India’s Macroeconomic Response to Global Shocks: Evidence from Oil Prices, Financial Crisis and COVID-19

Abstract

In past decades, the macroeconomic stability of India has been tested repeatedly by major global disruptions, including oil price shocks, the 2008 global financial crisis and the COVID-19 pandemic. Analysing how macroeconomic variables respond to these shocks is essential for evaluating external vulnerability and policy resilience in emerging economies. Our study provides a comprehensive empirical investigation of the dynamic responses of wholesale price inflation, industrial output, oil prices and exchange rates in India by employing monthly data from January 1993 to December 2024. To examine long-run equilibrium relationships along with short-run adjustment dynamics, the present study employs co-integration analysis within a Vector Error Correction Model (VECM) framework. Further, we applied impulse response functions and forecast error variance decomposition to track volatility spillover mechanisms. Quantile regression and ARCH–GARCH models were further estimated to account for distributional heterogeneity and time-varying volatility. The findings of our study suggested stable long-run linkages among the selected variables, where oil price shocks emerged as a key external source of macroeconomic fluctuations. Short-run dynamics suggested that shocks in oil prices are transmitted primarily through inflation and exchange rate channels and then affect industrial output. Distributional estimates revealed the effects were stronger during stress periods, indicating tail risks that were not captured by the mean-based models. Lastly, volatility analysis confirmed persistent clustering, especially during phases of crisis. Overall, the findings suggest that India’s macroeconomic system remains externally sensitive, with adjustment mechanisms that operate gradually but come under strain during global disruptions. These results underscore the importance of energy risk management and crisis-responsive macroeconomic stabilisation policies.

1. Introduction

India is among the fastest-growing economies across the globe and aims to achieve middle-income status by 2047 (Shanmugam & Odasseril, 2024).

As an Asian power, India has committed to achieving net-zero emissions by 2070 (World Bank, 2024). The economic health of any economy is assessed by the performance of its macroeconomic variables, with governments and policymakers striving to maintain stability (Magbondé et al., 2022; Bhardwaj et al., 2023), but global disruptions in the recent past in the form of COVID-19, the global crises of 2008 or oil price shocks have had a significant effect on the performance and functioning of global economies.

These global events have influenced macroeconomic variables such as inflation, money supply, the Index of Industrial Production (IIP), Foreign Direct Investment (FDI) and interest rates (Bhat et al., 2018).

The literature identifies macroeconomic indicators as crucial for achieving overall stability and evaluating economic functioning (Abbass et al., 2022). The present study investigates the effects of the 2008 financial crisis, the COVID-19 lockdowns, and oil prices on key economic indicators of India.

1.1. The Shock of 2008: A Macroeconomic Perspective

Increased bilateral trade and the formation of regional blocs have facilitated technological advancement while also introducing opportunities for volatility spillovers. During the 2008 financial crisis, the collapse of global financial institutions led to diminished cross-border trade, decreased industrial demand and elevated inflation rates (Bems et al., 2013; Board of Governors of the Federal Reserve System, 2009). Specifically, India’s GDP growth and industrial output declined by 3.1% and 2.4%, respectively, and the economy saw inflation exceeding 8% due to rising oil prices, which was followed by significant depreciation of the rupee (Singh, 2009; State Bank of India, 2018). The economy underwent substantial macroeconomic adjustments in response to external shocks, and due to stimulus measures, the central government deficit rose to 6% of GDP (Bajpai, 2011).

On the monetary policy front, the Reserve Bank of India reduced the repo rate from 9% to 4.7% by April 2009, maintaining a CRR of 5% and an SLR of 24% to enhance liquidity (Islam & Rajan, 2009; R. Kumar & Vashisht, 2009).

According to Government of India (2009) Economic Survey, merchandise exports amounted to US$168.7 billion in 2008–09, reflecting a 3.6% year-on-year growth, while imports increased by 14.3% to US$287.8 billion, thereby exacerbating trade imbalances. These developments highlight that India’s macroeconomic indicators are responsive and sensitive to global disturbances.

Table 1. Key macroeconomic indicators during the 2008–09 global financial crisis (India).

Macroeconomic Indicators	Description (%)
GDP Growth Rate (%)	6.7%
IIP Growth Rate (%)	2.6%
WPI Inflation (%)	8.4%
Exchange Rate (INR/USD)	47.5%

Sources: Economic Survey of India (2008–2009), (MOSPI), Office of the Economic Adviser, Ministry of Commerce and Industry, India.

below showcases the performance of macroeconomic indicators at the time of the global financial crisis of 2008.

As observed in Table 1, GDP growth was moderate and satisfactory, while the industrial sector witnessed a significant decline, with growth declining to about 2.4% and inflationary pressures rising to 8.4%. Further, the depreciation of the rupee exacerbated the external balance and raised the cost of imports. This is illustrated in Figure 1, where the GDP growth rate fell to roughly 2.4% during the 2008–2009 period.

These disruptions highlight how external shocks can destabilise India’s industrial production, inflation and value of currency as soon as the economy is exposed to global volatility. Further, movements in oil prices have emerged as a crucial transmission channel of volatility (Petroleum Planning & Analysis Cell, 2025; Kundu et al., 2025). Given India’s reliance on imported crude oil, any volatility in the global oil markets can have an effect on inflation, exchange rates and growth of the Indian economy, hence making it imperative to thoroughly examine their macroeconomic implications (Deheri & Sahu, 2024; Sreenu, 2018).

