This paper investigates the relationship between public investment and private investment to determine whether public investment crowds in or crowds out private investment. In the research, the Structural Vector Autoregressive (SVAR) model will be employed to analyze effects of structured variables in the model.
ECONOMIC DEVELOPMENT No 205, September 2011 DOES PUBLIC INVESTMENT CROWD IN OR CROWD OUT PRIVATE INVESTMENT IN VIETNAM? by Assoc Prof., Dr SỬ ĐÌNH THÀNH * Public investment is the primary channel for economic infrastructures During the economic transition, Vietnam’s government has promulgated fiscal expansion policies by increasing public investment in the hope that public investment might positively influence the economic productivity as a whole and crowd in private investment Vietnam, in recent years, has tried its best to control and cut back on public investment in an attempt to curb inflation Thus what is the effect of cutbacks in public investment on private investment? This paper investigates the relationship between public investment and private investment to determine whether public investment crowds in or crowds out private investment In the research, the Structural Vector Autoregressive (SVAR) model will be employed to analyze effects of structured variables in the model The research finds that public investment helps crowd in private investment in a long run Finally, some solutions will be recommended for improving public investment policies of Vietnam in future Keywords: public investment, private investment, crowding-in, crowding-out effect, FDI, gross investment Theoretical framework There are two economic schools concerning effects of public investment on private one The neoclassical economics hold the view that expansion of public investment can improve the investment climate and cut back on cost for private investment via an improvement in infrastructure, quality and supply of human resource When the economy operates below its potential level, raising the aggregate demand via expansion of public investment can crowd in private investment (IMF, 2007) If this theory would be correct, it is probable to assert that public investment facilitates private one Public investment, however, might also crowd out private investment The crowding-out effect can take place via three channels (IMF, 2007) Firstly, when the economy obtains full employment; public investment financed by tax rises and local borrowings can reduce private investment Secondly, the increased public expenditure financed by borrowings from the financial system or issue of bonds can result in a decrease in credit for the private sector and push up the market rate of interest, which hardly encourages private investment Thirdly, there are some preferential fields for the public sector or state-owned enterprises wherein private *University of Economics - HCMC RESEARCHES & DISCUSSIONS 27 ECONOMIC DEVELOPMENT No 205, September 2011 investment might work more effectively For economies in transition, mechanisms and policies in favor of state-owned enterprises together with the expansion of state-owned corporations have produced “crowding out” effect, which becomes a great obstacle to development of private sector It is possible to analyze the relationship between public and private investment via the Cobb – Douglas production function as follows: (1) Yt At Lt K t 1 KGt1 Where, t denotes time, Y is GDP, A expresses productivity, L represents the total labor, K is the capital from private sector and KG is public investment, and , , are parameters Equation (1) shows that the marginal product (output) of private investment is Yt / K t 1 and that of public investment is Yt / KGt 1 If the private sector and the government accumulate optimum capital, then the marginal product is equal to the real interest rate of each sector (Toshiya Hatano, 2010) If q and r represent the interest rate in the private sector and the public one respectively, we have: qt Yt / K t 1 , and r Yt / KGt 1 (2) Because both interest rates are under the market pressure, it is possible to assume that they go in parallel with each other even though there is a small difference originated from risk premiums Suppose that these two interest rates maintain the difference via the constant for qt rt , we have: K t ( / ) KGt (3) If the constant equals 1, the marginal product of public investment is equal to that of private investment Equation (3) means that a