International Journal of Economics and Empirical Research http://www.tesdo.org/Publication.aspx Determinants of Consumption Function, In Case of China and G7 Countries Khalid Khan, Chen FEI, Muhammad Abdul Kamal, Sarfaraz Ahmed Shaikh School of Economics, Huazhong University of Science and Technology, (HUST), Wuhan, P.R. China Highlights Estimate consumption function for China and G7 countries. The bounds testing approach is used. Private consumption determines economic growth. Abstract Purpose: The objective of the study is to estimate consumption function for China and G7countries. Methodology: The study applied ARDL approach to estimate consumption function over the period 1985 to 2013. Findings: The results of the study indicate that Gross Domestic Product (GDP) and wealth are the most important determinants of aggregate consumption in both short and long run. However, real interest rate, negatively affect the aggregate private consumption both in the long and short run, apart from Canada, similarly, unemployment rate have negative effect of aggregate consumption in all cases. Recommendations: This study open new directions for further research. Key worlds: Private Consumption, Gross Domestic Product (GDP), Wealth, Real Interest Rate JEL Classification: E2, F43
Corresponding Author:
[email protected] Citation: Khan, K., FEI, C., Kamal, M. A. and Shaikh, S. A. (2015). Determinants of Consumption Function, In Case of China and G7 Countries. International Journal of Economics and Empirical Research. 3(4), 202-210.
- 202 -
International Journal of Economics and Empirical Research. 2015, 3(4), 202-210. I. Introduction The understating of aggregate consumption is very important in case of China and G7 countries (Canada, Japan, France, Germany, Italy, UK and US), because in these countries aggregate private consumption constitutes a large share of Gross Domestic Product (GDP), which are 39%,59%,56%,65%,55%, 62%, 63% and 85%1respectively for each country. Due to globalization and financial liberalization, aggregate private consumption is not only important for the economic growth and business cycle of G7 countries but also for countries that are closely connected to G7 countries like China. Therefore, it is very important for policy makers to understand aggregate private consumption in China and G7 countries. Keynes (1936), Kuznets (1946), Duesenberry (1949), Friedman (1957), Hall (1978), Campbell and Mankiw (1990) and Davidson et al (1978) contributed to consumption literature and tried to explain the main factors that determine the aggregate private consumption. Similarly, the main objective of this study is to produce a better understating of aggregate private consumption and factors which affecting aggregate private consumption in short and long run, in case of China and G7 countries. We examine the dynamic relationship between real aggregate private consumption and its determinants by using Autoregressive Distributive Lag (ARDL) technique. In consumption literature the work of Davidson, Hendry, Srba, and Yeo, (1978) is widely acknowledged. He used time series data and estimated consumption function for UK. Later on a lot of empirical studies have documented using the methodology of DHSY, However, more recently Fakhraii and Mansuori (2008) used the ARDL approach to estimated consumption function for Iran; likewise, Khan et al. (2014) estimated the real private consumption model of Pakistan through ARDL approach. II. Methodology and Data The simplest formation of the consumption function based on Absolute Income Hypothesis (AIH), Permanent Income Hypothesis (PIH) and Life Cycle Hypothesis (LCH) is given below: CONS t f (Yt , W t )
(1)
Where (CONSt ) is aggregate private consumption and (Yt ) is disposal income and (Wt ) is wealth, equation-1 represents the long run households’ consumption function. However, some empirical work and alternative theories suggested additional determinants for aggregate private consumption such as real interest rate and income uncertainty. Therefore, the complete functional form of the households’ consumption function is: CONSt f (Yt ,Wt , Rt ,URt )
(2)
where, (Rt ) is real interest rate and (URt ) is unemployment rate. The study used annual data over a period of 1985 to 2013. Real households’ consumption and real GDP are used as proxies for aggregate private consumption and disposable income respectively. Moreover, for wealth we used quasi money as a proxy. Moreover, for real interest rate we used different proxies depending upon the availability of data for respective country. For instance, in case of Canada and USA we used bank rate, while in case of France, Italy, Japan and UK, we used lending rate, however, for China and Germany, we used call money rate and discount rate respectively. Moreover, to capture the effect of income uncertainty, we used unemployment rate which is very widely used in consumption literature. We obtained annual data from 1985 to 2013, from International Financial Statistics (IFS) and World Development Indicators (WDI). The variables of aggregate private consumption, wealth and GDP are deflated with Consumer Price Index(CPI) and GDP deflator respectively while the real interest rate is obtained by subtracting inflation rate from nominal interest rate. We used log-linear form of equation (2) for empirical estimation of coefficients for consumption function. The log-linear formulation of consumption function becomes:
ln CONS t 0 Y ln Yt W ln Wt R ln Rt UR ln URt i
(3)
where (ln CONSt ) is the natural log of real households’ private consumption and (lnYt ) is the natural log of GDP, (ln Wt ) is the natural log quasi money, (lnUR) natural log of unemployment rate and is random error.
