Inflation Rate Determinants in Saudi Arabia: A Non-Linear ARDL Approach

Abstract

Inflation across the globe after the COVID-19 pandemic has shown some persistence and followed an upward trend well above inflation targets and beyond normal historical movements. The Saudi inflation rate followed similar patterns of global trends, surging significantly and persisting well above the pre-pandemic levels. This paper examines determinants of inflation in Saudi Arabia, considering internal and external factors, and evaluates whether inflation responds to common global shocks or is largely influenced by macroeconomic variabilities within the economy. Findings and analyses in this paper are based on both conventional Auto Regressive Distributive Lag (ARDL) and non-linear ARDL (NARDL) models using quarterly level data to capture short-run dynamics and long-run relationships between inflation rate and examined macroeconomics variables, namely oil prices real effective exchange rate, money supply, and government spending. Reported results reveal an asymmetrical relationship between oil price fluctuations and inflation rate volatilities in Saudi Arabia. Inclines in oil prices lead to higher inflation, while the decline in oil prices does not alleviate inflationary pressures, and these results are consistent both in the short-run and the long run. The influence of pass-through real effective exchange rate is also evident in transmitting global shocks to local consumer prices in the long run, where a depreciation in real effective exchange rate results in a higher cost of imported goods, exerting additional stresses on local inflation. For factors within the economy, findings indicate a substantial long-term inflationary effect of money supply on the inflation rate in Saudi Arabia, where a one percent increase in the money supply led to more than one-third increase in inflation in the long run. On the other hand, while the influence of government spending on inflation was statistically significant, its impact is less pronounced in explaining the inflation rate’s variations. The analysis reveals that the evaluated variables exert a stronger influence on inflation in the long run. This underscores the critical need for policymakers to consider the cumulative effects of these determinants when formulating effective long-term inflation stabilization policies.

1. Introduction

Inflation worldwide following the COVID-19 pandemic has exhibited notable persistence and has consistently trended above inflation targets, surpassing typical historical patterns. Numerous efforts were made to comprehend the underlying causes behind the unprecedented surges in current inflation rates. One of the contributing factors is the significant increase in government spending, including the implementation of a stimulus package following the economic fallout after the lockdowns [1,2,3]. However, the persistence of the current inflation is primarily due to the initial presumption that it is only temporary and the delays from policymakers in implementing proper policy tools to control it [4].

The recent inflationary trends have also been analyzed from both demand and supply perspectives, shedding light on their role as contributing factors to the observed inflation. The work by [5,6] pointed to a shift of consumptions from services to goods that inflated the prices of consumables after the pandemic. On the other hand, work by [7] showed that the struggle originated from the supply side and the delays in factors of production known as “supply chain bottleneck”, such as shortages in silicon chips [8] and declines in labor rate [9,10]. Their work shows that the cause of the recent inflation could originate from internal factors within the economy or be observed from external shocks that spill over to the economy.

A growing body of literature on inflation spillovers shows that economies are not only affected by countries’ specific shocks but also by global common shocks that spill over internationally [11,12,13,14,15]. Numerous studies estimated the impact of global shocks and discussed external factors that significantly impact local inflation rate variations. One dominant source of inflation rate volatility is oil price shocks [16], and its effect is not only time-varying but also grows over time, as discussed in [17,18]. The ongoing conflict between Russia and Ukraine has added additional strain on inflation rates and expectations. Work by [19] shows that the Russian–Ukrainian conflict is one source of the global inflation surge, and the geographic proximity and trade interactions with the conflict-involved countries influence the extent of observed inflation. The conflict is projected to contribute an additional two percent to world inflation in 2022 and one percent in 2023, according to [20]. Further, work by [21] shows that a 10% incline in oil real price due to an oil supply disruption in the current conflict will lead to inflation, reaching a peak of 0.4% and 0.3% in the US and Euro area, respectively. The author also emphasized the important role of “second-round effects” of pass-through that shape a more persistent inflation in the Eurozone. This shows that the cause of the recent inflation trend could originate from internal factors within the economy or be observed from external shocks that spill over to the economy.

The inflation rate in Saudi Arabia followed similar patterns of global inflation trends after the pandemic. Figure 1 offers a comprehensive overview of the country’s historical inflation fluctuations from 1980 to 2023. After years of stable inflation, the Saudi economy experienced a period of escalated prices that originated from the oil prices boom that started in 2003 and peaked within the decade in 2007, as documented by [22]. The inflationary pressures were exacerbated due to the generous expansions in payrolls, as discussed in [23]. Following the global financial crisis (GFC), there was a clear downward slope to control inflation facilitated by the implementation of energy subsidies [24] and food subsidies [25,26] to alleviate inflationary pressures on consumer prices. The inflation rate hovered around 0–2% prior to the pandemic and recorded the lowest rate in decades during the lockdown. After the reopening of the economy, the inflation rate experienced a sharp reversal, surging significantly and persisting well above the pre-pandemic levels.

Several studies investigated key determinants of inflation in Saudi Arabia, incorporating internal factors within the economy and external shocks that could spill over to the economy. Work by [27] indicates that in the long term, the Saudi economy is significantly impacted by global inflation, depreciation of the domestic currency, and the output gap while increasing the money supply can boost inflation in the short term using the conventional ARDL model. Study [28] shows that inflation, money supply, oil prices, and import goods are cointegrated using the Johansen test and shows that Saudi inflation is mostly derived from money supply and imported inflation using multiple regression methods. Work in [29] uses the regression analysis method and shows that circulated money, pegging to the US dollar, trade balances, and oil prices notably influence inflation in Saudi Arabia. Study [30] similarly utilized the Johansen test, demonstrating that Saudi inflation, local prices, the money supply, rate of exchange, and oil prices exhibit cointegration. Moreover, the study indicates that only local demand and money supply have inflationary effects on inflation in the short term, as revealed by VECM analysis.

