Monetary policy transmission in India under the base rate and MCLR regimes: a comparative study

The study investigates how effectively changes in India’s monetary policy rate transmit to banks’ lending rates, focusing on internal lending benchmarks used by banks. India’s banking system plays a predominant role in financial intermediation, making the interest rate channel crucial for achieving monetary policy objectives of price stability and growth. Historically, India moved from PLR to BPLR (2003), to base rate (2010), and to MCLR (2016) as internal benchmarks to improve transparency and transmission. Despite reforms, transmission remained partial and uneven, with concerns about discretion in benchmark computation and price discrimination between old and new borrowers under the base rate regime. Against this backdrop, the paper examines the degree of pass-through to lending rates across two internal benchmark regimes—base rate and MCLR—using bank-level quarterly data, comparing how bank-specific characteristics shape the transmission. It aims to quantify long-run pass-through under each regime and identify factors behind differential transmission, providing new evidence by directly comparing benchmark regimes.

The literature spans nearly a century, covering traditional interest rate and credit channels (Keynes, Friedman and Schwartz, Ando and Modigliani, Tobin, Taylor, Obstfeld and Rogoff, Meltzer, Bernanke and Gertler). More recent works analyze transmission across advanced and emerging economies (Mohanty and Turner; Mishra et al.; Trichet), with mixed evidence on the relative importance of channels. For India, studies examine various channels and modeling approaches (RBI 2002; Patra and Kapur; Goyal; Anand et al.; Patra and Ray; Mazumder; Paul), and pass-through estimates using macro data (Khundrakpam; Kapur and Behera; Bhoi et al.). A growing literature uses bank-level data to capture heterogeneity (Gambacorta; Altavilla et al.; Holton and D’Acri; Abuka et al.; Sapriza and Temesvary). Indian bank-level studies (Bhaumik et al.; Das; John et al.; Mishra and Kelly) highlight roles of ownership, balance sheet strength, asset quality, and liquidity in transmission. However, no prior study directly compares monetary transmission across internal lending benchmark regimes, leaving a gap that this paper addresses.

Data and periods: The analysis covers April 2004–July 2019 descriptively; econometric estimates use quarterly bank-level data for all domestic banks from 2012–13:Q4 to 2018–19:Q2, with subperiods: base rate regime (2012–13:Q4 to 2015–16:Q4) and MCLR regime (2016–17:Q1 to 2018–19:Q2). Foreign banks and the BPLR regime are excluded from econometric models due to comparability and data constraints. All series are seasonally adjusted; variables in levels that are I(1) are differenced. Aggregate analysis (VAR/VECM): Variables include weighted average lending rate on fresh rupee loans (WALRF), monetary policy indicator (weighted average call rate, WACR), inflation (CPI), and real GDP. Unit root tests (ADF/PP) indicate I(1); Johansen tests indicate one cointegrating relation at 5%. A VECM is estimated to assess short- and long-run dynamics; impulse responses are identified via Cholesky decomposition. Bank-level dynamic panel (system GMM): The main specification models the quarter-over-quarter change in WALRF for bank i on lags of its own change, lags of the monetary policy indicator (WACR), interactions of WACR with bank-specific characteristics, and macro controls (inflation, and initially GDP growth). The dependent variable is the change in WALRF on fresh rupee loans, chosen over outstanding loans due to contemporaneous pricing to the prevailing benchmark. WACR is used as policy indicator (correlated 0.94 with the repo rate), reflecting both stance and liquidity conditions. Macroeconomic controls: CPI inflation; GDP growth was included initially but dropped in panel models due to multicollinearity with dummies. Two dummies are included: D1 for the taper tantrum period (2013–14:Q3–Q4) when MSF supplanted the repo rate; D2 for demonetization impact (2016–17:Q3 to 2017–18:Q1). Bank-specific characteristics: Nine models each include one characteristic interacted with WACR, normalized as deviations from cross-bank period averages: (1) weighted average term deposit rate (funding cost), (2) size (log total assets), (3) liquidity ratio (cash and balances measures to total assets), (4) CRAR (capital adequacy), (5) return on assets, (6) non-performing assets ratio, (7) non-interest income share, (8) operating expenses to assets, (9) SLR securities to assets. Normalization ensures interaction terms capture deviations from the average bank so that WACR coefficients can be interpreted as the average pass-through. Estimation and diagnostics: Dynamic panel system GMM (Arellano and Bover, 1995) is used to address endogeneity from lagged dependent variables and potential simultaneity. Instrument validity is assessed via Sargan tests; serial correlation via Arellano-Bond AR(1)/AR(2) tests (acceptable if AR(1) significant negative and AR(2) insignificant). Lags are selected empirically, retaining significant terms. Robustness checks for seasonality and integration are implemented.