1.2. Oil–Macroeconomy Relationship

According to Ahmad et al. (2022) and Alawadhi and Longe (2024) for developing economies, crude oil prices are a key determinant to economic activities, serving as essential components for economic development. Variations in oil prices are largely influenced by structural shifts in supply and global factors, further leading to either a deficit or surplus in the balance of payments (Brini et al., 2016). India, a major oil importer, is significantly dependent on imports for its industrial sectors. Oil price shocks have uneven impacts on economies and their macroeconomic variables such as inflation, exchange rate, and unemployment; this is particularly evident for those economies that are major importers of oil (Zulfigarov & Neuenkirch, 2020), meaning where price increases tend to dampen output, while price decreases do not necessarily boost growth. Hamilton (1983, 2003) indicates that oil-importing economies frequently experience weak macroeconomic performance and financial instability. Tiwari et al. (2022) and P. Sharma and Shrivastava (2021) illustrate that rising oil prices impact the macroeconomic stability of both developing and developed nations. In 2008, Brent crude oil reached a peak of USD 147 per barrel in July, contributing to India’s current account deficit, which rose to 11% of GDP. In contrast, during the COVID-19 pandemic, prices dropped to USD 20 per barrel in April 2020, which helped lower inflation and ease the current account deficit (IMF, 2021). These contrasting trends are evident in Figure 2.

Figure 2 depicts notable fluctuations in Brent crude oil prices from 2000 to 2024. As observed, prices notably surged before the 2008 financial crisis and then sharply declined during the crises period (2008–2009), followed by a huge dip in 2020 due to the COVID-19 pandemic. Recently, prices have stabilised between USD 70 and 90 per barrel, highlighting ongoing global economic uncertainty. It is crucial for policymakers to bolster the resilience of economies, especially emerging ones, against oil shocks (Antonakakis et al., 2017; Hailemariam et al., 2019; Hajebi et al., 2022).

1.3. COVID-19 and Macroeconomic Fluctuations

The Indian economy, recognised as one of the world’s fastest-growing economies, encountered significant challenges during the COVID-19 pandemic in 2020, affecting every sector (Khalid, 2021). The pandemic essentially brought the entire economy to a halt (Jawad & Naz, 2023; Jawad et al., 2021), causing disruptions in trade, industry, tourism and agriculture. Beyond the health implications, the pandemic had serious implications for macroeconomic fundamentals (Jawad & Naz, 2023). According to the Economics Observatory, India’s GDP contracted by 24.4% from April to June 2020 due to nationwide lockdowns, whereas national income figures for 2021–2022 indicated a 7.4% contraction in the second quarter (July–September 2020). Although there was a modest recovery in subsequent quarters, with GDP increasing by 0.5% and 1.6%, the overall decline for 2020–2021 was −7.3%. The second wave in 2021 further affected key macroeconomic indicators, including oil prices and exchange rates (Bhama, 2022), and prolonged lockdowns further influenced inflation, unemployment rates and GDP rates (Khalid, 2021). In response to these crises, the government launched the “Atmanirbhar Bharat” program, introducing a ₹20 lakh crore stimulus package, equivalent to 10% of GDP, to stimulate demand and support sectors like MSMEs, agriculture and vulnerable populations (Ministry of Finance, 2020), hence reviving the economy. Figure 3 below demonstrates the downswing of the Indian economy during crises.

As showcased in Figure 3, India witnessed a severe economic contraction during the COVID-19 pandemic, with a downswing more pronounced than what was observed in many other nations. This disturbance exposed India’s structural vulnerabilities, making the economy more susceptible to global shocks through channels such as depreciation of currency, widening trade imbalances, mounting inflationary pressures and supply-side constraints (IMF, 2025; World Bank, 2024; Eichengreen & Gupta, 2024). According to Chauhan et al. (2025), key macroeconomic variables such as the Index of Industrial Production, wholesale price inflation and exchange rates become key parameters of the economic health of an economy; therefore, examining these indicators is crucial for formulating effective policies.

While a large body of literature has examined the effects of oil price shocks and global crises on India, most studies have analysed these episodes over a short time span or without a unified framework. The present study provides a detailed analysis of India’s macroeconomic adjustment to external shocks and global crises by employing advanced econometric models. The contribution of our study is not just the methodological innovation but also the country-specific evidence and extended dataset covering major global crises in past decades. Additionally, the employment of a long monthly dataset enabled us to assess whether macroeconomic responses remained stable or changed under extreme conditions. By combining VECM-based impulse responses and variance decompositions with quantile regression and volatility models, our study demonstrates how mean-based estimates mask substantial heterogeneity in the distribution tails, specifically at times of severe crisis episodes. While doing so, it provides India-specific evidence on how core macroeconomic variables interact over time and across regimes. This study offers insights that directly serve macroeconomic stabilisation, energy policy and preparedness for crises.

Motivated by recent disruptions in the global economy and their potential effects on emerging markets such as India, this study sets two core objectives:

• To analyse the asymmetric and volatility-related responses of India’s key macroeconomic indicators (industrial production, inflation, and exchange rate) to major global disruptions: the 2008 global financial crisis, the COVID-19 pandemic and persistent oil price shocks.
• To derive policy implications aimed at enhancing India’s macroeconomic resilience and strategic preparedness against future global disruptions.

2. Methodology

2.1. Data and Sources

The present study relies on monthly data spanning from 1993 to 2024. This period captures the ripple effects of three key global disruptions: the 2008 financial crisis, the COVID-19 outbreak and recent oil price fluctuations. By employing this timeline, we were able to trace how India’s macroeconomic variables responded across different phases of volatility. All selected series are captured on a monthly basis and compiled from official sources (Reserve Bank of India, MOSPI, U.S. Energy Information Administration (EIA)). Base year (2011–12 = 100) was considered for Industrial Production (IIP) and Wholesale Price Index (WPI). Oil prices were evaluated using Brent crude prices (USD per barrel), and the exchange rate was assessed as INR per USD. Prior to estimation, all variables were transformed into natural logarithms. Unit root test for stationarity was conducted on first differences of logged series.

Seasonal adjustment was calculated where required using standard filtering procedures. Short data gaps (up to two consecutive months) were interpolated linearly, while longer gaps were excluded from the sample. The sample time period for analysis after preprocessing and alignment to a consistent monthly frequency was January 1993 to December 2024.

