Problem statement
After the official joint to the World Trade Organization (WTO) in 2007, Vietnam has been opening the financial market, economy and trade With the competitive advantages of lower production cost and investment risk than in other Southeast Asian countries, Vietnam has become an attractive destination of foreign capitals
Vietnam is making a good impression on international investors because it is expanding rapidly in emerging market and obtaining dramatic growth after the global financial crises in 2008 Due to government's gradual relaxing macro policies on foreign investment restrictions in the stock market, Vietnam has further enhanced attraction to international equity investors
About the regulations on foreign share holding rate, Vietnam has raised the percentage of foreign holding rate from 30% up to 49% These regulations are rather suitable for Vietnam Stock Market (VSM) in this developing period
About the regulations on profit transferring outflows, in order to encourage foreign investors we have offered duty-free on this kind of outflows since 2004 for Vietnamese foreigners and foreign residents
On July 28, 2012 Vietnam stock market (VSM) was 12 years old with some notable achievements when it reached more than 1.2 million transaction accounts, 1,690 public companies, 105 securities companies, 4 7 fund management companies and 23 stock investment funds Market capitalization accounted for
27% of GDP Till May 31, total loans for securities of whole banking systems were about 12 trillion VND, bad debt at 485 billion VND
The July average trading volume reached 41.83 million shares per session with the average value of 626.3 billion VND (down 3 7% in volume and -43,9% in value
I frbm June) On July, VN-Index ended at 414.5 points, down 10.9 points from June (see Economic Financial report Jul2012)
Additionally, foreign investors' transactions were still remammg gloomy On Hose, they bought 56.4 million shares valued at 1,377 trillion VND and sold 50 million shares valued at 1,292 trillion VND Net purchase value in July was 86 billion VND while in June, foreign investors posted net sale of 650 billion VND
On HNX, foreign players also posted net purchase of nearly 60 billion VND (See Economic Financial report Jul 20 12)
In fact, FPI flows into Vietnam have been increasing rapidly in recent years
Especially, since Vietnam officially joined WTO in 2007, FPI flows have increased strongly, accounting for more than 50% of total foreign investment capitals In the 2008 global financial crisis, the FPI flowed out; the stock market fell 66%, from 921 points to 316 points and caused bad effects on macro economy
Then, along with the economic recovery, the stock market witnessed a net inflow of FPI capitals but the stock market rarely crossed 500 points, with mini-recoveries inevitably followed by lengthy slumps On theory, FPI flows can benefit an economy in three broad ways First, FPI inflows can provide a non-debt capital source of foreign investment for a developing country and supplement domestic savings for improving the investment rate Moreover, FPI also reduces the pressure of foreign exchange gap for the less-developing countries Second, rises in foreign capital inflows can increase the allocated efficiency of capital in a country
Therefore, FPI can induce financial resources to flow from capital-abundant countries to capital-scarce ones Consequently, resource flows into the capital- scarce countries reduce their capital cost, increase investment and raise output
Third, through its various linkage effects via the domestic capital market, FPI
affects the economy by giving an upward thrust to the domestic stock market prices, impacting on the price-earning (PIE) ratios of the firms and making these ratios become higher which lead to a lower cost of finance and in tum attract a higher amount of investment Consequently, the lower cost of capitals can encourage new equity issues with higher premium On the other hand, FPI flows also stimulate the domestic stock market's development when it opens the entry for foreign investors
In Resolution No.01/NQ-CP dated March 01, 2012 on key solutions to realize the socio-economic development plan and state budgeting for 2012, the main targets for economic development in 2012 have been set up such as about 6% to 6.5 %in the GDP growth rate, 13% in total export growth , import surplus accounted for 11% -12% of total export turnover , controlling trade deficit under 10%, the overspending in state budget controlled less than 4.8% of GDP, total capitals invested in social development accounting for 33.5% of GDP, the expansion in the consumer price index less than 10%, 1.6 million employed workers, the urban unemployment rate remaining at 4%
So, they aim to these objectives: prioritizing curbing inflation by applying tight, cautious and flexible monetary policy in accordance with the tight and effective fiscal policy, stabilizing the macro-economy, maintaining growth rate at a reasonable level by reinforcing the inspection of market and prices, well- organizing the domestic market, encouraging exports, controlling imports and reducing trade deficit On the other hand, they also object to growth model renovation, the national economy restructure and improvement in the quality performance and competitiveness of the national economy and enhancement in the performance of external relations and international integration by these solutions: restructuring investments focused on public investments, restructuring the financial and banking system, especially on commercial banks, restructuring enterprises focused on SOEs
Despite a huge amount of empirical researches on stock market behaviors, most
~tudies have focused on the major well- established markets or on the other macro factors' influences on stock markets Thus, an increased knowledge of how foreign portfolio investments influence on Vietnam stock market (VSM) is a practical interest to investors and financial researchers
Therefore, finding the impacts of FPI flows on Vietnam Stock Market, especially in the long-run, will help policy makers improve policies to attract more FPI flows and achieve national economic objectives.
Research objectives
• To examine the impacts ofFPI on Vietnam Stock Market (VSM)
• To recommend general policies for sustainable development in VSM to encourage more FPI flows into Vietnam
Research questions
The thesis aims to answer the following questions:
• Is there a long-run or short-run impact ofFPI flows on VSM?
• Is it correct to say that VSM takes time to be fully adjusted after any changes in FPI flows? So how long does it take for the change to take effect?
Research scope
The research focuses on finding the impacts of FPI flows on Vietnam Stock Market
So we only investigate monthly time series data of net FPI flows and VN-Index from July 2000 to June 2012.
Structure of the thesis
1fhis thesis has five chapters which are organized as follows:
• Chapter 1 explains reasons why to choose this topic for research, the research's significance, main objectives, some research questions and the scope of the research
• Chapter 2 provides an overview on theoretical background of FPI impacts on VSM which are expressed in general through the conceptual framework
• Chapter 3 presents the research methodology Based on the other empirical researches, we draw out the study framework This chapter also describes the general econometric models, variables, data collection which are explained the reasons for choosing variables included and reliable sources to collect data
• Chapter 4 shows the statistic results from adopting the econometric model above Findings are analyzed to answer the research questions in the chapter
• Chapter 5 obtains the main findings and recommends sustainable policies to improve VSM in order to encourage FPI flows.
