Stock prices and macroeconomic variables in vietnam

Stock Prices and Macroeconomic Variables in Vietnam: An Empirical Analysis Nguyen Trong Hoai1 Nguyen Thi Bao Khuyen2 The article employs the cointegration and error correction version o

Stock Prices and Macroeconomic Variables in Vietnam:

An Empirical Analysis

Nguyen Trong Hoai1

Nguyen Thi Bao Khuyen2

The article employs the cointegration and error correction version of Granger causality tests to investigate whether the Vietnamese stock market exhibits the publicly informational efficiency The test results strongly suggest informational inefficiency in the Vietnamese stock market Specifically, the results from bivariate analysis suggest that the Vietnamese stock market is not informationally efficient in both short- and long-run In addition, the stock market seems to even divorce from the most part of the economy Therefore, it is still possible for a

“professional” trader to make abnormal returns by analyzing good or bad news contained in some macroeconomic variables The findings re-assure that the Vietnamese stock market is not well functioning in scarce resource allocation and not attractive enough to encourage foreign investors Since the market is not informationally efficient, especially with respect to monetary variables, it may be dangerous for policy makers to realize the role of monetary policies, especially the so-called demand stimulus packages In terms of the investors’ point of view, fundamental analysis is still significant for their investment decisions Thus, companies with strong equity analysts would have higher comparative advantages in this inefficient market Furthermore, instead of becoming more efficient over time, as one might expect, the Vietnamese stock market appears to have become increasingly divorced from reality This also reveals that the last financial crisis has serious impact on the Vietnamese stock market

1 Introduction

Efficient market hypothesis has been at the center of debates in financial literature for several years The term efficiency is used to describe a market in which all relevant information is immediately impounded into the price of financial assets If the capital market is sufficiently efficient, investors cannot expect to achieve superior profits from their investment strategies As a result, capital asset pricing models could be useful for various investment decisions In the economic perspective, the efficient market is even more important because it implies that the stock market is well functioning in scarce resource allocation However, this is not always the case, especially in the emerging stock markets

1 Professor of economics, University of Economics, Ho Chi Minh City, Vietnam; Dean, Faculty of Development Economics, University of Economics, Ho Chi Minh City, Vietnam

2 Senior Analyst, VietFund Management Company, Vietnam

Islam and Khaled (2005) calls developing countries as ‘capital starved economies’, so efficient allocation of scarce resources and encouragement of private foreign investment are both of vital importance They also stated that the success of an increasing privatization of these economies will depend crucially

on the presence of an active and efficient stock market Indeed, rational investors expectedly drive their investments into the most profitable projects, given acceptable risks The efficient market can address the ‘mixed feelings’ problem, which investors are always skeptical about the intrinsic value of any stock under consideration This may lead their decisions based on others In other words, this phenomenon is commonly considered as herding behavior For foreign investors, inefficient markets are usually equivalent to high risky markets when making their investments abroad Hence, they tend to apply higher hurdle rates, which in turn underestimate investment opportunities in developing countries Eventually, it’s hard for any developing country with inefficient/weak stock market to attract foreign portfolio investment flows

The efficient market hypothesis is theoretically viewed in three common forms, depending on the kind of available information embodied These are commonly classified into weak-form, semi strong-form, and strong-form efficiency The weak form is the lowest form of efficiency that defines a market

as being efficient if current prices fully reflect all information contained in past prices only The semi-strong form efficiency suggests that the current price fully incorporates all publicly available information Semi-strong efficiency requires the existence of market analysts who are not only financial economists able to comprehend implications of vast financial information, but also macroeconomists, experts adept in understanding processes in product and input

markets (Ross et al., 2006) The strong form efficiency states that the current

price fully incorporates all existing information, both public and private (also called inside information) The main difference between the semi-strong and strong efficiency hypotheses is that, in the latter, nobody should be able to systematically generate profits even if trading on information not publicly known

at the time The rationale for strong-form market efficiency is that the stock market anticipates, in an unbiased manner, future developments and therefore the stock prices may have incorporated all relevant information and evaluated in a

much more objective and informative way than the insiders According to Ross et

al (2006), a very strong assumption of this form is that inside information cost is

always zero However, this assumption hardly exists in reality, so the strong form efficiency is not very likely to hold