2.2. Data and Variable Construction

The analysis employed monthly data for selected key macroeconomic determinants: the Index of Industrial Production (IIP), Wholesale Price Index (WPI), nominal exchange rate (INR/USD) and Brent crude oil prices (US$/barrel). Our study updated historical series to the latest base year. Prior to transformation and estimation, we seasonally adjusted IIP and WPI series using the X-13 ARIMA-SEATS. Further, all series were converted into natural logarithms.

• Index of Industrial Production (IIP): Δlog(IIP), a proxy for industrial performance and overall economic activity.
• Wholesale Price Index (WPI): Δlog(WPI), represents inflationary trends and cost pressures in the economy.
• Exchange Rate (USD/INR): Δlog(INR/USD), captures external sector sensitivity and currency volatility.
• International Crude Oil Prices (Brent): Δlog(Brent), serves as the external shock variable, given India’s dependence on oil imports.

Additionally, we measured the Index of Industrial Production (IIP) with the official base year as reported by MOSPI. The Wholesale Price Index (WPI) was expressed using the latest available base revisions. The exchange rate was measured as the nominal INR per US dollar, and lastly, crude oil prices were represented by Brent benchmark prices expressed in US dollars per barrel. For robustness, oil prices were expressed in rupee terms as a log (Brent × INR/USD) to account for the effects of oil prices and exchange rates on the domestic economy.

2.3. Data Cleaning and Crisis Controls

This step deals with handling missing values conservatively due to employment of extensive data. For gaps that are short (two months or less), we interpolated linearly. Longer gaps were dropped to maintain time-series properties. As a robustness check, extreme observations were winsorised at the 1 percent tails, and key models were re-estimated, excluding extreme crisis months. The results remained unchanged, highlighting that the findings were not driven by outliers or data artefacts.

For the global financial crises and COVID-19 crisis, dummies were included as exogenous regressors in the sensitivity checks, although the main analysis relied on the structural identification of oil shocks and formal break tests.

2.4. Pre-Testing and Structural Breaks

We employed ADF and KPSS unit root tests to determine the integration order of data properties. A structural break analysis was conducted using the Zivot–Andrews tests for single breaks and Bai–Perron tests for multiple breaks, with break dates evaluated against the timelines of the global financial crises and COVID-19. Gregory–Hansen tests were utilised as a robustness check for co-integration in the presence of structural breaks.

2.5. Sources of Data Collection

Data on macroeconomic variables such as IIP, WPI and exchange rates were sourced from the RBI and MOSPI, while crude oil price data were obtained from the EIA.

2.6. Econometric Techniques

This section highlights the empirical framework employed to examine distributional and volatility responses to global crises. The empirical framework is structured around the VECM framework to capture long-run equilibrium and short-run adjustment among oil prices, exchange rates, inflation and industrial output. Impulse response functions and variance decompositions are derived from this model to trace transmission mechanisms and assess the importance of shocks. Further, we applied quantile regression to analyse distributional heterogeneity that cannot be captured through mean-based models, specifically during crisis periods when tail risks become economically relevant. Further, ARCH and GARCH models were used separately to examine volatility persistence and clustering rather than as substitutes for the mean dynamics. This structured approach assigned each technique a distinct analytical role, thereby avoiding redundancy while maintaining internal consistency across the empirical results. Lastly, to preserve clarity, diagnostic tests, structural break analysis and robustness checks were reported selectively, with supporting outputs presented in a condensed form.

2.7. Stationarity and Integration Tests

The analysis begins by testing the time-series properties of the selected variables: Index of Industrial Production (IIP), Wholesale Price Index (WPI), exchange rates and global oil prices. Since meaningful econometric modelling requires stationarity, we employ a battery of tests: augmented Dickey–Fuller (ADF) and Kwiatkowski–Phillips–Schmidt–Shin (KPSS).

If the variables are found to be integrated of order one [I(1)], we proceed to the Johansen co-integration framework. The existence of a co-integrating relationship implies a long-run equilibrium between macroeconomic indicators and oil prices.

2.8. Long-Run and Short-Run Dynamics: VECM and ECM

When co-integration is confirmed, we estimate the Vector Error Correction Model (VECM). Before applying co-integration, test lag length criteria were determined by employing AIC, SIC and HQ criteria. The optimal lag length of 2 was selected for further analysis. We estimated the Johansen co-integration test under a specification including an unrestricted intercept and no deterministic trend in the VAR. To allow for mean adjustment, the constant term was restricted to co-integrating space without incorporating deterministic linear trends in the system.

Next, to capture how quickly the deviations from the long-run equilibrium are corrected after a shock, we employed ECM. A negative and statistically significant ECT indicates ineffective adjustment. Model adequacy is assessed through diagnostics, including serial correlation (LM), normality (Jarque–Bera), heteroskedasticity (ARCH–LM) and stability (inverse AR polynomial roots).

The generic form of the error correction model is:

Δy_t = α + βΔx_t + λ(y_t−1 − γx_t−1) + ε_t

(1)

• Δy_t is the short-term change in a dependent variable such as IIP;
• Δx_t is the change in independent variables (e.g., oil prices);
• λ is the speed of the adjustment;
• (y_t−1 − γx_t−1) is the long-run co-integrating relationship.

This structure allows us to examine both long-term co-integration across variables and short-term deviations triggered by global shocks such as the 2008 Global Financial Crisis (GFC) and the COVID-19 pandemic. Impulse response functions and forecast error variance decompositions were obtained from the estimated VECM model to assess the shock transmission dynamics and their relative contribution to macroeconomic fluctuations. Further, IRFs and FEVD were computed using a Cholesky decomposition of the variance–covariance matrix. The variables were ordered according to an external-to-internal transmission mechanism, placing oil prices first, followed by the exchange rate, inflation and industrial output. This ordering reflects that, within the monthly horizon, global oil prices are exogenous to the Indian economy. Further, due to import bill and external balance effects, exchange rates are allowed to respond quickly to movements in oil price. Inflation is influenced by both oil and exchange rate movements. On the other hand, industrial production is considered an endogenous variable, responding with a lag to external and price shocks. Within this identification framework, global oil price shocks can immediately influence domestic macroeconomic variables, but those domestic variables do not contemporaneously affect global oil prices during the same period. Robustness checks using alternative orderings were assessed, and it was found that they do not materially alter the qualitative impulse response patterns.

