The Impact of Oil Price Shocks on Economic Growth: The Case of Taiwan

Most studies continue to analyze oil shocks. Earlier authors recognize that oil price volatility plays a critical role in the economy. There is accordingly evidence that oil price shocks negatively impact real gross domestic product (GDP) growth rates and cause higher inflation. However, this paper uses different perspectives to investigate whether it is beneficial to Taiwan’s overall economy based on a low oil price event. This study’s results reveal that an increase in the oil price leads to an increase in the consumer price index (CPI), causing higher inflation. Moreover, a long-term rise in oil prices would negatively impact the GDP growth rate. Alternatively, in the event of falling oil prices, there may not be an immediate decline in the price of goods. However, firms’ reductions in the cost of goods resulted in declining CPI, due to the decreasing oil price over the past few months. Furthermore, it was observed that GDP would decrease when there is a long-term decline in the oil price. All the previously mentioned results are almost consistent with those of previous studies of high oil price events. In addition, the economy would be negatively impacted by a long-term decline in oil prices.


INTRODUCTION
According to Blanchard and Gali (2007), who examined high oil prices as the main reason for the recession, the oil price is an important source of economic fluctuation. Previous literature indicated that high oil prices would cause several phenomena: increased costs of producing goods and services could lead to lower aggregate demand (Cunado and De Gracia, 2005;Cologni and Manera, 2008) in household consumption, thereby affecting firms' production activities and earnings. In addition, oil price shocks affect corporate performance, further impacting financial markets. Hamilton (2005) found that oil price shocks lead to increased costs of production, worsening firms' earnings and market valuations. On the other hand, higher energy costs decrease household demand, reducing firms' output and further influencing the labor market (Al-Tai, 2015;Vizek et al., 2020). Furthermore, the higher energy cost reduces disposable income and increases households' precautionary savings. Thus, reduced consumption adversely affects corporate profits (Tsai, 2013).
The high oil price event would cause low investment, influenced mainly by declining corporate profits, economic instability, and inflation, leading to diminished consumer purchasing power. Moreover, some studies claimed that oil shocks negatively impact real GDP growth rates and cause increased inflation (Adebayo, 2020;Bachmeier, 2007;Cunado and De Gracia, 2005;Darby, 1982;Gershon et al., 2019;Gao et al., 2014;Lacheheb and Sirag, 2019;Hamilton, 2005;Nusair, 2016;Shaari et al., 2012;Zhao et al., 2016). Notwithstanding the above, a decline in oil prices would have positive benefits, such as increasing the GDP growth rate, corporate profits, and consumer spending. Sadorsky (1999) suggests that depressed oil prices will cause positive shocks in stock returns, interest rates, and industrial production. Furthermore, Mohaddes and Raissi (2016) found that a drop in oil prices brought global growth to 0.4%. However, under different market conditions, the decline in oil prices affects investor sentiment, causing stock returns to fall (You et al., 2017).
A few years ago, oil prices fell by about 50% between June 2014 and February 2016. The initial gradual slide was from $110 to $25 for each barrel of Dubai crude oil ( Figure 1). In considering the period for which oil prices continued to decline, next, this paper will examine whether the low oil prices may have positively impacted Taiwan's economy, such as increased household budgets and improved GDP.
The remainder of the paper is organized as follows: Section 2 describes the sample and data. Section 3 describes the methodology. Section 4 reports the analyses the results, and Section 5 concludes.

SAMPLE AND DATA
The data were collected mainly from Taiwan Economic Journal (TEJ), and the study period is 2006-2015. Table 1 describes the contents of each variable.

METHODOLOGY
The study proposes a test of the null hypothesis where an observable time series is stationary around a deterministic trend (Kwiatkowski et al., 1992). In fact, most macroeconomic variables were probably nonstationary. The economic variables need to do the unit root test and co-integration test to avoid spurious regression, and it is important to examine the stability of this process. However, this testing procedure would apply the unit root of the Augmented Dickey-Fuller (ADF) test and Philips-Perron (PP) test, co-integration, the Granger causality test, and impulse response.

