INTRODUCTION
Problem statement
Recent discussions among economists and policymakers have highlighted the significant link between institutional frameworks and economic performance This renewed interest stems from a growing consensus that institutions play a vital role in determining economic outcomes The exploration of this relationship has been propelled by the emergence of New Institutional Economics, which traces its roots to Coase's 1937 work, underscoring that "when it is costly to transact, institutions matter."
Since 1995, research has highlighted the importance of institutions—defined as the formal and informal "rules of the game" (North, 1994)—in driving economic growth The central thesis is that without efficient institutions, traditional factors of production struggle to foster rapid growth, particularly in transition economies (Eicher, Garcia-Penalosa & Teksoz, 2006) Both theoretical and empirical studies indicate that institutions influence economic outcomes directly by altering costs or indirectly by shaping investment incentives Despite the wealth of research on the relationship between institutions and growth, institutional analysis remains an evolving field, indicating a need for further investigation (Brousseau & Glachant, 2008; Chang, 2006).
“before the institutional perspective can be fully operationalised” (Pelikan, 2003; Rodrik, Subramanian & Trebbi, 2004)
Vietnam, as a transition economy undergoing significant institutional reforms, provides a valuable context for examining the challenges faced by its non-state sector, particularly at the provincial level Since the implementation of the "Doi moi" economic reforms in 1986 and the introduction of the Enterprise Law in 2000, the non-state sector has experienced notable growth However, this sector continues to encounter various obstacles related to local government interpretations of central policies, which differ widely across provinces The complexity of laws and the necessity for numerous sub-law gazettes for enforcement contribute to these challenges, as local officials often exercise considerable discretion in interpreting regulations This high level of local authority discretion is a distinctive aspect of Vietnam's institutional reforms compared to other developing nations.
Vietnam's provinces exhibit significant disparities in economic performance, particularly between 2007 and 2011, with GDP growth rates ranging from a low of 2.92% to an impressive 29.52% Notably, Ba Ria-Vung Tau recorded the highest GDP figures of approximately VND122,000 billion in 2007 and VND170,000 billion in 2008, while Ho Chi Minh City consistently outperformed with GDPs of VND131,000 billion, VND151,000 billion, and VND158,000 billion from 2009 to 2011 In contrast, Bac Can and Lai Chau struggled with GDP figures below VND3,000 billion during the respective periods This stark contrast raises important questions about the role of institutions in driving the economic variations among Vietnam's provinces.
The concept of institutions varies among scholars, making it challenging to measure this multi-dimensional notion (Nelson & Sampat, 2001) The Provincial Competitiveness Index (PCI) is an annual composite index that reflects the local business community's perspective on governance quality across Vietnam's provinces, based on survey responses from private enterprises and hard data from published sources It evaluates nine governance aspects through corresponding sub-indices: Entry costs, Land access and security of tenure, Transparency and access to information, Time costs of regulatory compliance, Informal charges, Proactivity of provincial leadership, Business Support Services, Labor and Training, and Legal institutions Launched in 2005 by the Vietnamese Chamber of Commerce and Industry (VCCI) and the USAID-funded Vietnam Competitiveness Initiative (USAID/VNCI), the PCI serves as a vital tool for assessing the local regulatory environment from the private sector's viewpoint Schmitz, Dau, Pham, and McCulloch highlight the PCI's role in demonstrating the private sector's influence on provincial economic reforms, emphasizing a dynamic where the government proactively seeks private sector input, leading to effective governance.
Since 2012, provinces in Vietnam have engaged in a competitive effort to enhance their Provincial Competitiveness Index (PCI) scores, aiming to foster a more conducive environment for economic growth This situation prompts an essential inquiry into the correlation between improved PCI scores and actual economic performance in these provinces This study utilizes PCI and its sub-indices as proxy variables for institutional quality to analyze the effects of institutional implementation on the economic performance of Vietnam's provinces during the period from 2007 to 2011.