Despite its impressive growth, the Vietnamese stock market is really struggling with various typical weaknesses of an emerging market (Truong, 2006) As a result, trading behavior in the Vietnamese stock market may be much different from that in developed/newly emerging stock markets Investors, especially those who have just experienced the economic downturn, may base their actions on the decisions of others who are well informed about market developments by following the market consensus (Nguyen, 2009) Thus, the

question is whether the Vietnamese stock market is informationally efficient? And whether the market pattern is seriously affected by the financial crisis?

The mixed evidence from the study of Truong (2006) in the 2002-2004 period may imply that the Vietnamese stock market is, to which extent, characterized by the weak-form efficiency However, this lowest form of efficiency cannot assure the Vietnamese stock market is well functioning in scarce resource allocation and attractive enough to encourage foreign investors (Nguyen, 2006) Both investors and policy makers mostly concern if the current market prices reflect all publicly available information, such as information on inflation, economic growth, money supply, exchange rates, interest rates, annual earnings, stock splits, etc

Ibrahim (1999) states that the significant lagged effects of macroeconomic variables on stock prices indicate informational inefficiency of the stock market

If this is the case, individual investors can earn abnormal profits by exploiting past macroeconomic information As a result, this exploitable opportunity would seriously distort the market’s ability to efficiently allocate scarce resources The reverse effects of stock prices on macroeconomic variables imply that stock market movements anticipate future economic conditions Accordingly, they may

be employed as a leading indicator in helping formulating current economic stabilization policies This article will investigate these dynamic interactions for the case of the Vietnamese stock market

The organization of the paper is as follows The next section briefly reviews the literature of semi-strong market efficiency Then, Section 3 outlines the analytical framework Section 4 describes the data and examines their temporal properties using integration and cointegration tests Section 5 presents the results

of bivariate causality tests Finally, our concluding remarks are contained in Section 6

2 Literature Review

Since its introduction into the financial economics literature over almost 50 years ago, the efficient markets hypothesis has been examined extensively in numerous documents Most previous studies of semi-strong form efficiency have been based on the analysis of the causal relationship between macroeconomic variables and stock prices

The types of relationships between stock market returns and macroeconomic variables can be varied As Mahdavi and Sohrabian (1991) report, there was an asymmetric causal relationship between those variables when they explored the relationship between changes in stock prices and GDP growth in the U.S markets using the standard version of Granger causality tests The stock market growth rate caused GDP growth rate, yet no reverse causation was found Chen (1991), however, finds that the current and future economic growth could be revealed by several domestic variables, such as the market dividend-price ratio, short-run

interest rates, production growth rate, term premium, and the default premium In addition, Rousseau and Wachtel (2000) reveal that equity markets have been key institutions in promoting real economic activity in 47 countries However, it is worth noting that this finding may be different in countries with less financially developed markets (Minier, 2003)

Mauro (2003) suggests that the developments in stock prices should be taken into account in forecasting output However, the relationship between stock returns and economic growth has not been stable over time (Stock and Watson 1990) Cheng (1995) argues that a number of systematic economic factors significantly influenced the U.K stock returns This result seems to contradict with that of Poon and Taylor (1991) who also observe the interrelationship between macroeconomic factors and stock prices in the United Kingdom

The relationship between stock prices and economic activity is not only limited to the relationship between stock prices and economic growth, but also extends to other economic factors (Fama, 1981) Abdullah and Hayworth (1993) argue that stock returns are positively related to inflation and growth of the domestic money supply in the United States, but negatively related to domestic interest rates By the same manner, Beenstock and Chan (1988) find that interest rates, input costs, money supply, and inflation are the significant risk factors of the London stock market

