Domestic Direct Investment is one of the contributors in Malaysia economic growth. The purpose of this paper is to investigate what determine the domestic direct investment in Malaysia by using time series data from year 1980 until 2007 (28 observation) based on the OLS regression which is Linear Log Multiple regression method and causality testing. Many study shows that the financial factor is the most important determinant to domestic direct investment. But the result shows that only credit to private sector is react positively to domestic direct investment. Nonfinancial also affect domestic direct investment except inflation variable. By combining the entire variable, it shows the strong relationship between them. This finding implies that credit both financial and nonfinancial can stimulate higher investment in Malaysia.

Keyword: Domestic Investment, Malaysian economy, Financial and nonfinancial factor, Ordinary Least Square (OLS), Granger Causality Test.

## 1 Introduction

Investment is a kind of activities that require not only money but something that have value. In term of economic or in economic view, it is connected with both saving consumption. The primary motive behind the investment is only to make profit. In general view, investment means the investor use some of their money in hoping to get or making more money in the future.

For many countries, whether they are developed or less develops countries, they depend on the investment made by the investors both from inside and outside of the country. They depend on the Foreign Direct Investment to measure their economic growth. According to Blomstom, Lipsey, and Zejan (1994) in their study among developing countries, they found a positive relationship between economic growth and ration of Foreign Direct Investment inflow to GDP. But, recently, some country facing a problem with a downward sloping (negative) in foreign direct investment, for example Malaysia.

So, the Prime Minister, Dr Mahathir, in the Malaysian Insider paper about “Dr M wants tax holiday for domestic direct investment” suggests the country to focus more on the Domestic Direct Investment so that the wealth is only flow within the country. As both foreign and domestic investment is important to economic growth, thus, there is a need to examine and analyze the domestic direct investment contribution to economic growth and what determine the domestic direct investment. The main objective of this study is to examine and analyze the determinants of domestic direct investment that contribute to economic growth in Malaysia. The specific objectives of this study include i) to determine the relationship between domestic direct investment and independent variable which is domestic credit to private sector, M3, credit provided by the bank sector, GDP per capita, gross domestic saving, gross domestic product deflator growth rate, and the interest rate charge by bank on loans ii) to examine the granger causality among the variable and iii) to determine what factors should we focus more.

To achieve this objective, this study used 28 observations for analysis purposes which cover the data for a certain period from 1980 to 2007. Malaysia is one of the members of developing country was used in this study. The data used are domestic credit to private sector, total liquidity liability of financial intermediaries (M3), credit provided by the bank sector, gross domestic product per capita, gross domestic saving, gross domestic product deflator growth rate, and the interest rate charge by bank on loans.

This study will use the linear log multiple regression model which is the ordinary least square method to determine and examine the relationship between the dependent and independent variable. To measure the causality between them, the Granger Causality test approach will be use. The limitation of this study is most of the previous study is using the dynamic panel data analysis and the OLS is said give inconsistent result [Hsiao 1986] because the study involve many country. Since this study is only focus on one country which is Malaysia, OLS approach which is linear log multiple regression method and the Granger causality test applied to achieve the objective of this study.

## 2 Literature Review

Other than the foreign direct investment, the domestic direct investment also contributes to the economic growth in Malaysia. There are so many factors that can determine the domestic direct investment that can lead to economic growth in a country. The determinants are like domestic credit to private sector, total liquidity liability of financial intermediaries as a proxy of M3, domestic credit provided by bank, gross domestic product per capita, growth rate of gross domestic product deflator, gross domestic saving, and interest rate charge on loans by using the lending rate.

Actually, these factor determinants domestic direct investments can be divided into two groups as stated by Ndikumana, L. (2000) in his paper financial determinants of domestic investment in Sub-Saharan Africa: evidence from panel data, which is financial determinants and the macroeconomic determinants. The three factor which are domestic credit to private sector, total liquidity liability of financial intermediaries (M3), and the credit provided by banks fall under financial determinants, while the other factor fall under the macroeconomic determinant.

The financial determinants with the “credit” statement is said to be the most important determinants on investment [Schumpeter (1932), Keynes, 1937, 1973]. This idea is supported by Gurley and Shaw (1955). They relate the financial development with the economic growth for a country. Fisher (1933) also state that the poorly financial market performance can result in economic downturn. Financial intermediary is also one of the important factors under the financial determinants. The financial intermediaries give better liquidity to the savers at the same time reduces the liquidity risk [Levine, 1997; Pagano, 1993; Gertler, 1988]. Liquidity is the ability of asset to be converted into money or cash. When people want to invest, they need to make sure that their investment has high liquidity.