Research objective and research questions
This thesis aims to investigate the contribution of institutional factors to Vietnam’s provincial economic performance through answering the following research question:
- How much of the variation in cross-provincial GDP of Vietnam could be explained by institutional quality, both in a broad sense and as specific factors of economic governance?
This thesis examines the economic disparities among Vietnam's provinces by analyzing the impact of institutional factors, specifically the Provincial Competitiveness Index (PCI) and its sub-indices Key areas of focus include Entry Costs, Time Costs, Regulatory Compliance, Informal Charges, Land Access and Tenure Security, Legal Institutions, Transparency and Access to Information, Labor Policy and Training, Business Support Services, and the Proactivity of Provincial Leadership.
Contribution of the study
This study enhances the existing literature by providing the first empirical evidence of the impact of institutional factors on Vietnam's provincial economic outcomes, addressing a gap in research Unlike previous studies that focused on firm-level analysis within a single year, this research examines provincial-level data over a five-year period, using the Provincial Competitiveness Index (PCI) as a measure of institutional quality Additionally, the study incorporates the de facto overstatement of growth data into the estimation models, which contributes to more reliable regression results for the analysis.
Organization of the study
This thesis is organized into several key chapters: the second chapter examines the theoretical and empirical literature on the relationship between institutions and economic growth Chapter 3 outlines the data, research methodology, and empirical models used to assess how provincial institutional quality affects economic performance Chapter 4 presents and analyzes the regression results Finally, the concluding chapter summarizes the study's findings, discusses policy implications, and identifies limitations and avenues for future research.
LITERATURE REVIEW
Theoretical literature: The New Institutional Economics
In recent decades, there has been a renewed focus on institutional literature, particularly through the lens of New Institutional Economics (NIE) This approach has significantly advanced economic analysis by integrating institutional perspectives, leading to an extensive body of research that examines how the quality of institutions affects economic performance.
2.1.1 A renewed interest – some distinct features
Institutional economics, initially highlighted by Adam Smith, regained significance in the 1990s due to the shortcomings of neoclassical economics in addressing economic growth and development (Nye, 2008) The transition struggles of ex-Soviet regimes underscored the importance of incorporating institutional factors into mainstream economics (North, 1994) Furthermore, globalization has exacerbated inter-country inequality, prompting the World Bank to identify the establishment of effective market-supporting institutions as a key challenge for development policy at the turn of the twenty-first century (World Bank, 2002).
New Institutional Economics (NIE) builds on the foundational analyses of Williamson (1985) and North (1990), defining institutions as "governing structures" and "rules of the game." NIE posits that individuals operate with incomplete information and limited rationality, leading to uncertainty and transaction costs (Menard & Shirley, 2008) Consequently, humans establish both formal institutions, which are documented rules, and informal institutions, comprising social norms and beliefs Additionally, they create organizational frameworks that facilitate transactions and enhance cooperation in production and exchange, as noted by Brousseau and Glachant.
Since 2008, significant research has focused on the interplay between institutions and organizational arrangements, examining how these arrangements can, in turn, transform the established rules of the game (Menard & Shirley, 2008, p.12).
New Institutional Economics (NIE) distinguishes itself from Traditional Institutional Economics, originally defined by Hamilton in 1919, in two key aspects Firstly, NIE is founded on the neo-classical principles of scarcity and competition Secondly, while Traditional Institutional Economics lacked systematic theoretical foundations and empirical support, NIE is characterized by significant original contributions and is supported by extensive empirical research, as noted by Menard and Mary (2008).
New Institutional Economics (NIE) is not a unified theory but rather a synthesis of various concepts from different academic traditions (Brousseau & Glachant, 2008, p.2) Its development is the result of an evolutionary process, allowing for contributions from scholars across diverse social and scientific disciplines This openness fosters an evolutionary perspective and contributes to the heterogeneity of academic input within NIE.
The New Institutional Economics (NIE) recognizes the value of neoclassical economic frameworks while also identifying their limitations By utilizing established analytical tools, NIE expands the scope of economic analysis to include a wider range of institutional factors that influence economic performance, thereby enhancing our understanding of these complex dynamics.