stable long-term relationship does exist between the marginal product of public investment and that of private investment with a fixed risk premium After taking the logarithm of (3), such the long-term relationship can be illustrated in linear form With c0 = log – log – log wherein c1 is the estimated parameter and et is error, Equation (2) can be rewritten as: 28 RESEARCHES & DISCUSSIONS logKt = co + c1logKGt + et (4) Equation (4) implies that if c1 is positive, public investment crowds in private investment Contrarily, if c1 is negative, public investment will have the crowding-out effect on private investment Public and private investment in Vietnam since 1990 to the present time Over the past two decades, the gross investment has played a crucial part in expansion and growth of Vietnam’s economy The 1997 financial crisis had curbed the economic growth rate that then rose a little higher, around 8.5% in the 2005-2007 period The gross investment also showed the same trend, that is, a steep fall to 32% of GDP in the period 19981999 and a strong rise up to 43% of GDP in 2007 After the 2008 global financial crisis, the ratio of gross investment to GDP was maintained at some 40%; yet the economic growth rate showed a downward trend, plunging to 6% (see Figure 1) Figure 1: Growth rate of Vietnam’s economy and investment Source: GSO Figure shows that the investment share of the private sector including FDI and domestic private investment fell around 40% in the late 1990s due to effects of the 1997 crisis; meanwhile, the share of public investment jumped to 60% The trend has recently reversed in part That is, the share of private investment seems a little higher than public investment which can be proven through the recent recovery of domestic private investment and FDI The domestic private investment rose from 8% of ECONOMIC DEVELOPMENT No 205, September 2011 Table 1: Output share in GDP by sectors (%) 1995 1997 1998 1999 2000 2001 2003 2005 2006 2007 2008 2009 Public sector 40.18 40.48 40.00 38.74 38.52 38.40 39.08 38.40 37.39 35.93 35.54 35.13 Domestic private sector 53.52 50.45 49.98 49.03 48.20 47.84 46.45 45.61 45.63 46.11 46.03 46.54 6.30 9.07 10.03 12.24 13.28 13.76 14.47 15.99 16.98 17.96 18.43 18.33 FDI Source: GSO GDP in the period 1996-2001 to 14% - 15% in the period 2006-2010 FDI, after increasing rapidly during the economic reform program in the late 1980s, sharply fell from 10% in 1997 to 5% in post-crisis years, and then were kept stable at 10% of GDP in 2006-2010 period The foreign sector has emerged as the economic dynamic for the recent decade The contribution of FDI to GDP increased from 9% in 1997 to 18% in 2009 whilst the contribution of public investment went from 40% in 1997 down to 35% in 2009 Figure 2: Public and private investment and FDI in Vietnam Source: GSO Although private investment has quickly regained its speed, public investment still represents about 50% of gross investment Public investment includes budget investments, state credits (for financing projects via issuing bonds and not included in the budget income), and investments from state-owned enterprises After the 1997 crisis, public investment increased to around 20% of GDP in the period 2000-2005 when the government concentrated on infrastructure improvement (i.e traffic roads, bridges, electric and water supply system, schools, irrigation and drainage, and rural communications), and poverty alleviation The amount of public investment for traffic works constituted a substantial ratio, from around 30% in the late 1990s to 40% in recent years ODA and government bonds have become vital capital sources supporting public investment In addition, the banking system also becomes an important source of finance for state-owned enterprises (IMF, 2007) In the period 2006-2010, public investment was cut down to 15% - 17% of GDP with a view to weathering the high inflation rate after the 2008 global financial crisis Under such circumstance, public investment might have generated crowding-out effects on private one through capital transmission channels of the financial system Table shows that net credit for the government was approximately zero in the period 2000-2001, yet then went up to 4% of GDP in the period 2005-2006 At the same time, the government also issued more bonds to finance budget deficits As a result, the gross public debts increased by five times, from 5% of GDP in 2000 to 18.7% in 2006 Furthermore, credits