Testing for Causality between Oil Prices and Money Supply in Saudi Arabia

The impact of oil price shocks on a country’s economy has been well studied in the economic literature. However, until now, articles analyzing the impact of these shocks on the Saudi Arabian economy have been relatively sparse. This paper attempts to shed some light on this important topic by examining the causal relationship between oil prices and an important monetary variable, the M3 (broad-based) money supply. Monthly data going back to 1982 were used in this study, which employs unit root tests and Granger causality analysis to test whether there is a causal relationship or not. No discernable causality relationship was found; this lack of such link leads the authors to conclude that this may be due to the prudent and stable fiscal and monetary policy on the part of the Kingdom’s government and its central bank, the Saudi Arabian Monetary Authority (SAMA).


Introduction
Since the pioneer work of Hamilton (1983), there has been a growing and concentrated literature probing the link between oil price shocks and various macroeconomic indicators such as economic growth (e.g., Hamilton, 1996Hamilton, , 2003Hamilton, , 2008Hamilton, , 2009Hamilton, , 2010Mork, 1986;Hooker, 1986;Barsky & Kilian, 2004;Kilian, 2008aKilian, , 2009Al Rasasi & Yilmaz, 2016), inflation (e.g., Bachmeier & Cha, 2011; Barsky price shocks on various economies across the globe, there is still much less research focusing on the Saudi Arabian economy. Indeed, the literature analyzing the consequences of oil price swings on the Saudi economy is very scarce, and it mainly evaluates the impact of oil price shocks on some of the key macroeconomic variablese.g., economic growth, inflation, the exchange rate, and the stock market (Note 1). This in turn indicates the need for further research analyzing how changes in global oil prices impact the Saudi economy through various channels. Thus, the main purpose of this research paper is to investigate whether changes in oil prices are able to predict changes in a key economic variable, the M3 money supply, in the case of Saudi Arabia.
Before proceeding with our investigation, it is important to understand how changes in oil prices can influence the money supply in oil exporting economies. Typically, oil-exporting countries enjoy elevated oil export revenues, with a majority of such revenues being deposited in the governments' banks accounts. For instance, in the case of Saudi Arabia, oil revenues represented an average of 80.03 percent of total government revenues over the 2010-17 period; although, the recent decline in oil prices starting in mid-2014 resulted in lower oil revenues as shown in Table 1 and Figure 1. However, the real economic impact of these deposits (which are increases in the central bank's liabilities) on the domestic economy depends on whether or not these deposits are spent by the central government and how the government spends themon domestic goods and services, or on imports. Thus, the fiscal stimulus of this spending determines the impact on both nominal and real economic growth, which in turn affects the money supply.  Against this background, this paper attempts to enrich the prevailing literature on the Saudi economy by assessing whether oil price volatility causes changes in the M3 money supply. In addition, this paper relies on the method introduced by Qualls et al. (2017) to convert money supply data from the Hijra calendar to the Gregorian calendaran important addition to the pre-1988 monetary data available to SAMA for analytical purposes (Note 2).

Data
The data utilized in the analysis of the causal link between money supply and oil prices in Saudi Arabia are based on monthly frequencies spanning from January 1982 to April 2018. The data for the money supply, measured by M3 (the broad measure of money supply), and the oil price, measured by the Arab Light price for crude oil, are extracted from SAMA databases (Note 3). It is also important to emphasize that we have converted the money supply data prior to June 1988 from a Hijra calendar basis to that of the Gregorian calendar based on the methodology employed by Qualls et al. (2017).
Furthermore, it is vital to note that all variables are expressed in natural logarithms.

Unit Root Tests
Testing economic time series for unit roots is a requirement to ensure the stationarity of these economic variables and to avoid spurious results. To do so, the common unit root tests were conducted; namely, the Augmented Dickey-Fuller (1984) and Phillips-Perron (1988). The null hypothesis of both tests is the presence of a unit root against the alternative hypothesis stating the stationarity of the data. The results of both tests are summarized in Table 2 and confirm the non-stationarity of the employed data; in other words, the variables used in the analysis are integrated of order one, I (1).

