THE IMPACT OF OIL FACTOR ON THE CAR IMPORT IN AZERBAIJAN

This paper analyzes the impact of oil exports and oil prices on car imports in Azerbaijan for the time span of 2010-2019 (monthly basis). There has been no study evaluating the direct effect of oil prices on import, especially, on car imports. However, similar issues have been studied in the context of the impact of oil prices on GDP or on many macroeconomic indicators of individual countries in general. The methodology used in this study is based on econometric methods which were used to analyse time series data. Stationary tests of variables (ADF, PP, and KPSS) were done. ARDL model was used as an research methodology. To investigate more specific aspects of the long run causality relationship between oil exports and oil imports, cointegration relationships are reassessed by using different econometric models such as FMOLS, DOLS and CCR. These estimates are consistent with the estimates obtained from ARDL model. Some remarkable contributions can be derived from this study toward the regulation of car imports in Azerbaijan. In general, it is concluded that in the long run, there is a positive effect of oil exports and oil prices on car imports in Azerbaijan.


INTRODUCTION
As import and export transactions are carried out in USD, automobiles imported to Azerbaijan are also purchased in USD. However, as the convertible currency (US dollars) flowing into our country is directly related to oil exports and oil prices in the world market, it can be stated that imports, namely, automobile imports are highly dependent on oil. There is no doubt that all factors affecting the import of automobiles and the car market are ultimately dependent on the conjuncture of the global oil market. Thus, after the devaluation in 2015, the import of vehicles into Azerbaijan has fallen more than 12 times compared to 2014-2016. Generally, the main underlying reasons for the decline in imports of cars are the following: -Reduction in public expenditures by government bodies.
-To adjust the exchange rate of the national currency, the central bank of Azerbaijan conducted a "sharp" devaluation in February and December 2015. As a result, the official exchange rate of 1 Azerbaijani manat was $ 1.05 in February and $ 1.60 in December. Declining purchasing power of manat due to almost twice depreciation of the national currency against the foreign currency; -Decreases in car purchases on credit after devaluation. Imports of cars have also been affected by the suspension of consumer loans by banks for 2-3 months after devaluation. During that period, two out of every 3 cars sold in Azerbaijan were purchased on credit. After a while, a new law emerged and it stated that the initial payment for new cars had to be 50% of the price and 80% for the old ones. However, before this new legislation, the initial payment did not exceed 10% of This Journal is licensed under a Creative Commons Attribution 4.0 International License the price. Therefore the import of cars decreased by 30-40% after implementation of the new law; -The motives of banks to issue loans in US dollars prevented people from taking loans. This was because people were afraid that Azerbaijani manat would fall again, the US dollar would rise, and they would not be able to repay the loan. Therefore, the demand for credits fell sharply. -Due to the rise of the US dollar against manat, "administrative reduction" of prices in the local market was observed which influenced the import of foreign cars produced abroad; -Except for Russian cars, the other automobiles are mainly bought in dollars and euros. So importing cars at the previous prices was not profitable. It was impossible to buy cars in dollars and sell them at affordable prices in the local market; -The situation in Russia was also an important factor that could influence. The worsening of the situation in Russia reduced remittances flow from there to Azerbaijan by 2 times. Russia is the main source of income for some regions of Azerbaijan. That is why these regions were highly affected by the bad situation in Russia. The strict implementation of sanctions in Russia, the devaluation of the ruble affected the citizens of Azerbaijan who were sending large sums of money to their families in Azerbaijan.
The strengthening of the national currency thanks to targeted monetary and fiscal measures undertaken by the state, as well as the increase in the average salaries in the country, improved the automobile market. Starting from 2017, import of transport vehicles, especially cars, began to increase and its overall value was close to the figures of 2005-2010, while the value of vehicle imported in 2017 was unlikely to reach the level just before the devaluation. It is not only related to the financial resources of the government and the population but also the fact that the automobile market in Azerbaijan is saturated with cars so that during off-peak hours traffic jams can be observed on the roads.
Increasing traffic congestion in the capital city, Baku, is directly linked to the rise in the number of vehicles in the city. Many experts offer a variety of methods to prevent this. Some advocate restrictions on the purchase of cars based on the year of release, others support imposing restrictions according to their type (truck, car). Currently, there are no serious restrictions in this area, and our citizens are making a significant contribution to the import of cars. So supply for automobiles decreased with regard to switching to the standard and demand for cars dropped due to low willingness to take loans from banks.
The downturn in the car market is directly proportional to several factors such as the decline in oil revenues which is the outcome of falling oil prices, cuts in budget expenditure thanks to worsening oil revenues, and, ultimately, emerging unemployment stemming from the cessation of bridges and road construction at the expense of the budget spending.
The main reason for declining automobile imports is decreasing oil prices which lead to lower oil income flowing into the economy and, eventually, lower disposable income of individuals.
There are very serious ongoing economic processes in the world. The sharp decline in oil prices, devaluation of the national currency in the countries, including the devaluation process in Azerbaijan, certain dependence of Azerbaijan economy from oil, the sharp drop in oil prices, of course, caused certain problems and declines in people's incomes.

