The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach

74 Tạp chí Khoa học & Đào tạo Ngân hàng Số 217- Tháng 6. 2020 © Học viện Ngân hàng ISSN 1859 - 011X The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach Anh Thi Lam Pham Banking Academy of Vietnam Ngày nhận: 19/03/2020 Ngày nhận bản sửa: 15/05/2020 Ngày duyệt đăng: 19/05/2020 After the 2008 global financial crisis, financial stress index- an indicator measuring the instability and risk in financial markets has become one of the crucial indicators to forecast the financial crisis. Besides the models to estimate this index, the effect of financial stress on economic variables is the main topic that the economists focus on researching recently. For Vietnam, the economy experienced a double crisis in the period from 2008 to 2012. Besides the high inflation and the sharp decline in economic growth, the financial market also experienced a high risk and uncertainty period. Thus, whether there is a link between financial stresses and the decrease in economic growth in Vietnam is a big question. The study employs threshold vector auto-regression for the monthly data from 2008 to 2018 to find the answer to this question. The result indicates the existence of a threshold of financial stress index and the unusual association between financial stress and economic growth in Vietnam Keywords: financial stresses, economic growth, threshold vector auto- regression, TVAR Tác động của chỉ số áp lực tài chính đối với tăng trưởng kinh tế của Việt Nam-cách tiếp cận thông qua mô hình ngưỡng tự hồi quy Tóm tắt: Sau khi cuộc khủng hoảng tài chính toàn cầu năm 2008, chỉ số căng thẳng tài chính một chỉ số đo lường sự bất ổn và rủi ro trên thị trường tài chính đã trở thành một trong những chỉ số quan trọng để dự báo các cuộc khủng hoảng tài chính. Bên cạnh các mô hình để tính toán chỉ số này, tác động của căng thẳng tài chính đối với các biến số kinh tế là chủ đề chính mà các nhà kinh tế tập trung nghiên cứu. Đối với Việt Nam, nền kinh tế trải qua khủng hoảng kinh tế suốt trong giai đoạn 2008 đến 2012. Bên lạm phát cao và kinh tế suy giảm, thị trường tài chính việt nam trong giai đoạn này cũng trải qua rất nhiều rủi ro và bất ổn. Do đó, một câu hỏi đặt ra ở đây là liệu có mối liên hệ giữa căng thẳng tài chính và sản lượng của nền kinh tế tại Việt Nam hay không. Nghiên cứu sử dụng hồi quy véc tơ ngưỡng và dữ liệu hàng tháng từ năm 2008 đến hết năm 2018 để kiểm chứng mối quan hệ giữa hai biến số này. Kết quả cho thấy sự tồn tại của ngưỡng chỉ số căng thẳng tài chính và mối liên hệ bất thường giữa căng thẳng tài chính và tăng trưởng kinh tế ở Việt Nam. Từ khóa: Căng thẳng tài chính, Tăng trưởng kinh tế, mô hình ngưỡng tự hồi quy, TVAR Phạm Thị Lâm Anh Email: lamanh@hvnh.edu.vn Học viện Ngân hàng ANH THI LAM PHAM Số 217- Tháng 6. 2020- Tạp chí Khoa học & Đào tạo Ngân hàng 75 1. Introduction After the 2008 global financial crisis outbreak, financial stress – a concept measures the instability and risks in financial markets- have attracted more and more attention from the researchers. This is because; financial stress index- the indicator measuring financial stress did forecast quite precisely the appeal of the 2008 financial crisis. Unlike other financial concepts, different authors give a different definition of financial stress. llling and Lu(2006)- the pioneer in the studying of financial stress interpreted financial stress as the force affecting economic activity through instability and risk in financial markets and institutions. Hakkio and Keeton (2009) characterized financial stress as the increase in uncertainty about the fundamental value of assets and asymmetry of information as well as the decline in the demand for risky and illiquid assets. Balakrishnan et al. (2011) defined financial stress as a period of weakened financial intermediation. Aboura and Royeb (2017) described financial stress as a combination of uncertainty and risk perception. Although the researchers defined financial stress in various ways, they all measure the degree of risk of financial stress through four markets including: the banking sector, equity market, bond market, and foreign exchange market for financial stress index. Therefore, the financial