2.9. Distributional Effects: Quantile Regression

To account for heterogeneity across the outcome distributions, we applied quantile regression for industrial growth and inflation. To capture heterogeneous responses, we employed quantile regression (Koenker & Bassett, 1978). This method estimates the relationships at different points of the conditional distribution rather than focusing only on the mean.

The model is specified as

Qτ(y_t | x_t) = x_t′βτ

(2)

where Qτ(y_t | x_t) is the conditional quantile τ (e.g., 0.10, 0.50, 0.90) of the dependent variable y_t (IIP, WPI or exchange rate) given covariates x_t (oil prices, global shocks). By estimating these quantiles, we could distinguish between weak economic conditions, normal periods and stress episodes without overfitting extreme observations.

Standard errors were obtained through bootstrapping, and the Koenker–Xiao test confirmed significant slope heterogeneity across quantiles. The results revealed that oil price shocks and global crises exerted disproportionately stronger effects at the lower and upper tails of the distribution. This helped us in capturing severe output contractions and inflation surges that are observed in mean-based models. By focusing on economically meaningful quantiles, this framework differentiates normal adjustment dynamics from crisis-driven responses characterised by supply disruptions, exchange rate pressures and cost-push inflation.

2.10. Structural Stability and Break Analysis

To ensure the estimated relationships remain robust across crisis and non-crisis periods, thus avoiding spurious inferences, we employed CUSUM and CUSUMSQ tests so as to detect gradual drifts or sudden shifts in the estimated coefficients.

Additionally, the Bai–Perron multiple break was employed to help identify structural shifts associated with major episodes such as the 2008 global financial crisis and the COVID-19 pandemic.

2.11. Volatility Modelling: ARCH–GARCH Family

Macroeconomic and financial series often display volatility clustering, such as periods of high fluctuations and calm intervals. To capture this, we employed the ARCH (Engle, 1982) and GARCH (Bollerslev, 1986) models. The conditional variance equation for ARCH(1) is as follows:

σ²_t = α₀ + α₁ε²_t−1

and, for GARCH(1,1), is as follows:

σ²_t = α₀ + α₁ε²_t−1 + β₁σ²_t−1

(3)

where σ²_t denotes conditional variance at time t, ε²_t−1 is the lagged squared residual, and β₁ captures persistence in volatility. The GARCH(1,1) specification has been adopted as a benchmark that captures volatility clustering and persistence in macroeconomic and financial time series. Preliminary comparisons were conducted using information criteria, which indicated no systematic improvement from asymmetric alternatives, and the analysis, therefore, focused on assessing volatility persistence rather than leverage effects. These choices confirmed comparability across variables while avoiding over-parameterisation.

For univariate series, we estimated GARCH(1,1) models with Student-t innovations to capture fat tails, which allowed us to track co-volatility through time and link spikes in correlation to global shocks.

2.12. Diagnostic and Robustness Checks

All the models were validated using a comprehensive set of diagnostics. Residual serial correlation was checked using LM tests, heteroskedasticity with ARCH–LM and normality with Jarque–Bera. The stability was confirmed using inverse AR roots and recursive estimations. Koenker–Xiao slope tests were employed to validate distributional heterogeneity of quantile regression.

Overall, this multi-layered approach enabled a detailed assessment of the dynamic linkages between oil prices, financial crises and India’s macroeconomic variables. These were captured in both the mean and variance dimensions while incorporating structural breaks and distributional asymmetries.

3. Results and Discussions

The results section is framed around VECM as the core framework, with distributional and volatility analysis being employed as complementary tools for interpreting the underlying transmission mechanisms.

This study applies descriptive statistics in

Table 2. Descriptive statistics of key macroeconomic variables.

Variables	Mean	Standard Deviations	Minimum	Maximum	Skewness	Kurtosis (Excess)
IIP	91.83	37.46	28.03	160.0	−0.22	−1.36
Exchange rate	55.46	15.52	30.83	83.87	0.17	−1.27
WPI	91.26	36.47	33.14	155.4	0.01	−1.28
Oil	54.97	28.34	11.35	133.88	0.29	−0.86

Source: Author’s calculations, Note: Kurtosis values represent excess kurtosis (normal distribution = 0).

, offering insights into the behaviour of macroeconomic variables. The results highlight oil prices exhibiting the highest standard deviation at 28.34, with values ranging from $11.35 to $133.88 per barrel. These values capture events such as the 2008 spike and the COVID-19 crash in 2020. Moderate variability is seen in exchange rate data, with a mean of ₹55.46 and a standard deviation of 15.52. These values reflect shocks like the 2013 taper tantrum.

While IIP and WPI appear more stable, the negative skewness of IIP (−0.22) suggests longer industrial slowdowns than compared to booms. WPI patterns reveal the asymmetric effects of the inflation shocks. These patterns, as noted in earlier studies (e.g., P. Kumar & Singh, 2022), support the use of both symmetric and asymmetric modelling frameworks.

Before engaging in time-series modelling, it is crucial to assess the stationarity of the variables. The ADF test results in

Table 3. Augmented Dickey–Fuller (ADF) unit root test.

Original Level			First Difference
Variables	p-Value	Results	p-Value	Results
IIP	0.8	Non-stationary	0.00	Stationary
Exchange rate	0.9	Non-stationary	0.00	Stationary
WPI	0.9	Non-stationary	0.00	Stationary
Oil Price	0.2	Non-stationary	0.00	Stationary

Source: Authors calculation.

indicate that none of the four macroeconomic indicators is stationary in their level forms, as evidenced by p-values exceeding 0.05. However, after the first differencing, each series achieves stationarity, with a p-value of 0.00, confirming that they are integrated of order one, I(1). The KPSS test was applied in

Table 4. KPSS unit root test results (null hypothesis: stationarity).