This paper addresses some research gaps on determinants of inflation in Saudi Arabia. First, while prior evaluations on Saudi Arabia provide essential insights into Saudi’s inflation using annual data, further understanding of the relationship between inflation rate dynamics and macroeconomic variabilities could be captured when using quarterly level data, particularly for detecting short-term fluctuations more accurately and allowing for a more precise trailing of macroeconomic trends on inflation. Work in [31], for example, signifies the importance of increasing granularity. They show that the impact of exchange rate movements on inflation is apparent only in the first four months, while its effect is over in one year in Turkey. Such short-term volatility is averaged when utilizing annual data. Second, recent work did not elaborate on how fiscal policy influences inflation variabilities in Saudi Arabia. Study [32] discussed the following countercyclical fiscal policy to mitigate domestic growth volatility and the implemented policy mix that entails restrictive monetary policy during expansionary fiscal policy in normal times, which allows for better inflation control. In such normal cases, expansionary fiscal policy does not impact inflation, as discussed in [33]. However, after the GFC in 2008, policymakers in Saudi Arabia activated both expansionary monetary and fiscal policies to support domestic demand and sustain economic growth, as discussed in [32]. Thus, it is imperative to reconsider the effect of implemented fiscal policy in case of global economic turmoil, where both fiscal and monetary policies were activated simultaneously in the same direction. Third, this work also extends other work covering recent major global economic events such as the expansion of shale oil, COVID-19, and the current Ukraine–Russia conflict, all of which may potentially reinforce the effect of common global disruptions on Saudi inflation. Finally, earlier work assumes a linear correlation between the prices of oil and inflation in the Saudi Arabian economy. However, recent work [34,35,36,37], among others, highlighted the significance of relaxing this assumption to understand the inflation rate behavior in response to an incline or a decline in oil prices and affirmed the asymmetry between inflation and oil prices. The modeling technique using the NARDL model offers a rich analysis of the short-run dynamics and long-term relations between the inflation rate and its determinants evaluated in this paper.

In what follows, the “Literature Review” section provides a comprehensive review of prior studies investigating key inflation determinants. Then, the “Data and Descriptive Analysis” is presented, followed by performing “the Preliminary Test” to choose the proper model based on the data structure. The “Methodology” section presents the selected model to investigate the research question. The “Results and Discussions” presents the findings in this paper, and the “Conclusions” section summarizes the main findings, discusses policy implications, state research limitations, and proposes avenues for further exploration in this study.

2. Literature Review

Numerous studies investigated major causes of inflation and identified various economic factors contributing to its emergence and escalation. Table 1 provides a comprehensive overview of determinants of inflation that were investigated in different economies. Oil prices rise as a key source of global common shocks, and its effect on local inflation is not only time-varying but also grows over time [17]. In fact, the effect of oil prices on local inflation is well documented using different series and employing various models. Multiple studies have assumed a linear relationship and have documented a significant influence of oil prices on local inflation. Specifically, Hassan et al. [38] observed this impact in both the short and the long term in Pakistan, while Aliyev et al. [39] found that it affects the long-term inflation in Azerbaijan using the ARDL model. Additionally, the cointegration between oil prices and local inflation was identified in Azerbaijan by [40] and in India by [41] utilizing the Vector Error Correction Model (VECM). The significant impact of oil price movements on inflation in Turkey was also reported by [42] using the Structural Vector Autoregression (SVAR) model.

Other studies explored the potential asymmetry between oil prices and inflation, aiming to assess how inflation reacts to inclines and declines in oil prices. Work by [43] confirmed the asymmetry and reported the significant relation between increases in oil price and inflation rate appreciations, while the influence of reduction in oil prices on inflation was not statistically significant in Algeria using non-linear ARDL (NARDL). Moreover, work by [13] also confirmed the asymmetry between oil prices and inflation rate, where higher oil price shocks have a higher influence on inflation compared to negative oil shocks, and the effect is aggravated post-crisis compared to pre-crisis periods using NARDL. Similarly, Bawa et al. [44] evaluated this question in the case of Nigeria and showed that inflation responds positively to oil price shocks, whereas negative oil prices lead to a decline in the cost of production and local inflation. The asymmetry was also documented in Turkey using NARDL. According to Altunöz [45], the long-term effect of oil price increases on Turkey’s inflation is higher compared to declines in oil prices. Work by [37] utilized a multi-threshold NARDL model across BRICS countries and observed significant asymmetries in China in the short term with a particularly pronounced inflationary influence following declines in oil prices. Additionally, strong asymmetries following supply shocks were identified in China and Russia in the short run, while long-term asymmetries were reported in South Africa.

Using a panel of high and low oil-dependent economies (“High oil dependency index: Singapore, Portugal, Philippines, India, Greece, Belgium, South Korea, Italy, Pakistan, Spain. Low oil dependency index: Bulgaria, Canada, Norway, Denmark, Ecuador, United Kingdom, Mexico, Malaysia, Brazil, Venezuela.” (source: geocommons.com. http://geocommons.com/datasets?id=11950 (accessed on 27 July 2024)), Sek et al. [46] show that fluctuations in oil price impact domestic inflation for low-dependent economies, but its effect on high-dependent economies is captured indirectly by variations in production cost using panel ARDL. Work in [34] also utilized the same modeling technique on a set of countries that are importing or exporting oil and found a long-term increase in inflation associated with higher oil prices for both groups of countries while having no distinctive asymmetry in the short run. Study [35] also utilized panel NARDL and showed that both the declines and inclines in oil prices have an escalating impact on inflation in African OPEC member economies.

Prior studies also investigated the influence of pass-through on consumer prices, examining how volatilities in exchange rates can affect the overall price levels in the economy and contribute to inflationary pressures. Research conducted by [31] demonstrated that exchange rate volatility significantly affected price levels in Turkey using the vector autoregression (VAR) model. Their findings were confirmed by [42], which indicated that exchange rate changes significantly influence inflation in the short term, but this effect fades away over time. Studies [38,47] utilized the ARDL model and showed that the exchange rate plays a substantial role in inflation volatility in both the short term and the long term in the case of Pakistan and Egypt, respectively. Additionally, Gadau (2021) [40,48] also reaches a similar conclusion and confirms the cointegration in the case of Azerbaijan and Nigeria, respectively. The effect of exchange rate volatilities on inflation expectation was also documented in [49], which showed that the exchange rate is the key driver of inflation expectation in the case of the Czech Republic.