The nonfinancial determinants or the macroeconomic indicator also give impact on the domestic direct investment in a country. They are the GDP per capita, growth rate of GDP deflator, gross domestic saving, and interest rate charge on loans by using the lending rate. According to Fielding, 1997, 1993; Greene & Villanueva, 1991; Wai & Wong, 1982, the investment rate is high when the output growth is also high.

For the gross domestic saving relationship with the domestic investment, Bayoumi (1990); Dooley, Frankel & Mathieson (1987); Feldstein & Horioka (1980) found there is positive relationship between saving and investment. The lower the domestic saving, the lower the domestic investment in a country.

## 3 Data and Methodology

To investigate the relationship between the variables and the causality between them, there are many empirical studies have been done on the impact of domestic direct investment on economic growth in a country. Many of the researchers apply the dynamic panel data analysis to examine such high impact economic activities. But, there are some other researchers that use Ordinary Least Square method (OLS) to examine the relationship and the Granger causality test to measure the causation between the variables. Previous study has been done by using the dynamic panel data analysis to examine the financial and non financial factor that determines the domestic investment. There is also some study conducted by using causality test and cointegration test. Hsiao 1986 state that the estimation using OLS is inconsistent. By using panel data, financial indicator is said to be react positively with the domestic investment [Ndikumana L, (2000)].

## Unit Root Test

Before the regression, the unit root test should be done to measure the stationary and the order of integration of each of the data. This test also used by Chan, Tze-Haw and Baharumshah, Ahmad Z. [2003] in their paper measuring capital mobility in the Asia Pacific Rim. The unit root test they used are Augmented Dickey-Fuller (ADF) test (Dickey & Fuller, 1981) and the Phillips-Perron (PP) unit root test [Phillips (1987) and Phillips and Perron (1988)]. The test of level of stationary is at level, first different, and the second different. After the unit root test are taking, then the available and suitable testing will conducted. The ADF test estimating the following example regression:

âˆ†Yt = Î²1 + Î²2t + Î´Yt-1 + (1)

Source: Basic Econometrics by Gujarati D N. and Porter D C.

The unit root test is conducted only to determine the level of stationary of the data whether it is stationary at level, first different, second different or not stationary. It is to consider whether the data can be use in regression or not.

## Logarithm Transformation of The Data

Since the data is not in the same unit, there is a need to conduct logarithm transformation of the data. The logarithm of data also used by Chan, Tze-Haw and Baharumshah, Ahmad Z. (2003) before the unit root test is done. It is important so that the data and also the result are not biased and easy to estimate the data later. Logarithm transformation is use in the nonlinearities relationship between variables. The importance of this method is the elasticity coefficient can be gain directly from the regression as said by Ndikumana, L., (2000) in his paper of financial determinants of domestic investment in Sub-Saharan Africa: evidence from panel data. In this study, the data that not in percentage will be transform into log to make it in the form of percentage and equal as the other form of data. The data transformed in this study are GDP and M3.

## Diagnostic Testing

Diagnostic testing is important step to determine whether there is an error or residual occur in the data being investigation. The errors that may be occurring are such as heteroscedasticity, multicollinearity, or autocorrelation. There is some hypothesis develop under this study whether the error estimation occurred.

Hypothesis 1:

H0: There are autocorrelation between the observations

H1: There is no autocorrelation between the observations

Hypothesis 2:

H0: There is multicollinearity between the observations

H1: There is no multicollinearity between the observations

## OLS Estimating

After the unit root test is applied, then the ordinary least square method is conducted to know the relationship between the dependent and independent variables. This study is use the linear log multiple variable regressions because the dependent variable which is domestic direct investment may be influence by more than one factor of economic such as are domestic credit to private sector (CPS), total liquidity liability of financial intermediaries (M3), credit provided by the bank sector (CPBS), gross domestic product per capita(GDP), gross domestic saving (GDS), gross domestic product deflator growth rate (INF), and the interest rate charge by banks on loans (INT). The dependent variable is in linear form and all the independent variables were transforming into log. The OLS test the following regression:

DDI = Î± + Î²1LNCPS + Î²2LNCPBS + Î²3LNM3 + Î²4LNGDP + Î²5LNGDS + Î²6LNINF + Î²7LNINT +â„°i

Source: Basic Econometrics by Gujarati DN. and Porter D C.

The DDI is the dependent variable and the other variables are the independent variable. The error term there means that there is other factors that may be affect the variables [Gujarati, D.N and Porter, D.C (2009)]

## Granger Causality Test

Although the regression shows there is relationship between dependent and independent variable, it is not necessary that they imply any causality or the direction of influence. Granger causality test is adopted to determine the causality between variable. It is applied to examine whether the variable X causes the variable Y, whether the variable Y causes the variable X, or both of them affect each other.