LITERATURE REVIEW
The impact of oil prices on imports of oil-producing countries has been analyzed in studies specific to each country. According to a study by the International Organizations (IMF, 2007;IMF, 2008), most of the oil-exporting countries have experienced fluctuations in demand (increases and decreases) since the 1970s. Error Correction Model (ECM) was operated in some of the oil-exporting countries to detect whether the import trend changed or not. After the selective forecasting, it was revealed that actual imports in OPEC countries were slightly lower than the projected import expenditure.  Mwega (1993Mwega ( ) 1964Mwega ( -1989 Kenya ECM Import demand is less elastic than relative prices and income. Currency reserves are the main factor in determining imports Senhadji (1998Senhadji ( ) 1973Senhadji ( -1998 77 countries OLS FMOLS

Monte Carlo
In terms of profitability, elasticity is relatively low for oilexporting countries. Because export revenues take a significant part of national income in these countries Lim and Kim (2002) 1962 North Korea Cointegration Some non-market factors are important determinants of imports Tang and Nair (2002) 1970-1998 Malaysia UECMBounds Test Import demand, income and relative prices are cointegrated -Oskooee, and Kara, (2003) 1973Q1-1998Q2 9 industrialized countries.

ARDL
Long-term income elasticity is higher than the import demand function; trade flows from different countries react differently Metwally (2004Metwally ( ) 1968Metwally ( -2001 GCC Countries (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and the United Arab Emirates)

VAR
The drop in oil prices led to a sharp drop in imports of all oilproducing countries Changes in GDP have a strong impact on import demand in GCC countries. However, changes in relative prices do not have a significant impact on imports in most of these countries In all the studied GCC countries (except Oman) over the past 30 years, the import demand against GDP has been very elastic Dutta and Ahmed (2004) 1971-1995 India UECM There is a cointegration relationship among the volume of imports, relative import prices and real GDP GDP significantly affects import demand Islam and Hassan (2004) 1974Q1-1998Q2 Bangladesh VAR LM ARCH Import demand is mainly determined by income and relative prices Chang et al. (2005) 1980-2000 South Korea ARDL UECM The volume of imports, income and relative prices -these are all interrelated Katsimi and Moutos (2006) 1948 USA VECM VAR

Income inequality affects import demand
There is no sufficient evidence to support the existence of long-term import relationships (including imports, income and relative prices) There are cointegration equations for import, income and relative prices in the VAR specification. The level of income has a remarkable and positive impact on the import demand in the United States Rehman (2007Rehman ( ) 1975Rehman ( -2005 Pakistan VAR Johansen-Juselius There is a long-term equilibrium relationship between the variables The import demand function remains constant throughout the sampling period Ziramba (2008) 1970-2005 South African UECM Import volume, relative prices and real income (GDP) are integrated Oteng-Abayie and Frimpong (2008)

1970-2002
Ghana ARDL There is an inelastic and positive relationship between three expenditure components and import demand. Relative price is also inelastic but has a negative impact on the overall demand Adam et al. (2008Adam et al. ( ) 1970Adam et al. ( -1997 59 countries, developing and developed ones

OLS
Inequality has a great impact on import demand It affects positively in high-income countries and negatively in low-income countries Shareef and Tran (2008)

Australia
ARDL UECM The demand for imports does not depend on price or income. In the short run, the price is more elastic than income In the short run, price and income are the key factors of demand for imports Ozturk and Acaravci (2009) 1975-2005 Latin American and Caribbean countries

Dynamic panel data methods
The volume of imports demanded is negatively correlated with relative prices and positively correlated with real income Alam and Ahmad (2010).