- Aggregate VECM: There is a stable long-run relationship between WALRF and WACR. Long-run coefficient indicates that a 1 percentage point increase in WACR is associated with a 0.36 percentage point increase in WALRF. - Benchmark panel (no bank characteristics): Long-run pass-through from WACR to WALRF is 0.12 for the whole sample, 0.12 for the base rate subperiod, and 0.21 for the MCLR subperiod. Inflation positively and significantly affects lending rates. - Whole sample with bank characteristics (nine models): Long-run pass-through ranges from 0.13 to 0.24. Inflation typically has a positive significant effect. Significant interactions indicate: - Deposit rate: positive with WACR (higher funding costs raise lending rates). - Size: positive with WACR (larger banks adjust lending rates upward more when policy tightens, and reduce less in easing). - CRAR: negative with WACR (better-capitalized banks lower lending rates more in easing, aiding transmission). - Operating expenses: positive with WACR (higher costs hinder pass-through in easing). - Base rate regime (2012–13:Q4 to 2015–16:Q4): Long-run pass-through ranges from 0.11 to 0.19 across models; one-quarter lag effects of WACR are 0.22–0.35 and significant. Inflation generally raises lending rates. Significant interactions: - CRAR: negative with WACR (higher capital facilitates pass-through in easing). - Non-interest income: positive with WACR; higher NII is associated with less reduction in lending rates during easing, impeding transmission. - MCLR regime (2016–17:Q1 to 2018–19:Q2): Long-run pass-through is higher in all models, ranging from 0.26 to 0.47 per 1 percentage point change in WACR. Significant interactions: - Deposit rate: positive (higher funding cost raises lending rates). - Size: positive (larger banks reduce lending rates less in easing). - Liquidity: negative (greater liquidity facilitates stronger declines in lending rates in easing and dampens increases in tightening). - NPA: negative (contrary to expectations); attributed to risk-averse strategy post-AQR and PCA, with banks shifting toward lower-rate, collateralized retail lending; thus, rising NPAs did not obstruct pass-through during easing. - Regime comparison: For a 100 bps policy change, estimated long-run increase in WALRF is 26–47 bps in MCLR vs 11–19 bps in base rate across models, indicating materially stronger transmission under MCLR. - Event dummies: Demonitization dummy generally not significant for enhancing transmission; taper tantrum dummy captured MSF episode effects as intended. - Contextual drivers: Better alignment of liquidity management with policy stance, adoption of flexible inflation targeting, and weaker credit demand likely supported improved MCLR-era transmission.

The research question was whether monetary policy pass-through to bank lending rates differs across internal benchmark regimes and what bank-specific factors condition transmission. The findings show that pass-through is consistently stronger under the MCLR regime than the base rate regime across all specifications. This suggests that the more transparent and formula-based MCLR improved the responsiveness of banks’ lending rates to policy impulses compared with the base rate, which suffered from discretion and heterogeneity in calculation and pricing discrimination. Bank characteristics significantly condition transmission: higher funding costs, larger size, and higher operating expenses tend to dampen reductions in lending rates during easing, while stronger capitalization and greater liquidity facilitate transmission. The unexpected negative association between NPAs and lending rates in MCLR reflects balance sheet repair and regulatory constraints leading banks to pivot toward lower-risk retail lending, limiting pricing freedom but not obstructing pass-through during easing. Macro conditions—alignment of liquidity operations with stance and the FIT framework—also contributed to improved transmission. Overall, the results reinforce the importance of benchmark design, liquidity alignment, and bank balance sheet health for effective interest rate channel transmission in a bank-dependent economy like India.

The paper contributes by providing the first direct comparison of monetary policy pass-through to lending rates across India’s base rate and MCLR internal benchmark regimes using bank-level dynamic panel methods. It finds that a 100 bps policy rate change translates into 26–47 bps changes in lending rates under MCLR versus 11–19 bps under the base rate, with aggregate VECM corroborating a meaningful long-run relationship between policy and lending rates. Bank-specific factors—funding costs, size, capitalization, liquidity, operating costs—systematically shape pass-through. Despite improvements under MCLR, transmission remained partial, motivating the RBI’s subsequent shift to external benchmark-based pricing for retail and MSME loans from October 2019 (extended to medium enterprises from April 2020). Future research could examine the external benchmark regime’s effectiveness, explore nonlinearities and asymmetries across tightening/easing cycles, and assess heterogeneity by ownership and business models as longer post-2019 data become available.

- Data constraints prevented inclusion of the BPLR regime in econometric models and conducting nonlinearity tests. - Foreign banks were excluded due to differing business models and small credit share. - GDP growth was dropped from the panel regressions due to multicollinearity with event dummies, potentially omitting some demand-side variation. - WACR is used as a proxy for the policy stance; while highly correlated with the repo rate, it may also reflect liquidity conditions. - Sample period for MCLR is relatively short (2016–2019), which may limit inference on long-run dynamics.