stress index (FSI) characterizes for the change in the instability and risk of financial markets. Besides playing the role of the early warning signs of the financial crisis, financial stress affects economic activities. Aboura and Roye (2017) referred that financial stress caused to changing the behavior of private sector investment and consumption. Paries et al. (2011) indicated that increases in money market spreads would decline bank lending, which directly reduced economic output. David and Hakkio(2010) explained that financial stress would cause companies to postpone their decision of new investment in order to observe how uncertainty is overcome. Further, these authors also referred that the increase in financial stress would make companies’ financial condition worse because of the tightening in financial resources. As a result, firms would reduce their investment and profit. Thus, the rising of financial stress would lead the negative impacts on economic growth In the case of Vietnam, since 2008 and 2012, the Vietnamese economy faced a double crisis: inflation crisis and financial crisis. For inflation, this rate rocketed to about 20% in 2008 and 2011 and over 10% in 2007 and 2010. Besides inflation, in the period 2008 to 2011, Vietnam also faced a financial crisis with severe issues in banking sectors and the exchange rate market. From March to August of 2008, the interest rate climbed to nearly 20% while the interbank rate also rocketed to nearly 40%. In 2011, the banking sector also faced a crisis regarding liquid risk and the increase in the bad debt of the banking system. In this year, the exchange rate market also suffered the big shock that forced the Vietnamese State Bank had to depreciate Vietnam Dong 11%. As a consequence, the growth rate of the economic decline to 5% while this number for the period between 2004 and 2007 is 7% to 8%. This fact causes the authors to question whether or not there is a linkage between the instability in the financial market and economic The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach Tạp chí Khoa học & Đào tạo Ngân hàng- Số 217- Tháng 6. 202076 activities in Vietnam. In other words, whether the financial stress affects the output of the Vietnamese economy? To reach the final answer, firstly, this research will summarize some empirical studies on financial stress and economic growth. In the next part, employing the threshold vector auto-regression model, the paper will estimate the threshold of financial stress index before assessing the relationship between financial stress and output in each regime. Based on the result of the previous part, the last part will give some recommendations and end with some concluding points. 2. Literature review The impact of financial stress on economic growth was studied in various aspects. For the advanced economies, Liu and St-Amant (2010) used a threshold vector auto-regression for quarterly data from 1981Q4 to 2006Q4 to assess the effect of monetary policy on the real economy in the different scenarios of financial stress in Canada. The findings pointed out that in the high financial stress regime, the Canadian economy would experience weaker output growth, higher inflation, and higher interest rates. For the US economy, Hubrich and Tetlow, (2015), Davg and Hakkio (2010), Ferrer, et al. (2018), Galvao and Owyang (2018) used the different methods to estimate as well as employed the different model to analyze the relationship between financial stress and economic growth. However, they all reached the same finding that financial stress had negative effects on US economic growth. Roye (2013) used a dynamic approximate factor model to estimate FSI for Germany and examined the link between financial stress and economic growth through the threshold vector autoregression model. The results also indicated that high financial stress had significant adverse effects on output in Germany. Aboura and Roye (2017) applied different models- Markov- Switching Bayesian vector autoregression (MS-BVAR) for the French financial stress data; also found that episodes of high financial stress would lead to lower economic activity. Mittnik and Semmler (2013) indicated further finding in studying the