1982Q1-2008Q2
Pakistan ARDL There is a long-term relationship between the demand for imports, the real economic growth, the relative price of imports, and the real effective exchange rate Narayan and Narayan (2010) 1960-2005 Mauritius and South Africa ARDL There is a long-term relationship between imports, income and prices Domestic income and relative prices have a significant impact on import demand in both countries, and income is the most important factor Serge and Yaoxing (2010) 1970 Cote d'Ivoire ARDL In the long run, investment and exports are key factors in imports. In the short run, both components of expenditure are key determinants of import demand. However, import demand is not sensitive to price changes The long-term relationship between imports, national income and the relative price is proved Alam and Ahmad (2010)

1982Q1-2008Q4
Pakistan ARDL There is a long-term relationship between import demand, real economic growth, the relative price of imports, and the real effective exchange rate General import demand has a positive impact on GDP. The relative price of imports does not reduce import demand The devaluation of the local currency has no effect on the reduction of import demand The import demand is inelastic against the exchange rate Change of import demand is a short-term event Rashid and Razzaq (2010) 1975 Pakistan ARDL DOLS There is a long-term certain relationship among variables included in the import demand model Ozturk and Acaravci, (2011)

1993Q1-2003Q3
Slovakia ARDL Real import, relative price and real GDP are cointegrated. There exists a stable import demand function Moutos and Katsimi (2011) 1948 USA VECM VAR A long-term relationship exists between imports and income level There is cointegration among import, income and relative prices. The level of income has a large and positive impact on the import demand in the United State Ozturk and Acaravci, (2011)

1993Q1-2003Q3
Slovakia ARDL There is a cointegration link among real imports, relative prices and real GDP. The import demand function is stable Wang and Lee (2012)

1992M01-2011M07
China ARDL Import is related to internal economic activity, efficient exchange rates and global risks Domestic income has a significant positive impact on imports Real effective exchange rates have a negative impact Declining foreign market competitiveness (rising prices) will lead to lower imports Knobel (2013) 2000-2010 Russia OLS The demand for imports is highly sensitive to real effective exchange rates and import prices changes Gozgor and Oktay (2013) 1989Q1-2012Q2 Turkey ARDL The decline in the value of the Turkish lira has a limited impact on all imports GDP has a greater impact on fixed assets in the short run and consumer products in the long run Durmaz and Lee (2015) 1980-2011 Turkey ARDL There is a long-term relationship between the dependent variable and the independent (explaining) variable in the import demand function. All explanatory variables are statistically significant both in the long-term and the short-term. All independent variables have an inelastic effect on imports, except for total consumption Mishra and Mohanty (2017) 1980-19812013-2014 India ARDL There is a link between import demand, relative prices of imports, domestic activities and foreign exchange reserves In the long run, the reaction of import demand to relative import prices is negative and less than the unit Mohamed (2017)

1970-2014
Egypt OLS ECM There is a positive and significant relationship between demand for imported goods and real GDP in both the long-term and short-term There is a negative and significant relationship between demand for imported goods and real effective exchange rates Hor et al. (2017Hor et al. ( ) 1993Hor et al. ( -2015 Cambodia ARDL Relative prices and exchange rates have a negative impact on import demand, both in the long-term and short-term The volume of exports has a positive impact on import demand Umoru et al.

2000-2017
Nigeria GLS Import demand is heavily dependent on the availability of currency reserves, tariff policies, and final consumption expenditure Olcay et al.