2.1.2 Institutions matter for economic performance
Institutional factors play a crucial role in explaining economic growth disparities both between countries and within regions of a single country (Smith, 1976) As highlighted by North (1990), institutions serve as fundamental determinants of long-term economic performance, prompting extensive research into the question of how institutions influence economic outcomes.
The research on institutional issues within New Institutional Economics (NIE) can be categorized into three main theoretical approaches: The Historical Perspective Approach, Comparative Institutional Analysis, and Imperfect Information Theory.
The Historical Perspective Approach, introduced by North in 1990, emphasizes the importance of analyzing institutions within their historical context Recognizing that institutions are specific to their time and place, it is crucial to consider historical factors when addressing institutional change, as highlighted by Alston in 1996.
In his influential 1990 book, "Institutions, Institutional Change and Economic Performance," the author presents a theory that intertwines human behavior with transaction cost economics The core argument emphasizes the significance of institutions in driving economic performance by reducing uncertainty and minimizing transaction costs associated with social interactions, ultimately fostering economic growth.
The second strand of the theoretical framework, "Comparative Institutional Analysis," is extensively developed by Aoki (2001) Aoki integrates both Evolutionary and Repeated game theory into his analysis, defining institutions as "self-sustaining systems of shared beliefs" that alter the incentive structures of games This, in turn, impacts the strategic choices and interactions among agents.
A key aspect of New Institutional Economics (NIE) is the "Imperfect Information Theory," which posits that institutions arise from strategic behaviors influenced by asymmetric information among involved parties (Bardhan, 2000) This framework highlights how information and enforcement costs can lead to the absence of certain markets and a lack of competitiveness in others In this context, institutions serve as a solution to address missing markets and can also help mitigate information issues in existing ones (Hoff, Braverman, Stiglitz, & Arnott, 1993).
The fundamental distinction among the Historical Perspective, Comparative Institutional, and Imperfect Information approaches lies in their analytical tools for investigation The Historical Perspective seeks to integrate economic theory with economic history, while the Comparative Institutional approach effectively employs game theory alongside historical data In contrast, the Imperfect Information framework is the most mathematically focused of the three methodologies (Nabli & Nugent, 1989).
2.1.2.2 Mechanism and channels of influence
The above-mentioned theoretical approaches lay fundamental analytical frameworks for analyzing the “institutions matter” issue, specifically the institutional mechanism and channels of influence on economic performance
Institutions that govern property rights and contract enforcement play a crucial role in shaping economic performance As noted by Coase (1937), well-defined property rights can address issues like the prisoners' dilemma and enhance collective action These institutions establish norms and rules that determine control over returns from investments and outline exchange procedures, fostering internal security and trust By reducing the risks of expropriation, they significantly influence economic agents' decisions regarding savings and investments in both physical and human capital In essence, a lower risk of expropriation and higher security levels correlate with increased investment and economic growth, assuming other factors remain constant.
Theoretical model: the aggregate production functions
2.2.1 History of the production functions
The production function is widely recognized by scholars as a crucial analytical tool in neoclassical economics Schumpeter (1954) traces its origins to 1767, when Turgot, in his Observations on a Paper by Saint-Peravy, implicitly formulated the production function by discussing how variations in the proportions of standard factors of production influence marginal productivities This foundational concept was further developed by Philip Wicksteed.
In 1894, it was widely recognized that the economist who first algebraically formulated the input-output relationship was P = f(x₁, x₂, , xₘ) However, Humphrey (1997) presents evidence suggesting that Johann von Then actually developed the first production function formulation in the 1840s.
Between the 1950s and 1970s, the production function garnered significant interest from economists, leading to various analyses of input-output relationships and their implications (Mishra, n.d) Stigler (1952) highlighted that the logarithmic production function was originally introduced by Malthus, while Humphrey (1997) revealed that the Cobb-Douglas function was first presented in a "disguised form" by Von Then in The Isolated State, vol – II Additionally, Velupillai (1973) pointed out that Wicksell formulated a production function identical to Cobb-Douglas in his work from 1900-1901, and further elaborated on this in his 1923 review of Gustaf Akerman’s doctoral dissertation.
