Impact of Inflation Uncertainty on the Prices of Industrial Products

Transcription

1 World Applied Sciences Journal 20 (8): , 2012 ISSN IDOSI Publications, 2012 DOI: /idosi.wasj Impact of Inflation Uncertainty on e Prices of Industrial Products M. Taslimi, M. Goudarzi and R. Rostamian 1 Agricultural Economics, Islamic Azad University, Qaemshahr Branch 2 Department of Agricultural Economics, Islamic Azad University, Qaemshahr Branch, Iran Abstract: Inflation rate in Iran has been a growing concern and gradually has become e country s economic problems. Despite e intense struggle, e rate of inflation in Iran has been growing during is time. The grow of Industrial production rate has been increased. This study compared e influence of price indices for agricultural and industrial products by inflation uncertainty using time series data The inflation uncertainty is estimated by using GARCH models. The desired model using vector auto regression (VAR) was evaluated. Impulse response functions (IRF) and variance decomposition has also been checked. Results from e VAR test in e industry section indicate at e industrial added-value, real exchange rate variable, e volume of liquidity and inflation uncertainty are in positive relation wi e price index industrial products. Impulse response functions in industrial sector have been studied to e incoming shocks of all variables. The variance decomposition results indicate at in a short, middle and long period of time, most contributions of fluctuations for industrial products' price index are followed by e industrial products price and ey were about respectively, 72%, 55% and 50%. The results show at e inflation uncertainty variable along wi oer variables has a significant impact on industrial products' price, especially in a long term. Thus, in studies which analyze e behavior of industrial products' prices and affecting factors, e inflation uncertainty variable should be considered along wi oer variables. JEL: C13 C53 E31 Key words: Inflation uncertainty Industrial prices GARCH model Vector Auto Regression model (VAR) Johansson integration INTRODUCTION economic grow [3]. Inflation is often being considered as a measure of overall macroeconomic situation and Industry is one of e important and vital parts of macroeconomic instability and uncertainty [4]. Therefore, each economy for grow and development, which is e several studies have been performed in is field. Such as target of most communities including developing [5] discussed e effect of increased production of countries. For many years, economic grow and agriculture, industry and services on inflation in Iran. development in developing countries was followed up The results show at for e control of inflation in Iran, rough is sector to solve problems and to expend many it is essential at government s economic policies pay costs [1]. In most developing countries, e fundamental more attention to e prices and expand production in and structural constraints in e supply side economics agriculture and services sectors [6] examined e effect of are major concerns at limit supply grow and cause e government's credit policy in industry sector along wi emergence of inflation pressures [2]. Inflation imposes monetary and fiscal policies. The results suggest at costs on a society. The harmful effects of inflation ere is a strong and stable positive relationship between include: redistributing e income in favor of asset owners bank credits and industry value added at represents and detriment of wage and salary, increasing e inflation excessive dependent industrial production to is rates and volatility in macroeconomic, shortening e time governmental policy variable. [7] examined e link horizon of decision making, reducing capital investment between inflation and inflation uncertainty using in manufacturing activity and, as a result, decreasing e Granger causality test indicates at inflation leads to Corresponding Auor: Mahsa Taslimi, Islamic Azad University. Qaemshahr, Mazandaran, Iran. Tel:

2 higher inflation uncertainty in Iran but e reverse y t= x' t + t (1) relationship is not significant [8]. Used quarterly data in multivariate EGARCH models. This study found at bo t = + i t i + j t j (2) inflation uncertainty and output uncertainty had negative and significant effects on output grow. The study also y tis e dependent variable in period t, x tis e dependent found at inflation uncertainty and e inflation level had variable in period t and t is e residual in e period t. bo declined since e adoption of a formal inflation- The equation (1) is a conditional mean model, as a targeting monetary policy in Australia. [9] Studied e function of exogenous variables wi e disturbance. causal relationships between inflation, output grow and