2003-2018
Turkey LS 2SLS The alteration in total imports is mainly due to changes in income and relative prices. The elasticity of income and expenditures over time decreases in total imports. Relative price elasticity remains almost unchanged for investment and import of consumer goods Sharif and Abedin (2019) 1980-2016 8 frontier countries, 8 emerging countries, and 10 developed countries

Cointegration
There is a long-term relationship between import demand, relative prices, exchange rates and real GDP in all countries papers examining the link between oil exports and demands for import consider the reaction of current account balance to changes in oil prices or trade conditions.
Our research covers the impact of oil prices on the import of cars in Azerbaijan. There has been no study evaluating the direct effect of oil prices on import, especially, on car imports. However, similar issues have been studied in the context of the impact of oil prices on GDP or on many macroeconomic indicators (Table 1 )  Ghalayini (2011), by examining oil price fluctuations, concluded that price shocks affect macroeconomic indicators in different ways. Other economists, such as Hamilton (1983), Bruno and Sachs (1985), studied the impact of oil prices on economic development, financial instability and inflation in Great Britain during 1950-1979 and concluded that these variables were closely connected. Increase in oil prices leads to higher prices in the economy, lower employment and productivity (Dornbusch, 2001).
The impact of prices on macroeconomic indicators has been widely studied by Hamilton. Hamilton was one of the first scientists to demonstrate the importance of changing energy prices for the US economy. He has proven (Hamilton, 2008) that rising oil prices are more important than their fall.  analyzed the US economy in 1948-1980 using the Sims et al. (1990) method and the VAR method, and then he concluded that oil prices and GDP in the United States were strongly correlated. Hamilton and other researchers (Gisser and Goodwin, 1986;Mork, 1989;Lee et al., 1995;Hamilton, 1996;Hamilton, 2003) concluded that oil prices had a negative impact on US GDP.
In addition, the impact of oil prices on the exchange rate of currencies has been the subject of research. Thus, some researchers have reported that oil prices had impact on the exchange rate (Amano and Van Norden, 1998;Akram, 2004;Benassy-Quere, 2005;Lizardo and Mollick, 2010). Others have proven the opposite of this statement (Brown and Phillips, 1986;Cooper, 1994): The exchange rate affects oil prices.
Interest toward oil price volatility and its role in macroeconomics revived in the early 2000s with the sharp rise in oil prices and the immediate fall in 2008 (caused by the Lehman crisis) Hamilton, 2013;Yoshino and Taghizadeh-Hesary et al., 2016). A study by Peersman and Van Robays (2012) and Taghizadeh-Hesary et al. (2015) identified the winning and losing economies after the recent shock of oil prices. Aydoğan et al. (2017) assessed the relationship between oil prices and stock markets. It was revealed that that the correlation between oil prices and stock markets varied depending on whether the country was an oil exporter or an oil importer.
As a rule of thumb, oil price fluctuations have a significant impact on the oil importing countries' production costs and, consequently, the price level of those nations (Michael and Jeffrey, 1982). In the countries that are energy exporters, the change in oil prices has a major impact on revenues from oil export and state budget revenues. However, it is widely acknowledged that volatility in energy prices is not only an important cause of macroeconomic shocks, but also affects the fiscal and monetary policies of various countries.

Data Descriptions
The datas used in the study were taken from the sites of Azerbaijan State Customs Committee (www.customs.gov.az) and OPEC (www.opec.org) ( Table 2). Descriptive statistics was in Table 3. World oil prices, dynamics of oil exports and car imports (2010-2019) were described in Figure 1.

Methodology
The methodology used in this study is based on econometric methods which were used to analyse time series data. In this situation, we are considering two important steps in econometric methodology. The first step involves the investigation of stationarity of variables included in the model and the utilization of Augmented Dickey-Fuller (ADF), Phillips-Perron (PP), Kwiatkowski-Phillips-Schmidt-Shin (KPSS) tests for this  purpose. The purpose of using three different unit root tests is to compensate for the possible weakness in any of them and ensure the reliability of the test results which can be potentially affected by the limited quantity of data. Due to unit root test results, it will be revealed whether the variables used in the model are nonintegrated (I(0)) or integrated of order 1 (I(1)). The second step includes the use of cointegration methods. More specifically, it is necessary to analyze existence of short and long term relationships among used variables. In this case, Johansen's multidimensional coordinate approach or the Auto Regressive Distributed Lags Model (ARDL) and the Pesaran and Yongcheol (1999) boundary test will be used. To test the long term relationship among variables, ARDL models and boundary tests for cointegration approach will be utilized.