LITERATURE REVIEW
The role ofFPI on economic development
~apital flows including short-term portfolio flows and long-term investments have related to economic development and even to infrastructure development To boost
1 economic growth and expand resources for development finance, governments usually promote international capital inflows, strengthen capital markets in order to encourage efficient financial markets According to Bakardzhieva et al (2000), capital flows were clarified into several types "Three distinctive flows appear in the financial account of balance of payments, namely foreign direct investments (FDI), portfolio investments and other investments" In this paper, we just focus on portfolio investments into the Vietnam stock market These flows are referred as the foreign portfolio flows (FPI) in Vietnam
Foreign Portfolio Investment (FPI) represents passive holdings of securities such as foreign stocks, bonds or other financial assets, none of which entails active management or control of the securities' issuers by the foreign investors However, they can sell off easily the securities and pull out the portfolio investment
Therefore, FPI is much more volatile than FDI For developing countries, FPI can bring rapid development, helping an emerging economic opportunity, job creations, and significant wealth When an economic takes a downturn, or fails to meet the expectations of international investors, the huge capital inflows can be withdrawn grammatically According to World Bank (2001), the external problem of excessive capital outflows were as follows: the capital outflows above critical threshold levels might impact adversely on the domestic economy by draining foreign exchange reserves, reducing the resources available for domestic investment, and slowing the developing of the financial sector However, the World Bank (2001) report also found that there was an existence of a strong relationship between FPI with domestic investment In the other word, in a research on some East Asia economies during 1990s of Henry (2000), stock market liberalization on trading might lead to investment booms But, capital inflows might not lead to economic growth because it occurs in conjunction with a set of domestic complementarities for capital absorption, retentive capabilities, and consequent impact on production and consumption In fact, global financial integration only allows greater ease in the entry and exit of capital _
The role of Vietnam Stock Market
The State Securities Commission, part of the Ministry of Finance, has been managing the Vietnam stock market They have issued a Law on Securities in June
2006 to facilitate the development of the securities market speedily and sustainably by covering the regulation of listing and trading securities, the State's roles in administering and inspecting the securities market There are two stock exchanges in Vietnam, one in Hanoi and one in Ho Chi Minh City The Ho Chi Minh stock exchange was inaugurated in July 2000 and became a main Vietnam stock exchange with approximately 280 companies listed We use the measure of VSM expressed as VN-Index quoted in the Ho Chi Minh stock exchange It is a capitalization- weighted index of all the companies listed on the Ho Chi Minh Stock Exchange
The index was created with a base index value of 100 as of July 2000 In order to estimate the growth rate of VN-Index, we take the logarithm of VN-Index over lag one ofVN-Index and named this variable as Delta-VN
Stock markets bring benefits to corporations, individual investors and governments
For corporations, by making an Initial Public Offering (IPO) on the stock exchange, ' a corporation can gain access to a huge amount of investors, raise capitals by attracting abundant capital resources for their business Moreover, access to the stock markets also facilitates growth by merger or acquisition through share purchases For investors, stock markets are able to help them improve returns by diversifYing their choices of different corporations and industries to invest Equities cannot ensure a fixed rate of return Thus, they become a riskier investment than money markets or bonds What equities provide is the prospect of a combination of income and capital gains, plus a superior rate of return For the economy, stock markets can put people's savings to work, the economy cannot get benefits or just a little from individual cash savings or bank accounts Stock investment is a direct method in the success of businesses and helps promote stronger economic growth
In the other word, stock markets are also a measure of the economy's performance
In general, the performance of share prices is a good indicator of its current condition and of the confidence of individuals within that economy So, in some extent, the performance of the stock market is correlated with the health of the economy Moreover, the strict regulations and requirements for corporation's stock to be listed on the stock exchange and maintained on trading are a good way for investors to ensure corporate governance because management standards and standards of record keeping within that corporation are maintained at a high level
For governments, stock markets can give access to funds because the stock exchange allows individuals to lend money to their government when government may issue bonds quoted on the stock market to raise money for infrastructure or major projects.
Theoretical framework
7here are many researches on the role of FPI flows through the stock market into emerging market economies
2.3.1 Foreign Portfolio and stock market
Theoretically, many economists and researchers have the different viewpoints about the relationship between FPI and stock returns in domestic economies
The first theory is to support the unilaterally impacts from FPI flows on stock returns According to Clark and Berko (1997) who investigated the economically and statistically significant positive correlation between monthly foreign purchases of Mexican stocks and Mexican stock returns, a percent increase in foreign inflows led to 13 percent increase in Mexican stock prices Additionally, Choe et al (1999), who examined foreign investors' impacts on the Korean stock index from November 1996 to the end of 1997, found evidences of foreign investors' positive trading and herding before the Korea economic crisis However, there was no evidence that the foreign investorsã transactions had a destabilizing impact on the Korean stock market during their sample period The Korean stock market was adjusted quickly and efficiently by large sales of foreign investors On the other hand, Bose and Condoo (2004), who studied the impact ofthe FII policy reforms on FII portfolio flows to the Indian stock market through a multivariate GARCH regression model, strongly suggested that liberalization policies had had the desired expansionary effect and obtained a sensitive impact of FII inflows to a change in BSE returns
The second theory is to illustrate the viewpoint that there is no correlation between FPI flows and stock returns According to Singh and Weisse (1998) who examined two major components of financial liberalization : stock market development and portfolio capital flows in the scenarios of less developed countries (LDC), LDCs should pay attention on strengthening their banking systems rather than stock markets because their banks could promote long-term economic growth and
1industrialization Moreover, they were able to suffer the burden of globalization 1without speculative portfolio inflows In addition, Pal (2006), who aimed to
1examine the impact of Foreign Portfolio Investment on Indian economy through the
I stock market, showed that the perceived benefits of foreign portfolio investment had not been utilized in India The prediction that the foreign portfolio investors would boost economy through a country's stock market did not work in India
The third theory is to demonstrate the impacts from stock returns on FPI flows
According to Ko et al (2005) who investigated the characteristics of the stock ownership by institutional and foreign investors in both Japan and Korea, foreign investors had more advantages in preferences to large capitalization and low book- to-market ratio stocks than institutional investors in both stock markets
Furthermore, foreign investors prefered high-return stocks, especially in Korea
Moreover, the preferred stocks of both institutional and foreign investors had statistically significant positive abnormal profits in both markets while favored ones by either institutional or foreign investors had statistically significant positive abnormal only in Korea In other word, Liljeblom, E and Loflund, A (2005), who investigated determinants of foreign equity investment flows after the deregulation of Finnish stock market, indicated that the Finish stocks owned by foreigners were found to deviate clearly from the Finnish stock market Portfolios of foreign investors were significantly titled to additional withholding tax on low dividend- yield Moreover, large- capitalization and liquidity stocks were preferred with a record of strong profitability (measured by past ROI) Additionally, Thapa and Poshakwale (2012), who found the answer for the question whether national equity market characteristics explained specific differentiation in distribution of foreign equity portfolios by using panel data of comprehensive foreign portfolio holdings and different measures of national stock market factors for 36 host countries, showed that foreign investors prefered larger and more visible developed markets with higher liquidity, higher efficiency and lower transaction costs
However, only a few studies like Froot et al (200 1) found a bilateral impact ,between FPI flows and stock returns They studied the behavior of international
I portfolio flows and their relationship with equity returns by employing panel data of
144 countries and found that international portfolio flows were strongly influenced by past returns while foreign portfolio inflows had positive impact for future equity returns Moreover, the sensitivity of local stock prices to foreign inflows was positive and large
Based on the reality scenario in Vietnam, we support the theory about the unilateral impacts from FPI flows on the VSM
Through theoretical frameworks, we can present the impact of FPI flows on stock market as follows:
Test the impacts of FPI flows on Vietnam stock market
(Delta-VN) in the long-run and short-run
There has been the growing role of foreign portfolio investment in international financial markets over the last decade The increased flows to securities investment from industrialized countries to emerging markets are able to lead to possibility for development in all involved countries The percentage of FPI holding in a stock is an important factor to analyze When this rate increases, the stock price goes up and when it drops, the share price comes down If foreign investors invest in a company,
I jt is the good signal for this company's growth rate because they have seen the potential growth in the recipient However, in case, foreign holding rate is too large, tt means that this company's stock price is very volatile and risky because it's easier for foreigners to move out of a stock But no one can deny some positive impacts of FPI on stock markets First, foreign portfolio investors are professionals on stock markets So, they always purchase stocks on the basic of fundamentals It means that they require more information to evaluate This leads to growing demands on companies to become more transparent and more disclosure in order to be more attractive to investors Thus it helps reducing information asymmetric on stock markets Second, the globalization on stock markets require to reform securities trading and transaction systems, nurture securities brokers and liquid markets