Since e variance of each period is forecast by a uncertainty for e Turkish economy based on e prior period it is called e conditional variance. system-garch meodology. The estimation results The conditional variance equation (2) is a function of e revealed at policies aiming at reducing inflation would followings: lead to a more efficient functioning of e price system and is would contributed to e real output grow. [10] Average investigated e behavior of Japanese stock market News about volatility in e last period, which is volatility wi respect to a few macroeconomic variables delayed by residual of squares variable obtained from including gold price, crude oil price and currency 2 e equation t i. This statement is called ARCH. exchange rates (Yen/US$). The results showed at Forecasting variance of last period 2 t i. This macroeconomic variables used in is study had no component is called a GARCH. impact on e volatility of Japanese stock markets and e simplest GARCH (1, 1) model yielded e best result. The necessary condition for positivity of conditional The aim of is study is to investigate e impact of variance is at all e coefficients should be positive. industrial products prices on inflation uncertainty and demonstrate at e industrial product prices are affected a i > 0 i = 1,2,...,p (3) by inflation uncertainty. Therefore, e macroeconomic variables and vector autoregressive model (VAR) are used j > 0 i = 1,2,...,q (4) for is study. The data used are annual dating from 1974 to The results are estimated by EVIEWS6.0 And also should have: > 0. Sufficient conditions for e GARCH (p,q) for being weak stationary is at: MATERIALS AND METHOD p q i= 1 i + i= 1 j < 1 (5) Structural and eoretical models are generally used to predict e time series. Structural models are based on If e shocks are unstable en e model will provide e eory of single-equation regression models and a sufficient condition [13]. In order to evaluate e simultaneous equations, while non-eoretical models are inflation uncertainty variable, e model is estimated in not based on e eory and e future behavior of different states using e Box-Jenkins meod and e variables and are determined by eir past behavior plus best optimal lag is obtained by using Akaike information e error term at is not predictable. These models criterion. The optimal lag is chosen by taking into account include e ARIMA, AR, MA (Box-Jenkins technique e lowest obtained Akaike information criteria. Vector 4 is used) and VAR models [11].In order to measure and autoregressive meod is an option along Box-Jenkins s estimate e inflation uncertainty (IU) variable, e meods at is similar to simultaneous equations. Generalized Autoregressive Conditional Heterokedasiticy The simultaneous equations meod was severely (GARCH) model has been used. The conditional variance criticized by Christopher Sims and VAR option was of e error term follows an ARIMA process. Here q is e proposed by Sims (1980). Sims believes at is eory rank of moving average ARCH and p is e rank of cannot provide enough limitation for identification of Autoregressive GARCH. The GARCH (p,q) can be written structural models [14]. In VAR model, ere are some as follows: [12] endogenous variables at each of em was explained by 1 Auto Regressive Integrated Moving Average. 2 Auto Regressive. 3 Moving Average. 4 Vector auto regressive. 1140

3 k t =δ+ 1 t k t k + t =δ+ j=1 j t j + t Y A Y... A Y v A Y v (8) IM and EX and GDP, respectively, represent e total k Yt = j=1 AY j t j + v (9) amount of imports, exports and GDP. t RER: real exchange rate. In is study, e country's To simplify e formula, e intercept is removed. real exchange rate is determined by notion's purchasing Also assumed at all variables have one or zero order power parity (PPP). For is purpose, e following cointegration. This sample: equation is used: [17] k 1 t t 1 j=1 j t j t Y = BY + B Y + v World Appl. Sci. J., 20 (8): , 2012 eir past values and e values wi interruption of all been due to e influence of variables from various oer endogenous variables models. The two time series sectors of e economy including e price indices of x t and Y tfor two variables will be as follows: [15] products and from e state of economic instability and fluctuation in prices. The study investigates e k n Xt = a0 + j=1 BX j t j + i=1δ jyt i + u (7) relationship between e price index of industrial products 1t and uncertainty. A remarkable aspect of is study is at k n Yt = a0 + j=1 AX j t j + i=1λ jyt i + u (8) such a study has never been conducted before. Wi 2t regard to research goals and internal and external reviews of studies performed, e desired pattern can be specified VAR model is estimated by OLS model and e as follows: results of is model depend on entered variables and lag