group of advanced economies (the US and the five largest EU economies) with multi- regime vector auto-regression (MRVAR). Conducting a size-dependent response, the authors proved that stress –increasing shocks harmed economic activity in the high- stress period, then during low stress, which was only right for a small shock. When shocks are sufficiently large, in the high regime, the effect of large negative shock in financial stress on real activity is positive and sizable. Despite employing the different method for the different economies, these studies all found that high financial stress index would lead to lower output growth For the developing economies, Cevik et al. (2013) measured the financial stress index and studied the relationship between this index and economic activities in five transition economies, namely Bulgaria, the Czech Republic, Hungary, Poland, and Russian. The result refers that there is a moderate relationship between financial stress and some variables of economic activity. Cevik et al. (2016) concluded that financial stress had caused significant economic slowdowns after analyzing the effect of financial stress on economic ANH THI LAM PHAM Số 217- Tháng 6. 2020- Tạp chí Khoa học & Đào tạo Ngân hàng 77 activity in five emerging Asian economies. Tnga and Kwekb (2015) employed a structural vector autoregression (SVAR) for ASEAN- 5 economies and found that an increase in financial stress led to tighter credit conditions and lower economic activity in all these countries. The impact of financial factors on Vietnamese economic activities also has been studied in numerous researches. For the stock market, Vo et al. (2016) studied the linkage between financial structure and economic growth in Vietnam. The authors point out that the stock market development had litter impact on economic growth, and this relationship is a one-way effect from the stock market capitalization to economic growth. For the banking sector, Pradhan et al. (2014) show that in Vietnam, economic growth led to banking sector development or economic growth determines the level of banking sector development. Le and Pfau (2008), and Vo and Nguyen (2016) both concluded that banking credit is the primary monetary transmission channel in Vietnam. For the exchange rate market, Le and Pfau (2008) indicate that the exchange rate channel is one of the monetary transmissions in Vietnam, and the real effective exchange rate had to impact the change in the output of the Vietnamese economy in the period of 1996Q2 to 2005Q4. By contrast, Vo and Nguyen (2016) argued that the exchange rate channel would be weak and almost non-existent in Vietnam as a consequence of the government’s intervention in foreign exchange markets. For stock market volatility, although Vo (2015), Vo (2017), and Nguyen and Nguyen (2013) studied the volatility of the Vietnamese stock market, these authors did not show any evidence for the relationship between the stock market volatility and economic growth in Vietnam. Although the effect of financial factors on economic growth in Vietnam has been examined, the impact of financial stress index on economic activity has not investigated yet. This reason motives the author to employ threshold vector auto regression- model.- TVAR to examine the relationship between these two factors. 3. Methodology 3.1. Model selection Besides examining the effect of financial stress on economic growth, the study also looks for the threshold of financial stress index for the Vietnamese economy. Thus, following Liu and St-Amant (2010) and Roye (2013), this study uses threshold vector auto regression- TVAR model. The threshold VAR model with two regimes is Zt = α1 + A1Zt + B1(L)Zt-1 + (α2 + A2Zt + B2(L)Z t-1 ) I(C t-d , γ) + εt (1) Where the vector of variables (Zt) includes the Zt = (gapt, f sit, intt, cpit) gapt is output gap, f sit is the financial stress index (FSI), intt is the interest rate, cpit is the consumer price index I is an indicator that equals 1 if the threshold variable C t-d is larger than the FSI threshold value γ and 0 otherwise. When I = 0, the relevant coefficients are α1, A1 and B1(L) whereas represents the vector of constant, B1(L) represents the matrix of contemporaneous interaction The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach Tạp chí Khoa học & Đào tạo Ngân hàng- Số 217- Tháng 6. 