for state-owned enterprises incessantly went up, from 10% of GDP in 1996 to 22% of GDP in 2006 When the public sector absorbs more and more savings, private investment might be crowded out However, Table shows that bank credit provided for the private sector was higher than that for the public sector, i.e from some 9% of 1996 GDP to 49% of GDP in 2006 Yet during this time, the money supply also soared up, from 25% of GDP in 1996 to 95.5% of GDP in 2006 RESEARCHES & DISCUSSIONS 29 ECONOMIC DEVELOPMENT No 205, September 2011 Table 2: Credit structure of Vietnam’s financial system (as % of GDP) 1996 Domestic credit 1997 1998 1999 2000 2001 2003 2005 2006 20.3 21.3 22.4 28.9 35.1 39.7 51.7 69.9 75.6 Government 1.6 1.4 2.3 0.7 -0.1 0.4 3.3 3.9 3.8 State-owned enterprises 9.9 9.9 10.5 13.6 15.8 16.6 15.9 21.6 22.6 Private sector 8.9 10 9.6 14.6 19.4 22.7 32.4 44.4 49.2 Domestic / / / / 5.1 6.2 13.7 17.4 18.7 Foreign / / / / 27.9 29.8 27.3 26.6 25 23.8 26 28.4 40.1 50.5 58.1 67 74.4 95.5 Public debt backup Money supply (M2) Source: IMF (2007) That is to say, the financial system could boost the supply of credit for both private and public sector Deposit in saving accounts surpassed the credit supply, and thereby causing no rise in the interest rate and reducing the risk of financial crowding-out effects (IMF, 2007) In sum, the aforementioned data set could not illuminate whether or not public investment would crowd out private one Although the healthy growth of the financial system in recent years has met the increasing demand of both private and public sector for credits, some private enterprises have faced certain difficulties in accessing to bank borrowings due to the fact that the government and state-owned enterprises have borrowed a great deal from the local financial market In order to identify exactly the existence of crowding-in or crowding-out effects, it is crucial to evaluate changes and interaction between public and private investment over the past years Research model In order to determine whether public investment has crowding-in or crowding-out effects on private one, the SVAR model will be employed for GDP, public (or government) investment (GI), private investment (PI) and FDI The VAR model is not employed because it does not help evaluate impacts of public investment on private one and GDP (Pritha 30 RESEARCHES & DISCUSSIONS Mitra, 2006) Assume that the VAR model will be: C ( L) X t u t (5) where C(L) is a polymonial matrix with the constrained lag time, L is the lag operator, Xt is the set of variables ((Xt = [logGI, logGDP, logPI, logFDI]) and ut is the vector of residues In the VAR model, residues are assumed to be correlated, and thus a shock on a certain variable may affect simultaneously on others It implies that it is impossible to extract shocks of a certain variable (public investment for example) on private investment The SVAR model identifies the relationship between residues of the VAR model (unexpected shocks) and structural shocks (i.e shocks from independent or exogenous variables which are not related to each other) The dynamics of the variables are explored with the following VAR structure for the SVAR model (IMF, 2007): l l l l GI t GI iGI GI t i iGI FDI t i iGI PI t i iGI GDPt i utGI i 1 i 1 l l i 1 l i 1 l FDIt FDI iFDI GI t i iFDI FDI t i iFDI PI t i iFDI GDPt i utFDI i 1 l i 1 i 1 i 1 l l l PI t PI iPI GI t i iPI FDI t i iPI PI t i iPI GDPt i utPI i 1 i 1 l l i 1 i 1 l l GDPt GDP iGDPGIt i iGDP FDI t i iGDP PIt i iGDPGDPt i utGDP i 1 i1 i 1 i 1 The SVAR model that is run to determine the relationship between residues of the VAR model and structural shocks can be summed up as follows: ECONOMIC DEVELOPMENT No 205, September 2011 Aet But (6) Where ut is the vector of structural shocks, A and B are the matrix that establishes the relationship between structural shocks and residues of the VAR model (et) This system is defined via imposing constraints on A and B (as per economic theories) and needs evaluating Through (6), the relationship between residues of the VAR model and structural shocks of the SVAR model is specified as follows: etGI C (1)utGI (7) etFDI C ( 2)etGI C (3)utFDI (8) etPI C ( 4)etGI C (5)etFDI C (6)utPI GDP t e GI t FDI t C (7)e C (8)e PI t C (9)e C (10)u (9) GDP t (10) Equation (7) reflects the supplementary relationship between GI and FDI Public investment creates an attractive climate for FDI whilst FDI prospects encourage the government to