leng. In connection wi e stationary of variables, e LN(IND) = 0 + 1LN(INVID) + 2LN(OPEN) + non-stationary variables existence intensifies e 3LN(RER) + 4LN(M 2) + 5LN(CPI) + IU + u likelihood for e existence of spurious regression and (13) cointegration relationships. Therefore, in e VAR model at contains a series of non-stationary, existence of The variables are: cointegrated vectors should be tested. In order to IND: Industrial product price index, e indicator measures estimate e model by using vector autoregressive (VAR), e amount of output from e manufacturing, mining, e stationary state must be considered first: 1. if e electric and gas industries, INVID: value added of variables are stationary at e level, variables of e model industries, The value added of an industry, is e will be explained. 2. If e variables are stationary at e contribution of a private industry or government sector to first difference, en e VAR model, as was said earlier, overall GDP and OPEN: degree of trade openness index. will be explained. Then, e cointegration will be reviewed To determine e degree of openness index, e basic by Johansson test [16]. For is purpose, we use indicators have been used as follows: [17] Johansson cointegration meod. Model VAR in (8) wi m variables is written like is: OPEN = IM + EX GDP (14) is written as: (10) WPI RER = ER CPI The WPI is e wholesale price index out of e B = (I A 1 A 2... A k) (11) country and as a close approximation; e wholesale price index of e America has been used instead. CPI is e B j = (A j+1 A j+2... A j+k) j = 1,2,3,...,k 1 (12) consumer price index and ER is e exchange rates in e market. RER indicates e real exchange rate. M2: e Sample (12) is like ECM model where if all variables volume of money liquidity, where liquidity is e sum of have cointegration order one, Y t j variables will be static. volume of money and quasi money. CPI: consumer price Now by assuming e existence of cointegration index. IU: e inflation uncertainty variable. Related data between variables and being BY t i as stationary, a have been calculated and estimated using time series data consistent pattern can be estimated. In recent years, of industrial products price index (IND). For is purpose, inflation uncertainty has been used as an indicator to e Generalized Autoregressive Conditional show e status of macroeconomic instability. This has Heteroskedastic model (GARCH) has been used. (15) 1141

4 Table 1: The results of stationary of variables Variable status Observed value at level Critical value at level st Observed value at 1 difference st Critical value at 1 difference Log ind intercept Log invid intercept Log open intercept Log m2 intercept Log rer intercept Log cpi intercept , and indicate critical numbers in e level 1%, 5% and 10%, respectively Table 2: The stationary of inflation uncertainty of industrial products prices by Dicky Fuller test Variable Status Values Results IU level intercept Observed value Critical value RESULTS The above equation is GARCH (1, 0). The estimated equation provides bo e necessary and sufficient The first step in e analysis of time series variables condition for GARCH model based on eoretical is to study e stationary of variables. Therefore, e principles and foundations, because e necessary stationary of all variables was done by Augmented Dicky condition for weak stationary GARCH model is at e Fuller Test for which e results are given in Table 1. sum of GARCH coefficients needs to be less an one. The null hypoesis in Dicky Fuller Test assumes This is 0.98 in e evaluated equation and is less an one. at e variable is stationary, yet e alternative In oer words, e necessary condition for e incoming hypoesis is vice versa. If e value of statistics is shocks is not stable in disturbance, at is, e sum of greater an e critical Mackinnon, e null hypoesis, GARCH coefficients is less an one. The sufficient e variable at is stationary is accepted. If e value of condition for e GARCH model is at e intercept be statistics is smaller an e critical Mackinnon, e null positive and e conditional variance of e disturbance hypoesis is rejected and e variable is introduced coefficient be positive and significant. The estimated non-stationary. It should be noted at because of equation, us, provides e conditions. After estimating Dicky Fuller Stationary Test is a tailed test for stationary e inflation uncertainty variable, e stationary of e or non-stationary comparison of variables, e absolute variable is tested by Dicky Fuller Test. The results are value is not used. As Table 1 shows, all e industrial shown in Table 2 and indicate at e variable is products price index variables in Dicky Fuller Stationary stationary at e level: Test are stationary wi e first order difference. By examining e variables, it was found at The estimation results for inflation uncertainty variable variables are stationary at first difference. In order to from industrial products price index are in equation 16: explain e model, e maximum lag should be considered at first. Then wi e use of LR, Akaike, Schwartz, 2 2 = t 1 (16) Schwarz Bayesian and Hanan-Quinn criteria, e suitable 1142