For the Pacific region, in Australia, there was a unidirectional relationship (in negative fashion) between inflation and the nominal stock returns during the 1965-1979 period, with price levels leading the equity index (Saunders and Tress 1981) Other researchers (Leonard and Solt, 1987; Giovanini and Jorion, 1987; Kaul and Seyhun 1990; Randal and Suk, 1999) also support a significant relationship between inflation or expected inflation and stock market prices

In terms of the relationship between stock market returns and exchange rate, Johnson and Soenen (1998) state depreciation may cause the cost of imports to increase, leading to domestic price level increases, which would expectedly have

a negative impact on stock prices Morley and Pentecost (2000) also confirm that stock markets and exchange rates are linked, and note that this connection is through a common cyclical pattern rather than a common trend

For the Asia-Pacific region, Hamao (1988) found a significant relationship between the Japanese stock returns and several factors, such as the changes in expected inflation and the term structure of interest rates Ibrahim (1999) also observed the Malaysian exchange rate by using bivariate and multivariate cointegration as well as Granger causality tests and found cointegration when the M2 measure of money supply and reserves are included, but no long-run relationship between the exchange rate and stock prices was found using bivariate models These suggest that in the short run the exchange rate might play

a significant role in the domestic economy, and that the Malaysian stock exchange is informationally inefficient

However, in some cases, macroeconomic factors cannot be reliable indicators for stock market prices movement in the Asian markets because of the inability

of stock markets to fully capture information about the change in macroeconomic fundamentals (as is cited in Wongbangpo and Sharma, 2002)

Ibrahim (1999) investigates the dynamic interactions between seven macroeconomic variables (the industrial production index, consumer prices, M1, M2, credit aggregates, foreign reserves and exchange rates) and the stock prices for an emerging market, Malaysia, using Granger causality tests and error correction mechanism tests This analysis is conducted using monthly data series for the period from January 1977 to June 1996 To smooth possible volatility, all data series are expressed in logarithmic forms Generally, the results suggest informational inefficiency in the Malaysian market

Using the bivariate causality tests, Ibrahim (1999) suggests three important points First, the results largely indicate that the lagged changes in macroeconomic variables have no significant predictive ability for the movements in stock prices Second, the stock market movements could help anticipate variations in the industrial production, the M1 money supply, and the exchange rate From this finding, he says that the causal link from stock prices to the M1 money supply may reflect the importance of the stock market on the M1 money demand Third, there exists the cointegration between the stock prices and three macroeconomic variables – consumer prices, credit aggregates and official reserves The results suggest that deviations from the equilibrium path are adjusted by about 5%–8% the next month through the movements in stock prices Thus, the adjustment toward the long-run relationship is extremely low in Malaysian stock market

Hanousek and Filer (2000) examine the possibility that newly-emerging equity markets in Central Europe exhibit semi-strong form efficiency such that

no relationship exists between lagged values of changes in macroeconomic variables (M1, M2, exports, imports, trade balance, foreign capital inflow, budget deficit, government debt, CPI, PPI, exchange rate, and industrial production) and changes in equity prices using Granger causality tests They find that while there are connections between real economy and equity market returns in Poland and Hungary, these links occur with lags, suggesting the possibility of profitable trading strategies based on public information and rejecting semi-strong efficiency hypothesis For Czech Republic and Slovakia, the situation is more complex In the early years of their existence, these markets may have possessed elements of semi-strong efficiency, with both lagged and contemporaneous relationships between real variables and equity markets However, these links have disappeared over time In other words, these stock markets appear to have become increasingly divorced from reality In the same manner, Azad (2009)

conducts a cross-market test including China, Japan, and South Korea and concludes that while Chinese stock market is inefficient, Japanese and South Korean stock markets are semi-strongly efficient These empirical studies, along with other studies about the U.S markets, suggest that developed equity and newly-emerging stock markets exhibit semi-strong form efficiency, while this is not a case in developing stock markets