	At Level			First Difference
Variables	Level-Statistic	Critical Value (5%)	Results	First Difference	5% Critical Value	Results
IIP	0.82	0.46	Non-stationary	0.14	0.46	Stationary
Exchange rate	0.64	0.46	Non-stationary	0.19	0.46	Stationary
WPI	0.91	0.46	Non-stationary	0.15	0.46	Stationary
Oil Prices	0.57	0.46	Non-stationary	0.12	0.46	Stationary

Source: Author’s calculation.

with the null hypothesis if stationarity provided supportive evidence. At level, the KPSS test concluded with rejecting the null hypothesis, as KPSS statistics exceeded the 5% critical value. However, after first differencing, series were found to be stationary, as the KPSS statistics were less than the corresponding critical values. Together with the ADF results, it was confirmed that all selected variables are first order stationary, i.e., integrated of order one, I(1). The results confirm the application of the Johansen co-integration framework. P. Sharma and Shrivastava (2021) and Upadhyaya et al. (2023) observed similar I(1) behaviour for the exchange rate, inflation and industrial output. These findings have both methodological and economic implications, supporting the application of co-integration analysis, vector autoregressive and ECM frameworks in subsequent sections, as endorsed by S. Rizvi et al. (2022).

To confirm our results were not influenced by data issues or outliers, we again conducted the analysis. This time, effort was made to interpolate short gaps, excluding longer missing data and winsorizing at the 1% tails. The results for unit root, co-integration and other models remained consistent, confirming the robustness of the results. For the Johansen co-integration analysis, we determined the optimal lag length, as showcased in

Table 5. Lag length selection criteria.

Lag	AIC	SIC	HQIC
0	−2.45	−2.20	−2.33
1	−5.92	−5.40	−5.70
2 *	−6.20 *	−5.55 *	−5.90 *
3	−6.15	−5.30	−5.75
4	−6.10	−5.00	−5.60

Source: Author’s calculation. NOTE: * denotes selection of lag length.

, using the following standard information criteria: Akaike Information Criterion (AIC), Schwarz Information Criterion (SIC), and Hannan–Quinn Information Criterion (HQIC). AIC criteria indicated that a lag of 2 was optimal, which we used in the Johansen co-integration tests to capture the relevant dynamics without over-parameterisation.

Before proceeding to the co-integration analysis, we examine whether structural breaks affected the time-series properties of the variables. Given that crises can alter the data-generating process, we apply the Zivot–Andrews and Bai–Perron tests endogenously. These tests help ensure that the long-run relationships are not biased by regime shifts during episodes such as the global financial crisis, the taper tantrum and COVID-19.

The Zivot–Andrews test, with results shown in

Table 6. Zivot–Andrews test and Bai–perron test.

Variable	Zivot–Andrews	Test Statistic	Critical Values (5%)	Break Specification	Bai–Perron Break Dates
IIP	2009 (Global financial crises)	−5.12	−4.80	Level and Trend	2009, 2020
Exchange rate	2013 (Taper tantrum)	−5.34	−4.82	Level Shift	2008, 2013, 2020
WPI	2008 (Global financial crises)	−5.01	−4.78	Trend Shift	2008, 2014, 2020
Oil Prices	2020 (COVID-19 pandemic)	−5.80	−4.85	Level and Trend	2008, 2014, 2020

Source: Author’s calculation.

, detected significant structural breaks in the IIP (2009), WPI (2008), exchange rate (2013) and oil prices (2020), consistent with the timing of global crises. Bai–Perron multiple-break tests confirmed additional breakpoints in 2008–09 (global financial crisis), 2013 (taper tantrum episode) and 2020 (the COVID-19 pandemic). These results validate that crisis episodes materially altered the data-generating process, yet the co-integration and VAR estimates remain robust when accounting for regime shifts.

Given that the Zivot–Andrews test permits an endogenous structural break in the unit root testing framework, the integration properties of the variables were re-evaluated under break-adjusted models. The findings supported I(1) behaviour, hence reinforcing that, even after accounting for structural shifts, the order of integration remained unchanged. Next, we explore long-run equilibrium relationships using a co-integration test.

Table 7. Johansen co-integration test.

Hypothesised Number of Co-Integrated Equations	Trace Stats	(95%) Critical Values (Trace)	Max-Eigen Stats	5% Critical Value (Max-Eigen)
None *	89.00	55.25	35.00	35.01
At most 1 *	54.00	35.01	28.91	24.25
At most 2	25.08	18.4	14.50	12.32
At most 3	10.57	3.84	10.57	3.84

Source: Author’s calculation. NOTE: * denotes co-integration.

presents all variables with p-values below 0.05, indicating valid co-integrating links. For the null hypothesis of no co-integration (r = 0), both the trace statistic (89.00) and the max-eigen statistic (35.00) exceed their 95% critical values (55.25 and 35.01), leading to the rejection of the null hypothesis and confirming at least one long-run relationship. Oil price and exchange rate exhibit strong connections, suggesting that India’s external sector dynamics are tied to global commodity movements. This finding is consistent with Alam et al. (2020), who identified causality between oil prices and exchange rates, and Sultan et al. (2020), who established a co-integration between oil price, inflation, and exchange rate. These results support the use of models that accommodate long-run linkages, such as the ECM.

In addition to the trace statistics, the maximum eigenvalues highlight and confirm the presence of at least one stable long-run co-integrating relationship among the variables, which is consistent with the selected lag structure. To ensure that the long-run relationship among the variables is not biased by regime shifts such as the 2008 financial crisis or the COVID-19 pandemic, we apply the Gregory–Hansen test, which allows for a single endogenous structural break in the co-integration relationship (Guliyev, 2022). The Gregory–Hansen results, as depicted in

Table 8. Gregory–Hansen test for co-integration with structural breaks.