In panel-based studies, Sek et al. [46] reported that real exchange rate and imported goods are the leading sources of domestic inflation. Study [35] shows that exchange rate appreciations cause higher inflation in the African OPEC Countries using panel NARDL. Study [50] also employed NARDL and found verifications of the non-linear relationship between the rate of exchange and inflation expectations in Gulf Cooperation Council (GCC) countries. Work in [51] also presented links of pass-through effects using NARDL, and their impact is more apparent in the short term for the ASEAN-5 region countries Exchange rates are also perceived as a transmitting channel of shocks in energy prices to local inflation. Based on [52], the exchange rate pass-through is pivotal in transmitting inflated pressures from oil prices into domestic prices, particularly for developing countries and during sudden oil supply shocks using DCC-GARCH modeling. This finding is also confirmed by [21], who showed that the exchange rates transmit more persistent inflation due to shocks in oil prices. These reveal that the exchange rate volatilities and oil prices could cause a substantial impact on inflation rate movement and form external shocks to local inflation.

The impact of internal shocks initiated within the economy on inflation was investigated in the literature. In fact, recent work on inflation spillover shows that a country’s specific shocks outweigh common global shocks in certain economies. Evaluations on inflation spillovers by [53] among the G7 countries show that the inflation rates of all developed economies, except France and the US, are mainly affected by idiosyncratic shocks. Similarly, authors in [54] evaluated spillovers across the G20 group using the Diebold and Yilmaz method and showed that inflation in developing nations majorly stems from shocks originating within the economy. Thus, evaluations of inflation using internal factors within the economy remain essential in understanding inflation rate behavior in certain economies.

Indeed, recent work elaborated on the influence of activated fiscal and monetary policies on the unprecedented spikes in inflation after the COVID-19 pandemic. Study [55] employed the Ordinary Least Square (OLS) model and utilized cross-sectional data (the year 2022) across OECD countries and showed that both fiscal and monetary expansions were the reasons behind the spikes in inflation post-pandemic. Work by [56] utilized a semi-parametric (LP) model on a panel of 139 countries and showed that shocks originating from fiscal policy significantly influence inflation in developing nations. Their analyses indicate that the degree of fiscal intervention effect on inflation is subject to implemented monetary policy, followed exchange rate, and whether a country has an outlined fiscal rule. Work by [57] investigated determinants of inflation post-pandemic between January and July (the year 2020) among EU members and showed that exchange rate ratios and money supply were key sources of inflated prices. Study [35] shows that growth in the circulated money led to higher inflation, while higher food production alleviated inflation pressures in African OPEC countries using panel ARDL modeling. In the US, Gharehgozli and Lee [58] show that the expansionary fiscal and monetary policies were the sources of core inflation due to excessive money supply from implemented fiscal policies, while work by [59] argues that government spending does not significantly impact US inflation, especially in cases where the economy experiences a liquidity trap. Study [60] evaluated how shocks to monetary policy shocks affect inflation using the VAR model and showed that contractionary monetary policy has a significant impact on all main CPI components, with the exception of shelters. Work by [39,40] show that an expansion in the money supply affects inflation in the short and the long run in Azerbaijan and Egypt, respectively, using ARDL.

Table 1. Overview of determinants of inflation in the literature.

Author(s)	Sample Data	Country(s)	Method	Results
Yasin and ORHAN (2023) [55]	2022	OECD countries	OLS applied to cross-sectional data	- Fiscal and monetary are the sources of high inflation across countries.
Gharehgozli and Lee (2022) [58]	1960–2021 *	US	Structural VAR	- The excessive money supply from implemented fiscal policy is the main cause of observed. - Velocity of money, along with money supply shocks, majorly impacted core inflation.
ERDOĞAN et al. (2020) [57]	2020 **	EU members	Spatial panel data	- The rise in inflation was attributed to fluctuations in exchange rates and changes in the money supply.
Köse and Ünal (2021) [42]	1988–2019 **	Turkey	(SVAR)	- The exchange rate accounted for the most significant variation in inflation, although its impact gradually diminished over the course of time. - The reaction of inflation to changes in oil prices is significant and shows a gradual increase over time.
Reda and Nourhan (2020) [47]	2005–2018 *	Egypt	ARDL	- Money supply along with exchange rate are the leading sources of inflation
Aliyev et al. (2023) [39]	1997–2021	Azerbaijan	ARDL	- Oil prices exert significant long-term influences on inflation - Money Supply’s effect is observed in the short and long term.
Jørgensen and Ravn (2022) [59]	1966–2008 *	US	(SVAR) model	- Government spending does not cause inflation. - Government spending multiplier is very low at the stage of liquidity trap
Cevik and Miryugin (2023) [56]	1970–2021	Panel of 139 countries	local projection (LP) method	- Fiscal policy shocks cause inflation in developing countries. - Fiscal policy shocks are triggered by monetary policy frameworks, exchange rates, and if the economy follows explicit fiscal rules.
Sek et al. (2015) [46]	1980–2010	Set of high and low oil dependents	Panel ARDL	- Low oil dependency countries: Direct influence of oil prices on inflation. - High oil dependency through indirect throughout the exporter’s production cost. - Real exchange rate is the key source of inflation
Salisu et al. (2017) [34]	2000–2014	Oil export vs. import countries	Panel NARDL	- Long-run: The influence of oil on inflation was both substantial and positive for both categories. - Short-run: the results are mixed.
Bala and Chin (2018) [35]	1995–2014	African OPEC	Panel NARDL	- Both decreases and increases in oil prices inflate prices. - Gross domestic product, exchange rates, and money supply all contribute positively to inflation.
Li and Guo (2022) [37]	2000–2021	BRICS	MTNARDL	- Changes in oil supply and demand exert substantial influence on short-term inflation. - Declines in oil prices exacerbate inflationary pressures.
Gadau (2021) [48]	1985 to 2019	Nigeria	VECM	- Long-run causality between exchange rate equation and inflation rate.
Hassan et al. (2016) [38]	1976–2011	Pakistan	ARDL	- Short-term and long-term effects of oil prices and exchange rates on inflation.
Ding et al. (2023) [52]	2010–2022 **	China	DCC-GARCH model	- The pass-through effect exerts increased pressure on local prices, which is particularly noticeable in emerging economies and during periods of economic expansion.
Mukhtarov et al. (2019) [40]	1995–2017	Azerbaijan	VECM	- Exchange rate, oil price, and inflation are cointegrated.
Lacheheb and Sirag (2019) [43]	1970 to 2014	Algeria	NARDL	- Significant oil price increases lead to inflation, while lower oil prices do not impact inflation.
Sultan et al. (2020) [41]	1970 to 2017	India	NARDL	- Oil prices affect inflation in the short and the long-run
Pham et al. (2023) [51]	2000–2019 *	ASEAN-5	NARDL	- Oil price is the biggest contributor to inflation in the five economies. - Output growth and Money supply also affect inflation.