Third, the stock markets' openness for FPI makes it more attractive to foreign capital either direct or indirect flows Fourth, FPI inflows boost financial innovation 'and development in trading instruments Not only does it enhance competition in financial markets but also improves the alignment of asset prices to fundamentals
:However, the other side of the coin is that there are some dangers if certain limits are exceeded First, FPI flows are free and unpredictable Moreover, foreign investors always look for profits When FPI flows move investments, they are likely to cause severe price fluctuations resulting in risky volatility Second, increased holding rate from overseas may lead to loss of control in domestic firms Third, when FPI flows into stock market in huge amounts, they can create great influence on the way the stock market behaves, going up or down Fourth, the effect ofFPI on the currency appreciation may lead to the un-competitiveness in the export industries
In this thesis, we aim to investigate the impacts of FPI flows on Vietnam Stock Market (VSM) in the long run and the short run Then, we make some recommendations on policies for sustainable development in VSM to attract more FPI flows i •
Empirical studies
Recently, there have been a lot of researches about the relationship between FPI flows and stock returns By applying various methods, most recent studies demonstrated the complementary impact of FPI on stock returns
Chakrabarti, R (200 1) analyzed FII flows in India and their relationship with other economic variables including stock returns by applying Granger Causality test on monthly data from January 1993 to December 1999 of FII flows to India, market capitalization data and other financial data like the exchange rate, short term interest rate in India, return on the MSCI world index, S&P 500, BSE national index and country credit rating data He arrived the conclusions related to stock returns that while the flows were closely related to Idian equity returns, they were effected by these returns rather than the cause of that Moreover, the Asian crisis marked a regime change in the determinants of FII flows to Indian equity returns becoming the sole driver of flows since the crisis Moreover, when Fils were compared to the local investors, they did not face an informational disadvantage in India
On the other hand, Mukherjee et al (2002) used two types of variables to explore the relationship of foreign institutional investment (FII) flows to the Indian equity market with its possible covariates for the period January 1999 to May 2002 The first included variables reflecting daily market return and its volatility in domestic ã and international equity markets The second were macro variables affecting foreign investors' expectation about return in Indian equity market like exchange rate, short-term interest rate and index of industrial production (liP) The data-set was combined with day-to-day variations, thus, it was suitable to test various correlations, including Granger causality for equity market operations After relating daily FII flows, they modified the model specification to include the variables' short-term past history over different time frames, like a week or fortnight Later, they tried to relate FII flows to Indian macroeconomic indicators
Therefore, their results indicated that FII flows in the Indian equity market had
I tendency to be caused by stock market returns but not vice versa It meant that Indian equity return was the most important factor affecting the FII flows into India
Moreover, while FII sale and FII net inflows were strongly affected by Indian stock market's performance, this market performance did not cause FII purchase In other words, FII investors did not diversify their investments by Indian stock market
Also, returns from exchange rate differentiation and the Indian economic performance might have affected on FII inflows , but such effects did not seem to be strong Finally, FII flows were automatically daily correlated and this correlation could not be calculated for the all
In addition to the international evidences of the role of FPI on stock returns, there are numerous evidences in a group of specific countries Syriopoulos (20 1 0) employed weekly stock index closing prices in the Balkan countries such as the BET -C of Romania, SO FIX of Bulgaria, CROBEX of Croatia, ISE-1 00 of Turkey, GR-GI of Greece, CYP-GI of Cyprus, S&P500 of the US and DAX of Germany expanding from April 27, 1998 to September 10, 2007 to examine the risk and :return of international portfolios allocated by investors to major Balkan equity markets They applied linear method like error-correction vector autoregressive model and non-linear method like switching regime error correction model to test for the potential linkages between Balkans and developed stock markets The results illustrated the presence of co-integration vectors indicating a stationary long-run relationship among the Balkan equity markets On the other hand, the Balkan equity markets were affected by both domestic and external forces and shaped their long- run equilibrium path However, inflows of international portfolio investments and trading activity in the Balkan equity markets were growing rapidly
Saxena and Bhadauriya (20 11) collected daily data series on Fils inflows and S&P CNX NIFTY computed by logarithmic returns on daily closing prices for 7 financial years from April 2003 to March 2010 to explore the causal relationship between FII inflows and volatility in indices of NSE by adopting unit root test, Granger Causality test in a hi-variable VAR framework and vector auto regression
For applying the test, they had to convert all variables into a stationary process before including them into a V AR system In order to test the variables' stationary, they used Augmented Dickey Fuller test and Phillip Peron test They aimed to find the answer for the question that whether movements in Fil inflows had an effect on stock market returns or movements in stock market returns had an effect on direction of Fil inflows in India and got the results that there was no bidirectional causal relationship between stock market volatility and Fil inflows They found that stock market volatility was a cause to foreign institutional investment inflows and the trends of foreign institutional investment inflows did not have that much impact on stock market volatility Also, they found that the past data of stock market returns could forecast the present and future trend of foreign institutional investment inflows to India
The most recent research is Kumar et al (2012) who studied empirically dynamic interaction between Foreign Institutional Investor (Fil) flows and Indian stock market returns through aggregate daily Fil data comprising three components purchases, sales and net purchases along with the S&P CNX Nifty market index taken by the log difference from 7th January 2000 to 6 1 h August 2009 using ordinary least square regression , vector auto regression and impulse response function along with Granger Causality test to illustrate a sharp and significant impacts between Fil flows and Indian equity market returns The results showed strong evidence of positive feedback trading of Fils with an adjusted R square of eleven percent Also the Granger Causality test led to rejection of both null hypothesis lending strong support to a bidirectional relation between Fils and equity market returns in India
But, the overall response function of institutional investors to a one standard error shock revealed a sharp and significant impacts dying out in four to five days.
Suggested research model
Theoretical framework shows that there are strong impacts from FPI flows on stock returns As a paper of Saxena and Bhadauriya (20 11) and Syriopoulos (20 1 0), we adopt unit root test, co-integration test, Granger Causality test and vector error
~orrection model to study the impact of FPI flows on Vietnam stock market Due to
6ur support to the theory of unilateral impacts from FPI flows on stock returns, we
I prefer Delta-VN as dependent variable and FPI as independent variable in our
I model in order to estimate the role of FPI flows on the VSM
Therefore, our suggested general model for this thesis is as follow:
In summary, most studies employ daily data to investigate the relationship between FPI flows and stock market returns However, we shall expand the data into monthly data like the study of Chakrabarti (200 1) to get a general view about the 'impacts from FPI flows on stock returns in month by month Moreover, we shall learn the methods of Saxena and Bhadauriya (20 11) in testing time series data such ,as unit root test using both Augmented Dickey Fuller (ADF) test and Phillip Peron (PP) test, Granger Causality test in a hi-variable VAR framework Additionally, we also extend to co-integration test on these monthly data to examine a long-term relationship between them In other words, we shall adopt a lesson from Syriopoulos (20 1 0) in using error correction model to investigate a short-run relationship between two variables Finally, we infer a general model for this thesis
The beginning of this chapter is to discuss and justify the methods to analyze: stationary & unit root test, co-integration, Granger Causality, error correction model Then, we describe our data collection and go ahead to data analysis Finally, we summarize this chapter into the chapter remark
3.1.1 Stationary and unit-root tests
According to Gujarati and Porter (2009) a time series is stationary when its mean and variance are constant over time and the value of the covariance between the two time periods depends only on the distance or gap or lag between the two time periods and not the actual time at which the covariance is computed Our thesis uses a set of time series data to generalize the behavior of foreign portfolio investment (FPI) and VNindex (adjusted as Delta-VN to obtain the growth rate of VNindex) Thus the measurement of stationary time series becomes an important issue
Because if a time series is non-stationary, we can study its behavior only for the time period under-consideration As a consequence, it is not possible to generalize it to other time periods So the regression becomes spurious and lead to incorrect conclusions
By applying the augmented Dickey-Fuller (ADF) test for stationary (see Dickey and Fuller 1979, 1981) we obtain the estimated equation for the ADF test as follows:
• m AYt= ao + ~~Yt-I+Ot: + IeiAYt-I +Et (1) i=1
Where A is the first different operator, Yt is Delta-VN or FPI, t is the time trend, E is
I the stationary random error and m is the maximum lag length The null hypothesis is that the series contains a unit root which implies that ~ =0 The null hypothesis is rejected if~ is negative and statistically significant
When we regress two non-stationary time senes, we may produce a spunous regression Suppose we consider two time series data X &Y and subject these time series individually to unit root analysis, we find that they contain a stochastic trend