To examine the causality, some researcher has done this test. According to Engle and Granger (1987), there is causal relationship between variable X and variable Y if they are cointegrated. If X cause Y it means unidirectional function. If the X and Y is affected each other it means there is bidirectional function.

## 4 Empirical Results

The result and evidence from the test or regression will be elaborate in this section.

## Stationary Test

The existence of the unit root in each time series data, both of the Augmented Dickey Fuller (ADF) and Phillips – Perron (PP) tests is conducted in this study. Both tests have given almost the same result. The result shows in table 1 below.

## Table 1: Stationary Test

## variables

## ADF test (intercept)

## PP test (intercept)

CPS

-4.035146 (0.0047)***

-4.034534 (0.0047)***

CPBS

-4.573188 (0.0013)***

-4.575517 (0.0013)***

LNM3

-2.645653 (0.0971)*

-2.645653 ( 0.0971)*

LNGDP

-3.871691 (0.0069)***

-3.748933 ( 0.0092)***

GDS

-5.198757 (0.0003)***

-5.246528 (0.0002)***

INF

-5.743415 (0.0001)***

-5.737964 (0.0001)***

INT

-5.340167 (0.0002)***

-5.341039 (0.0002)***

Source: EViews7

Notes:

The figure within parentheses indicates the p-value. All variable except INF is stationary at first different.

significant level

*significant at level 0.1

***significant at level 0.01

The result shows that only M3 is significant at level 0.1 by using both ADF and Phillips-Perron test statistic and the other variable significant at level 0.01. For the stationary test to determine whether there is unit root test in the time series, the CPS, CPBS, LNGDP, GDS, LNM3 and INT is stationary at the first different which means lagged one time period. The INF is stationary at level which means no lagged of time period is needed.

## OLS Estimation

The linear log multiple regression shows that, the dependent variable, DDI, equals to domestic credit to private sector (CPS), total liquidity liability of financial intermediaries (M3), credit provided by the bank sector (CPBS), gross domestic product per capita(GDP), gross domestic saving (GDS), gross domestic product deflator growth rate (INF), and the interest rate charge by banks on loans (INT). The regression is done with the lagged time and the result show in table 2 below.

## Table 2: Ordinary Least Square (OLS) estimating

## Variable

## Result

DDI(constant)

t-stat

p-value

131.8437

2.103347

(0.6562)

LNCPS(t-1)

t-stat

p-value

82.86427

5.632050

(0.0002)***

LNCPBS(t-1)

t-stat

p-value

-56.57944

-6.658117

(0.0000)***

LNGDP(t-1)

t-stat

p-value

38.34651

5.098421

(0.0003)***

LNGDS(t-1)

t-stat

p-value

17.89063

0.602438

(0.5591)

LNINF

t-stat

p-value

-1.837128

-2.139562

(0.0556)**

LNINT(t-1)

t-stat

p-value

-39.18221

-3.562907

(0.0045)***

LNM3(t-1)

t-stat

p-value

-40.98774

-5.545882

(0.0002)***

R-squared

0.94

Adjusted R-squared

0.90

F-stat

25.63

Durbin Watson

2.45

observation

28

Source: EViews7

Notes:

Figures in the parentheses indicate the p-value.

significant level

**significant at 0.05

***significant at 0.01

Based on the result, all independent variable are significant except one independent which is gross domestic saving is not significant. All the LNCPS, LNCPBS, LNGDP, LNINT, and the LNM3 are significant at 0.01 levels. The LNINF is significant at 0.05 levels and the LNGDS is not significant at any level. There are three variables that give a positive impact on the DDI. The LNCPS, LNGDP, and LNGDS have a positive relationship with the DDI. The other four react negatively with the DDI which is LNCPBS, LNINF, LNINT and LNM3. The R-square is also high which means there is strong relationship between dependent and independent variables.

## Diagnostic Result

The result on error term shows below using the developed hypothesis which is hypothesis 1 and hypothesis 2. For autocorrelation, the test is using the Chi-Square. For multicollinearity, the test is using the Variance Inflation Factor (VIF) to examine whether there is multicollinearity occur.

Hypothesis 1

To test the autocorrelation occurring, since the Chi-Square is not significant which is 0.1299, the null hypothesis cannot be acceptable which means there is no autocorrelation occurred.

