial corporations, the ratio of debt securities issued by non-financial corporations to GDP is used, capturing the availability of alternative funding sources. For household loans, non-bank loans to households are considered, and the importance of mutual funds (as a percentage of GDP) is used to measure alternative savings options. These analyses provide a more nuanced understanding of how the presence and relative size of alternative financing mechanisms affect the bank interest rate pass-through process for different customer segments and product types.

The study explores the impact of financial innovation on interest rate pass-through, using various indicators, all as a percentage of GDP: turnover in single-currency interest rate derivatives (BIS data); securitization and issuance of covered bonds (European Securitization Forum data); and venture capital investments (European Venture Capital Association data). The findings indicate that countries with relatively high levels of securitization experience a higher pass-through for long-term loans to non-financial corporations and mortgages, although the magnitude of the effect and its statistical significance varies depending on the specific product category. A strong relationship is observed between interest rate derivative turnover and the pass-through to long-term loans to non-financial corporations and mortgages, mainly affecting the speed of adjustment in these segments and is likely explained by derivatives acting as a hedging mechanism against interest rate risk. The study observes no discernible effect of venture capital market development on the pass-through in most segments. However, a significant effect is noticed for long-term loans to non-financial corporations, particularly in the first quarter.

To ensure the robustness of the findings, the researchers conducted several checks using alternative model specifications. The primary model was re-estimated using various approaches. These included models incorporating fixed effects across countries with product dummies for both broad markets (model R1) and individual product segments (model R2); a two-way random effects model (model R3); and separate seemingly unrelated regression models for loans and deposits, each incorporating country dummies (model R4). The purpose of this multifaceted approach was to test the sensitivity of the results to different assumptions about the data generating process and to assess the consistency of the findings across alternative model structures. This rigorous approach increases confidence in the reliability of the key results presented in the paper's earlier sections, because the consistency of results across multiple model types suggests the findings are not artifacts of a specific modeling approach or assumptions.

The results obtained from these alternative model specifications are reported to be essentially the same as those obtained using the baseline models (models 1 and 2). This consistency across multiple econometric approaches serves to strengthen the confidence in the main findings of the study. The unchanging nature of the results suggests that the core conclusions regarding the bank interest rate pass-through dynamics, any asymmetries identified, and the impact of financial market characteristics and innovation are not sensitive to the specific econometric choices made in the initial estimations. The robustness checks provide strong evidence that the patterns identified in the baseline models are not a matter of statistical chance or specific modeling limitations, but rather represent genuine relationships within the data.