The equation P = AL^αC^β, where α + β = 1, represents a fundamental concept in production theory The Leontief production function, recognized for its significance, was collaboratively developed by notable economists such as Jevons, Menger, and Leon Walras It is important to note that the academic contributions of individual scholars often reflect the collective research efforts of their time, as highlighted by Russel (1984), emphasizing that these achievements are the result of a broader intellectual endeavor rather than isolated accomplishments.
The Constant Elasticity of Substitution (CES) production function, formulated by Arrow et al (1961), allows for a fixed elasticity of substitution between capital and labor, varying from zero to infinity, across isoquants regardless of output size This framework encompasses special cases such as Cobb-Douglas, Leontief, and linear production functions Classical forms of these functions assume Hicks-neutral technological progress, meaning the marginal rate of substitution between production factors remains unaffected by technical advancements or output levels Brown and Cani (1963) expanded the CES function to include non-constant returns to scale, while Uzawa (1962) and Kazuo Sato (1967) contributed to the integration of multiple production factors Additionally, Nervole (1963) and Ringstad (1967) generalized the Cobb-Douglas function to allow for variable returns to scale Diewert (1971) introduced a production function that accommodates variable elasticities of substitution, leading to a near-complete generalization of the classical Cobb-Douglas and CES functions by the mid-1970s.
The empirical estimation of the aggregate production function is a crucial aspect of macroeconomic analysis, yet it faces scrutiny from two main perspectives: the Cambridge capital controversy and the aggregation literature, as noted by Felipe and Fisher (2003) The Cambridge controversy questions how capital goods are measured in production functions, while the aggregation literature investigates the conditions under which micro-level neoclassical production functions can be combined into an aggregate form Given the discrepancies between the characteristics of individual firms and entire industries, as well as the broader economy, researchers must approach the identification of macroeconomic production functions with considerable caution.
Knut Wicksell (1923) supported the use of aggregate linearly homogeneous functions by demonstrating that non-homogeneous production functions for individual firms can align with a linear homogeneous function for the entire industry (Humphrey, 1997) This research advanced significantly with Koopmans' activity analysis (1979) and Georgescu-Roegen's aggregate linear production function.
The separation theorems and the generalization of Von Neumann’s model introduced in 1951, along with Nikaido's solid proofs from 1968, significantly reinforced the foundation for utilizing the aggregate production function in economic analysis.
The Cobb-Douglas production function remains a cornerstone in the analysis of growth and productivity, as noted by Felipe and Adams (2005) Its origins trace back to the groundbreaking research conducted by Cobb and Douglas in 1928, where they first econometrically estimated an aggregate production function using data from the U.S manufacturing sector spanning 1899 to 1922, presenting their findings to the economic community.
2.2.2 Properties of the production function
The production function reflects the physical relationship between inputs and outputs, which can be mathematically defined as
The production function has the following properties
There is no such thing as a "free lunch" in economics, emphasizing that resources are limited Additionally, the production function is characterized by monotonicity, which necessitates that all inputs yield positive marginal products in a single-output production scenario.
Thirdly, the production function is quasi-concave Notably, the monotonicity and quasi- concavity property of the production function cannot be imposed globally
Finally, for non-negative and finite x, f x( ) is, finite, continuous, non-negative and single-valued
Various functional forms are utilized in production function modeling, with the Cobb-Douglas and Translog functions being the most prevalent The Cobb-Douglas function is characterized by its linearity in logs, ease of estimation, and interpretation, and it operates under the assumption of constant proportionate returns to scale In this model, the elasticity of substitution between capital and labor is fixed at one, indicating a one-to-one percent relationship Conversely, the Translog function, which generalizes the Cobb-Douglas model, offers greater flexibility by providing a second-order approximation with fewer restrictions on production and substitution elasticities, making it a more widely adopted choice in empirical research.
Empirical literature
Over the past ten years, numerous empirical studies have utilized various forms of the production function to explore the relationship between institutions and economic growth across different countries Specifically, the production function is frequently employed in the context of analyzing this critical issue.
The "extended production function" integrates institutional proxies with traditional growth factors, as highlighted by Glaeser et al (2004) This specification also encompasses various growth determinants, including trade openness, geography, and macroeconomic policy, which influence institutional effects on economic performance (Jacob & Osang, 2007) The choice of dependent variables introduces further heterogeneity, categorized into output-level and output-growth related variables Eicher and Leukert (2006) found that output levels exhibit more robust effects than output growth Additionally, the selection of institutional proxies and the treatment of their potential endogeneity contribute significantly to this heterogeneity Institutional factors are often represented by standardized indices and survey-based indicators at the national level Aidis et al (2009) and Shirley (2008) note that the multi-dimensionality and disparity among these proxies raise concerns about their validity in representing institutions accurately Overall, empirical research on institutions is marked by substantial heterogeneities, making it difficult to derive a consistent empirical impact across the literature (Efendic, Pugh & Adnett, 2011).
Dias and Tebaldi (2012) emphasize the crucial role of institutional interactions and human capital accumulation in driving economic growth Utilizing panel OLS and GMM dynamic panel estimation techniques, their study analyzes data from 61 countries spanning 1965 to 2005 The research incorporates democracy and autocracy as indicators of political institutions, while the share of educated labor serves as a measure of structural institutions These institutional factors are integrated into a formal growth model alongside a human capital variable developed based on Hall and others.
Jones (1999) introduced a piecewise function, while Easterly and Levine (2001) calculated capital stock using the Perpetual Inventory Method Empirical evidence supports the idea that strong structural institutions enhance productivity and contribute to sustained economic growth.
Siddiqui and Ahmed (2013) analyze the impact of institutional factors on economic performance across 84 countries from 2002 to 2006, utilizing the Institutionalized Social Technologies (IIST) index and its sub-indices, including Institutional and Policy Rents, Political Rent, and Risk Reducing Technologies Their extended growth model incorporates human capital, savings, and trade openness, applying OLS and GMM methods The introduction of interactive variables reveals significant complementarities between institutions that safeguard property rights and political institutions The findings indicate that robust institutions play a crucial role in enhancing economic outcomes.
Atul and Sal (2012) utilize Solow’s growth accounting model alongside World Bank regulation indicators to examine the influence of regulatory frameworks on economic performance By applying fixed and random effects estimation to data from 23 OECD countries between 2002 and 2008, their findings reveal that high-quality regulation positively impacts growth by enhancing total factor productivity.
Gagliardi (2008) reviews key theoretical and empirical literature on how institutional frameworks influence economic performance, utilizing New Institutional Economics approaches such as historical perspective, institutional comparison, and imperfect information theory The concept of institutional complementarities is emphasized, highlighting its significant policy implications Despite challenges in measuring institutions, empirical evidence supports the notion that a country's institutional framework is essential for its economic development.
In his 2008 study, Sobel utilizes state-level cross-sectional data from the U.S and OLS regression to explore the connection between institutional quality and entrepreneurial productivity, measured through the economic freedom index and net entrepreneurial productivity index The research incorporates geographic and demographic control variables, including the median age, population density, male population percentage, and the proportion of college-educated individuals Sobel's findings provide the first empirical support for Baumol’s theory, demonstrating that strong institutions facilitate productive entrepreneurship, ultimately explaining the variations in economic performance among different states.
Hasan, Wachtel, and Zhou (2007) explore the relationship between China's institutional development and economic growth, focusing on financial deepening, legal institution development, and political pluralism They use proxies such as the size of the private sector, rule of law, and property rights awareness, measured through the ratio of private sector fixed investment, the number of lawyers per 10,000 people, and the ratio of domestic trademark applications to firms Employing Barro's (1991) growth equation, the study analyzes an annual dataset from 31 provinces between 1986 and 2002 using OLS and GMM estimation methods The findings indicate that institutional development significantly contributes to the variations in provincial economic growth.