Granger Causality Analysis
To assess whether changes in oil prices cause changes in the M3 money supply in Saudi Arabia, the most popular causality test, which was developed by Granger (1969) was implemented. The basic intuition behind this test is that the past changes in values of one of the bivariate variables, in this case the price of oil, in predicting the current changes of the other variable, the money supply.
Put simply, the vector autoregressive (VAR) framework is the suitable model regressing a certain variable to its own past values and the past values of other variables. Therefore, to evaluate if changes in oil prices are able to predict the movement of money supply, we estimate the following bivariate model: (1) where ∆ and ∆ represent the changes in oil prices and money supply respectively at time t, while 1 and 2 are the error terms in both equations. The coefficients needed to be estimated are δ 1i and φ 1i where i = 1, … . , s, in which s is the lag length to be determined based on the Akaike information criterion (AIC) for the VAR model.
Once we estimated the VAR model with difference data, we would be able to test whether changes in the oil price cause changes in money supply. If we found evidence suggesting that variations in the money supply would be captured by past variations in oil prices, then we would have a unidirectional causality running from oil prices to money supply. The null hypothesis needs to be tested stating that changes in oil prices do not "granger cause" changes in the money supply; in other words, we need to test 1 = 2 = 0. Table 3 presents the Granger causality results revealing the absence of not only the bidirectional causality running from changes in oil prices to changes in money supply, but also the unidirectional causality running from both variables to each other. As can be seen, there is no statistically discernable causal relationship between the oil price and the Saudi M3 money supply. The lack of a causal relationship between the oil price and the M3 money supply could be due to several factors. First of all, if the oil revenues are not spent immediately, the transmission mechanism of domestic government spending would be absent. It is important to note that the impact on liquidity can be neutralized by appropriate liquidity measures on the part of SAMA (e.g., the liquidity injection operations). In other words, a combination of fiscal policy on the part of the government and appropriate liquidity measures on the part of SAMA can also effectively decouple oil prices and the M3 money supply (Note 4).
Thus, the lack of a discernable causal relationship between oil prices and the broad money supply may be evidence of the success of prudent and stable fiscal and monetary policy on the part of the Kingdom's government and its central bank. Indeed, the success of such macroeconomic policies has led to maintaining the stability of its exchange rate participating in stabilizing price levels during the previous decade as well as having a financial sector that is resilient to external shocks such as those stemming from the oil market.

Conclusion
The main objective of this research paper is to evaluate whether changes in oil prices can result in money supply volatility in an oil-exporting economy such as Saudi Arabia. To do so, we utilize monthly observations for the broad money supply (M3) and the Arab Light crude oil price from January 1982 to April 2018. It is important to note that this paper has utilized revised money supply data prior to June 1988, which corrects the distortions resulting from the old Hijra data by adopting the approach of Qualls et al. (2017), in order to convert the data from a Hijra calendar basis to that of the Gregorian calendar basis. By doing this, our analysis should be more accurate and robust. Once the issue of data calendar compatibility has been dealt with, we are able to apply the most popular causality test known as Granger's (1986) causality test, using a VAR framework. The empirical results reveal the absence of causality between the two variables. The lack of causality running between these variables can be attributed to several reasons, as discussed in the previous section. Nonetheless, some credit should be given to the prudent and stable fiscal and monetary policy on the part of the Kingdom's government and its central bank.
The empirical findings of this research paper open the door for future research that could explore the application of new econometric techniques such as non-linear causality tests or non-linear cointegration tests. In the same vein, it provides an opportunity to investigate whether the financial sector is resilient to oil price shocks or not; for instance, this could be reached by assessing the impact of oil shocks on financial markets or banking system profitability in Saudi Arabia.