boost public investment Therefore, an independent shock of GI (an increase in GI for example) is expected to affect FDI Similarly, an independent shock of FDI (such as an increase in FDI), as shown in Equation 8, expectedly influences GI Equation (9) takes into account the relationship between PI and GI, and between PI and FDI PI may respond to the positive independent shock of GI either positively (for example, an increase in GI may attract PI) or negatively in case there is a crowding-out effect The spillover effects of FDI on PI are also included with a view to mulling an overall shock of FDI on PI Eventually, as per the definition of GDP, an overall shock of GDP is the function of GI, PI and FDI (equation 10) Data set, estimate and research results a Data set: The annual ratio of GDP, GI, PI and FDI in the period 1990-2010 are collected from the GSO website for estimation (Table 3) Then, we get the logarithm of variables GDP, GI, PI and FDI so as to determine their range over time Table 3: Data for the model: GDP, GI, PI and FDI in 1990-2010 (as %) Years GDP GI/GDP PI/GDP 1990 FDI/GDP 0.05094 0.06075 0.07035 0.01890 1991 0.05809 0.05871 0.08383 0.02511 1992 0.08700 0.06845 0.08546 0.05974 1993 0.08078 0.11865 0.09270 0.07573 1994 0.08834 0.10290 0.09581 0.09163 1995 0.09541 0.13302 0.08738 0.09612 1996 0.09340 0.15768 0.08014 0.08345 1997 0.08152 0.17081 0.07812 0.09661 1998 0.05765 0.18014 0.07701 0.06731 1999 0.04774 0.19242 0.07887 0.05669 2000 0.06787 0.20246 0.07833 0.06152 2000 0.06787 0.20246 0.07833 0.06152 2001 0.06895 0.21187 0.08002 0.06236 2002 0.07080 0.21416 0.09447 0.06495 2003 0.07341 0.20631 0.12126 0.06243 2004 0.07790 0.19548 0.15344 0.05780 2005 0.08442 0.19260 0.15538 0.06089 2006 0.08229 0.18999 0.15807 0.06734 2007 0.08457 0.17311 0.17898 0.11314 2008 0.06311 0.14076 0.14615 0.12839 2009 0.05324 0.17338 0.14478 0.10925 2010 0.06700 0.15962 0.15114 0.10825 Source: GSO b Stationary test and lag test: Testing the SVAR model requires a unit root test or a stationary test of time-series data The Augmented Dickey-Fuller (ADF) test is employed to test whether or not time-series data of all variables are stationary in accordance with the following hypothesis: H : => There exists a unit root and the time-series data is non-stationary H1 : 0 => There is no unit root and the time-series data is stationary The important point is that if t-stat for yields a negative value larger than 5% of RESEARCHES & DISCUSSIONS 31 ECONOMIC DEVELOPMENT No 205, September 2011 tabulated critical value, H0 will be rejected; and accordingly there is not a unit root and the timeseries data is stationary; and vice versa The testing results are reported in Table Table 4: ADF testing results Variables Lag LogGDP LogFDI LogPI LogGI D(logPI) D(logGI) D(DlogGI) 1 0 t- stat for -3.45 -3.33 -0.787 -2.5 -3.06 -2.5 -4.4 5% of tabulated critical value -3.00 -3.00 -3.00 -3.00 -3.00 -3.00 -3.00 Unit root Absent Absent Present Present Absent Present Absent Table shows that logFDI and logGI are stationary with the significant level of 5% Meanwhile, logPI is stationary at first difference and logGI at second difference with the significant level of 5% Thus, the first and second difference of these variables is included in the SVAR model After running a unit root test, we test sufficient conditions for stability of VAR model and find that the VAR model satisfies stability conditions and no root lies outside the unit circle Testing for cointegration using Johansen method proves that variables are cointegrated with the significant level of 5%, and thus the lag time will be tested next Based on AIC (Akaike Information Criterion), SC (Schwarz Information Criterion), and HQ (Hannan-Quinn Information Criterion), the optimal lag time opted for the SVAR model is c Research results: Based on stationary time-series data including d(dlogGI), logGDI, dlogPI, and logGDP, the Eviews7 is employed to run the VAR model and then estimate primary parameters of the SVAR model The estimation results are shown in Table Table 5: Evaluation results of the SVAR model Coefficient Std Error z-Statistic Prob C(2): GI -> FDI 1.348414 0.718334 1.877141 0.0605 C(4): GI -> PI 1.412444 0.430131 3.283754 0.0010 32 RESEARCHES & DISCUSSIONS C(5): FDI ->PI -0.237501 0.132174 -1.796881 0.0724 C(7): GI -> GDP -0.933317 0.823345 -1.133568 0.2570 C(8): FDI ->GDP 0.548043 0.215885 2.538583 0.0111 C(9): PI->GDP 0.307661 0.363154 0.847192 0.3969 With the correlation