Augmented
Where y t is variable, a denotes the floating items, t is time trend and ɛ t is error term and then ɛ t ~iid (0,σ 2 ), p is the lag length.
The null hypothesis is H 0 :α 1 =0, when reject the null hypothesis, it indicates that this sequence is stationary time series, also known as I (0) series. However, if the original series was unable to reject the null hypothesis, it may be used the difference to stationary of time series. After the first difference, we observed the sequence is stationary, it denotes I (1).

Co-Integration Model
Two (or more) time series with long-term trends, it means that the time series have co-integrated relationship. Model is as follows: The variables of Y t , X 2t and X 3t are nonstationary, thus regression function as follows:

Granger Causality Test
The main is that predict the relationship of variables. In order to study the causal relationship between X and Y, then consider the following regression equation:

EMPIRICAL RESULTS
We used the database of TEJ and collected data from 2006 to 2015. This study applied E-views as the analysis tools. The empirical results, as follows:  50,145.98, 4,247.97, and 38,189, respectively. In addition, from the perspective of standard deviation, the study found sharp fluctuations in the wage level, and fluctuations in the minimum United States Dollar (USD) exchange rate to the New Taiwan Dollar (NTD). Skewness and kurtosis coefficients are used to determine data patterns, for which the skewness for normal distribution is zero and kurtosis is three. In this regard, the study observed that the skewness coefficient is less than zero for the CPI, GDP, TAIEX index, and oil price; moreover, there is a tendency to the left side. On the other hand, there is a tendency to the right side for the exchange rate, Taiwan's household expenditure, and wage level; however, the maximum skewness coefficient is 3.02 for the wage level. The elements for which the kurtosis coefficient is less than three include the CPI, exchange rate, Taiwan's household expenditure, and the oil price.

Quantile-Quantile (QQ) Plot
Following the descriptive statistics, the study will examine the normal distribution of the variables. The QQ plot ( Figure 2) was accordingly used to observe changes in the variables. Zivot and Wang (2007) examined a scatterplot of the standardized empirical quantiles of y t against the quantiles of a standard normal random variable. In this regard, if y t is normally distributed, the quantiles will lie on a 45-degree line. However, as depicted in Figure 2, the results of the QQ plot reveals a are non-linear 45-degree for each variable, reflecting a non-normal distribution.

Analysis the Results of Unit Root Tests
The study will examine the stability of the variables. To this end, it used the ADF and PP tests. If the P-value rejects the null hypothesis, the time series is stationary. However, for this study, it was non-stationary.
In statistics, a unit root test examines whether a time series variable is non-stationary and possesses a unit root. Based on the reported estimates, the null hypothesis was unable to be rejected for the CPI, oil price, exchange rate, household expenditure, and the TAIEX index. However, the null hypothesis of GDP and wage were rejected. The test results are shown in Table 3. Table 4 shows that the first difference of each variable is stationary. Therefore, it is suggested that it is best described as being stationary in the first differences for each series.

Co-integration Test
Zakrajsek (2009) indicates that economic theory often suggests that economic variables should be linked by a long-run economic relationship. Thus, if two or more I (1) variables are co-integrated, they must obey an equilibrium relationship in the long run, although they may diverge substantially from that equilibrium in the short run. Therefore, this paper used the Johansen cointegration test to estimate the trace test and the max-eigenvalue test. The study will use different perspectives to investigate the different reactions of CPI and GDP to the high oil price (April 2006 to November 2011) and low oil price (November 2012 to December 2015) events.
Tables 5-8 indicate that the trace test and max-eigenvalue test of CPI and GDP during the high oil price event have a co-integration     −9.7778 (0.0000)*** −11.4324 (0.0000)*** ***, **, and * indicate significance at the 1%, 5%, and 10% levels. relationship; therefore, reflecting a long-term stable equilibrium relationship. On the other hand, the results of Tables 9-12 indicate that the trace test and max-eigenvalue test of CPI and GDP during the low oil price event have a co-integration relationship; therefore, reflecting a long-term stable equilibrium relationship.