Hypothesis 2

To test whether the multicollinearity occurred or not, we look at the VIF. The result shows that there is multicollinearity for the CPS at 29.79821, GDS at 29.35209, GDP at 15.95545, INT at 14.20043 and M3 at 78.24159. It is because the VIF result is more than 10. But for the CPBS at 8.678442, and INF at 2.217933, there is no multicollinearity because the VIF is less than 10. So, we cannot reject the null hypothesis of there is no multicollenarity between the observations. The result of multicollinearity testing presented in table 3 below.

## Table 3: Residual Diagnostic for Multicollinearity problem: Variance Inflation Factor

## Variable

## Centered VIF

LNCPBS(-1)

8.678442

LNCPS(-1)

29.79821

LNGDS(-1)

29.35209

LNINF

2.217933

LNINT(-1)

14.20043

LNGDP(-1)

15.95545

LNM3(-1)

78.24159

C

NA

Source: EViews7

Notes: The centered VIF shows whether the variable have multicollinearity problem or not. If the VIF is bigger than 10, it means there is a multicollinearity problem.

## Granger Causality Test

The result of causality test shows that there is only unidirectional function occurred between DI and CPS and GDP. The table 4 below show that the null hypothesis of DI does not granger cause CPS is rejected where DI is granger cause CPS but the null hypothesis on CPS does not granger cause DI is acceptable. For GDP, the null hypothesis on GDP does not granger cause DI is not acceptable because result shows that GDP does cause DI but the hypothesis on DI does not cause GDP is acceptable. The null hypothesis of DI does not granger cause INF is also not acceptable because the result shows that DI actually cause the INF. DI is granger cause to INT which means that the null hypothesis of DI does not granger cause INT is not acceptable. Meanwhile, the other variable does not imply causation to each other.

## Table 4: Pairwise Granger Causality Tests with lags 2

Null hypothesis

F-Statistic

Prob.

LNCPBS does not Granger Cause DI

1.03345

0.3732

DI does not Granger Cause LNCPBS

1.17515

0.3283

LNCPS does not Granger Cause DI

0.59206

0.5622

DI does not Granger Cause LNCPS

## 3.70010

## 0.0420

LNGDP does not Granger Cause DI

## 3.44332

## 0.0509

DI does not Granger Cause LNGDP

1.36662

0.2767

LNGDS does not Granger Cause DI

0.41730

0.6642

DI does not Granger Cause LNGDS

1.60712

0.2242

LNINF does not Granger Cause DI

0.27698

0.7621

DI does not Granger Cause LNINF

## 12.6583

## 0.0007

LNINT does not Granger cause DI

0.83573

0.4541

DI does not Granger Cause LNINT

## 14.1109

## 0.0004

LNM3 does not Granger Cause DI

0.61312

0.5511

DI does not Granger Cause LNM3

0.61193

0.5517

Source: EViews7

Notes: The word in bold shows that the null hypothesis of granger cause is not acceptable.

## 5 Conclusions

The main objective of this paper is to analyze the determinants of domestic direct investment in Malaysia by using the OLS regression and Granger causality test. The data are collected from the World Bank Data which focus in Malaysia. The observation time series is from 1980 to 2007. The econometrics approaches used are ADF and Phillip-Perron, OLS and the Granger causality test. The evidence in this paper shows that the result support the previous that the financial indicator or factor are important determinants to domestic investment in Malaysia. The result shows that the CPBS, CPS, and M3 have significant at 0.01 level and only CPS has positive relationship with domestic investment. For nonfinancial factor (macroeconomic), not all variable is significantly determining the domestic investment in Malaysia. The gross domestic saving factor is not significant in this evidence. The domestic investment is strongly determined by its independent variable based on the R-square (94%). There is no autocorrelation problem occurred with its regression result on error estimation by testing using the residual diagnostic on autocorrelation that show the Chi-square is 0.1299 which reject the null hypothesis. The Multicollinearity does occur between the variables (show in table 4). To determine who cause to whom, the granger causality test is applied. The result shows that the domestic investment influences the credit to private sector. The GDP also influence the domestic investment. The GDP represent the growth and as the economy growing the investment also increase. DI gives influence to INF because in real situation, as the inflation rate increase, the domestic investment is less. As well as the DI that causes INT where if the interest rate charge by bank on loans is high, the people who want to make loan is less and the investment also less. Overall, both financial and nonfinancial factor is important but most important is financial factor. So, Malaysia should focus both financial and nonfinancial but more on the financial to increase their investment.

## Appendix

## Notes on variables:

M3 = total liquidity liability of financial intermediaries

CPS = domestic credit to private sector

CPBS = credit provided by the bank sector

GDP = gross domestic product per capita

GDS = gross domestic saving

INF = gross domestic product deflator growth rate

INT = interest rate charge by banks on loans

DDI = domestic direct investment