202078 coefficient, represents the matrix of lag polynomials. When I = 1, the relevant coefficients are α1 + α2, A1 + A2 and B1(L) + B2(L). εt represents the vector of structural innovations. This paper employs the Tsay (1998) method to test for the threshold nonlinearity of the model. This approach generates the C (d) test statistic following by the estimation of an arranged regression. The null hypothesis that the model is linear: α2 = 0, A2 = 0, B2 = 0. C (d) follows a chi-squared distribution with k(pk+qv+1) degree of freedom. In this case, k and v represent the number of endogenous and exogenous variables; p and q are their corresponding lag orders. When the null hypothesis of linearity is rejected, this research utilizes a grid search method and Akaike Information Criteria (AIC) to find the thresholds. We utilize the Cholesky ordering for the shock identification in the VAR model. The first order is the GAP since GAP is the slow-moving variable. The second one is the FSI, and the interest rate is placed last. This structure is consistent with the empirical literature, which suggests that financial stress and monetary policy indicators are fast-moving market-based variables (Saldias, 2017). 3.2. Data We use the monthly data from 2008M2 to 2018M2 on Gap, FSI, CPI, and policy interest rate. Among these variables- output gap(GAP) represent for economic growth and economic output, as the monthly data of GDP is hard to measure, the study use industrial production index (IIP) to replace for GDP to account for economic growth. GAP is calculated through the change of industrial production index(IIP)and HP filter in Eviews software. In this study, we use financial stress index data for Vietnam calculated by the Asian Development Bank (ADB), which is based on Park and Mercado (2014) methodology. According to ADB (2019), the FSI for the Vietnamese economy is computed using measures for four major financial sectors with the equation presented as follows: FSI = β + Stockreturns + Stockvolatility + Debtspreads + EMPI The five FSI components in the equation come from five sectors and variables : banking Sector withincluding banking sector price index and stock price index; equity market returns including the current period’s equity return and its lag; equity market volatility; debt markets with sovereign debt spreads(=long-term Table 1. Variables and source Variables Source Financial stress index (FSI) Asian Development Bank website (www.adb.org) Consumer price index (CPI) General Statistics Office of Vietnam website (www.gso.gov.vn) Output gap( GAP) General Statistics Office of Vietnam website (www.gso.gov.vn) Policy interest rate (INT) State bank of Vietnam website (www.sbv.org) ANH THI LAM PHAM Số 217- Tháng 6. 2020- Tạp chí Khoa học & Đào tạo Ngân hàng 79 (10-year) local government bonds- US Treasuries in basis points) ; foreign exchange market with the exchange market pressure index (EMPI). All series are seasonally adjusted by using X-12 method and then taken in natural logarithm (except for the policy interest rate) before estimation. We also conduct the unit root test by using the Augmented Dickey-Fuller (ADF) test and Phillips- Perron (PP) test (Table 2). The results suggest that all series are stationary at first difference. After that, we set up the VAR estimation in the first difference. 4. Empirical results and discussion 4.1. Estimation of the inflation threshold In this section, we employ the method of Tsay (1998) to decide the financial stress threshold for Vietnam. Our objective is separating to the high and low financial stress regime using distinct sets of model parameters. Based on the value of the finan- cial threshold, the times series can be split into two different cases. When the financial stress threshold variable is higher than the critical value, the time points are classified as the high regime. Otherwise, the time points are classified as a low regime. Table 3 indicates the results of test statistic C (d) rejecting the null hypothesis of the linear relationship in all cases of different starting numbers for recursive estimates (m0 = 30 and m0 = 50). This implies that FSI is a suitable threshold variable, and that is worthwhile to split into two regimes. We decide the threshold lag delay (d) is 1, corresponding to delay by a month. For two-regime models, we assume the threshold γ ϵ [-4, .05] and use 300 grid points. The interval determination is based on the value of the threshold variable. The estimated threshold value for the output gap is 0.35, with the smallest AIC of (-143.00724). The low regime is active when the FSI is below the estimated threshold, 0.35. It presents the standard period of the economy, which is described by the economic growth and low financial market stress. In converse, the high regime indicates the economy moves to the slowdown situation characterized by high financial market stress. Figure 1 shows a plot of the estimated output threshold value and the threshold Table 2. Unit root test Variables ADF test PP test P-value Conclusion P-Value Conclusion GAP 0.0000 stationary 0.0001 Stationary FSI 0.094 Non stationary 0.0000 Stationary FSI at 1st difference 0.0000 Stationary 0.0000 Stationary CPI 0.0000 stationary 0.0001 Stationary Interest_ rate 0.0992 Non-stationary 0.2362 Non-stationary Interest_ rate at st difference 0.0001 stationary 0.0000 Stationary Source: Author’s computation in Eview 10 The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach Tạp chí Khoa học & Đào tạo Ngân hàng- Số 217- Tháng 6. 202080 variable. The estimated threshold value di- vides our sample into two regimes that are highly consistent with the economic devel- opment in Vietnam. The high episode dominates the period from 2008 to 2009. During this time, the Vietnam economy experienced a decline in economic activi- ties, the struggle of the banking system, and the reduction in the stock market. The low episode is captured by some period from 2010 to 2015 and the time after 2016. At that time, the economic grew and stabled; the reconstruction of the banking system has some progress. The empirical model endogenously selects the separation of the sample. 4.2. Impulse response analysis After splitting the sample into the high and low regime, we estimate the VAR model in each regime. We assume two lags in both cases. Figure 2 to 4 indicates the estimated impulse response functions over 12 months horizon in linear VAR, high regime, and low regime. In the case of the line VAR, FSI is the unique variable that effects on GAP, although this influence is not significant. CPI and IR do not show their impact on the growth of the industrial production index. On contrast, GAP is seen to have an impact on CPI and IR. In the high regime, FSI is shown to have a positive effect on GAP while GAP harms FSI, but this is not significant. GAP has a positive response to the increase in CPI, but CPI is not seen to not react to GAP. In this episode, GAP will increase in the short term with the positive IR shock then decrease, but IR is understood to not respond to the GAP shock. In the low regime (Figure 4), GAP will go Table 3. Result of the threshold test FSI threshold d C(d) p-value 1 30 57.36 0.013 1 50 66.76 0.001 γ 0.35 AIC -143.00724 Source: Author’s computation in RAT pro 8.0 *Note: d is a delay for the threshold variable, is starting number for recursive estimates, C(d) tests statistic based on the method of Tsay (1998). AIC is the Akaike Information Criterion 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 -6 -4 -2 0 2 4 6 Figure 1. FSI Source: Author’s computation in RAT pro 8.0 Note: The solid line illustrates FSI, the dotted line indicates the threshold value (0.35), and the shaded area is the high FSI period. ANH THI LAM PHAM Số 217- Tháng 6. 2020- Tạp chí Khoa học & Đào tạo Ngân hàng 81 up when it faces to positive IR and CPI shock, but GAP is seen to have no impact on both these variables. GAP almost has no- reaction to the FSI shock in this scenario. For linear VAR, high regime, and low regime case, GAP is seen to have a response to all other variables in senior regime while the reaction of GAP to other shocks in linear VAR and low regime case is quite the same. The positive response of GAP to the positive trauma of IR and CPI in all scenario shows that monetary policy seems to have a litter effect on output. 