Note: * shows work that utilizes quarterly data, ** represents work that studies that used monthly data.

Table 1. Overview of determinants of inflation in the literature.

Author(s)	Sample Data	Country(s)	Method	Results
Yasin and ORHAN (2023) [55]	2022	OECD countries	OLS applied to cross-sectional data	- Fiscal and monetary are the sources of high inflation across countries.
Gharehgozli and Lee (2022) [58]	1960–2021 *	US	Structural VAR	- The excessive money supply from implemented fiscal policy is the main cause of observed. - Velocity of money, along with money supply shocks, majorly impacted core inflation.
ERDOĞAN et al. (2020) [57]	2020 **	EU members	Spatial panel data	- The rise in inflation was attributed to fluctuations in exchange rates and changes in the money supply.
Köse and Ünal (2021) [42]	1988–2019 **	Turkey	(SVAR)	- The exchange rate accounted for the most significant variation in inflation, although its impact gradually diminished over the course of time. - The reaction of inflation to changes in oil prices is significant and shows a gradual increase over time.
Reda and Nourhan (2020) [47]	2005–2018 *	Egypt	ARDL	- Money supply along with exchange rate are the leading sources of inflation
Aliyev et al. (2023) [39]	1997–2021	Azerbaijan	ARDL	- Oil prices exert significant long-term influences on inflation - Money Supply’s effect is observed in the short and long term.
Jørgensen and Ravn (2022) [59]	1966–2008 *	US	(SVAR) model	- Government spending does not cause inflation. - Government spending multiplier is very low at the stage of liquidity trap
Cevik and Miryugin (2023) [56]	1970–2021	Panel of 139 countries	local projection (LP) method	- Fiscal policy shocks cause inflation in developing countries. - Fiscal policy shocks are triggered by monetary policy frameworks, exchange rates, and if the economy follows explicit fiscal rules.
Sek et al. (2015) [46]	1980–2010	Set of high and low oil dependents	Panel ARDL	- Low oil dependency countries: Direct influence of oil prices on inflation. - High oil dependency through indirect throughout the exporter’s production cost. - Real exchange rate is the key source of inflation
Salisu et al. (2017) [34]	2000–2014	Oil export vs. import countries	Panel NARDL	- Long-run: The influence of oil on inflation was both substantial and positive for both categories. - Short-run: the results are mixed.
Bala and Chin (2018) [35]	1995–2014	African OPEC	Panel NARDL	- Both decreases and increases in oil prices inflate prices. - Gross domestic product, exchange rates, and money supply all contribute positively to inflation.
Li and Guo (2022) [37]	2000–2021	BRICS	MTNARDL	- Changes in oil supply and demand exert substantial influence on short-term inflation. - Declines in oil prices exacerbate inflationary pressures.
Gadau (2021) [48]	1985 to 2019	Nigeria	VECM	- Long-run causality between exchange rate equation and inflation rate.
Hassan et al. (2016) [38]	1976–2011	Pakistan	ARDL	- Short-term and long-term effects of oil prices and exchange rates on inflation.
Ding et al. (2023) [52]	2010–2022 **	China	DCC-GARCH model	- The pass-through effect exerts increased pressure on local prices, which is particularly noticeable in emerging economies and during periods of economic expansion.
Mukhtarov et al. (2019) [40]	1995–2017	Azerbaijan	VECM	- Exchange rate, oil price, and inflation are cointegrated.
Lacheheb and Sirag (2019) [43]	1970 to 2014	Algeria	NARDL	- Significant oil price increases lead to inflation, while lower oil prices do not impact inflation.
Sultan et al. (2020) [41]	1970 to 2017	India	NARDL	- Oil prices affect inflation in the short and the long-run
Pham et al. (2023) [51]	2000–2019 *	ASEAN-5	NARDL	- Oil price is the biggest contributor to inflation in the five economies. - Output growth and Money supply also affect inflation.

Note: * shows work that utilizes quarterly data, ** represents work that studies that used monthly data.

3. Data and Descriptive Analysis

This work utilizes external and internal measures and examines their influence on Saudi inflation. Oil Prices and real exchange rate are considered external factors where fluctuations in the former are determined by the supply and demand in the global market and the latter is known as a transmission channel of global shocks to local inflation [21,52]. Money supply (M3) and government expenditure are used to represent internal factors that are expected to have a major influence on Saudi’s inflation.

If variabilities in the inflation rate are explained mostly by these exogenous series, then it could be concluded that the inflation rate in Saudi Arabia is sensitive to global shocks and spillovers from external factors. On the other hand, if the variations in the inflation rate mainly stem from internal factors, then the conclusion is that inflation responds to domestic macroeconomic factors, and policymakers should form the proper tools to stabilize inflation to counter internal shocks. It is expected that external and internal factors will significantly influence the inflation rate in Saudi Arabia.

This paper uses quarterly data starting from 2005(Q1)–2023(Q1) of consumer price index (CPI), Broad Money (MS), government expenditure (GE), real effective exchange rate (RE), and oil prices (OP). CPI data were retrieved from the International Monetary Fund (IMF). The log difference of the CPI is calculated to compute inflation (${\pi }_t)$ as follows ${\pi }_t=\mathrm{l}\mathrm{n}\left({CPI}_t\right)-{\mathrm{l}\mathrm{n}(CPI}_{t-1})$. Datasets on considered fiscal and monetary policy tools were gathered from the General Authority of Statistics in Saudi Arabia. Real government final consumption expenditure is used as a proxy for the fiscal policy followed, and Broad Money (MS) is used to represent the monetary policy implemented in Saudi Arabia. Data on oil prices were collected from the U.S. Energy Information Administration (EIA), and Saudi real effective exchange rate data were gathered from the IMF database.

Table 2. Descriptive statistics. Source: author calculation.

	CPI	OIL	OIL+	OIL−	REER	MS	GE
Mean	2.028	1.832	0.029	−0.027	2.026	6.132	3.727
Median	2.058	1.833	0.015	0.000	2.013	6.213	5.110
Maximum	2.119	2.093	0.167	0.000	2.085	6.416	5.252
Minimum	1.885	1.444	0.000	−0.305	1.958	5.709	0.000
Std. Dev.	0.070	0.138	0.038	0.054	0.034	0.195	2.306
Skewness	−0.805	−0.307	1.597	−2.971	−0.024	−0.648	−1.012
Kurtosis	2.349	2.563	5.333	12.92	1.637	2.294	2.026
Jarque–Bera	9.181	1.727	46.94	401.3	5.652	6.633	15.343
Probability	0.010	0.421	0.000	0.000	0.059	0.036	0.000
Sum	148.0	133.75	2.129	−1.944	147.9	447.63	272.0
Sum Sq. Dev.	0.3562	1.379	0.1073	0.213	0.085	2.753	382.88
Observations	73	73	72	72	73	73	73

shows a summary statistic on the sampled data. All series were transformed into logarithms, and interpreted results are elasticity measures in percentage change.

4. Preliminary Tests

Before evaluating the influence of selected variables on inflation, it is mandatory to perform some preliminary tests to understand the nature of the data at hand. It is also essential to select the proper modeling of the research question based on the structure of the data. The initial step involves conducting unit-root tests to investigate each series’ stationarity and degree of integration to control for potential spurious regression. When all series are integrated of order zero I(0), it could be concluded that they are stationary and model the data in their levels, such as OLS. However, if the series is I(1) and no cointegration exists, then the VAR model in the first difference form should be implemented [61]. However, there might be a case where a mixed order of integration is encountered, and some of the series are stationary while others contain unit roots. This is when ARDL is the proper model to use [62], which will be explained in detail in the methodology section.

To explore the stationarity of the series, both traditional unit root tests are utilized, including the Augmented Dickey–Fuller test and the Phillips–Perron test. Further, unit root testing with structural breaks is implemented to consider the possibility of structural breaks in the examined series. Study [63] asserted that researchers should use unit root tests with structural breaks to ensure that none of the series are I(2). Thus, the preliminary testing also investigates the series using [64] unit root test that considers the possibility of the existence of a structural break.

Table 3. Results of unit root tests.

Augmented Dickey and Fuller Test
Variable Name	Level			First Difference
	None	Intercept	Intercept and Trend	Intercept	Trend	Intercept and Trend
CPI	5.722	−2.533	−1.027	−3.174 ***	−6.941 ***	−7.464 ***
Oil	0.091	−2.551	−2.660	−7.224 ***	−7.176 ***	−7.127 ***
Oil+	−3.441 ***	−8.131 ***	−8.072 ***	−8.202 ***	−8.140 ***	−8.083 ***
Oil−	−5.330 ***	−6.267 ***	−6.228 ***	−8.876 ***	−8.810 ***	−8.473 ***
RE	0.772	−0.834	−2.696	−6.697 ***	−6.719 ***	−6.699 ***
MS	2.791	−3.985 ***	−2.063	−3.514 **	−3.5146 **	−6.927 ***
GE	0.052	−1.603	−1.562	−8.344 ***	−8.414 ***	−8.437 ***
Phillips and Perron Test
CPI	4.432	−2.306	−1.110	−5.341 ***	−7.030 ***	−7.472 ***
Oil	0.722	−2.787 *	−2.897	−7.200 ***	−7.187 ***	−7.069 ***
Oil+	−5.697 ***	−8.133 ***	−8.074 ***	−45.890 ***	−47.769 ***	−47.257 ***
Oil−	−5.330 ***	−6.108 ***	−6.064 ***	−40.298 ***	−42.198 ***	−42.24 ***
RE	0.984	−0.8276	−2.336	−4.843 ***	−4.792 ***	−4.732 ***
MS	5.630	−3.227 **	−1.935	−3.305 ***	−6.556 ***	−7.240 ***
GE	0.053	−1.603	−1.578	−8.344 ***	−8.414 ***	−8.437 ***
Zivot and Andrews Test
Variable Name	Level			First Difference
	None	Intercept	Break Date	None	Intercept	Break Date
CPI	−4.605	−3.90	2007 Q3	−7.923 ***	−9.115 ***	2014 Q4
Oil	−3.97	−5.15 **	2014 Q3	−8.27 ***	−8.132 ***	2008 Q4
Oil+	−8.702 ***	8.97 ***	2020 Q2	−15.128 ***	−15.01 ***	2006 Q2
Oil−	−7.985 ***	−7.858 ***	2008 Q4	−12.081 ***	−11.98 ***	2009 Q2
RE	−5.735 ***	−5.900 ***	2014 Q3	−8.279 ***	−8.159 ***	2008 Q4
MS	−5.038 ***	−4.579	2019 Q3	−7.75 ***	−8.188 ***	2015 Q3
GE	−139.99 ***	−134.62 ***	2009 Q4	−178.02 ***	−172.51 ***	2010 Q1

Note: ***, **, and *, respectively, denote the 1, 5, and 10 percent significance levels.

shows the results from both the traditional and structural break unit root tests. Each series is evaluated in level and in first difference and in cases where the series contains intercept, trend, or both. The null hypothesis for all tests is that the data contain a unit root, while the alternative hypothesis is that the series is stationary.

Results from unit root tests fail to reject the null hypothesis in all series and use different specifications, pointing to the fact that the data contain a unit root in levels using traditional tests. However, when considering the possible structural break in the series, the real exchange rate and government expenditure are stationary in levels. Nevertheless, when estimating the series at their first difference, we were able to reject the null hypothesis at the 1% confidence level for all series and accept the null hypothesis as integrated of order one, i.e., I(1). Thus, the conclusive results from both traditional and structural unit root tests suggest that the results were mixed between I(0) and I(1), which suggests that ARDL is the proper form to use to model the research question.