Granger Causality Test
The result from Table 13 indicates a causal relationship between the CPI, Taiwan's household expenditure, GDP, the TAIEX index, and the high oil price event. Moreover, a significant causal relationship between the exchange rate and oil prices and Taiwan's        However, during the high oil price event, the oil price shocks impact the economy, including the CPI, GDP, Taiwan's household expenditure, and the TAIEX index. On the other hand, changes in oil prices impact the CPI and GDP during the low oil price event.

Impulse Response Function
The study uses different perspectives to investigate whether high and low oil prices impact the economy differently. Therefore, this section aims to use impulse response to observe the relationship between them.
The test results from Figure 3 show the response of the high oil price event to shocks to the CPI, Taiwan's household expenditure, and wage, respectively. The first four period curves are raised, and then a flat curve occurred after the fifth period of CPI and wage.
The results indicate that the high oil price event causes the cost of producing goods and services to increase, leading to higher inflation. In addition, most studies show that an increase in the oil price affects production activity and corporate earnings, thus affecting salary. However, there are different theories from the past generation. It was observed that during a high oil price event, the salary levels rose sharply. Possibly, the resulting price increase was passed on by the firm to the consumers. After the increase in corporate profits, employees earn a pay increase. Moreover, there are negative implications to households. In this regard, Hamilton household expenditures and wages was discovered. Furthermore, Table 14 presents the Granger causality tests from the low oil price event. Based on these results, a significant causal relationship was observed between the GDP, CPI, and oil prices.
The results suggest that oil price significantly affects both GDP and CPI, and the exchange rate significantly affects the oil price.  Wage 0.0000*** ***, **, and * indicate significance at the 1%, 5%, and 10% levels.
(2005) refers to oil price shocks leading to reduced household disposable income. Furthermore, the results reflected in Figure 3 reveal that a high oil price event shock the GDP, exchange rate, and TAIEX index, respectively. The first four period curves are raised, following which the curve began to decline after the fifth period of GDP. This result means that GDP growth rates are negatively impacted during the long-term rise in oil prices. In addition, high oil prices have little effect on the stock market. Generally, when the oil price rise caused the USD to fall, in other words, depreciate, this also results in an appreciation of NTD. Therefore, it was observed that when the oil price rose, there was a sharp appreciation of the NTD. Figure 4 shows that a low oil price event caused shocks to the CPI, Taiwan's household expenditure, and wages, respectively. The first three period curves are raised, and then the curve began to decline after the fourth period of CPI. A decline in the oil price generally helps to reduce inflation, leading to reductions in the price of goods. However, based on the falling oil price event, although the price of goods did not decline immediately, when firms reduced costs, the resultant decline in CPI due to the declining oil price continued over the past few months. In addition, there are negative responses to wages and household demand. Theoretically, the fall in oil price leads to increased firms' earnings, and their market valuation leads to wage increases. Conversely, salary levels decline. The study investigates the period of falling oil prices during which global economic growth slowed. In this regard, although corporate costs decreased, profits and wages did not increase. Surprisingly, the drop in oil prices coincided with a decline in household spending. It was observed that the slowdown in global growth resulted in limited consumer demand and increased precautionary savings -people facing uncertainty regarding future income delay consumption, lead household incomes not to rise. Moreover, Figure 4 reports the results and indicates that a low oil price event resulted in shocks to GDP, the exchange rate, and the TAIEX index, respectively. The first three period curves are raised, and the curve began to decline after the fourth period of the TAIEX index. In further investigating this possibility, it was revealed that investors who are inclined to pay more on the stock market during the period of falling oil prices engender a rise in the stock market. However, long-term falling oil prices cause investor confidence to decline, leading the stock market to fall. Theoretically, the oil price decline leads to increased GDP, but this is a short-term phenomenon because the GDP declined after the third period. Moreover, the first two period curves reflect decline, after which, after the third period of exchange, they began to rise. The estimated results suggest that enduring the long-term effects of low oil prices lead to the depreciation of the NTD.

Regression Analysis
The impact of each variable on the CPI and GDP can be understood from the regression analysis. High oil price events are significant for both CPI and GDP. On the other hand, the same results apply to low oil price events. Particularly in Equation (4.7.3), the changes in the CPI and oil prices move significantly in the same direction, which means that when oil prices fall for a sustained period, CPI also declines: a result that applies to the impulse response.
The results are analyzed as follows: