Monetary Policy Pass-Through in Nepal: An SVAR Analysis of Reserve Ratios and Policy Rates

Research Motivation

How effective are the Nepal Rastra Bank's (NRB) monetary policy instruments in influencing macroeconomic outcomes? This question has critical implications for a developing economy highly sensitive to external shocks and remittance inflows.

Policy Context: Nepal's Unique Monetary Challenges

Nepal's monetary transmission mechanism operates under distinct structural constraints:

• Exchange Rate Peg: Fixed exchange rate with Indian Rupee limits monetary independence
• Shallow Financial Markets: Limited secondary market development constrains interest rate transmission
• High Informal Sector: ~60% of economy operates outside formal banking channels
• Remittance Dependence: Worker remittances constitute ~25% of GDP, driving liquidity shocks

Given these constraints, quantifying the effectiveness of Cash Reserve Ratio (CRR) adjustments versus Bank Rate changes became a priority research question for the NRB Research Department.

Theoretical Framework & Literature Context

Monetary Transmission Channels

Central bank actions affect the real economy through multiple transmission channels:

Research Gap Addressed

Prior studies on Nepal's monetary transmission (Paudel & Timilsina, 2013; Dhungana, 2016) focused primarily on interest rate channels. Our contribution:

Data Sources & Variable Construction

Dataset Characteristics

Data Cleaning & Stationarity Testing

library(tidyverse)
library(tseries)
library(urca)

# Load and prepare time series
macro_data <- read_csv("nepal_macro_data.csv") %>%
  mutate(
    date = as.Date(paste0(year, "-", quarter*3, "-01")),
    m2_growth = c(NA, diff(log(m2))) * 100,
    credit_growth = c(NA, diff(log(credit))) * 100,
    inflation = c(NA, diff(log(cpi))) * 100,
    gdp_growth = c(NA, diff(log(gdp))) * 100
  )

# Augmented Dickey-Fuller tests for unit roots
adf.test(macro_data$bank_rate, k = 4)      # Stationary in levels
adf.test(macro_data$crr, k = 4)            # Stationary in levels
adf.test(macro_data$inflation, k = 4)      # Stationary after differencing

Stationarity Results: All variables achieved stationarity after appropriate transformations, satisfying VAR estimation requirements. Policy variables (BR, CRR) entered in levels; flow variables entered as growth rates.

Structural VAR Methodology

Model Specification

The reduced-form VAR(p) model:

Where \(\mathbf{Y}_t = [BR_t, CRR_t, M2_t, Credit_t, CPI_t, GDP_t]'\) and \(\mathbf{u}_t\) represents reduced-form residuals

Structural Identification

To recover structural shocks from reduced-form residuals, we impose recursive (Cholesky) restrictions:

Where \(\boldsymbol{\epsilon}_t\) represents orthogonal structural shocks. The recursive ordering reflects causal contemporaneous effects:

Lag Selection & Model Diagnostics

library(vars)

# Prepare VAR dataset
var_data <- macro_data %>%
  select(bank_rate, crr, m2_growth, credit_growth, 
         inflation, gdp_growth) %>%
  ts(start = c(2012, 1), frequency = 4)

# Optimal lag selection
VARselect(var_data, lag.max = 8, type = "const")
# AIC suggests 4 lags; BIC suggests 2 lags
# Selected p = 4 based on economic rationale

# Estimate reduced-form VAR
var_model <- VAR(var_data, p = 4, type = "const")

# Diagnostic checks
serial.test(var_model, lags.pt = 8, type = "PT.asymptotic")  # No serial correlation
normality.test(var_model)  # Residuals approximately normal
stability(var_model)       # All roots inside unit circle

Model Validation: Selected VAR(4) specification passed all diagnostic tests. Inverse roots of characteristic polynomial lie inside unit circle, confirming stability. Portmanteau test fails to reject null of no serial correlation (p = 0.43).

Empirical Results & Policy Insights

Impulse Response Functions (IRFs)

IRFs trace the dynamic response of each variable to a one-standard-deviation shock in policy instruments:

Figure 1: Impulse Responses to Bank Rate Shock - Inflation peaks at quarter 3 (-0.38%), credit growth contracts by -1.2% at quarter 4

Figure 2: Impulse Responses to CRR Shock - Stronger immediate effect on credit (-2.1% at quarter 2) but dissipates faster than rate shock

Key Finding 1: CRR More Potent Than Bank Rate

Key Finding 2: Delayed Inflation Response

Forecast Error Variance Decomposition (FEVD)

FEVD Interpretation: NRB policy instruments explain substantial medium-term variance in credit and inflation but remain modest contributors to GDP variance. This suggests real output primarily driven by external factors (remittances, Indian economic cycle, monsoon agricultural shocks)—consistent with small open economy characteristics.

Robustness Checks & Alternative Specifications

Tests Conducted

• Alternative Orderings: Swapped BR and CRR positions in Cholesky decomposition → Results robust to ordering
• Lag Length Sensitivity: Re-estimated with VAR(2) and VAR(6) → Qualitative findings unchanged
• COVID-19 Dummy: Included binary indicator for Q1 2020 onward → Policy effectiveness remains significant
• Sign Restrictions: Alternative identification using sign restrictions (Uhlig, 2005) → Confirms main results

Limitations & Caveats

Policy Implications & Recommendations

For Nepal Rastra Bank

Broader Implications for Developing Economies

Nepal's experience offers lessons for similar economies:

• Bank-dependent financial systems may benefit from quantity-based instruments over interest rate channels
• Exchange rate pegs complicate monetary transmission through import price effects
• Structural constraints limit monetary policy effectiveness—complementary reforms (financial deepening, fiscal coordination) essential

Research Impact & Dissemination

Publication Details

Policy Uptake

• NRB Internal Briefing: Presented findings to Monetary Policy Committee (March 2025)
• Integrated into Policy Framework: FEVD results cited in NRB's Monetary Policy 2025/26 white paper
• Training Materials: IRF charts incorporated into NRB staff training on monetary transmission

Presentation History

• Nepal Economic Association Annual Conference (December 2024) - Paper presentation
• NRB Research Department Seminar Series (February 2025) - Invited speaker

Technical Skills Demonstrated