5. Methodology

5.1. The Symmetric ARDL Model

This paper utilizes the ARDL model proposed by [62] to evaluate the influence of selected macroeconomic variables and oil prices on inflation. The ARDL model offers certain benefits compared to conventional cointegration models and allows for the estimation of short-term dynamics and estimate of the convergence speed to the long-run equilibrium, in addition to its capability to estimate the cointegration of the estimated coefficients [65]. The initial step in the ARDL model is to assess whether cointegrations exist between estimated variables. The cointegration test, also known as the bound test, is derived from the unconstrained ARDL:

$${\Delta CPI}_t={\alpha }_0+\sum _{i=1}^p{\beta }_{2i}\Delta {CPI}_{t-i}+\sum _{i=0}^{q_1}{\beta }_{3i}\Delta {OP}_{t-i}+\sum _{i=0}^{q_2}{\beta }_{4i}\Delta {RE}_{t-i}+\sum _{i=0}^{q_3}{\beta }_{5i}\Delta {MS}_{t-i}+\sum _{i=0}^{q_4}{\beta }_{6i}\Delta {GE}_{t-i}+{\theta }_1{CPI}_{t-1}+{\theta }_2{OP}_{t-1}+{{\theta }_3{RE}_{t-1}+{\theta }_4{MS}_{t-1}+{\theta }_5{GE}_{t-1}+\epsilon }_t$$

where ∆ denotes the first difference operator, $\beta$’s are the short-run coefficients, and $\theta$’s captures the long-run estimates. The p and q represent the optimal lag selection for each variable, which can be determined using one of the information criteria, such as the Schwarz information criterion (SIC). The choice of the selected variables is based on the prior work that investigates determinants of inflation in a specific country. The significance of oil prices (OP) on inflation is reported on refs. [21,34,35,36,37,38,39,40,43], among others. The exchange rate (RE) volatilities were shown to be determinantal and were documented in [42,46,47,48,52,57], among others. The links between money supply (MS) and inflation are evident from the quantity theory of money and are well documented in the literature. The extent to which fiscal policy (GE) could significantly impact inflation rate movements is reported in [55,56,58,59]. Thus, this work reinvestigates the impact of these macroeconomic variables on the inflation rate in Saudi Arabia.

Work by [62] derived bound tests, which are based on the F-statistics, which estimates whether long-run relationships exist. The null hypothesis is no cointegration, while rejection confirms cointegration across examined variables and is expressed in the following:

$$H_0:{\theta }_1={\theta }_2={\theta }_3={\theta }_4={\theta }_5=0$$

$$H_1:{\theta }_1\ne {\theta }_2\ne {\theta }_3\ne {\theta }_4\ne {\theta }_5\ne 0$$

Study [62] also derived the critical values to estimate asymptotically for the existence of cointegration. Calculated lower and upper bound is based on the integration order of regressors, the number of included variables, and whether the equation contains intercept, trend, or both. If the value is less than the lower bound, then the conclusion is that no cointegration exists, while a value higher than the threshold confirms cointegration among examined variables. The test is inconclusive if it falls between the lower and the upper bound. Once cointegration is confirmed, the influence of the independent variables across examined variables across the short run and in the long run is estimated.

Long-run coefficients are calculated from

$${CPI}_t={\alpha }_0+\sum _{i=1}^p{\beta }_{1i}{CPI}_{t-i}+\sum _{i=0}^{q_1}{\beta }_{2i}{OP}_{t-i}+\sum _{i=0}^{q_2}{\beta }_{3i}{RE}_{t-i}+\sum _{i=0}^{q_3}{\beta }_{4i}{MS}_{t-i}+\sum _{i=0}^{q_4}{\beta }_{5i}{GE}_{t-i}+{\epsilon }_t$$

And the short-term is estimated from the conditional error correction model (ECM):

$${\Delta CPI}_t={\alpha }_0+\sum _{i=1}^p{\beta }_{1i}\Delta {CPI}_{t-i}+\sum _{i=0}^{q_1}{\beta }_{2i}\Delta {OP}_{t-i}+\sum _{i=0}^{q_2}{\beta }_{3i}\Delta {RE}_{t-i}+\sum _{i=0}^{q_3}{\beta }_{4i}\Delta {MS}_{t-i}+\sum _{i=0}^{q_4}{\beta }_{5i}\Delta {GE}_{t-i}+\phi {ECM}_t+{\epsilon }_t$$

The (ECM) coefficient should be negative and significant, which signifies the speed of adjustment toward the long-run equilibrium. In other words, it captures the short-run deviations due to macroeconomic events (i.e., shocks) and the convergence speed to the long-run equilibrium.

5.2. Asymmetric ARDL or Non-Linear ARDL (NARDL)

Work by [66] extended the conventional ARDL model by introducing the non-linear ARDL model. The asymmetric ARDL allows us to estimate the cointegration when asymmetry exists between the dependent variable and one or more explanatory variables. The conventional ARDL model assumes asymmetry between the dependent variable and the regressors, while the NARDL model allows testing whether the changing behavior of the independent variable would have a variant impact on the dependent variable. Following [34,35,36,37,43], this work evaluates whether asymmetrical relations exist between oil prices and inflation.

The oil price variable can be represented as follows:

$${OP}_t={OP}_t^{+}+{OP}_t^{-}$$

where ${OP}_t^{+}$ and ${OP}_t^{-}$ are the positive and negative partial sums:

$${OP}_t^{+}=\sum _{i=1}^T\Delta {OP}_i^{+}=\sum _{i=1}^T\mathrm{m}\mathrm{a}\mathrm{x}({\Delta OP}_i,0)$$

$${OP}_t^{-}=\sum _{i=1}^T\Delta {OP}_i^{-}=\sum _{i=1}^T\mathrm{m}\mathrm{i}\mathrm{n}({\Delta OP}_i,0)$$

Thus, the ECM in NARDL is derived as follows:

$${\Delta CPI}_t={\alpha }_0+\sum _{i=1}^p{\beta }_{1i}\Delta {CPI}_{t-i}+\sum _{i=0}^{q_1}{\beta }_{2i}\Delta {OP}_{t-i}^{+}+\sum _{i=0}^{q_1}{\beta }_{3i}\Delta {OP}_{t-i}^{-}+\sum _{i=0}^{q_2}{\beta }_{4i}\Delta {RE}_{t-i}+\sum _{i=0}^{q_3}{\beta }_{5i}\Delta {MS}_{t-i}+\sum _{i=0}^{q_4}{\beta }_{6i}\Delta {GE}_{t-i}+{\theta }_1{CPI}_{t-1}+{\theta }_2{OP}_{t-1}^{+}+{\theta }_3{OP}_{t-1}^{-}+{{\theta }_4{RE}_{t-1}+{\theta }_4{MS}_{t-1}+{\theta }_5{GE}_{t-1}+\epsilon }_t$$

where ${\beta }_2$ and ${\beta }_3$ demonstrate the short-run asymmetry and ${\theta }_2$ and ${\theta }_3$ show the asymmetry in the long run. The implementation of NARDL is similar to the conventional ARDL model and uses the same F-statistics in the original work.