Jalilian, Kirkpatrick and Parker (2007) investigate the role of regulatory framework on economic growth in developing countries Data set covers 117 countries for the 1980 –
Between 1980 and 2000, a study analyzed data from 96 countries using cross-section and panel regression methods to evaluate the impact of regulatory quality and government effectiveness on GDP per capita growth The analysis controlled for factors such as initial GDP per capita, gross capital formation, education, trade, inflation, and government spending The findings support the theoretical literature, demonstrating a strong correlation between a robust regulatory framework and economic growth.
Using Vietnam’s firm-level data in 2005 and the Provincial Competitiveness Index
In their 2009 study, Tran, Grafton, and Kompas examine the impact of Vietnam's 2000 institutional reforms on the economic performance of the non-state sector, using 2006 as a proxy variable for local governance practices Their model incorporates control variables, including provincial initial endowments like human capital and market size, alongside dummy variables to account for firm-specific characteristics such as size, age, capital intensity, and ownership type The findings reveal that enhanced institutional quality—particularly in areas like market information, land tenure security, and labor training—significantly boosts firms' labor productivity.
Several studies challenge the idea that institutions drive economic growth Notably, Glaeser et al (2004) reexamined this proposition using cross-country data from 1960 to 2000 and found no supporting evidence Their research highlights that existing literature on the relationship between institutions and growth often suffers from poorly measured institutional variables and the use of irrelevant instrumental variables.
In her 2007 study, Jenny Minier analyzes the indirect effects of institutions on economic growth using panel data from 70 countries between 1960 and 2000, challenging the assumption of parameter invariance prevalent in most empirical research The study employs executive constraint as a proxy for institutions and includes control variables such as physical capital accumulation, education, geography, and initial income Utilizing the endogenous sample splitting technique proposed by Hansen (2000), Minier assesses the varying parameters of growth determinants, acknowledging the existence of thresholds The findings indicate minimal support for the threshold effect of institutions on production factors through parameter heterogeneity; instead, they suggest that institutions primarily affect the marginal impact of policy variables on economic growth.
DATA AND RESEARCH METHODOLOGY
Data and variable measurement
This study utilizes secondary data from various published sources, focusing on investment and GDP data from the Provincial Statistics Year Book for the periods of 2005-2009 and 2010-2011 Labor force data is sourced from the national General Statistics Office (GSO) at www.gso.gov.vn Additionally, scores for the Provincial Competitiveness Index (PCI) and its sub-indices are obtained from PCI Annual Reports spanning 2007 to 2011, along with information available on their official website.
The Provincial Competitiveness Index (PCI) is developed through a collaborative effort between the Vietnam Chamber of Commerce and Industry (VCCI) and the United States Agency for International Development (USAID), managed by Development Alternatives, Inc (DAI) The PCI creation involves three key steps: collecting survey-based and hard data, constructing sub-indices, and calibrating these sub-indices to accurately reflect their impact on private sector development It utilizes perception indicators from surveys alongside hard data from published sources, aiming to balance the subjective nature of perception data with the objectivity of hard data By integrating both types of data, the PCI research team seeks to enhance the reliability of the index while minimizing potential biases.
To collect and process "soft" data, a stratified sampling method is utilized at the provincial level, focusing on business age, sector, and legal form to ensure a representative sample of the local business community with a 3% sampling error The sample size varies between 7,000 and 10,000 private domestic firms due to fluctuations in response rates over the years Since 2010, the survey has also included over 1,000 foreign-invested firms Additionally, the survey questions, tailored for domestic and foreign-invested firms, are regularly updated to reflect significant changes in the provincial business and regulatory landscape.
It is worth noticing that alterations were implemented to PCI methodology in 2009 Complying with the principle set up from the outset that construction measures could be
The PCI Annual Report (2009) highlights that while "tweaked at the margins," significant methodological revisions should be approached with caution to maintain consistent over-time comparisons Key changes include the removal of the SOE bias sub-index amid the equitization of local state-owned enterprises, minor adjustments to indicators within each sub-index, and a new method for calculating sub-index weights to prevent accidental correlations Consequently, the subsequent data analysis and estimation model will be designed to effectively capture the institutional impacts on provincial economic performance resulting from these methodological changes.
This study utilizes annual data from Vietnam's 58 provinces for the period of 2007 to 2011, focusing on the Provincial Competitiveness Index (PCI) Although PCI data is available from 2005 to 2011, the years 2005 and 2006 were excluded due to inadequate provincial statistics for key variables like GDP and investment Additionally, six provinces—Ha Tay, Cao Bang, Hung Yen, Thanh Hoa, Long An, and Ninh Binh—were removed from the analysis due to insufficient data for the variables of interest.
Real Gross Domestic Product (RGDP) is a key economic indicator that represents the annual outcomes of production and business activities within a province It is calculated at constant prices by adjusting nominal GDP values, based on current prices, using the provincial annual Consumer Price Index (CPI), with 2005 as the base year RGDP is expressed in millions of Vietnamese Dong (VND).
Due to the unavailability of annual provincial Consumer Price Indices (CPIs) for five out of 58 provinces—specifically Dac Nong, Hau Giang, Hoa Binh, Lang Son, and Phu Yen—national CPIs are utilized as substitutes when calculating real GDP and investment figures for these regions.
LABFO refers to the labor force, encompassing individuals aged 15 and older who are actively engaged in business and production activities within the province This metric is quantified in thousands of people.
CAP denotes capital input variable Applying the Unified approach of the Perpetual
The Inventory Method (PIM) introduced by Berlemann and Wesselhửft (2012) utilizes provincial nominal investment values to generate the capital input variable As defined by the General Statistics Office (GSO), investment refers to the expenditures made to enhance and sustain material assets over a specific timeframe, typically one year This investment can be assessed through various projects and national programs designed to boost both fixed and liquid assets.
In this study, real values of investment are obtained by deflating nominal values using provincial annual CPIs, with 2005 as the base year
The net capital stock at the beginning of period t, let sayK t , could be mathematically expressed as
Where K t 1 is the net capital stock at the beginning of the previous period t1
D t is the consumption of fixed capital of the previous period
In the current period, gross investment is analyzed under the assumption of a constant depreciation rate of 5% Consequently, the capital stock can be recalculated to reflect these conditions.
Repeatedly substituting this equation for the capital stock at the beginning of period t1,
To determine the initial value of the capital stock (K t-1) for the base year, we utilize the steady-state approach proposed by Harberger (1978) Assuming our economy operates at a steady-state, we calculate the GDP growth rate accordingly.
Thus making it possible to derive the function of K t 1 as
It is worth-noticing that in computingg GDP , we use average GDP growth rate with the omission of g GDP outliers
In our analysis of the initial value of investment, we utilize the method proposed by Nehru and Dhareshwar (1993) in their application of the Steady State Approach This approach involves regressing the time-series of log investments on time using Ordinary Least Squares (OLS), represented by the equation lnI_it = α + β_i + i_t + ε_it.
And then the fitted value for the first period investment is obtained
By applying the exponential function to the fitted values, we obtain a complete time series of investment values essential for calculating capital stock The initial fitted value from this investment time series is utilized to determine the capital stock for the base year according to formula (3.5).
This procedure offers a significant advantage over the traditional steady-state approach by allowing the initial value of the investment time series to be generated without depending solely on the investment figure from any single year, resulting in a more reliable initial value.