It is possible that the two series have the same common trend so that the regression of one on the other will not be spurious We say that two variables are co integrated
'It means that two variables will be co-integrated if they have a long term relationship between them, on equilibrium (Gujarati and Porter, 2009)
Among various approaches to test co-integration, Engle and Granger approach is the most conventional technique But it is very restrictive because it can be applied only on a single co-integrating relation Thus, we should employ additionally the , most commonly used method, say, Johansen approach based on the autoregressive
I representation which provides two different likelihood ratio tests: the trace statistic
, and the maximum eigenvalue (see Johansen 1988, Johansen and Juselius 1990) ã 3.1.3 Granger causality tests
In our thesis, we use the Granger Causality Test to check the existence of causal relationship between two variables: FPI and Delta-VN According to Gujarati and Porter (2009), the Granger causality test assumes that the information relevant to
Chapter remark
Qn these variables A Granger causality test involves regressing a variable X on lagged values of itself and on lagged values of the other variable of interest Z, as specific as the following model:
Where a is the intercept, K and L are the order of auto regression for the variable X and Z, respectively, Bi is the parameter associated with the ith lag of variable X, A:i is the parameter associated with the jth lag of variable Z and £ 1 is white noise The null hypothesis that X does not Granger cause Z is determined by comparing the F- test of this full model 2 versus a restricted model in which X is regressed on only its own lagged values (A:i=O, j=l,2 L)
In this thesis, X may be Delta- VN and Z may be FPI By interchanging the roles of Delta-VN and FPI, reverse causality as well as the existence of feedback between the two variables can be examined
In the short-run, the variables may be disequilibrium, after confirming that the variables are co-integrated and have Granger causality relationship, we conduct the ECM test to treat the error term as the "equilibrium error" We use this error term to tie the short-run behavior of the variables to its long-run value (see Gujarati and Porter, 2009) as the ECM below:
~ Yt = 9o + k-1 Iei~ Yt-1 + a~ãyt-k +Et (3) i=l Where ~ is the difference operator, Yt is (Delta-VN, FPI), 90 represents the intercept, and Et represents the vector of white noise process The matrix ~ consists of r (r ~ n-1) co-integrating vectors Similarly, the matrix a contains the error correction parameters In equation (3), the null hypothesis is that the matrix (rr a~) has a reduced rank of r ~ n-1 The alternative hypothesis, on the other hand, is that the matrix (n =a~) has full rank
In this study, the error correction models are based on the following regression
~ Delta-VNt = azt-1+ L ~i~Delta-VNt-i + LDj~FPit-1 +Jlt (4) i=1 j=1 c d
~FPit = OZt-1 + L ei~Pit-i + L~~Delta-VNt-1 + Et (5) i=1 j=1
Where Zt-J represent the error correction term lagged by one period, Delta-VN is the growth rate of VNindex, FPI stands for the foreign portfolio investments, and a, b, c and d represent the optimal lag lengths obtained from the Akaike Information Criterion (see Akaike, 1973) In equation 4, the null hypothesis the growth rate of FPI does not cause the growth rate of VNindex is rejected on the condition that
~either the sum of E 1 or a is statistically significant Similarly, in equation (5), the
1null hypothesis the growth rate ofDelta-VN does not cause the growth rate ofFPI is
1rejected provided either the sum of Aj or o is statistically significant
The analysis is conducted on monthly data from July 2000 to June 2012 Thus, we have 144 observations All variables are presented in table 3 1
VARIABLE DEFINITION SOURCE pependent variable
VN-Index Vietnam Stock Index HOSE
Delta-VN The growth rate of VN- Author's calculation
Index equals logarithm of (VN-Index/VN- Index(-1)
Investment (Net FPI inflows) ãThe main objective of the thesis is to explore the impacts of FPI flows on VSM The study is carried out by using monthly data series of FPI flows into VSM and
'Vietnam market returns (expressed through VN-Index) from July 2000 to June
12012 We have to limit the study within this time range as the VSM was inaugurated in July 2000 Both variables have been taken from the official website of Ho Chi Minh City Stock Exchange (HOSE: www.hsx.vn) Aggregate monthly FPI data comprising two components purchases and sales, then, the net FPI inflows are computed by taking purchase flows less sales flows and simply named FPI ã variable in this thesis
On the other hand, we take the last indices of the month as monthly indices for VN- Index variable, then, we compute the growth rate of VN-Index by taking logarithm ã the VN-Index using the following formula:
VNindext Delta-VN=logVNI de n
• Delta-VN =the growth rate ofVN-Index
By using Eviews, we conduct analysis procedures in order to answer the research questions whether there is a long run or short run impacts of FPI flows on VSM or
' hot and how long it takes for the change to take effect
, Graph3.3-l: Delta-VN=log (VN-lndex/VN-Index (-1)) Source: Hose & Author's
Delta-VN is a variable that denotes the growth rate of VN-Index Through graph3.3- 1, we can see that over the sample period, the growth rate of VN-Index
' has been fluctuating around its mean which suggests that these series are stationary
They seem to be I (0) so they have stationary stochastic process This can be
1 confirmed by unit root analysis as shown in Table 4.2-1
(iraph 3.3-2: Foreign portfolio investment flows (FPJ) to Vietnam from July 2000
]n this thesis, FPI variable denotes the net inflows of foreign portfolio investment over the sample period As can be seen in the graph, the series have been fluctuating around its mean Before the late of 2006, there seems to have no investment from tbreigners into VSM After that, thanks to the opening in the equity market, foreign portfolio capitals started to flow rapidly into VSM
3.3.3 Interaction between FPI flows and VN-Index
~raph 3.3-3: FPI & VN!ndexfrom July 2000-June 2012 Source: IFS & Hose f'.fter signing the Bilateral Trade Agreement with United State in July 2000 and officially joining World Trade Organizations (WTO) in 2007, Vietnam had committed to reform and opened the economy for foreign investors Then, we have got some achievements in development, especially in private sector and trade liberalization The VSM officially traded on July 28th 2000 starting with the start at
100 points and the first two trading shares were REE (Refrigeration Electrical Ip:ngineering Joint Stock Company) and SAM (Cable and Telecommunication
Materials Joint Stock Company) Then, the VN-Index had been continuously
I increasing during the following 12 months and peaked at 570 points on June 25th lOO 1 Due to the scarcity of trading goods in this period, the fact that demand was greater than supply many times led to investors competing to buy at the ceiling prices and boosted the VN-Index increase continuously After peaking the mark of
5170 points, VN-Index had gone down quickly during the next 3.5 months, VN-
Ipdex lost 64% of its value, and got 203 points on October 5th 2001 Before 2001,
I there was nearly no portfolio inflows from foreigners into VSM Only some investors paid interest in the stock market The net inflows were always lower than
5 billion dongs because just a little bit foreign capital flows ran into Vietnam The
• difficulties that foreign investors had to face in Vietnam were unstable economic
I environment, the out-of-date legal framework which did not catch up with the
1 market growth and the restrictions on foreign investment while other emerging markets were more attractive to investors than Vietnam During this period, the role of foreign investors did not stand out on the VSM Due to the long stand-by stage in the VN-Index, foreign investors did not appreciate the possibility of recovery and
~profitability of the share prices Moreover, the listed companies was too small and the legal framework was so strict
Then, there was a rebound on the VN-Index up to 301 points before going into the decline stage lasting for 2 years from November 2001 to 2003 The VSM was going sideways during the two years 2004 and 2005 On March gth 2005, the Hanoi Securities Trading Center, where the small chartered capital companies were able to be listed and traded, officially operated with less strict listing conditions to encourage small joint stock companies to list shares on the stock market So, it helped increasing the share supply and diversified options for investors to trade
Market boomed from the mark of 300 points in the earlier 2006 The period from the end of 2003 to 2006 was the recovery term and booming in FPI inflows when the government issued the opening and transparency policies for FPI investors livforeover, the government aimed to promote privatization, loosen the foreign investors' stock holding ratio.Since 2004, foreigners had some positive reactions to VSM, the flows fluctuated around 10 billion dongs and reached the peak in this period at 189.96 billion dongs on March 2004 In 2006, the interest of foreigners to VSM was increasing dramatically The net inflows had risen to over 500 billion dongs and hit the peak of2006 at 2,323.67 billion dongs on December
Before 2007, Vietnam economy performed regularly the low growth rate of CPI which fluctuated around 0.5% and kept the CPI over these years under two digits
Since 2007, as committed with WTO members, the government had allowed the
• privatization in some sensitive fields such as banking, insurance, petroleum, telecommunication sectors and relaxed many restrictions for foreigner investors
As a consequence of the openness in the economy, the story of inflation started in
2007 and boomed in 2008 While the economy was gradually absorbing huge cash
Chapter remark
This chapter expresses all results of empirical analysis At the end of this chapter,
I we summarize these results into the chapter remark
Table 4.1-1: Summary of Structural Breakpoint Test
Conclusion: By using the Quandt-Andrews breakpoint test, we recognize that the Delta VN is not broken while the FPI is broken at July 2006 The breakpoint of FPI variable is depicted in graph 3.3-2 The answer about the causal relationship between this pair variables will be clear after performing Granger causality tests presented in the following empirical results This figure also indicates that the series have increasingly grown since the breakpoint
,Unit root test results usingADFva PP
,In order to establish the order of integration of the variables, we apply the conventional unit root tests widely known as augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests Table 4.2-1 reports the results of the standard unit root tests (ADF and PP) on the integration properties of the Delta-VN and FPI
EMPIRICAL ANALYSIS
Structural Break Point test
Table 4.1-1: Summary of Structural Breakpoint Test
Conclusion: By using the Quandt-Andrews breakpoint test, we recognize that the Delta VN is not broken while the FPI is broken at July 2006 The breakpoint of FPI variable is depicted in graph 3.3-2 The answer about the causal relationship between this pair variables will be clear after performing Granger causality tests presented in the following empirical results This figure also indicates that the series have increasingly grown since the breakpoint.