Final notes on ARDL models are that, unlike traditional cointegration techniques such as [13], derived conclusions and generated results from the ARDL model are less subject to small sample bias [67,68]. ARDL estimation is unbiased because of the proper inclusions of previous lags that eliminate residual correlation and the endogeneity problem, as pointed out in [69,70]. However, the series should not have a higher order than one since it will invalidate calculated F-statistics and critical values calculated by the original work, as discussed in [63,71,72]. If, however, nonstationary variables are regressed on nonstationary series, we end up with spurious regression outcomes [73].

6. Results and Discussion

This section presents the findings concerning the influence of oil prices, exchange rates, monetary policy, and fiscal policy on the inflation rate in Saudi Arabia. After performing the unit-roots testing and confirming that the ARDL models are the appropriate model to address the research question, the cointegrations across the examined series are evaluated.

Table 4. Cointegration Tests of ARDL Models.

Model	Optimal Lag	F-Statistics	Critical Values (1%)		Critical Values (5%)		Conclusion
			Lower Bound I(0)	Upper Bound I(1)	Lower Bound I(0)	Upper Bound I(1)
Linear ARDL	(1, 1, 1, 0, 0)	7.16	3.74	5.06	2.86	4.01	Cointegration
Non-Linear ARDL	(1, 1, 1, 1, 0, 0)	7.97	3.41	4.68	2.62	3.79	Cointegration

Notes: Optimal lag selection is based on SIC. Source: Author Calculations.

shows the cointegration test results of symmetric ARDL and NARDL models. Specified models and optimal lag selections were based on (SIC). The F-statistics value needs to be bigger than the upper bound I(1) value to confirm the existence of cointegration in the models. The results show that the F-statistic is bigger than the asymptotic critical values at the 1% confidence interval, confirming that long-run relationships exist across examined variables using the conventional ARDL and NARDL models.

After the cointegration is confirmed, the ARDL models offer rich inferences on both long-relationship and short-run dynamics. The estimation and analysis of inflation determinants are based on linear and non-linear ARDL model findings. The choice of using quarterly data allows us to more appropriately evaluate the influence of macroeconomic variables on the inflation rate movements in Saudi Arabia both in the short run and in the long run.

Table 5. Inflation determinants results from ARDL models.

	ARDL Model	NARDL Model
Variable	Coefficients (Std. Error)	Coefficients (Std. Error)
Long Run Results
Oil	−0.021 (0.022)
Oil+		0.243 ** (0.096)
Oil−		−0.030 (0.054)
RE	−0.375 *** (0.022)	−0.250 ** (0.106)
MS	0.359 *** (0.024)	0.315 *** (0.0258)
GE	0.003 (0.002)	0.005 *** (3.185)
Short Run Results
Δ OIL	0.017 *** (0.006)
Δ OIL+		0.035 *** (0.009)
Δ OIL−		0.006 (0.007)
Δ RE	0.113 * (0.060)	0.093 * (0.051)
Constant	0.157 *** (0.025)	0.15 *** (0.020)
ECT (−1)	−0.254 *** (0.041)	−0.260 *** (0.036)
LM (1)	0.478	0.804
Heteroskedasticity	0.536	0.104
J-B	0.000	0.000
RESET	0.040	0.107

Notes: “OIL” denotes oil prices, “RE” is the real effective exchange rate, “MS” is the money supply, “GE” is the government expenditure, and “ECT” is the error correction term. ***, **, and * denote the 1, 5, and 10 percent significance levels, respectively. Source: author calculations.

presents the modeling results from both ARDL and NARDL models.

The results from the conventional ARDL model show that the influence of oil prices on inflation is only short-lived, while there is no evidence of cointegration between oil and inflation in Saudi Arabia. Nevertheless, variations on the real effective exchange rate influence inflation rate movements in the long run, where depreciation on the real effective exchange rate leads to a higher cost of imported goods, and this result aligns with [46] for low oil-dependent economies (Oil Dependency index used in [46] is defined as “a function of percent oil demand out of all energy demand, energy diversity, economic dependency on oil (oil demand out of GDP) and share of net imports to oil consumption” (source: geocommons.com, accessed on 27 July 2024)). Countries such as Norway, Canada, Mexico, and Brazil are considered low oil-dependent economies, which are also major oil-exporting economies like Saudi Arabia). In Saudi Arabia, a deterioration in the real effective exchange rate by 1% is expected to contribute around 0.37% of observed inflation in the long term, while its impact is less significant in the short term.

Reported results using the symmetric ARDL show a variant influence of evaluated fiscal and monetary policy tools on inflation. The money supply has a substantial and positive effect on observed inflation in the long run, and its influence is statistically and economically significant. This outcome aligns with the findings of [35], who showed that the influence of money supply on inflation is only evident in the long run in the case of African OPEC member countries. On the other hand, the government expenditure does not seem to put further pressure on prices using the conventional ARDL model.

To further evaluate the influence of oil prices on inflation, the existence of the non-linear relationship using the NARDL model is evaluated. The results reported in

Table 6. Wald test for short-run and long-run asymmetry.

Test	Short-Run Asymmetry	Long-Run Asymmetry
F-statistic	9.842 ***	4.019 ***
Chi-square	9.842 ***	16.07 ***

Source: author calculations. *** denote the 1 percent significance levels.

highlight the importance of distinguishing between positive and negative oil price movements in explaining inflation rate variabilities using the Wald test. These findings necessitate the assumption of asymmetry to properly evaluate the impact of oil prices on the inflation rate. This framework has been employed in recent work by [37,41,51], among others, and presumes the nonlinearity between oil prices and inflation.

The results from the NARDL model shown in Table 5 confirm the asymmetry and cointegration between oil prices and inflation in Saudi Arabia. The results demonstrate that elevated oil prices lead to increased inflation rates in Saudi Arabia in the short and long run. However, the inflation rate does not respond to a decline in oil prices, and lower oil prices do not alleviate inflationary pressures on consumer prices in Saudi Arabia. This finding is consistent with earlier findings that discussed the influence of oil price variations on the inflation of oil-producing economies. In particular, Bala and Chin [35] and Lacheheb and Siraj [43] investigated the same question on African OPEC Member Countries and Algeria, respectively, and showed that rises in oil prices result in increased inflation, whereas decreases in oil prices do not affect inflation. Similarly, Choi et al. [74] also confirmed the asymmetry and showed that positive oil price shocks have a higher influence compared to negative oil price shocks using a panel from 72 developed and developing economies. One reason for the insignificance of the decline in oil prices on local inflation is that Saudi Arabia follows a countercyclical fiscal policy, as discussed in [32]. In such cases where the economy faces downturns and lower capital inflows from oil revenues, expansionary fiscal policy is activated that could leave economic growth and inflation rate levels less affected.