Atmadja (2005) examines the existence of Granger causality among stock prices indices and macroeconomic variables in five ASEAN countries, Indonesia; Malaysia; the Philippines; Singapore; and Thailand with particular attention to the 1997 Asian financial crisis and period onwards Using monthly time series data of the countries, a Granger-causality test based on the vector autoregressive (VAR) analytical framework was employed to empirically reveal the causality among the variables This research finds that there were few Granger causalities found between the country’s stock price index and macroeconomic variables This indicates that the linkages between domestic stock price movements and macroeconomic factors were very weak Due to that, the ASEAN stock markets were relatively unable to efficiently capture changes in economic fundamentals during the observation period in most of the countries in accordance to the literature in emerging stock markets, and that the influence of specific macroeconomic factors on the domestic economies differ across countries This also implies that the stock markets do not seem to have played a significant role

in most countries’ economies, and macroeconomic variables are unlikely to be appropriate indicators to predict not only the future behavior of other macroeconomic variables, but also that of the stock market price indices

While causality analysis is widely discovered in both developed and newly emerging markets, only a few studies have been conducted for emerging markets such as Vietnam A recent notable study for the Vietnamese stock market is that

of Nguyen (2006) Applying the Engle-Granger cointegration tests, she find no evidence for cointegration between the VN-Index and macroeconomic variables such as industrial output, inflation rates, money supply, exchange rates, imports and exports using data that span from July 2000 to June 2006 Based on these findings, she concludes that the Vietnamese stock market is informationally inefficient with respect to all selected macroeconomic variables

In order to contribute to this line of literature for emerging markets, this paper would like to extend existing studies on the informational efficiency of the Vietnamese stock market on the following ways First, we examine the market efficient hypothesis using a wider range of macroeconomic variables3 In particular, we use twelve macroeconomic variables, namely, consumer price, industrial production, imports, exports, exchange rates, M1 money supply, M2 money supply, lending rates, deposit rates, domestic credit, foreign reserves, and

3 Because the time span is now longer than that of the previous study

money reserves Second, we make every effort to compare the informational efficiency situation of the Vietnamese stock market with that of other emerging markets Third, we will take the impact of the last financial crisis into account so

as to check the robustness of market efficiency exhibition

3 Analytical Framework

The analytical framework of this paper is to employ the Granger causality tests

In doing so, we will first examines whether the variables of concern are stationary and cointegrated As widely accepted, if all variables under

non-consideration are integrated of order 1, I(1), and they are not cointegrated, we

must apply the standard version of Granger causality test using the first differences of the variables If this is the case, we are just able to test whether the stock market exhibits the short-run efficiency By contrast, if the variables under

consideration are not only I(1), but also cointegrated, then we should employ the

cointegration and error correction (ECM) version of Granger causality tests According to Ihrahim (1999), the ECM conveniently combines the short-run dynamics and long-run equilibrium adjustments of the variables In the efficient market hypothesis literature, this allows us analyse whether the stock market exhibits both short-run and long-run efficiency

=

1 i

t i t i 1

+α

=

1 i

t i t i 1

t 0

+ α + α

=

1 i

t i t i 1

t 1

because it can, to which extent, indicate the presence or not of deterministic regressors If the test equation is characterised by the serial correlation, the PP test, conducted in a similar manner of ADF tests, becomes the most appropriate alternative

3.2 Cointegration Tests

One the order of integration of each variable is established, we then evaluate whether the variables under consideration is cointegrated According to Ibrahim

(1999), the cointegration of the time series such as the stock prices (P t) and a

macroeconomic variable (M t) suggests the existence of a long-run relationship that constrains their movements That is, although the variables are individually nonstationary, they cannot drift farther away form each other arbitrarily According to Asteriou and Hall (2007), if two variables are non-stationary, then

we could normally expect that they would combine to produce another

non-stationary process However, in the special case that two variables, say P and M,

are really related, we would expect the two stochastic trends could be very similar to each other and the combination of them could eliminate the nonstationarity In this special case, we say that the variables are cointegrated If the variables do not cointegrate, we usually face the problems of spurious regression and econometric work becomes almost meaningless (Granger and Newbold, 1974) Inversely, if the stochastic trends do cancel to each other, then

we have cointegration To test for cointegration, this paper employs the Engle and Granger (1987) approach4 This can be briefly summarised as follows:

Step 1: Test the variables for their order of integration

By definition, cointegration necessitates that the variables be integrated of the same order Thus the first step is to test each variable to determine its order of integration This study will apply the ADF tests There are three possible cases:

a) If both variables are stationary (I(0)), it is not necessary to proceed since

standard time series methods (OLS) can be correctly applied

b) If the variables are integrated of different orders, it is possible to conclude that they are not cointegrated

c) If both variables are integrated of the same order, then we proceed with step two

Step 2: Estimate the long-run (possible cointegrating) relationship

4 The Johansen cointegration approach is also widely used in empirical studies

If the results of step 1 indicate that both P t and M t are integrated of the same

order (usually in economics I(1)), the next step is to estimate the long-run

equilibrium relationship of the form:

t t 2 1

and obtain the residuals of this equation

If there is no cointegration, the results obtained will be spurious In this case,

we will apply the standard version of Granger causality tests for investigating short-run efficicieny However, if the variables are cointegrated, then OLS regression yields “super-consistent” estimators for the cointegrating parameter

2

ˆ

β In this case, we will apply the ECM version in order to analyse both short-run and long-run efficiency

Step 3: Check for (cointegration) the order of integration of the residuals

In order to determine if the variables are actually cointegrated, denote the estimated residual sequence from the equation by uˆt Thus, uˆt is the series of the estimated residuals of the long-run relationship If these deviations from long-run

equilibrium are found to be stationary, the P t and M t are cointegrated Because uˆt

is a residual, we do not include a constant nor a time trend in the ADF equation

3.3 Standard Version of Granger Causality Tests

Engle and Granger (1987) developed a relatively simple test that defined

causality as follows: a (stationary) variable ∆P t is said to Granger-cause

(stationary) variable ∆M t , if ∆P t can be predicted with greater accuracy by using

past values of the ∆M t variable rather than not using such past values, all other terms remaining unchanged

The standard version of Granger causality test for the two stationary variables

∆P t and ∆M t, involves as a first step the estimation of the following VAR model:

pt s

1 j

j t j r

1 i

i i 1

1 j

j t j p

1 i

i t i 2

where it is assumed that both εpt and εmt are uncorrelated white-noise error terms

In this model, we can have the following different cases:

Case 1: The lagged ∆M terms in (5) may be statistically different from zero as a

group, and the lagged ∆P terms in (6) not statistically different from zero In this case, we have that M t causes P t

Case 2: The lagged ∆P terms in (6) may be statistically different from zero as a

group, and the lagged ∆M terms in (5) not statistically different from zero In this case, we have that P t causes M t

Case 3: Both sets of ∆P and ∆M terms are statistically different from zero as a

group in (5) and (6), so that we have bi-directional causality

Case 4: Both sets of ∆P and ∆M terms are not statistically different from zero in

(5) and (6), so that P t is independent of M t

According to Hanousek and Filer (2000), a market is semi-strongly efficient, if

two following conditions must hold First, a contemporaneous relationship must

exist between stock prices (∆P t ) and macroeconomic variable (∆M t) Second,

lagged values of ∆M must not be enabling a potential investor to predict current returns (∆P t) in the market (Case 2 or Case 4) Both of these relationships are important Although the first is often ignored in empirical research, if it fails to hold, then the fact that the second does is not proof of efficiency It may simply

be that the variable under examination is irrelevant in determining prices in the equity market These relationships can be expressed as follows:

t r

1 i

i t i 0

1 i

i t i 0

∆β+

∆β+

∆α+α

=

1 j

t j t j t

0 r

1 i

i t i 0

returns than equation (7) In summary, results support market efficiency if both of null hypotheses hold:

(A) H0: β0 ≠ 0 (in equation 8) (B) H0: βj = 0, ∀ j (in equation 9)

A finding that hypothesis (B) does not hold suggests that the market is strongly) inefficient A finding that hypothesis (B) holds, but (A) does not suggest that efficiency considerations are irrelevant (Hanousek and Filer, 2000)

(semi-3.4 Error Correction Mechanism

If two variables are cointegrated, it is suggested that the ECM version of Granger causality tests would be the most appropriate specification The ECM can be specified as follow:

ε+π

+

∆β+

∆β+

∆α+α

=

1 j

t 1 t j

t j t

0 r

1 i

i t i 0

where ECT is the error correction term obtained from the cointegrating regression

or the linear long-run relationship of the variables (equation 4) In addition, the coefficient π in equation (10) is the error-correction coefficient and is also called the adjustment coefficient In fact, π tells us how much of the adjustment to equilibrium takes place each period, or how much of the equilibrium error is corrected According to Asteriou and Hall (2007), it can be explained in the following ways:

1 If π = 1, then 100% of the adjustment takes place within the period, or the adjustment is instantaneous and full

2 If π = 0.5, then 50% of the adjustment takes place each period

3 If π = 0, then there is no adjustment

Note that, with this specification, the changes in the stock prices will depend not only on the changes in the macroeconomic variable but also on the long-run relationship between them The latter allows for any previous disequilibrium,

measured by the error correction term ECT, to exert potential influences on the

movement of the stock prices According to Toda and Phillips (1994), the former may be termed ‘short-run causality’ from the macroeconomic variable while the latter may be termed ‘long-run causality’ In order to test whether the stock market is informationally efficient in the long run, we can test the following null hypothesis:

(C) H0: π is significant, but small (in equation 10) The reverse causation from the stock prices to the macroeconomic variables of interest may also be evaluated by reversing the roles of the two variables in equation (10) From these tests, one of the following four patterns of causality

can be noted: (1) unidirectional causality from X to Y; (2) unidirectional causality from Y to X; (3) bidirectional causality; and (4) no causality

5 Empirical Results

5.1 Descriptive Statistics

This section will investigate the temporal properties of collected data series Instead of displaying our data in the traditional manner such as graph, correlogram, and other statistics, we will concentrate on ADF and PP tests for stationarity Although they are not reported in this paper, line graphs and correlograms were used as the first feeling of our data From these views, we believe that stock prices and all macroeconomic variables are non-stationary

Table 1 reports the results of the ADF tests, the tests are implemented with and without the time trend This table presents the results for the log-levels of the data series, while Table 2 reports the results of ADF and PP tests for their first

differences Figures in these tables are computed t (or exactlyτ) statistics Note

that test results from Eviews respectively report critical t values of 1%, 5%, and

10%

Table 1: Integration Tests (Log Levels)5

Variables ADF Tests

None No trend Trend

5 The test results for the December 2000 to June 2008 period are the same

-1.58 (a)

LR (Lending Rate) 0.44 -2.21

-5.54* -3.35*** (a) -1.59 (b)

DR (Deposit Rate)

1.44 -0.90 -4.94*

-3.47**(a) -2.13 (b)

From the table 2, the null hypothesis for a unit root is rejected for all first differenced series in most cases In particular, the evidence from the PP tests strongly supports the stationarity of some variables when their null hypothesis cannot be rejected by ADF tests This is similar to the case of Malaysian stock market Thus, the evidence seems consistently to suggest the stationarity of the first-differenced series In other words, these variables can be characterized as

I(1) variables Because each set of variables are integrated of the same order, then

we can proceed with step two of the EG approach

Table 2: Integration Tests (First Differences)6

Variables ADF Tests

None No trend Trend

6 The test results for the December 2000 to June 2008 period are the same