Unit root test
,Unit root test results usingADFva PP
,In order to establish the order of integration of the variables, we apply the conventional unit root tests widely known as augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests Table 4.2-1 reports the results of the standard unit root tests (ADF and PP) on the integration properties of the Delta-VN and FPI
!variables Because the actual values of these series exhibit no trends, so all unit root test regressions include no intercept and trend tenns -
Table 4.2-1: Summary of Unit root test results
Level (with no trend and no
Variables intercept) at 1 °/o ADF statistic PP statistic
Notes: *, **,***indicate significance at 1%, 5%,10% levels respectively
'Both ADF and PP tests show remarkably similar results, all variables Delta-VN and
1FPI are stationary at their levels Hence we conclude that all series are I (0) at the 1% level of significance Given that these variables share common integration ,properties, we can proceed to test for the presence of a long-run co-integrating ,relationship between both variables
Co-integration test
Unit root test (or residuals (Engle and Granger) usingADF and PP
Residuals obtained from OLS regressions between Delta-VN and FPI are then tested by using the ADF and PP tests
Table 4.3-1: Summary of Unit root test for residuals using ADF and PP: Engle &
ADF statistic PP statistic Yes/No
Notes: *, **, *** indicate significance at I%, 5%, I 0% levels respectively, refers to SIC criterion, stationary at 1(0)
1Residual series are stationary at its level Thus we conclude that this series are I (0)
I at the 1% level of significance The test result indicates that there seems to be
1cointegrating relationship between this pair
Johansen cointegration estimation results between Delta- VN and FPI
To be more convinceable, we then apply the Johansen approach for testing 1cointegration between Delta-VN and FPI variables Since the Johansen approach is sensitive to the lag length used, we conduct a series of nested likelihood ratio tests on the first-differenced V ARs to determine the optimal lag length prior to performing co-integration tests Given the sample size, we have considered a maximum lag length of 36 Based on the AIC criteria, the optimal lag length is determined at 12 We follow this lag structure for the rest of the estimation
1There are five alternative models in co-integration test such as: model 1 (No 1intercept or trend in CE or V AR), model 2 (intercept (no trend) in CE, no intercept ,in VAR), model 3 (intercept (no trend) in CE and test V AR), model 4 (intercept and trend in CE-no trend in VAR), model 5 (intercept and trend in CE-linear trend in
VAR) However, in our case, we only estimate models 2, 3, 4 because models 1, 5 occur very rarely
Table 4.3-2: Summary of Johansen co- integration test ã Rank test ( trace) Model2 (intercept (no trend) in CE, no intercept or trend in V AR) Number of co-integration None
1 Model 3 (intercept in CE and V AR, no trend in CE and V AR)
Model 4 (intercept in CE and V AR, linear trend in CE, no trend in V AR)
Based on the trace statistics for all three models as in table 4.3-2, we choose an
I appropriate model to test for co-integration First, we start with the smaller number
I bf co-integrating vectors r = 0 and check whether the trace statistic for model 2
I rejects the null, if yes we proceed to the right, checking whether model 3 rejects the
I null and so on If the trace statistic is greater than critical value, we reject the null hypothesis Otherwise, we do not reject the null hypothesis In this case, model 4 suggests that the trace statistic is smaller than the critical value, so we fail to reject the null hypothesis It means that model 4 shows a co-integration So we choose ,model4 to estimate the co-integration test
Table 4.3-3: Summary of Trace Statistic Value
, In line with the literature review on FPI and stock markets (see Ko et al, 2005; ã Choe et al, 1999; Bose and Coondoo, 2004 ), FPI may have positive impacts on stock markets So the selected model should be statistically significant and consistent with the theoretically expected positive sign It can be inferred that FPI has a positive impact on VSM in the long-run
Granger Causality test
We know that cointegration implies the existence of Granger causality but it does not indicate the direction of the causality relationship Therefore, we have to apply the bivariate Granger Causality tests to find the direction of the causality relationship
Table 4.4-1: Summary of Granger Causality Test
, Ho: Null hypothesis based on Granger test for Granger
I FPI does not Granger Cause 24.69 Reject Ho at
Delta-VN does not Granger Cause Fail to reject
.From the bivariate Granger Causality tests, we have an important conclusion The null hypotheses that FPI do not Granger cause Delta_ VN is rejected at 5% significance level However, the null hypotheses that Delta_ VN do not cause FPI in 'Granger sense is not rejected even at 10% significance level Thus, we can conclude that there is an unilateral effect from FPI on Delta VN.
Error Correction Model
D(DELTA_ VN) = C(l)*( DELTA_ VN(-1) + 1.65037471191E-05*FPI(-1)- 0.000131596905459*@TREND(OOM07) + 0.00333129697177) + C(2)
*D(DELTA_ VN(-1)) + C(3)*D(DELTA_ VN(-2)) + C(4)*D(DELTA_ VN(-3)) + C(5)*D(DELTA_ VN(-4)) + C(6)*D(DELTA_ VN(-5)) + C(7)
*D(DELTA_ VN(-6)) + C(8)*D(DELTA_ VN(-7)) + C(9)*D(DELTA_ VN(-8)) + C(lO)*D(DELTA_ VN(-9)) + C(l1)*D(DELTA_ VN(-10)) + C(12)
*D(DELTA_ VN(-11)) + C(l3)*D(DELTA_ VN(-12)) + C(I4)*D(FPI(-l )) + C(l5)*D(FPI(-2)) + C(16)*D(FPI(-3)) + C(l7)*D(FPI( -4)) + C(l8)*D(FPI(
-12)) + C(26) Where the coefficients are expressed in the table below:
Table 4.5-1: Summary of testing Vector Error Correction Model c Coefficients t-statistic Pro b
C ( 1) means actual error correction term (ECT) which is the residual of the co integration equation between Delta_ VN and FPI In this case, C (1) shows negative and significant because the p-value is smaller than 5% level So, ECT becomes , significant It means that there exists a long run causality from FPI to Delta_ VN ã The coefficient of the ECT of about -0.86 suggests that about 86% of the discrepancy between long-term Delta-VN is corrected within a year, suggesting a high rate of adjustment to equilibrium ã On the other hand, in order to check the short-run causality from FPI to Delta_ VN,
1 we apply the Wald test to the coefficients from C (14) to C (25) which are the
I twelve lags of FPI variable in the vector error correction model (VECM) We have
~en these coefficients jointly equal zero, it means that FPI can not influence to belta _ VN Otherwise, we could conclude that there is a short-run causality from
FPI to Delta_ VN In our case, the probability of chi-square is about 1.64% smaller
I than 5% level So, we fail to reject the null hypothesis Consequently, we can make
I the conclusion that FPI cause a short-run impact on Delta_ VN However, the impact ofFPI on Delta_ VN is gradually decreasing from the lag 3 ofFPI variable
Moreover, to identify the model as the best regression one, it might satisfy some features such as: high R square value, no Serial Correlation in the residuals, no Heteroskedasticity in the residuals and normal distributed residuals In our case, the R-squared of the model (about 55.98 %) is acceptable Therefore, we should apply other tests to check the characteristics of the model like Serial Correlation LM test, Heteroskedasticity test and Histogram Normality test
The results of these tests are reported in the table below:
Table 4.5-2: Summary of the tests for appropriateness of the estimated model
, About Serial Correlation LM test, the hypotheses are as follows:
• Null hupothesis : Residuals are not serially correlated
• Alternative : Residuals are serially correlated
In our model, the p-value (about 99%) is greater than 5% So, we fail to reject the null hypothesis It means that residuals are not serially correlated
About Heteroskedasticity ARCH test, the hypotheses are as follows:
• Null hupothesis : Residuals are not heteroskedastic, that is homoskedastic
• Alternative : Residuals are heteroskedastic lrj our model, the p-value (about 59%) is greater than 5% So, we fail to reject the null hypothesis ' It means that residuals are not heteroskedasticity
About Histogram Normality test, the hypotheses are as follows:
I • Null hupothesis : Residuals are nomally distributed
• Alternative : Residuals are not nomally distributed
In our model, the p-value (about 22%) is greater than 5% So, we fail to reject the null hypothesis It means that residuals are normally distributed
Conclusion: We have failed to reject all the null hypothesis So, our model is a good regression model.