Test	Value	5% Critical Value	Estimated Break Date
ADF *	−6.12	−5.45	April 2020
Phillips Zt statistic	−6.05	−5.23	September 2008
Phillips Za statistic	−64.8	−58.4	March 2020

Source: Author’s calculation. Notes: The null hypothesis of no co-integration with a structural break will be accepted or rejected on comparison with test statistic with corresponding critical value and denoted by *.

, confirm the existence of long-run co-integration among oil prices, exchange rates, industrial production and inflation, even when structural breaks are allowed. The test is implemented under the level-shift with trend specification, which permits an endogenous structural break in both the intercept and deterministic trend of the co-integrating relationship.

The breakpoints align with known crisis periods (2008 and 2020), validating the robustness of our long-run findings.

To capture the dynamic transmission of global shocks to India’s macroeconomic variables, we estimate impulse response functions from the VECM.

Table 9. Vector Error Correction Model (VECM) estimates.

Dependent Variable	Variables	Coefficient	Std. Error	T-Stat	p-Value
ΔIIP	L1(ΔOP)	−0.054	0.021	−2.57	0.011
ΔIIP	ECT(−1)	−0.132	0.045	−2.93	0.004
ΔER	L1(ΔOP)	0.031	0.011	2.82	0.005
ΔWPI	L1(ΔOP)	0.047	0.018	2.61	0.010
ΔWPI	ECT(−1)	−0.058	0.031	−1.87	0.062

Source: Author’s calculations.

below highlights the short-run dynamics from the VECM analysis. Findings observed error term is negative and statistically significant (−0.132, p = 0.004) for the industrial production equation, indicating approximately 13% of last year dis-equilibrium being rectified each month, showcasing convergence towards equilibrium.

Further, in the short run, changes in oil price are seen to exert a negative and statistically significant impact on industrial output (−0.054, p = 0.011), consistent with cost-push pressures. Oil prices also demonstrate a positive effect on wholesale inflation (0.047, p = 0.010), indicating inflationary pass-through, whereas for exchange rate, weaker and statistically less robust effects are observed, demonstrating that most of the changes occur through the long-run error correction mechanism instead of immediate short-run feedback (Sahu et al., 2014; P. Kumar & Singh, 2022; Upadhyaya et al., 2023; Ahmed & Kaur, 2025). The following tables are interpreted jointly with the impulse response functions to emphasise the dynamic economic significance rather than isolated coefficient magnitudes.

Impulse responses are evaluated using the Cholesky identification scheme and reflect the dynamic effects of a one-standard-deviation shock.

Table 10. Impulse response of (IIP) to a one-standard-deviation oil price shock.

Horizon	% Change in IIP	Lower Bound	Upper Bound
1	−0.18	−0.29	−0.07
3	−0.35	−0.52	−0.18
6	−0.28	−0.44	−0.12
12	−0.10	−0.22	0.02

Source: Author’s calculation. NOTE: IRF of IIP to one SD oil price shock with 95% confidence interval based on 1000 replications.

,

Table 11. Impulse response of Wholesale Price Inflation (WPI) to a one-standard-deviation oil price shock.

Horizon	% Change in WPI	Lower Bound	Upper Bound
1	0.10	0.03	0.17
3	0.25	0.12	0.38
6	0.19	0.08	0.30
12	0.07	−0.02	0.16

Source: Author’s calculation. NOTE: IRF of WPI to one SD oil price shock with 95% confidence interval based on 1000 replications.

,

Table 12. Impulse response of the exchange rate (INR/USD) to a one-standard-deviation oil price shock.

Horizon	% Change in Exchange Rate	Lower Bound	Upper Bound
1	0.40	0.22	0.58
3	0.80	0.55	1.05
6	0.55	0.32	0.78
12	0.20	−0.05	0.45

Source: Author’s calculation. NOTE: IRF of exchange rate to one SD oil price shock with 95% confidence interval based on 1000 replications.

and

Table 13. Impulse response of Industrial Production (IIP) to a one-standard-deviation exchange rate shock.

Horizon	% Change in IIP	Lower Bound	Upper Bound
1	−0.06	−0.12	0.00
3	−0.14	−0.24	−0.04
6	−0.11	−0.20	−0.02
12	−0.04	−0.12	0.04

Source: Author’s calculation. NOTE: IRF of IIP to one SD exchange rate shock with 95% confidence interval based on 1000 replications.

highlight how IRFs trace the response of Industrial Production (IIP), Exchange Rate (ER) and inflation (WPI) to a one-standard-deviation innovation in Oil Prices (OP). Shocks in oil prices are quick enough to reduce industrial output (peak ≈ −0.35% at three months) and raise wholesale prices (peak ≈ +0.25% at three months). Oil shocks also depreciate the INR (peak ≈ +0.8% at three months). A pure exchange rate-depreciation shock further dampens IIP modestly, consistent with the cost-push and imported-input channels. The results are supported by studies conducted by V. Sharma et al. (2024).

As witnessed, some short-run coefficients in the VECM appeared numerically small, but their economic relevance lied in their cumulative and dynamic effects. Monthly macroeconomic adjustments typically evolved gradually, with small coefficients accumulating over time through the error correction mechanism and impulse responses. The impulse response functions further provided a clearer economic interpretation by tracing these cumulative effects. For instance, a (−0.14%) response of industrial production to an exchange rate shock meant a persistent decline in output for subsequent months. This reflected higher cost of inputs, which are imported, and reduced investment confidence. These dynamic responses are economically meaningful when evaluated over the full adjustment horizon rather than at a single point estimate.