4.3. Discussion In general, the impacts of shocks in the Linear VAR R es po ns es o f gap cpi dfsi int gap gap cpi cpi dfsi dfsi int int 0 1 2 3 4 5 6 7 8 9 10 11 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0 1 2 3 4 5 6 7 8 9 10 11 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0 1 2 3 4 5 6 7 8 9 10 11 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0 1 2 3 4 5 6 7 8 9 10 11 -0.025 0.000 0.025 0.050 0.075 0.100 0.125 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 1 2 3 4 5 6 7 8 9 10 11 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Figure 2. Impulse response functions in linear VAR Source: Author’s computation in RAT pro 8.0 Note: The impulse responses (mid-solid line) are presented over a 12-month period along the horizontal axis. 68% confidence intervals based on Monte Carlo simulation are plotted around each response (as per Sims and Zha, 1995) High R es po ns es o f gap cpi dfsi int gap gap cpi cpi dfsi dfsi int int 0 1 2 3 4 5 6 7 8 9 10 11 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0 1 2 3 4 5 6 7 8 9 10 11 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0 1 2 3 4 5 6 7 8 9 10 11 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0 1 2 3 4 5 6 7 8 9 10 11 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0 1 2 3 4 5 6 7 8 9 10 11 -2 -1 0 1 2 3 4 5 0 1 2 3 4 5 6 7 8 9 10 11 -2 -1 0 1 2 3 4 5 0 1 2 3 4 5 6 7 8 9 10 11 -2 -1 0 1 2 3 4 5 0 1 2 3 4 5 6 7 8 9 10 11 -2 -1 0 1 2 3 4 5 0 1 2 3 4 5 6 7 8 9 10 11 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 Figure 3. Impulse response functions in high FSI regime Source: Author’s computation in RAT pro 8.0 Note: The impulse responses (mid-solid line) are presented over a 12-month period along the horizontal axis. 68% confidence intervals based on Monte Carlo simulation are plotted around each response (as per Sims and Zha, 1995) The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach Tạp chí Khoa học & Đào tạo Ngân hàng- Số 217- Tháng 6. 202082 two regimes are quite different. There is no doubt that the FSI threshold strongly affects the relationship among GAP, CPI, FSI, and interest rate under various states of the economy. However, the effect of FSI on the GAP in the case of Vietnam does not follow the economic theory, when in the high regime, the positive shock FIS would lead to the increase in GAP (economic growth). Although the reaction of GAP to FIS shock is not significant, the sign of reaction is still unusual, compared to previous studies. This adverse result can be explained by the following reasons. The first reason comes from financial stress data. The study used the ADB financial stress index data based on Park and Mercado (2014) methodology. For this FIS methodology, Park and Mercado’s (2014) estimated the instability in the banking sector through β = cov(r,m)/ var(m). In this case, r and m are the returns to the banking stock price index and the overall stock price index, respectively. The higher the banking sector β, the higher, the greater the banking sector stress. The advantage of this estimation is easy to collect the data of banking sector equity from the Vietnamese stock market database. However, the drawback of this measurement is that the number of banking equities in Vietnamese stock market in the period of 2008-2014 only accounted for small part in the number of bank in Vietnam. Thus, β might not represent fully the risk and instability in the banking sector as well as in the financial market in Vietnam. The second reason of the unusual the relationship between financial stress and economic growth may come from the lag choice of model. The study chose the lag for the model is only two, while Goktas and Hepsag (2011) indicated that the transmission of the stock market to economic activity would last within six months. Thus, two-month lags would not be enough time for the financial stress to transmit its negative impact to economic output and economic growth. However, 6 to 12 months lags may cause the model Low R es po ns es o f gap cpi dfsi int gap gap cpi cpi dfsi dfsi int int 0 1 2 3 4 5 6 7 8 9 10 11 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0 1 2 3 4 5 6 7 8 9 10 11 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0 