1
International Financial Statistics (IFS), 2013
- 203 -
Determinates of Consumption Function, In Case of China and G7 Countries To estimate the long and short run relationship among aggregate private consumption and other variables, we used ARDL approach, introduced by Pesaran and Shin (1995) and later extended by Pesaran, Shin and Smith (2001). As compared to conventional co-integration techniques, the ARDL approach has some advantages. First, in case of small samples, ARDL estimates are more robust than any other co-integration technique. Second, it simultaneously estimates the short and long run coefficients of the model. Third, it is applicable regardless of the order of integration (I(0) or I(1)) and solve the problem of endogeneity. Therefore, we applied ARDL approach to estimate equation (4). The equation of real aggregate private consumption based on ARDL approach to co-integration is: j
ln C t 0
i0
j
1 i ln Y t i
j
2 i ln W t i
i 0
j
3 i ln C t i
i 0
4i
ln rt i
i0
j
5i
ln UR t i Y ln Y t 1 W ln W t 1 C ln C t 1 R ln R t 1 UR ln UR t 1 u t ,( 4 )
i 0
where, i are short run coefficients while Y , W , C , UR long run coefficients. The null and alternative hypothesis for
co-integration
is:
H0 Y W C R UR 0
while
the
alternative
hypothesis
is
H1 Y W C R UR 0 . To test the null and alternative hypothesis of co-integration for this purpose we applied bounds test of Pesaran et al. (2001), once the co-integration appraise among the variables and we reject the null hypothesis of no co-integration then in next step we estimate the long run and short run coefficients of the above model with Error Correction Term (ECT), which shows the speed of converges from disequilibrium to equilibrium. III. Results and Discussion Before applying the ARDL approach we want to confirm, that no variable(s)in the model are integrated of I (2), because in such a case we cannot apply the bound test of ARDL due to unavailability of critical values. The values of F-statistics provided by Pesaran , Shin and Smith (2001). In order to test the stationarity of the variables we used two different tests; Augmented Dickey Fuller (ADF) and Dicky-Fuller Generalized Least Square (DF-GLS) tests. Table-1 offers the results of the ADF test which shows that incase of Japan and UK aggregate private consumption is stationary at I(0) with trend , while in case of China and Canada, France, Germany, Italy and US aggregate private consumption is stationary at I(1) with and without trend. Similarly, in case Japan GDP is stationary at I(0) with trend while in case of China and other G7 countries GDP is stationary at I(1) with and without trend. However, in case of UK, GDP is just stationary without trend. Wealth is stationary at I(0) for China and Japan without trend and for UK with trend while in case of France and Italy it is stationary at I(1) with and without trend. Likewise, in case of Canada, Japan and US wealth is stationary at I(1) with trend. Unemployment rate is stationary at I(1)in case of Italy and UK with and without trend, however, for all remaining countries unemployment is stationary at I(0) with and without trend. Finally, interest rate is stationary at I(1) with and without trend for Germany and Italy while stationary at I(0)for all other countries. The same results we confirmed from DF- GLS test. The results of DF-GLS test offers in table 2, in which aggregate private consumption and GDP are stationary at I(0) in case of UK while stationary at I(1) for other countries. However, wealth is stationary at I(0) in case of Japan, France and UK while stationary at I(1) for all other countries. For Italy and UK real interest rate is stationary at I(1) while for other remaining countries real interest rate is stationary at I(0). However, for China and all G7 countries the unemployment rate is stationary at I(0).
- 204 -
International Journal of Economics and Empirical Research. 2015, 3(4), 202-210.
VR InCon dlnCon LnGDP dlnGDP LnW dlnW dlnUr dlnUr dlnr Dlnr
MO c,t C c,t C c,t C c,t C c,t C c,t C c,t C c,t C c,t C c,t C
China -1.46 -2.144 -4.25** -3.87** 4.92 -0.007 -4.40** -2.54** -2.99 -3.45* -----0.323 -2.69 -3.66* -3.85** -4.03** -4.07** -----
Canada -1.15 -1.49 -3.84** -3.49** -2.22 0.34 -3.52** -3.56** -1.188 -1.46 -2.78* -2.72 -2.32 -3.29* -------3.36* -0.77 -------
Table-1.Results of ADF Test Japan France Germany -5.03** -2.17 -1.85 -1.18 -0.97 -1.36 ----5.02** -4.82*** ----5.13*** -4.92*** -3.19** -2.13 -2.00 -2.04 -0.97 -1.53 ----4.91*** -5.58*** ----5.01*** -5.22*** -3.03 -0.60 -0.48 -1.79 -3.64** -1.66 -2.63* ----5.00*** -2.72 ----5.08*** -3.31* -2.66 -2.70 -0.97 -2.67** -2.83* -------------------3.07 -5.54*** -2.40 -2.12 -1.51 -2.52 -3.20** ----5.22*** -1.98** ----4.99***
Italy -1.98 -2.10 -5.49*** -5.50*** -2.25 -1.30 -5.21*** -5.30*** 4.98 0.74 -4.84** -2.92* -2.32 -0.65 -3.91** -4.83*** -3.99 -1.69 -4.51*** -4.60***
UK -3.19* -1.28 -------2.95 -0.89 -0.99 -3.02* -3.47* -0.97 -------2.62 -1.07 -5.74*** -5.78*** -1.41 -3.13* -------
US -3.20 0.56 -3.24** -3.34** -1.56 1.97 -3.48* -3.96** 0.96 4.14 -4.87*** -0.04 -8.07*** -2.29 -----3.07 -3.09** -------
Note: Where; VR for Variables, MO for Model and ‘c, t’ represents model with trend while ‘c’ represents model without trend, however, ***p