We have applied forecast error variance decomposition to assess the relative importance of different shocks beyond the impulse responses to get more insightful results. Over a time frame of 1–2 years, oil price shocks have resulted in a major share of the forecast error variance in industrial production (approx. 28%); for inflation (wholesale), it stood around 35%, while exchange rate shocks held a relatively smaller share, explaining the variance around 15% in IIP and 12% in WPI in a time frame of a 24-month horizon. Oil price shocks attributed nearly 32% of exchange rate variance, demonstrating pressures caused by higher import costs and current account dynamics while explaining about 29% of the variability in inflation (Deheri & Sahu, 2024). While the direction of these responses is broadly consistent with earlier evidence, the magnitude and persistence of the effects vary substantially across crisis regimes, with markedly stronger transmission observed during periods of global stress.

Table 14. Forecast Error Variance Decomposition (FEVD) of Industrial Production (IIP).

Horizon	IIP	Oil	Exchange Rate	WPI
1	82	10	5	3
3	60	18	12	10
6	52	22	15	11
12	47	28	15	10

Source: Author’s calculation.

,

Table 15. Forecast Error Variance Decomposition (FEVD) of WPI.

Horizon	WPI	Oil	Exchange Rate	IIP
1	70	18	7	5
3	52	30	10	8
6	46	34	11	9
12	43	35	12	10

Source: Author’s calculation.

and

Table 16. Forecast Error Variance Decomposition (FEVD) of exchange rate (INR/USD).

Horizon	Exchange Rate	Oil	WPI	IIP
1	78	12	6	4
3	64	18	10	8
6	60	20	11	9
12	58	20	12	10

Source: Author’s calculation.

summarise the econometric results and are interpreted jointly with the accompanying discussion to emphasise economic relevance rather than isolated coefficient magnitudes.

In Figure 4 below, we illustrate a detailed perspective by employing quantile regression. The Koenker–Xiao quantile-process test concludes significant slope heterogeneity across quantiles, validating this approach. Oil prices reveal a statistically significant impact (coefficient = 6.83, p < 0.05), with stronger effects observed at higher quantiles of inflation and IIP volatility, indicating their dominant role in driving macroeconomic instability during extreme shocks. The conclusions align with those stated by S. K. A. Rizvi and Itani (2021) and Su et al. (2016). Further, the IIP and WPI reflect weak coefficients, suggesting indirect effects through oil-linked channels, whereas exchange rate had a moderate effect as a transmission channel for external shocks. These results suggest the use of precautionary policy tools at times of external volatility.

The results in

Table 17. Quantile regression coefficients across different quantiles.

Variable	0.10 Quantile	0.50 Quantile	0.90 Quantile
Oil	−0.15	−0.08	−0.27
Exchange rate	−0.34	−0.58	−0.37
WPI	0.97	1.35	0.96

Source: Author’s calculation.

report oil shocks to be negative, with stronger effects in the upper quantiles. Exchange rate impacts are also found to be negative, strongest at the median, while WPI is confirmed to be positive across all quantiles, with a peak at the median. The quantile regression results state that shocks in oil price exert disproportionately stronger effects at the upper quantiles of inflation and output volatility. These results state that, during extreme stress episodes such as the global financial crisis and the COVID-19 pandemic, oil price shocks intensify due to supply-side rigidities, exchange rate depreciation and delayed policy transmission.

In contrast, muted effects were observed at lower quantiles, reflecting greater macroeconomic stability during normal periods. These findings concludes that mean-based models mask economically meaningful tail risks, hence understanding vulnerabilities in periods of crisis, as shocks in oil price intensify disproportionately during phases of heightened stress. Conventional mean-based models often fail to capture these dynamics.

Robustness was ensured by incorporating policy dummies for the global financial crisis (2008–09) and the COVID-19 pandemic (2020–21). Their inclusion did not alter the baseline estimates, hence confirming that quantile effects were not influenced by the crisis periods. Bootstrapped standard errors were used to validate their significance.

The null hypothesis of equal coefficients across the 0.10, 0.50, and 0.90 quantiles (p < 0.05) was rejected by the Koenker–Xio slope test, suggesting that oil and exchange rate shocks have heterogeneous impacts on macroeconomic outcomes. The stronger coefficients were witnessed at the upper quantiles, reflecting phases of rigidities in supply-side, depreciation of exchange rate and delays in policy transmission. All these amplified the pass-through of oil price shocks to inflation and output. In contrast, the relatively muted responses around the median reflect more effective adjustment under normal macroeconomic conditions. This asymmetry explains why mean-based models understate risks during episodes of crisis.

The ARCH–LM test was applied to examine autoregressive conditional heteroskedasticity in residuals.

Table 18. ARCH–LM test results for conditional heteroskedasticity.

Variable	LM Statistic	p-Value	Conclusion
IIP	23.38	0.00	Presence of volatility
Exchange rate	41.92	0.00	Presence of volatility
WPI	35.75	0.00	Presence of volatility
Oil Price	42.36	0.00	Presence of volatility

Source: Author calculation.

indicates significant ARCH effects in the oil price and exchange rate equations, suggesting that volatility depends on the previous shock patterns. These results justify the use of GARCH models to capture dynamic variance structures, as reported in prior studies by Mahalwala (2022) and S. K. A. Rizvi and Itani (2021). For India, with its high energy import dependence, ARCH behaviour indicates lingering shock effects, underscoring the need for price stabilisation buffers during volatile periods. The ARCH–LM results provide a basis for applying GARCH models to capture time-varying volatility in India’s response to global shocks.

The GARCH(1,1) results presented in

Table 19. GARCH(1,1) estimates of volatility persistence.

Variable	α (ARCH)	β GARCH Term	α + β	ν (df)	Log-Likelihood
IIP	0.21	0.71	0.93	7.84	−612.45
Exchange Rate	0.18	0.75	0.94	6.52	−489.32
WPI	0.25	0.58	0.93	8.21	−530.87
Oil Price	0.29	0.62	0.92	6.93	−721.66

Source: Author calculation.

confirm the presence of volatility clustering across all variables, in line with the ARCH–LM findings. The sum of α and β is close to unity (0.92–0.94), suggesting high persistence of volatility, where shocks in variables have prolonged effects. These results align with the findings of Bhatia and Gupta (2021) and Maharana et al. (2024). The results present that oil prices showcase the strongest ARCH effect, with α = 0.30, indicating immediate and sharp reactions to global shocks, whereas the exchange rate and IIP displayed prolonged volatility persistence, as highlighted by the GARCH term.