of 1.34 and the statistical significance of 10%, it seems that public investment has crowding-in effects on FDI Public investment in infrastructure and human capital has had a certain additional contribution to FDI Regarding the impulse response function, the positive shock of GI affects FDI in one to two years (Figure 3) before being diminished This result implies that GI might have supported FDI in a short and medium term Response of LOGFDI to DGOI -.1 -.2 -.3 -.4 10 Figure 3: Response of FDI to GI With the correlation coefficient of 1.41 and the significant level of 5%, it is possible to assert that public investment crowds in private domestic investment Figure figures out the progression of the relation between those two variables The impulse response function of PI to GI shocks goes down in first 1.5 to years and then shows an onward trend This implies that GI crowds out PI in short term and crowds in PI in long term As far as the relationship between PI and GDI is concerned, response of PI to FDI shocks is negative and has the correlation coefficient of -0.23 and statistical significance of 10% Figure illustrates the negative spillover effects of FDI on PI in term of 1.5 to years Although the impulse response function of PI shows an onward trend, it is still below zero point This result is different from conclusions proposed by Tueä Anh et al (2006) who supposed that FDI had positive spillover effects on local small and medium-sized enterprises ECONOMIC DEVELOPMENT No 205, September 2011 Response of DPI to DGOI 20 15 10 05 00 -.05 -.10 -.15 -.20 10 10 Figure 4: Response of PI to GI Response of DPI to LOGFDI 20 15 10 05 00 -.05 -.10 -.15 -.20 Figure 5: Response of PI to FDI Response of GDI to unexpected shocks of GI (0.93) and PI (0.3) is not statistically significant In other words, unexpected shocks to GDP not simultaneously go with unexpected shocks to PI and GI Response of GDP to unexpected shocks of FDI (0.54) has the statistical significance at 5% Conclusion and policy implications a The relationship between public investment and private one has been the matter of concern of many economists Some recent researches on crowding-out effects of public investment on the private sector of developing economies (i.e Khan et at [2009] on the case of Pakistan, and Majumder [2007] on the case of Bangladesh) show that government’s borrowings not crowd out yet crowd in private investment The case study of Mitra (2006) on India figures out that there is a short-term crowding-out effect and government’s investment has positive impacts on the economy in long term In Vietnam, over the past two decades, public investment has absorbed a large share of domestic saving, representing around 50% of the gross investment Within the scope of the research, estimation results show that it is not sure to assert whether public investment crowds out or crowds in private one In other words, public investment may have crowded in FDI at least in short term and PI in long term Therefore, (i) the development of financial system will have certain positive impacts on capital mobilization, and thereby meeting the growing demand for credit by both the public and private sectors and overcoming the crowding-out effect of public investment on private one; and (ii) increasing public investment to remove some of the infrastructure bottlenecks may facilitate flows of domestic private and foreign direct investments in long term b Our empirical results show that response of GDP to shocks of public and private domestic investment is not statistically significant It implies that the cutback on public investment in hopes of controlling high inflation at present does not sharply impinge on the economic growth rate, and that the government should reform its public investment policies with a view to facilitating the long-term sustainable economic development To so, firstly, the macroeconomic stabilization must be considered as the government’s functional target of top priority; and public investments, therefore, should be put in infrastructural facilities, institutional development, and human resources instead of businesses Simultaneously, it is necessary to clarify the function of the government as well as that of budget-financed investment and public investment (including preferential credit supplied to state-owned enterprises) The government functions as an administrator of a country and thus is not supposed to directly business