Chapter remark
We have presented the results of empirical analysis where we find out some
~ignificant short-comings First, we recognize that Delta-VN series is not broken
~hile FPI series is broken at July 2006 Second, the result illustrates the stationary
I of both variables at their first levels Third, the result indicates the co-integration
~elationship between these variables Moreover, we are likely to infer the model 4 in
'the Johansen Co-integration estimation test as a general model to investigate the error correction mechanism Fourth, the result only proves a unilateral effect from FPI on Delta-VN Finally, the result points out a specific error correction model to ãconclude long run causality from FPI to Delta-VN and suggests a high rate of 86% to adjust to equilibrium.
CONCLUSION AND POLICY RECOMMENDATIONS
Main findings 4 3
This thesis employs the unit root tests, co-integration tests both Engle Granger method and Johansen method, Granger Causality tests and vector error correction models to investigate impacts of FPI flows on Vietnam stock market both in the long-run and short-run The empirical analysis uses monthly time series data of VNindex and FPI variables from the period of July 2000 to June 2012 In general, ,the results support the theoretical and empirical literatures such as Choe et al (1999)
Studied Korean market, Froot et al (200 1) studied on 44 countries' stock markets,
Ko et al (2005) studied on Japan and Korean stock markets, Chakrabarti (2001) analyzed FPI flows on Indian economy, Syriopoulous (20 1 0) found the presence of co-integration vectors indicating long-run relationship on the Balkan equity markets, Kumar et al (2012) showed positive impacts ofFPI flows on Indian stock market returns and provide new evidences about the short-run and long-run impacts of FPI flows on VSM
The main findings of this thesis are summarized as follows:
First, there is only a unilateral effect from FPI flows on Vietnam stock returns
Second, there is the existence of a long-run impact from FPI flows to Vietnam stock returns This thesis illustrates that an increase in FPI flows can lead to an increase in Vietnam equity returns The error correction term shows a high rate adjustment of about 86% where the discrepancy between long-term and short-term would be ã adjusted fully to equilibrium after some months
Third, there is also a short-run impact from FPI to VSM This impact would be decreasing gradually since the third month
Policy recommendation
First, policy makers should set up appropriate management systems for FPI flows to :enhance competition power for the financial markets The government should be
'proactively to build up warning systems and efficient supervisions on the equity
I market in order to prevent the risks of financial crisis due to speculation, monopoly, rigging , information asymmetric on the stock market To achieve effective management, it is very necessary to develop information systems, consulting services, direct supports for foreign investors Moreover, accounting and audit
I systems in Vietnam should be taken seriously and transparently to provide accurate
1 information on the listed companies On the other hand, Vietnam should encourage
I the access from the prestigious credit rating corporations in the world such as
1 Standards & Poor's, Moody to enhance informational transparency and disclosure to
1 Second, the government should open the access for financial institutions to buy
1 shares of local banks and securities firms to increase the competitive power of our financial systems and securities companies and to serve both domestic investors and foreign investors better
Third,we should enhance the effectiveness of our financial and banking systems
The exchange rate policy should be extended its cap to be more flexible, more close to reality and accurately reflect supply and demand in the foreign exchange market
Also, the government should develop management on the inflows and make the accurate prediction about the amount of foreign exchange to the capital account
Moreover, the forex reserves of the country should be sufficient and stable to be able to intervene the market in time when crisis occurs because foreign reserves are an effective support for economic when it is under shocks and to regulate the volatility of capital flows.
Research limitation and suggestion for further study
In literature reviews, there are many macro indicators in various countries used to measure the causality relationship between FPI flows and stock returns However, in our research, we just employ two variables to run regression model because the limitations in accessing data have prevented us from collecting sufficient data to analysis Thus, it is the first limitation of our thesis
The second limitation is the insufficiency in the data series There are too many blanks in our data set because at the early period, there was nearly no foreigners' trading in VSM Thus, it's hard to show all features of the interaction between FPI and VN-Index under consideration term
The third limitation is analysis methodology application While other researchers employed diversified methods to investigate the impacts of FPI flows on economy and the behavior of stock markets such as quantitative method, multivariate GARCH regression model, Fama-French three factor model, international asset pricing models, uni-variable no-parametric Wilson-Mann-Whitney median test, Granger Causality test and Vector Error Correction ModeL we felt very confused in choosing an proper model for analysis Moreover, the gaps in knowledge of the chosen method also take us much more time to master and apply the methods fluently
The final limitation is the difficulty in comparing our findings with other empirical researches due to the lack of empirical studies on the case of Vietnam Most of them studied international scenarios or specific evidence in a country
Our suggestions for further researches is to utilize other macro indicators such as exchange rate, money supply, inflation rate, interest rate to develop our research to a higher level in order to get general sense of the interaction among macro indicators to equity market and FPI flows So, the further research will be more reliable and attract more interest
Bakardzhieva, D., Neceur, S., Kamar, B (20 10), "The impact of capital and foreign exchange flows on the competitiveness of developing countries", International Monetary Fund, No.lO, 154
Bose, S and Coondoo, D (2004), "The impact of FII regulations in India A time series intervention analysis of equity flows", Money and Finance, Icra Bulletin, Vol.2, No.l8, 54-83
Chakrabarti, R (2001), "FII flows to India: nature and causes", Money and Finance, Icra Bulletin, Vol.l2, No.28, 64-83
Choe, H., Kho, B and Stulz, R (1999), "Do foreign investors destabilize stock market? The Korean experience in 1997", Journal of Financial Economics, 54, 227-
Clark, J and Berko, E (1997), "Foreign investment fluctuation and emerging market stock returns: The case of Mexico", Reserve Bank of New York, Staff Reports, No 24
Dickey, D.A & Fuller, W.A (1979), "Distributions of estimators for autoregressive time series with a unit root", Journal of the American Statistical Association, vol 74, pp.423-431
Engle.R.F., Granger.C.W.J (1987),"Cointegration and error correction: representation, estimation and testing", Econometrica, Vol 55, 251-276
Froot, K., 0' Connell, P and Seasholes, M (2001), "The portfolio flows of international investors", Journal ofFinancial Economics, 59, 151-193
Gujarati, D and Porter, D (2009), "Basic econometrics", McGraw-Hill Higher Education
Henry, P (2000), "Do stock market liberalizations cause investment booms?"