ARDL bounds testing approach to cointegration
This research is based on ARDL models and boundary testing for the cointegration approach which was developed by Pesaran and Yongcheol (1999) and Pesaran et al. (2001). These models have recently been used extensively to test the existence of long-term relationship between various macroeconomic variables. The main advantage of this approach is that it does not require the same order of integration for each of the variables. In other words, this allows the inclusion of non-integrated time series data and time series data integrated of order 1 or more into the model simultaneously.
The implementation of the ARDL method consists of three stages. The first stage examines the existence of unit roots for utilized time series data by using ADF (Dickey and Fuller, 1979), PP (Phillips and Perron, 1988), and KPSS (Kwiatkowski et al., 1991). Three tests are used to check the reliability of the results.
In the second step, the following unrestricted ECM (Unrestricted
Lag p and q are selected based on the Akaike (AIC) information criterion. The Breusch-Godfrey Serial Correlation LM, Jarque-Bera Normality, ARCH and Breusch-Pagan-Godfrey tests are to be used to validate the estimated models. In addition, the following hypotheses are tested for each model: Н 0 :θ 0 =θ 1 =0 and Н 1 :θ 0 ≠θ 1 ≠0. The null hypothesis assumes that there is a cointegration relationship between variables. Wald test is also developed for decision-making procedure based on the F-test. Critical values for the F-test were given in Pesaran et al. (2001), but complemented by Narayan (2005) which included small recent additions. There are two asymptotic critical values bounds: one is lower bound and the other is upper bound. A lower critical value assumes that the regressors are all non-integrated (I(0)), while an upper critical value assumes that regressors are all integrated of order 1 (I(1)). Their values depend on the number of observations, the number of independent variables and the probability levels. The null hypothesis is rejected when the value of F-statistics exceeds the upper critical value. In this case the variables are cointegrated. However, when the value of F-statistics is lower than the critical value, we can't reject the null hypothesis. We understand that variables are not cointegrated at this time. Finally, it is not possible to draw conclusions when F-statistics are located between two critical values.  *, **and *** indicate significance at levels 5%, 1% and 0, 1% respectively. The optimal lag length for Augmented Dickey-Fuller (ADF) (Mackinnon, 1996) tests was determined using the Schwarz criterion. Phillips-Perron (PP) (MacKinnon, 1996) and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) (Kwiatkowski et al., 1992) tests was operated by using Bartlett kernel spectral estimation method and Newey-West Bandwidth. The maximum lag used in the test calculations were given in the brackets. Assessment period: 2010:01-2018:12 Unrestricted ECM: Impact of changing oil exports and oil prices on import of cars.

Long run granger causality test
The long-term relationship equations are evaluated, when the results indicate that the variables are cointegrated. In this case, the unrestricted ECM (UECM) which is given in equations (3) and (4) is analyzed in order to determine the short-term dynamics and correction rate.
Afterwards, long-term cause and effect relationship between dependent and independent variables is examined in each Unrestricted Error Correction Model (UECM). The negative sign (π) of the error correction coefficient indicates that there is a long-term causal relationship between the independent variables and the dependent variables.

Results of Unit Root Tests
As mentioned earlier, the stationarity of variables is tested using ADF, PP and KPSS tests. The results of the three unit root tests are shown in Table 4. Nearly all three tests give the same results, which confirm the validity of the test results. Therefore, it can be assumed that none of the variables are integrated at the second level.