5 Table 3: Optimal lag for VAR model Lags Schwarz Bayesian criteria (1,1) (1,2) (1,3) lag will be choosen. According to Ivanova and Kelly (2005), for models wi e sample size of less an 120, e most appropriate measure is Schwartz Bayesian [18]. The results are shown in Table 3 determining e optimal lag. According to Schwartz Bayesian criteria, e lag at has e smallest figures may be determined as e optimal lag. As can be seen in Table 3, e optimal lag is determined at (1, 3). After determining, e optimal lag is evaluated by using Johansen cointegration test. Johansen cointegration test has two matrices which represent Trace and Maximum Eigenvalue Matrix. The null hypoesis says at ere is no cointegration of vectors; it means at ere is not any long run relationship between e variables and e alternative hypoesis says at ere is cointegration of vectors. The level which e null hypoesis can be rejected at indicates e numbers of vectors at are co-integrated. Test results are presented below: As can be seen in e results of Trace and maximum Eigen value test, e null hypoesis of Trace is accepted at e first 6 levels, which means at ere is no cointegration vector. The null hypoesis of Trace is rejected at level 7 wi 0.05 probabilities, but e null hypoesis in e maximum Eigenvalue is rejected at level 4, which indicates at ere is long run equilibrium and cointegration vectors. In e VAR model to interpret e results, it should be noted at e estimated VAR meod basically, in e estimation of equations, coefficients and e percent of explanation of pattern parameters has not e importance of single equation meods. Therefore, e impulse response functions and variance decomposition analysis are used for analyzing [19]. The results of long run relationship between e variables and normalize vector in proportion wi e first endogenous variable are as follows in Table 6: As can be seen in Table 6, e degree of openness index (Log OPEN) and consumer price index (Log CPI) have a significant negative correlation wi e industrial products' price index variable as well as added-value of industrial sector (Log INVID), real exchange rates (Log RER). The volume liquidity (Log M2) and inflation uncertainty (IU) are positive and have a significant relationship wi e industrial products' price index.. Table 4: Trace results of Johansen cointegration test Null hypoesis e alternative hypoesis Observed value Critical value at 95% Probability at 95% r=0 r r 1 r r 2 r r 3 r r 4 r r 5 r r 6 r Table 5: Maximum eigenvalue results of Johansen cointegration test Null hypoesis The alternative hypoesis Observed value Critical value at 95% Probability at 95% r=0 1=r r 1 2=r r 2 3=r r 3 4=r r 4 5=r Table 6: Results of using VAR test for industrial price index Variable Coefficient Standard error Log invid Log open Log rer Log m Log cpi IU

6 Table 7: Impulse response functions result Period Log ind Log invid Log open Log rer Log m2 Log cpi IU Impulse Response Functions: This index shows a variable Industrial product price index response function response to e shock or impulse at is caused by e to e degree of trade openness index: e variable or oer variables over e course of time. industrial product price response to e incoming In oer words, it shows how a shock acts, causes a shocks in a short period of time has no reaction reduction and an increase in a variable and disappears at first but it decreases afterwards. However, in when a shock or impulse enters over time. To calculate, e mid-term period it gradually becomes positive e shocks are inserted according to e size of a standard and continues its trend in a long term period of deviation of a variable and en e reaction is observed time. That is, applying is trade policy has no over e time which will be done in 10 periods in future. effect on industry sector at first, but in e middle The evaluation of impulse response functions of of e period it causes e industrial price to drop. industrial product price index was done for e next Also, e implementation of is policy in a long 10 period, which will analyze e behaviors and reactions period of time will cause e industrial price to to e incoming shocks by e relevant variables in ree develop. periods of short term, midterm and long term. The results Industrial product price index response function to are as follows: real exchange rate: e industrial product price to e Industrial product price