in the market economy Public spending on a pure business objective should be gradually cut down and then expunged Government credits for investment that mostly come from public debts should be used for developing infant industries while public investment is spared for social RESEARCHES & DISCUSSIONS 33 ECONOMIC DEVELOPMENT No 205, September 2011 objectives and other vital fields that the private sector does not involve in Secondly, financial management mechanism of state-owned enterprises should be reformed and financial privileges of the public sector should be reduced State-owned enterprises that not operate for profit can be financially supported at the start, and those that business for profit must take out commercial loans Thirdly, it is necessary to tighten the borrowing and use of public investment capitals by allowing participation and supervision by local communities and professional organizations with suitable expertise Keynote projects should be provided with access to necessary information and rights to discuss, advise, and supervise the implementation of the projects Fourthly, investment from national budget should be sent only to major projects instead of being divided among too many projects; and decisions on investment should only be made when sources of finance are ensured In allocating investments from budget, top priority must be given to infrastructure projects in key economic zones to facilitate the even development among regions and bridge the gaps between them Finally, public-private partnership in development of infrastructures should be beefed up and perfected so as to assure the benefits of both public and private partners; and enable local private enterprises to participate in developing infrastructure via preferential policies The research also finds that FDI has positive effects on GDP with the correlation coefficient of 0.54 and the statistical significance level of 5% However, there is seemingly a negative spillover effect on PI (Sig = 10%) Therefore, in order to maximize benefits of FDI, it is necessary to work out appropriate policies to attract FDI FDI policies, in time to comes, should attend to the negative spillover effect on private domestic investment by FDI This is to say, the government needs to facilitate the development of small- and medium-sized enterprises and support them in collaborating with foreign- 34 RESEARCHES & DISCUSSIONS invested enterprises and accessing to state-ofthe-art technologies Simultaneously, the government should enhance “research and development” (R&D) competence of local enterprises to improve the acquisition of new technologies and expedite the technological transfer via R&D experts training and exchange programs among institutes, universities and enterprises for mutual sake References International Monetary Fund (2007), Vietnam: Selected Issues, IMF Country Report No.07/385 Khan, Rana Ejaz Ali & A.R Gill (2009), Crowding-out Effect of Public Borrowing: A Case of Pakistan, Pakistan: The Islamia University of Bahawalpur Majumder, B (2007), Does Public Borrowing Crowdout Private Investment? The Bangladesh Evidence, Policy Analysis Unit (PAU), Research Department, Bangladesh Bank, Working Paper Series: WP 0708, March 2007 Nguyễn Thò Tuệ Anh et al (2006), “The Impacts of Foreign Investment on the Economic Growth in Vietnam”, Research report within the SIDA Project: Capacity Building Project for Policy Research to Implement Vietnam's SocioEconomic Development Strategy in the Period 2001-2010 Pritha Mitra (2006), Has Government Investment Crowded Out Private Investment in India? available at www.aeaweb.org/annual_mtg_papers/ /0108_1015_0102 pdf Toshiya Hatano (2010), “Crowding-in Effect of Public Investment on Private Investment”, Public Policy Review, Vol.6, No.1, Policy Research Institute, Ministry of Finance, Japan www.gso.gov.vn ... public investment crowds out or crowds in private one In other words, public investment may have crowded in FDI at least in short term and PI in long term Therefore, (i) the development of financial... speed, public investment still represents about 50% of gross investment Public investment includes budget investments, state credits (for financing projects via issuing bonds and not included in. .. negative, public investment will have the crowding -out effect on private investment Public and private investment in Vietnam since 1990 to the present time Over the past two decades, the gross investment