Results of ARDL Models
Since all variables are found to be either I(0) or I(1), Johansen's multidimensional cointegration approach cannot be used. However, ARDL bounds test for cointegration can be used. Therefore, two ARDL models are presented and analyzed in equations (1) and (2). The results of model selection criterion are presented in Table 5.  −0.696575*** ∆Import gar (t-1) −0.508886*** −0.499663*** Import gar (t-1) 0.869924*** 0.893430*** Constant −3.232732* −7.023751*** ***, ** and *indicate rejection of the null hypotheses at the 0,1%, 1% and 5% significance levels respectively The stars *, **, and *** represent the 10%, 5% and 1% levels of the significance, respectively. The lower and upper boundaries at the significance level of 10%, 5% and 1% are determined by EVIEWS 9 software. The dependence of car exports on oil prices and oil exports were given as a linear dependence ( Table 6). Then ARDL model was drawn (Table 7). The results of the diagnostic tests applied to the models are shown in Tables 8 and 8a. The results of the Jarque-Bera Normality, Breusch-Godfrey Serial Correlation LM, ARCH, and Breusch-Pagan-Godfrey test show that in the two models given in (1) and (2), the residuals are normally distributed, homoscedastic, and there is no serial correlation among error terms at 5% significance level. Finally, the results of the tests of CUSUM and CUSUM squares are shown in Figure 2 respectively. Those outcomes indicate the effect of oil exports on car imports and the impact of oil prices on car imports. It has been shown that in 5% significance level calculated CUSUM and CUSUM of squares plots are between two boundary lines in all figures (Figure 2). Therefore, the coefficients of the models are dynamically constant. Thus, we can note the reliability of the ARDL models.
Thus, all ARDL models given in equations (1) and (2) pass all diagnostic tests smoothly. Bound test examined the existence of long-term dependency ( Table 9). The results of both models are given in Table 10. They show that there is a long-term relationship.
Due to test results, it can be claimed that there is a cointegration relationship between oil exports and car imports at the significance level of 5% (Table 11)   −9.412986*** −9.314620*** Stationarity S S ADF denotes the Augmented Dickey-Fuller single root system respectively. The maximum lag order is 2. The optimum lag order is selected based on the Shwarz criterion automatically; ***, ** and *indicate rejection of the null hypotheses at the 0,1%, 1% and 5% significance levels respectively. The critical values are taken from MacKinnon (1996)   −9.412986*** −9.314620*** Stationarity S S ADF denotes the Augmented Dickey-Fuller single root system respectively. The maximum lag order is 2. The optimum lag order is selected based on the Shwarz criterion automatically; ***, ** and *indicate rejection of the null hypotheses at the 0,1%, 1% and 5% significance levels respectively. The critical values are taken from MacKinnon (1996).  (1) and (2) are estimated by the indicated models (Tables 6 and  7). The long-term Granger-causality relationship among different variables is determined by by t and ECT t−1 -error correction term in each equation. The estimation of the correction coefficients of the models in Equations (3) and (4), as well as the long-term and short-term estimations, are given in Tables 10 and 12. The results show that the error correction coefficient is negative and significant at the 1% significance level in both of the models. These results confirm the existence of long-term relationships between different variables. They indicate that there is a long-term causal link between oil exports and car imports, and a long-term causal link between oil prices and car imports.
The results of the estimates show that oil exports and oil prices have a significant impact on the growth of automobile imports in the long run.

Robustness of the Results
To investigate more specific aspects of the long run causality relationship between oil exports and oil imports, cointegration relationships are reassessed by using different econometric models such as Fully Modified Ordinary Least Squares (FMOLS), Dynamic Ordinary Least Squares (DOLS), and Canonical Cointegrating Regression (CCR). Summaries of these reassessment and its results, as well as some diagnostic tests, are given in Table 13. In general, the results of the Jarque-Bera, Phillips-Ouliaris, and Engle-Granger tests indicate that all obtained estimates are valid. In addition, FMOLS, DOLS, and CCR estimates are consistent with the estimates obtained from ARDL model. Therefore, it can be claimed that the results obtained in this research are reliable.

CONCLUSIONS AND RECOMMENDATIONS
This paper analyzes the impact of oil exports and oil prices on car imports in Azerbaijan for the time span of 2010-2019 (monthly basis). Some remarkable contributions can be derived from this study toward the regulation of car imports in Azerbaijan. Likewise, this paper can play a valuable role for recent important economic topics such as the effect of oil revenues on economic growth. In this study, FMOLS, DOLS, and CCR are used to test the reliability of long-term ARDL results. The outcomes are similar for different cointegration methods and techniques. In general, it is concluded that in the long run, there is a positive effect of oil exports and oil prices on car imports in Azerbaijan.
As a result of the research, the following should be noted: First and foremost, economic growth in Azerbaijan, income of people, and the increase in car imports caused by them are mainly dependent on oil income. Secondly, more attention should be paid toward the utilization of oil revenues in Azerbaijan and automobile imports. Oil revenues should be directed to more productive projects that accelerate economic growth. Third, Azerbaijan needs to strengthen its diversification policy to reduce its dependence on oil revenues.  ***, ** and *indicate rejection of the null hypotheses at the 0,1%, 1% and 5% significance levels respectively