These results are consistent with this study’s objectives, as they reinforce the vulnerability of India’s macroeconomic variables to external disruptions.

As evident in the results shown in Table 19, the ARCH and GARCH coefficients are positive and statistically significant across all series. Volatility persistence remained quite high, with α + β equal to 0.93 for IIP, 0.94 for exchange rate, 0.83 for inflation (WPI) and 0.92 for oil prices, reflecting gradual decay of shocks. The estimated Student-t degrees of freedom ranged between 6.5 and 8.2, supporting fat-tailed residuals. Further, it can be concluded that Ljung–Box and ARCH–LM tests do not detect remaining serial correlation or ARCH effects, hence indicating that the GARCH(1,1) specification captured conditional heteroskedasticity.

Further, we have employed cumulative sum control to examine the temporal stability of the model coefficients through the CUSUM test. The cumulative sum of recursive residuals was plotted against critical bounds at a 95% confidence level (Bhattacharjee & Das, 2021). The CUSUM plots (Figure 5, Figure 6, Figure 7 and Figure 8) show that the lines for all macroeconomic indicators remain within the confidence bands throughout the selected time frame, indicating stability of the coefficient. In the early IIP series, around 1992–1993, fluctuations can be observed, likely economic restructuring in the post-liberalisation period. However, these deviations are brief, and the model stabilises thereafter. While previous studies confirmed structural breaks in exchange rates during global crises, our findings demonstrate relative stability. Further, our findings show that the WPI and oil prices remained stable within bounds. The stability of these macro-variables during volatile global conditions underscores the necessity for consistent policy signalling.

The CUSUM statistic for all selected variables remained within the 5% critical bounds, suggesting stability in the parameters for the selected sample period.

Lastly, in our analysis, we employed White’s test (

Table 20. Heteroskedasticity-robust standard error estimates.

Variable	Coefficient	Standard Error	T-Stats	p-Value
Constant	31.73	6.69	4.74	0.00
Exchange rate	−1.27	0.21	−5.95	0.00
WPI	1.56	0.09	15.92	0.00
Oil Price	−0.22	0.05	−3.81	0.00

Source: Author’s calculations.

), confirming heteroskedasticity, yet the key coefficients remain robust. The exchange rate shows a strong negative effect (–1.27, p = 0.00), WPI exerts a persistent positive impact (1.56, p = 0.00), and oil prices display a short-run negative effect (−0.22).

Robust standard errors improve the credibility of the model interpretation. S. Kumar et al. (2022) reported similar results.

The Durbin–Watson statistics in

Table 21. Durbin–Watson test for autocorrelation.

Variable	D-Stats	Lower Bound	Upper Bound	Results
IIP	1.984	1.5	2.5	No AC
Exchange rate	1.988	1.5	2.5	No AC
WPI	2.060	1.5	2.5	No AC
Oil Price	1.973	1.5	2.5	No AC

Source: Author’s calculations.

were employed to check for autocorrelation. All variables fell within the acceptable (1.5–2.5) range, confirming no autocorrelation in residuals. IIP (1.984), exchange rate (1.988), and WPI (2.060) indicate error terms are white noise. In our case, the stable d-stat values further reinforce the robustness of the modelling framework, particularly when used for forecasting or policy simulation under global uncertainty. Heteroskedasticity-robust standard errors yield qualitatively similar inferences, indicating that the results are not sensitive to variance misspecification.

4. Conclusions

The present study investigates how India’s key macroeconomic variables, industrial output, wholesale inflation, the exchange rate and oil prices, adjust to major global disruptions including the global financial crisis of 2008, the COVID-19 pandemic and frequently occurring oil price shocks. By employing a combination of co-integration analysis, a vector error correction framework, impulse response functions, forecast error variance decomposition, quantile regression and volatility models, our study demonstrates both long-run equilibrium relationships and short-run transmission mechanisms. The findings of our study state the presence of stable long-run co-integration among the selected variables. Oil prices are observed to play a crucial role in linking external shocks to domestic macroeconomic outcomes, whereas short-run dynamics indicate that shocks in oil price transmit primarily through inflation and exchange rate channels before hampering industrial activity. The results of quantile regression demonstrate that these effects are not uniform across the distribution; rather, they strengthen during phases of stress, revealing substantial tail risks that are not captured by mean-based estimates. Strong persistence and clustering in oil prices and exchange rates are confirmed by volatility analysis. In addition, structural break tests align major regime shifts with globally recognised episodes of crisis. Overall, the results of our study suggest that India’s macroeconomic adjustment process operates gradually under normal conditions but comes under substantial stress during periods of global shocks. External disruptions increase both volatility and persistence across key macroeconomic variables. Oil price shocks have emerged as the primary external transmission channel, specifically for inflation and exchange rate dynamics. Stronger responses were witnessed at upper quantiles, demonstrating that crisis-period risks are concentrated in the tails of the distribution and are not adequately captured by mean-based models. These findings confirm that macroeconomic stability in India is responsive to energy price movements rather than purely domestic demand fluctuations. Accordingly, policy frameworks need to explicitly account for crisis-period risks through counter-cyclical mechanisms that are activated under extreme conditions. Further, measures such as strategic oil reserves, targeted foreign exchange intervention and contingency-oriented inflation management will be more effective rather than uniform policy responses across all regimes. In addition, the persistence of volatility reported by the GARCH estimates suggests that shock effects are not transitory, underscoring the importance of oil price hedging and strengthened monetary–fiscal coordination during periods of global stress.