response function to incoming shock from real exchange rate in a short industrial product price index: e incoming shock time is negative and after at, temporary in e midfrom industrial product price in e first period causes term period is positive and in a long term period is e industrial product price to increase by 0.16 units. negative again. This fluctuation is caused by e real This in turn causes e industrial product price to exchange rate. increase by 0.17 units in e next period. This trend Industrial product price index response function to attains its peak at e four period and industrial e volume of money liquidity: e incoming shock product price rises by After at, e trend falls from money liquidity to industrial product price is and in e ten period it causes industrial product ineffective at first and has its negative effect by e prices to go up by 0.12 units. In general, industrial middle of e period. But in a long period of time, e product price responses to industrial product price industrial price response is positive and is shows shocks in a short period of time and middle of e at money liquidity grows and at money supply in period will be positive. Also is response will be a long term period causes e industrial prices to positive in a long term but wi decrement. In oer grow. words, e shock from industrial price index will Industrial product price index response function to cause e industrial price index to increase. consumer product price index: e industrial product Industrial product price index response function to price response to e incoming shock from consumer industrial added-value: The incoming shock from product price in a short period of time is negative and industrial sector s added-value to industrial product has a decreasing trend. In e middle of e period, price in a short time period is ineffective at first, but is trend is an ascending one and continues for a in e mid-term period, it is negative and in a long long period of term. In a long term, consumer price term period is positive and causes e industrial index shocks still have negative effects on industrial product to increase. product prices and make it decrease. 1144

7 Table 8: Variance decomposition of industrial products price Period Log ind Log invid Log open Log rer Log m2 Log cpi IU Industrial product price index response function to The results reveal at e effect of economic inflation uncertainty: industrial product price decisions is reflected effectively in e industrial sector. response to inflation uncertainty shocks has no The monetary policy is one of e most important reaction in a short time period but it has positive macroeconomic tools and awareness of how it influences effects on industrial product prices in e middle of planning. National and regional development is also an e term and a long term period of time. In oer important step. Therefore, planning and determining e words, inflation uncertainty causes e industrial appropriate economic policies and implementing ese product price to increase in e mid and long term policies can play important roles in price stability in is periods of time. sector. One of e effective factors in industrial products' Variance Decompositions Analysis: In is section e price is inflation which is caused by money supply results of analysis variance decomposition forecast for a grow. Thus, essential efforts should be made to period of 10 year can be interpreted. The function also as eliminate inflation and move toward price stability, using impulse response function is used in short term dynamic contractionary monetary policy. analysis. In is meod, fluctuations of different variables Due to e negative relationship between e degree can be divided into variables of e pattern and e of openness and industrial products' price, if Iran's trade relative importance of a variable can be seen in e policy moves to join e WTO for a wider economy, e behavior of oer variables [20]. Thus, e contribution of contribution of subsidies paid by e government will each variable can be measured based on e changes in reduce. As noted in e results, is will in turn increase oer variables over e time. The results are shown in e industrial products' prices in e middle and long Table 8. periods of time. According to e study, we can make Also, wi e help of variance decomposition, e several important observations as follows: contribution of each variable on changes of oer variables was evaluated. Results show at in a short Dicky Fuller Stationary test was used to determine period of time, most contributions of fluctuations for e stationary of variables. The results show at all industrial products' price index is about 70% followed by e variables are not stationary at level and were consumer price index wi 11%. In