Johansen S (1988), " Statistical analysis of cointegration vectors", Journal of Economic Dynamics and Control, Vol 12, 231-254
Johansen.S., Juselius.K (1990), " Maximum likelihood estimation and inference on cointegration with applications to the demand for money", Oxford Bulletin of Economics and Statistics, Vol 52, 169-209
Ko, K., Kim, K and Cho,S (2005), "Characteristics and performance of institutional and foreign investors in Japanese and Korean stock markets", Journal of the Japanese and International Economies, 21, 195-213
Kuma, S , Tavishi, Raju, C., and Khatua, A (2012), "Behavioral modeling of foreign institutional investor's in Indian equity market", Global Journal of Business Research, Vol.6, No.3
Liljeblom, E and Loflund, A (2005), "Determinants of international portfolio investment flows to a small market: Empirical evidence", Journal of Multinational Financial Management, 15, 211-233
Mukherjee, P., Bose, S and Coondoo, D (2002), "Foreign Institutional Investment in the Indian Equity Market", Money & Finance, 21-51
Pal, P.(2006), "Foreign portfolio investment, stock market and economic development: A case study of India", Globalize World, Sao Paulo Brazil
Saxena, S and Bhadauriya, S (20 11 ), "FII flows and stock market volatility: exploring causal link", Social Science Research Nentãork http:// ssm.com/abstract 63140
Singh, A and Weise, B (1998), "Emerging stock markets, portfolio capital flows and long-term economic growth: Micro and macro-economic perspectives", World Development, Vo1.26, No.4, 607-622
Syriopoulos, T.(20 10), "Financial integration and portfolio investments to emerging Bulkan equity markets", Journal of Multinational Financial Management, 21,40-
Thapa, C and Poshakwale, S (2012), "Country-specific equity market characteristics and foreign equity portfolio allocation", Journal of International Money and Finance, 31, 189-211
World Bank (2001), "Global Development Finance 2001", Washington DC: World Bank
Figure A-1: Structural Breakpoint Test for Delta-VN variable
Quandt-Andrews unknown breakpoint test Null Hypothesis: No breakpoints within trimmed data Varying regressors: All equation variables
Equation Sample: 2000M08 2012M06 Test Sample: 2002M06 2010M08 Number of breaks compared: 99
Maximum LR F-statistic (2007M03) 5.643179 Maximum Wald F-statistic (2007M03) 5.643179
0.2975 0.2975 Note: probabilities calculated using Hansen's (1997) method
Figure A-2: Structural Breakpoint test for FPI variable
Quandt-Andrews unknown breakpoint test Null Hypothesis: No breakpoints within trimmed data Varying regressors: All equation variables
Equation Sample: 2000M07 2012M06 Test Sample: 2002M05 2010M08 Number of breaks compared: 100
Maximum LR F-statistic (2006M07) 25.11271 Maximum Wald F-statistic (2006M07) 25.11271
Ave Wald F-statistic 8.928820 0.0000 Note: probabilities calculated using Hansen's (1997) method
Figure A-3: Unit root test for Delta- VN variable
Null Hypothesis: DELTA_ VN has a unit root Exogenous: None
Lag Length: 0 (Automatic based on AIC, MAXLAG)
Augmented Dickey-Fuller test statistic Test critical values: 1% level
Augmented Dickey-Fuller Test Equation Dependent Variable: D(DEL TA_ VN) Method: Least Squares
Date: 10/20/12 Time: 10:53 Sample (adjusted): 2000M09 2012M06 Included observations: 142 after adjustments Variable Coefficient Std Error DELTA_VN(-1) -0.593158 0.076565 t-Statistic
R-squared 0.298528 Mean dependent var Adjusted R-squared 0.298528 S.D.dependentvar S.E of regression 0.107294 Akaike info criterion Sum squared resid 1.623183 Schwarz criterion Log likelihood 115.9828 Hannan-Quinn criter
Null Hypothesis: D(DEL TA_ VN) has a unit root Exogenous: None
Lag Length: 7 (Automatic based on AIC, MAXLAG)
Augmented Dickey-Fuller test statistic Test critical values: 1% level
*MacKinnon (1996) one-sided p-values t-Statistic
Augmented Dickey-Fuller Test Equation Dependent Variable: D(DELTA_VN,2) Method: Least Squares
Date: 10/20/12 Time: 10:59 Sample (adjusted): 2001 M05 2012M06 Included observations: 134 after adjustments Variable Coefficient Std Error t-Statistic
D(DELTA_VN(-1)) -3.848659 0.517127 -7.442385 D(DEL TA_ VN(-1),2) 2.414885 0.478969 5.041843 D(DEL TA_ VN(-2),2) 1.930241 0.427741 4.512634 D(DEL TA_ VN(-3),2) 1.352062 0.369693 3.657254 D(DEL TA_ VN(-4),2) 1.051082 0.297798 3.529520 D(DEL TA_ VN(-5),2) 0.759331 0.217848 3.485605 D(DELTA_ VN(-6),2) 0.429470 0.150066 2.861869 D(DEL TA_ VN(-7),2) 0.258027 0.086796 2.972804
R-squared 0.718563 Mean dependent var Adjusted R-squared 0.702928 S.D.dependentvar S.E of regression 0.110303 Akaike info criterion Sum squared resid 1.533000 Schwarz criterion Log likelihood 109.3933 Hannan-Quinn criter
Null Hypothesis: DELTA_ VN has a unit root Exogenous: None
Bandwidth: 4 (Newey-West using Bartlett kernel)
Phillips-Perron test statistic Test critical values: 1% level
Residual variance (no correction) HAC corrected variance (Bartlett kernel)
Phillips-Perron Test Equation Dependent Variable: D(DEL TA_ VN) Method: Least Squares
Date: 10/23/12 Time: 22:08 Sample (adjusted): 2000M09 2012M06 Included observations: 142 after adjustments Variable Coefficient Std Error DELTA_ VN(-1) -0.593158 0.076565
-7.547058 -2.581349 -1.943090 -1.615220 t-Statistic -7.747132 R-squared 0.298528 Mean dependent var
Adjusted R-squared 0.298528 S.D dependentvar 0.128106 S.E of regression 0.107294 Akaike info criterion -1.619476 Sum squared resid 1.623183 Schwarz criterion -1.598661 Log likelihood 115.9828 Hannan-Quinn criter -1.611018 Durbin-Watson stat 1.889808
Figure A-4: Unit root test for FPI variable
Null Hypothesis: FPI has a unit root Exogenous: None
Lag Length: 11 (Automatic based on AIC, MAXLAG)
Augmented Dickey-Fuller test statistic Test critical values: 1% level
Augmented Dickey-Fuller Test Equation Dependent Variable: D(FPI)
Method: Least Squares Date: 10/20/12 Time: 10:56 t-Statistic
Sample (adjusted): 2001 M07 2012M06 Included observations: 132 after adjustments Variable Coefficient Std Error t-Statistic
FPI(-1) -0.218822 0.090261 -2.424328 D(FPI(-1)) -0.295215 0.115002 -2.567042 D(FPI(-2)) -0.386860 0.118246 -3.271659 D(FPI(-3)) -0.265469 0.123793 -2.144462 D(FPI(-4)) -0.038726 0.126186 -0.306894 D(FPI(-5)) -0.104438 0.126672 -0.824474 D(FPI(-6)) 0.097404 0.126775 0.768321 D(FPI(-7)) -0.066258 0.124123 -0.533812 D(FPI(-8)) 0.201415 0.122472 1.644580 D(FPI(-9)) 0.161526 0.115021 1.404311 D(FPI(-10)) 0.201123 0.104224 1.929707 D(FPI(-11)) 0.385354 0.095011 4.055913
R-squared 0.401647 Mean dependent var Adjusted R-squared 0.346798 S.D.dependentvar S.E of regression 725.6456 Akaike info criterion Sum squared resid 63187389 Schwarz criterion Log likelihood -1050.502 Hannan-Quinn criter
Null Hypothesis: D(FPI) has a unit root Exogenous: None
Lag Length: 10 (Automatic based on AIC, MAXLAG) t-Statistic
Augmented Dicke~-Fuller test statistic -3.160326 Test critical values: 1% level -2.582599
Augmented Dickey-Fuller Test Equation Dependent Variable: D(FPI,2)
Method: Least Squares Date: 10/20/12 Time: 10:58 Sample (adjusted): 2001 M07 2012M06 Included observations: 132 after adjustments Variable Coefficient
D(FPI(-1)) -2.330280 D(FPI(-1 },2) 0.848848 D(FPI(-2},2) 0.291185 D(FPI(-3},2) -0.123095 D(FPI(-4),2} -0.296988 D(FPI(-5),2) -0.522515 D(FPI(-6),2) -0.536323 D(FPI(-7),2) -0.703872 D(FPI(-8),2} -0.589072 D(FPI(-9),2} -0.499387 D(FPI(-10},2) -0.349596