1 2 3 4 5 6 7 8 9 10 11 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0 1 2 3 4 5 6 7 8 9 10 11 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0 1 2 3 4 5 6 7 8 9 10 11 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0 1 2 3 4 5 6 7 8 9 10 11 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0 1 2 3 4 5 6 7 8 9 10 11 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0 1 2 3 4 5 6 7 8 9 10 11 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 1.25 1.50 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 0 1 2 3 4 5 6 7 8 9 10 11 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 Figure 4. Impulse response functions in low FSI regime Source: Author’s computation in RAT pro 8.0 Note: The impulse responses (mid-solid line) are presented over 12 months along the horizontal axis. 68% confidence intervals based on Monte Carlo simulation are plotted around each response (as per Sims and Zha, 1995) ANH THI LAM PHAM Số 217- Tháng 6. 2020- Tạp chí Khoa học & Đào tạo Ngân hàng 83 to be unable to estimate the threshold. Therefore, the longer data for financial stress index is essential to access the effect of financial stress index on economic growth in Vietnam. These limitations may be a suggestion for further researches to develop a new method to calculate the volatility in the banking sector, as well as the new method to estimate financial stress for the bank- base economies. 5. Conclusion This paper extends the literature analyzing the impact of financial stress on economic activity. Earlier studies demonstrated that financial stress had a negative influence on the output of industrial in both advanced and developing economies. However, in the case of Vietnam, whether financial stress has an impact on economic activities has not been studied. Employing the threshold vector auto-regression, the paper finds that there exist financial stress index threshold which divides the Vietnamese financial stress index into two regimes: high regime. In the low regime, FSI has no influence on GAP measured by industrial production index (represent for economic activity) while in the high regime, the positive shock in FSI would follow by the increase in GAP. Although, this effect is not significant, the reaction of economic activity to the change in FSI does not follow the economic theory. This usual result might source from the method that is used to estimate FSI and the limitation of data. The study’s limitations suggest that researchers and policy-makers develop new methods to compute financial stress and conduct more research on the impact of financial stress on economic growth and other economic variables in Vietnam. The results of these studies will help Vietnamese policy-makers formulate macro-prudential policy for a financial system more comprehensively ■ References 1. Aboura, S., and Roye, B. 2017. Financial stress and economic dynamics: The case of France. International Economics, 149, pp.57-73 2. Balakrishnan, R., Danninger, S., Elekdag, S., and Tytell, I. 2011. The Transmission of Financial Stress from Advanced to Emerging Economies. Emerging Markets Finance and Trade, 47(sup2), pp.40-68 3. Cardarelli, R., Elekdag, S., and Lall, S. (2011). Financial stress and economic contractions. Journal of Financial Stability, 7(2), pp.78-97. 4. Cevik, E., Dibooglu, S. and Kutan, A. (2013). Measuring financial stress in transition economies. Journal of Financial Stability, 9(4), pp.597-611. 5. Cevik, E., Dibooglu, S. and Kenc, T. (2016). Financial stress and economic activity in some emerging Asian economies. Research in International Business and Finance, 36, pp.127-139. 6. Davig, T. and Hakkio, C. (2010). What is the effect of financial stress on economic activity? Economic Review, Federal Reserve Bank of Kansas City, issue Q II, pp 35-62. 7. Ferrer, R., Jammazi, R., Bolós, V. and Benítez, R., 2018. Interactions between financial stress and economic activity for the U.S.: A time- and frequency-varying analysis using wavelets. Physica A: Statistical Mechanics and its Applications, 492, pp.446-462. 8. Galvao.A and Owyang.M (2018). Financial Stress Regimes and the Macroeconomy. Journal of Money, Credit and Banking, 50(7), pp.1479-1505. 