e middle of e stationary at first difference. period, is amount reaches 55% for industrial products After reviewing e stationary of variables, e price and 18% for e real exchange rate. Also, in a long inflation uncertainty variable was estimated by using period of time, about 50% of fluctuations can be e GARCH model. The variable is estimated at represented by industrial products price index, 17% by e GARCH (1, 0) wi respect to e necessary and real exchange rate and about 10% by inflation uncertainty. sufficient conditions using GARCH process and Box-Jenkins meod. After estimating e inflation DISCUSSION uncertainty variable, e stationary of e variable was investigated and it was stationary at level. In is paper, e Impact of Inflation Uncertainty on The model was estimated by e VAR meod. It is e Prices of Industrial Products in Iran has been reviewed wor mentioning at e VAR model and its results over e period 1974 to For is purpose, e are not interpreted. But e pattern is used to GACRCH and VAR models have been used. investigate e cointegration and dynamics in 1145

8 e pattern. The results show at ere is a long term 9. Korap Levent, On e Links between Inflation, relationship between e variables. In industry, e Output Grow and Uncertanity: System-Garch value added of industry sector, e real exchange Evidence from e Turkish Economy, Iktisat Isletme rate, e liquidity volume and inflation uncertainty Ve Financo, 24(285): have positive and significant relationship wi 10. Choo, wei-chong, Lee, see-nie, Ung and Sze-Nie, industrial product price index. The consumer price Macroeconomics Uncertainty and Performance index and e degree of openness have negative and of GARCH Models in Forecasting Japan Stock significant relationship wi e industrial product Market Volatility, Center for promoting Ideas USA. price index. 11. Hamidepour, H., et al., Investigate e role of The results show at e inflation uncertainty Iran's oil revenues and monetary and fiscal policies in variable along wi oer variables has a significant e agricultural sector, Journal of Agricultural impact on industrial products' price, especially in a Economics, 4, (4): long term. Thus, in studies which analyze e 12. Heidari, H., et al., The effect of uncertainty on behavior of industrial products' prices and affecting economic grow in Iran (base on GARCH models), factors, e inflation uncertainty variable should be Journal of Economic Research, 14, (43). considered along wi oer variables. In is way, 13. Abunori, A., A. Khanaalipour and J. Abbasi, e null hypoesis based on e effects of inflation Effect of e news on e exchange rate fluctuations uncertainty on industrial products' price will be in Iran: applied ARCH, Journal of commerce accepted. Research, 50: Sims, C.A., Macroeconomics and Reality, REFERENCES Econometrica, 48: Noferesti Mohammad, Unit root and 1. Ezadi, H.M. and M. Ezadi, Effects of exchange co-addition in econometrics, october-november 2010, rate changes on e value of e industry using Rasa Publications. Kutany, Journal of Economic Research, pp: Saddighi Hamid Reza, K.A. Lawler and Shirinbakhsh, 2. Daie karimzadeh, S., Money inflation and shamsollah, Applied econometrics, Avayenoor causality, Master's esis, Shiraz University. institute. 3. Piraee, KH. and B. Dadvar, Impact of inflation 17. Karbasi, A. and M. Piri, Review e level of on economic grow in Iran wi an emphasis on agricultural prices and inflation uncertainty in Iran: uncertainty, Journal of Economic Research, 11, (1) , Journal of Business, (47): Golob, J.E., Does Inflation Uncertainty Increase 18. Salami Habibollah and Jahangard Halimeh, Time wi Inflation?, Economic Review, Federal Reserve series models for forcasting consumption of apple Bank of Kansas City, 79(3): and orange in iran, agricultural economics and 5. Naghdi, Y., Evaluation of anti-inflationary development, 17 year, (67). effects of increased production in various sectors of 19. Sabaghkermani, M. and V. Shaghaghishahri, industry, services and agriculture, Journal of Factors affecting e real exchange rate in Iran (vector Management, 5, (11). autoregressive model), Journal of Economic 6. Naderan, A., Credit policies of e value added Research, 5, (16): of industry sector, Journal of Economic References. 20. Abassinejad, H. and A. Shafiee, Is money really 7. Jafarisamimi Ahmad and Motameni Mani, neutral in e Iranian economy? Journal of Economic Inflation and Inflation Uncertainty in Iran, Australian Research, (68): Journal of Basic and Applied Science, 3(3): Gojorati, D. and H. Abrishami, Principles of 8. Girijasankar Mallik and Chowdhury Anis, Econometrics, Tehran University Institute. "Effect of inflation uncertainty, output uncertainty and oil price on inflation and grow in Australia, Journal of Economic Studies, 38(4):