Adjusted R-squared 0.729525 S.E of regression 740.1261 Sum squared resid 66282183 Log likelihood -1053.657 Durbin-Watson stat 1.948253
Null Hypothesis: FPI has a unit root Exogenous: None
Mean dependent var S.D.dependentvar Akaike info criterion Schwarz criterion Hannan-Quinn criter
Bandwidth: 7 (Newey-West using Bartlett kernel)
Phillips-Perron test statistic Test critical values: 1% level
Residual variance (no correction) HAC corrected variance (Bartlett kernel)
Phillips-Perron Test Equation Dependent Variable: D(FPI) Method: Least Squares Date: 10/23/12 Time: 22:08 Sample (adjusted): 2000M08 2012M06 Included observations: 143 after adjustments Variable Coefficient Std Error t-Statistic FPI(-1) -0.394015 0.066950 -5.885197
R-squared 0.196058 Mean dependent var Adjusted R-squared 0.196058 S.D dependent var S.E of regression 773.2237 Akaike info criterion Sum squared resid 84898244 Schwarz criterion Log likelihood -1153.438 Hannan-Quinn criter
Figure A-5: Co- integration test (Engle & Granger method) for residuals from the linear regression oftwo variables
Null Hypothesis: RES03 has a unit root Exogenous: None
Lag Length: 0 (Automatic based on SIC, MAXLAG)
Augmented Dickey-Fuller test statistic Test critical values: 1% level
Augmented Dickey-Fuller Test Equation Dependent Variable: D(RES03)
Method: Least Squares Date: 10/20/12 Time: 22:29 Sample (adjusted): 2000M09 2012M06 Included observations: 142 after adjustments Variable Coefficient Std Error t-Statistic
Adjusted R-squared 0.294700 S.E of regression 0.104857 Sum squared resid 1.550294 Log likelihood 119.2449 Durbin-Watson stat 1.899303
Null Hypothesis: RES03 has a unit root Exogenous: None
Mean dependent var S.D.dependentvar Akaike info criterion Schwarz criterion Hannan-Quinn criter
Bandwidth: 3 (Newey-West using Bartlett kernel)
Phillips-Perron test statistic Test critical values: 1% level
Residual variance (no correction) HAC corrected variance (Bartlett kernel)
Phillips-Perron Test Equation Dependent Variable: D(RES03) Method: Least Squares Date: 10/20/12 Time: 22:30
Sample (adjusted): 2000M09 2012M06 Included observations: 142 after adjustments Variable Coefficient Std Error t -Statistic RES03(-1) -0.585361 0.076256 -7.676256
R-squared 0.294700 Mean dependent var Adjusted R-squared 0.294700 S.D.dependentvar S.E of regression 0.104857 Akaike info criterion Sum squared resid 1.550294 Schwarz criterion Log likelihood 119.2449 Hannan-Quinn criter
THREE MODELS IN TESTING FOR COINTEGRATION
Figure A-6: Results of the Johansen test for model 2:
Included observations: 130 after adjustments Trend assumption: No deterministic trend (restricted constant) i Series: DELTA_VN FPI Lags interval (in first differences): 1 to 12 Unrestricted Cointegration Rank Test (Trace)
Trace test indicates 2 cointegrating eqn(s) at the 0.051evel
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug- Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue)
Max-eigenvalue test indicates no cointegration at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegrating Coefficients (normalized by b'*S11 *b=l):
Normalized cointegrating coefficients (standard error in parentheses)
Adjustment coefficients (standard error in parentheses) D(DEL TA_ VN) -0.784203
Figure A-7: Results of theJohansen test/or model3 :
Date: 10/22/12 Time: 00:32 Sample (adjusted): 2001M09 2012M06 Included observations: 130 after adjustments Trend assumption: Linear deterministic trend Series: DELTA_VN FPI
Lags interval (in first differences): 1 to 12 Unrestricted Cointegration Rank Test (Trace)
Trace test indicates 2 cointegrating eqn(s) at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue)
Max-eigenvalue test indicates 2 cointegrating eqn(s) at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegrating Coefficients (normalized by b'*S11*b=I):
Normalized cointegrating coefficients (standard error in parentheses)
(1.7E-05)Adjustment coefficients (standard error in parentheses)O(OELTA_VN) -0.784152
Figure A-8: Results of the Johansen test for model4:
Date: 10/22/12 Time: 00:33 Sample (adjusted): 2001 M09 2012M06 Included observations: 130 after adjustments Trend assumption: Linear deterministic trend (restricted) Series: DELTA_VN FPI
Lags interval (in first differences): 1 to 12 Unrestricted Cointegration Rank Test (Trace)
Trace test indicates 1 cointegrating eqn(s) at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue)
Max-eigenvalue test indicates no cointegration at the 0.05 level
* denotes rejection of the hypothesis at the 0.05 level
**MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegrating Coefficients (normalized by b'*S11 *b=l):
Log likelihood -884.8721 Normalized cointegrating coefficients (standard error in parentheses)
Adjustment coefficients (standard error in parentheses) D(DEL TA_ VN) -0.863239
Figure A-9: Results of the Granger Causality test:
VEC Granger Causality/Block Exogeneity Wald Tests Date: 10/23112 Time: 00:17
Dependent variable: D(DEL TA_ VN)
Figure A-10: Results of the vector error correction model:
Dependent Variable: D(DEL TA_ VN) Method: Least Squares
Date: 10/22/12 Time: 23:39 Sample (adjusted): 2001M09 2012M06 Included observations: 130 after adjustments D(DELTA_VN) = C(1)*( DELTA_VN(-1) + 1.65037471191E-05*FPI(-1)- 0.000131596905459*@TREND(OOM07) + 0.00333129697177) + C(2)
*D(DELTA_VN(-1)) + C(3)*D(DELTA_VN(-2)) + C(4)*D(DELTA_VN(-3)) + C(5)*D(DEL TA_ VN(-4)) + C(6)*D(DEL TA_ VN(-5)) + C(7)
*D(DEL TA_ VN(-6)) + C(8)*D(DEL TA_ VN(-7)) + C(9)*D(DEL TA_ VN(-8)) + C(10)*D(DEL TA_ VN(-9)) + C(11)*D(DEL TA_ VN(-10)) + C(12)
*D(DELTA_VN(-11)) + C(13)*D(DELTA_VN(-12)) + C(14)*D(FPI(-1)) + C(15)*D(FPI(-2)) + C(16)*D(FPI(-3)) + C(17)*D(FPI(-4)) + C(18)*D(FPI(
Coefficient Std Error t-Statistic Prob
Adjusted R-squared 0.453983 S.D.dependentvar 0.125916 S.E of regression 0.093043 Akaike info criterion -1.734659 Sum squared resid 0.900323 Schwarz criterion -1.161152 Log likelihood 138.7528 Hannan-Quinn criter -1.501624 F-statistic 5.290260 Durbin-Watson stat 2.000673 Prob(F-statistic) 0.000000
WALT TEST TO TEST THE SHORT RUN CAUSALITY
Figure A-ll: Results of the Wald test:
C(14) C(15) C(16) C(17) C(18) C(19) C(20) C(21) C(22) C(23) C(24) C(25) Restrictions are linear in coefficients
Figure A-12: Results of the Serial Correlation test:
Breusch-Godfrey Serial Correlation LM Test:
Ob$*R-squared 0.020024 Prob Chi-Square(2) 0.9900
Dependent Variable: RESID Method: Least Squares Date: 1 0/23/12 Time: 00:04 Sample: 2001M09 2012M06 Included observations: 130 Presample missing value lagged residuals set to zero
Variable Coefficient Std Error t-Statistic Pro b
Adjusted R-squared -0.264511 S.D.dependentvar 0.083542 S.E of regression 0.093943 Akaike info criterion -1.704043 Sum squared resid 0.900185 Schwarz criterion -1.086421 Log likelihood 138.7628 Hannan-Quinn criter -1.453083 F-statistic 0.000582 Durbin-Watson stat 1.992890 Prob(F-statistic) 1.000000
Figure A-13: Results of the Heteroskedasticity test:
Dependent Variable: RESID"2 Method: Least Squares Date: 10/23/12 Time: 13:11 Sample (adjusted): 2001 M11 2012M06 Included observations: 128 after adjustments
Variable Coefficient Std Error t-Statistic c 0.006191 0.001308 4.733449
R-squared 0.008180 Mean dependent var Adjusted R-squared -0.007689 S.D.dependentvar S.E of regression 0.011266 Akaike info criterion Sum squared resid 0.015866 Schwarz criterion Log likelihood 394.0940 Hannan-Quinn criter
F-statistic 0.515493 Durbin-Watson stat Prob(F-statistic) 0.598470