9. Goktas, O., and Hepsag, A. (2011).. Do stock returns lead real economic activity? Evidence from seasonal cointegration analysis. Economics Bulletin, 31(3): 2117–2127. 10. Hakkio,S.C. and Keeton, W.R. (2009). Financial stress: what is it, how can it be measured, and why does it matter?. Federal Reserve Bank of Kansas City Economic Review, Second Quarter, pp 5-50 11. Hubrich, K. and Tetlow, R., 2015. Financial stress and economic dynamics: The transmission of crises. Journal of The effect of financial stress index on the Vietnamese economic growth - A threshold auto regression approach Tạp chí Khoa học & Đào tạo Ngân hàng- Số 217- Tháng 6. 202084 Monetary Economics, 70, pp.100-115. 12. Illing, M. and Liu, Y. (2006). Measuring financial stress in a developed country: An application to Canada. Journal of Financial Stability, 2(3), pp.243-265. 13. Le, H. and Pfau, W. (2008). VAR Analysis of the Monetary Transmission Mechanism in Vietnam. Applied Econometric and International Development, 9(1), pp.165-179. 14. Li. F and St-Amant. P.(2010). Financial stress, monetary policy, and economic activity. Bank of Canada Working Paper 2010-12. May 2010. Avaliable at: https://www.bankofcanada.ca/wp-content/uploads/2010/05/wp10-12.pdf. 15. Mittnik, S. and Semmler, W. (2013). The real consequences of financial stress. Journal of Economic D ynamics and Control, 37(8), pp.1479-1499. 16. Nguyen, K. and Nguyen, D. (2013). The relationship between macroeconomic factors and stock market volatility: empirical evidence from the Vietnam stock market. Science and Technology Development Journal, 16(3), pp.86-100. 17. Paries, M.D., Sorensen, C.K., Rodriguez-Palenzuela, D., 2011. Macroeconomic propagation under different regulatory regimes: evidence from an estimated DSGE model for the euro area. International Journal of Central Banking, vol. 7(4), pages 49-113 18. Park, C. and Mercado, R., 2014. Determinants of financial stress in emerging market economies. Journal of Banking & Finance, 45, pp.199-224. 19. Pradhan, R., Tripathy, S., Chatterjee, D., Zaki, D. and Mukhopadhyay, B. (2014). Development of banking sector and economic growth: the ARF experience. Decision, 41(3), pp.245-259. 20. Saldias, M. (2017). The nonlinear interaction between monetary policy and financial stress. IMF Working Paper 17/184. 21. Sims, C. A., & Zha, T. (1995). Error bands for impulse responses. Working Paper No.95-6, Federal Reserve Bank of Atlanta. 22. Tng, B. and Kwek, K. (2015). Financial stress, economic activity and monetary policy in the ASEAN-5 economies. Applied Economics, 47(48), pp.5169-5185. 23. Tsay, R. S. (1998). Testing and modeling multivariate threshold models. Journal of the American Statistical Association 93, 1188-1202 24. Van Roye, B. (2013). Financial stress and economic activity in Germany. Empirica, 41(1), pp.101-126. 25. Vo, X. (2015). Foreign ownership and stock return volatility – Evidence from Vietnam. Journal of Multinational Financial Management, 30, pp.101-109. 26. Vo, X. (2017). Do foreign investors improve stock price informativeness in emerging equity markets? Evidence from Vietnam. Research in International Business and Finance, 42, pp.986-991. 27. Vo, X. and Nguyen, P. (2016). Monetary Policy Transmission in Vietnam: Evidence from a VAR Approach. Australian Economic Papers, 56(1), pp.27-38. 28. Vo, X., Nguyen, H. and Pham, K. (2016). Financial structure and economic growth: the case of Vietnam. Eurasian Business Review, 6(2), pp.141-154. trong tương lai, các doanh nghiệp hiện tại cần tăng cường các biện pháp nhằm đảm bảo khả năng tạo tiền, kiểm soát chặt chẽ mối quan hệ cân đối giữa sử dụng tiền và nguồn tiền, chủ động lập kế hoạch dòng tiền làm nền tảng tích cực cho việc khai thác nội lực và hiệu quả khi sử dụng các nguồn tài trợ từ bên ngoài nhằm tạo nên khoản thặng dư tiền trong kỳ, qua đó tăng số dư tiền nắm giữ. Ngoài ra, kết quả nghiên cứu còn gợi ý rằng các doanh nghiệp cần kiểm soát chặt chẽ quy mô doanh nghiệp, duy trì quy mô doanh nghiệp phù hợp với khả năng kiểm soát rủi tiếp theo trang 73 ro của bản thân doanh nghiệp; đồng thời các doanh nghiệp cần tăng cường học tập và tích lũy kinh nghiệm quản lý tài chính cũng như quản lý rủi ro kiệt quệ tài chính từ các doanh nghiệp có thâm niên hơn ■