The real exchange rate influence on inflation is also verified using the NARDL model, which exhibits a long-run cointegration in the inflation–exchange rate nexus in Saudi Arabia. This is consistent with earlier work on developing economies in the case of Pakistan [38], Egypt [47], Azerbaijan [40], Nigeria [48], Turkey [42], and emerging economies [52].

When it comes to internal factors, money supply significantly impacts inflation, and this finding is consistent across both models. Increasing the money supply by one percent will contribute approximately one-third percent to the observed inflation in the long run in Saudi Arabia. In addition, the influence of government spending on inflation turned out to be statistically significant, but its economic impact and contribution to inflation are negligible. This finding is consistent with results reported in [75], which show that the impact of continuous increases in the money supply will have an impact in the long run, while its influence in the short run is insignificant in the case of China and Vietnam. When comparing the results on the effect of implemented monetary and fiscal policy with the literature, Yasin and ORHAN [55] show that both expansionary fiscal and monetary policies are strongly associated with high inflation in OECD countries. Conversely, Jørgensen and Ravn [59] reported that government spending does not significantly impact inflation, especially when the economy is in a liquidity trap and during turmoil. Indeed, Gharehgozli and Lee [58] claim that the impact of expansionary fiscal policy is triggered by the excessive money supply that raises the core inflation in the US. Study [56] also emphasized the extent to which fiscal policy impacts inflation is prominently affected by exchange rate systems, monetary policy structures, and adherence to explicit fiscal regulations. These findings indicate that while the direct impact of fiscal policy on inflation is somewhat controversial, its inflationary impact could be captured indirectly through money supply and sensitive to the implemented monetary policy regime. Thus, future work should consider the interaction between fiscal policy and monetary policy to further understand the extent to which fiscal policy impacts inflation rate volatility in Saudi Arabia.

Finally, the ECM captures deviations due to short-term disturbance and estimates the adjustment speed for converging to long-term equilibrium. Results of the ECM from both models show that the speed of adjustment from short-term disturbance would take approximately four quarters to largely converge to long-run equilibrium.

7. Conclusions

This study investigates factors contributing to inflation in Saudi Arabia, including oil prices, exchange rates, money supply, and government spending. Both linear and non-linear ARDL models were utilized to evaluate the short-term dynamics and the long-term relationships between the inflation rate and its determinants. Results from both models show that the effect of external factors, namely oil prices and effective exchange rates, have a significant impact on inflation rate variations. Results from the NARDL model reveal the non-symmetrical relationship between oil prices and the inflation rate in Saudi Arabia. Elevated oil prices cause higher inflation both in the short and long run, while declines in oil prices do not alleviate inflationary pressures on consumer prices in Saudi Arabia, and these results are consistent both in the short and the long run. Evaluations of the influence of the real effective exchange rate also indicate its major impact on inflation, exerting additional pressure on the cost of imported goods over the long term. Thus, reported results demonstrate the important role of pass-through in shaping more persistent inflation, as pointed out by [21,52]. The initial short-run divergence from external shocks is expected to largely converge towards the long-run inflation rate equilibrium within one year.

For factors within the economy, the results from both models show that money supply substantially affects observed inflation; a one percent increase in broad money inflates consumer prices by approximately one-third percent in the long run. However, expansionary fiscal policy through higher government spending does not have a major contribution to inflation in the case of Saudi Arabia, and this finding is in agreement with [59], who showed that expansionary fiscal policy has no significant effect during economic turmoil in the case of the US. Nevertheless, the extent to which fiscal policy causes inflation should consider the excessive money supply associated with expansionary fiscal policy, as discussed in [58], along with the adopted monetary policy framework outlined by [56].

The findings of this study have several important policy implications. While earlier work shows that the impact of oil price volatility has a linear correlation with the inflation rate and recommends specific policy implications based on this linear relationship, this study reveals the asymmetrical relationship between oil prices and the inflation rate in the case of Saudi Arabia. Policymakers should implement effective measures to mitigate the effect of elevated prices, as they significantly influence inflation in the short run. The necessity for a policy response is emphasized since lower oil prices do not lead to significant inflation relief. Additionally, it should be recognized that the effects of inflated oil prices on inflation tend to persist in the long run. Thus, policymakers should consider the significant role that global economic shocks play in influencing local inflation rates brought by elevated oil prices and transmitted through the exchange rate. The Saudi government has a long-standing commitment to subsidizing energy [24] and food subsidies [25,26], and these policies proved to be important in stabilizing inflation over time. The influence of imported inflation is also evident in shaping local inflation. Thus, higher global inflation figures need to be carefully considered as they can exert upward pressure on local prices. When it comes to internal factors, findings in this paper demonstrate that circulated money supply has an evident influence on inflation rate variations in the long run. This shows that policymakers have accessible tools that could be utilized to combat the inflation rate. Adjusting the money supply can effectively influence inflationary pressures, and monetary policies can be used to stabilize the economy over time. The findings of this research are crucial to the sustainable economic development of Saudi Arabia.

Understanding the factors that drive inflation provides valuable insights for public authorities to implement effective inflation control measures that support sustainable economic development. Stabilizing input costs and localizing foreign products can be crucial in reducing external inflationary pressures, contributing to a more established economic environment. Additionally, implemented fiscal and monetary policies to combat inflation should also be cautioned on their influence on long-term economic growth.

The outcomes of this research, while insightful in understanding the determinants of inflation in Saudi Arabia, are accompanied by certain limitations. Firstly, this research focuses on Saudi Arabia, and the findings could be influenced by the specific characteristics of its economy and may not be generalized to all economies. Nevertheless, the outcome of this research suggests that examined factors are expected to have significant impacts on countries that share similar characteristics, namely oil-producing countries with comparable economic structures and dependencies. Second, results show that government spending does not majorly affect inflation rate fluctuations. However, the extent to which fiscal policy causes inflation should consider the excessive money supply associated with expansionary fiscal policy, as discussed in [58]. Thus, future work should consider the interaction between fiscal and monetary policies to provide deeper insights into the influence of fiscal policy implementation on inflation dynamics in Saudi Arabia.