For all the institution variables, relevant information is obtained via PCI Annual Report from
The Provincial Competitiveness Index (PCI) is a comprehensive 100-point scale that evaluates and ranks the provinces of Vietnam based on their economic governance quality This index is developed through a three-step process: Collection, Construction, and Calibration Initially, survey data is adjusted using reliable published sources to mitigate perception bias This data is then standardized into ten-point sub-indices In the final calibration step, these sub-indices are assigned weights of 15%-20% for high, 10% for medium, and 5% for low impact, reflecting their contribution to sector growth, investment, and profitability For example, the 2009 PCI report highlights that the Transparency & Access to Information and Labor Policy sub-indices receive the highest weight of 20% due to their significant influence on business outcomes, while Land Access, Legal Institutions, and Private Sector Development are assigned the lowest weight of 5% due to minimal score variation across provinces Additionally, the Time Cost of Regulatory Compliance sub-index is notable with a 15% weight, while other sub-indices receive 10%.
Research methodology
Choosing the right functional form is a crucial step in developing an estimation model According to Coelli et al (2005), an effective functional form should be flexible, linear in parameters, regular, and parsimonious A regular functional form adheres to the economic regularity properties of production functions, either inherently or through simple restrictions The two most commonly used production functions are Cobb-Douglas and Translog While the Cobb-Douglas function is parsimonious, it lacks flexibility; conversely, the Translog function offers flexibility but at the cost of parsimony A parsimonious function conserves degrees of freedom and represents the simplest solution to a problem, whereas a flexible functional form imposes fewer assumptions and restrictions on the characteristics of the production function.
The Cobb-Douglas production function is generally expressed as follows: i i i y x (3.8)
The Cobb-Douglas functional form is characterized by constant proportionate returns to scale and constant elasticity of factor substitution, with the assumption that all pairs of inputs are complementary These features contribute to its restrictive nature.
The Translog production function is a generalization of the CD function It is a flexible functional form providing a second order approximation, which is generally expressed as below:
ln i ln i 0.5 ij ln i ln j i i j y x x x
The Translog production function offers unrestricted monotonicity and factor dependence, lacking constant proportionate returns to scale Unlike the Cobb-Douglas function, it allows for testable properties and is deemed more realistic due to its flexibility However, its complexity arises from the inclusion of cross and squared terms, which increases parameter counts and may lead to high correlations among them, particularly problematic with limited observations.
Applying the Cobb-Douglas and Translog functional form to model the institutional effects on GDP, the model specification is constructed as
2009*ln it it it it it it
Y cap labfo INSTITUTION t dum INSTITUTION
(3.10) for the Cobb-Douglas functional form and
11 ln ln ln ln 2009*ln
0.5 ln( ) 0.5 ln( ) 0.5 ln( ) 0.5 ( ) 0.5 ln *ln
0.5 ln *ln it it it it it it it it it it it it
Y cap labfo INSTITUTION t dum INSTITUTION cap labfo INSTITUTION t cap labfo cap INSTITUTION
0.5 ln * 0.5 ln * it it it it it it cap t labfol INSTITUTION labfo t INSTITUTION t
The INSTITUTION variable encompasses PCI and its nine sub-indices: Encost, Timeco, Informal charges, Landacc, Leinsti, Infoacc, Pridevelop, Proact, and Labpol, which are integrated into both models Additionally, the variable 't' is utilized to represent technological change.
The variable Dum2009 is a dummy variable coded as 1 for the year 2009 and later, and 0 for prior years To account for a potential structural break in the relationship between institutions and growth, the interaction term Dum2009*lnINSTITUTION is incorporated into both models.
In the context that provincial growth data have been so far assumed to be target-driven, we conduct a check for such a de facto overstatement as illustrated in Table 3.4
Table 3.4: Overstatement of provincial growth data
Source: Calculated from Provincial Statistical Yearbooks
We calculated the GDP growth rate for each province using the proportion of provincial GDP to national GDP as weights, comparing our findings with data from the General Statistics Office (GSO) The results, presented in Table 3.3, reveal a significant discrepancy due to the overstatement of provincial data Consequently, we adopted various estimation methods to better assess the institutional impacts on provincial GDP, acknowledging the proven exaggeration in growth statistics.
We begin by utilizing Ordinary Least Squares (OLS) regression on cleaned panel data, following outlier detection Additionally, we implement the Stochastic Frontier Analysis model to effectively address both statistical noise and data overstatement through its component error terms.
3.2.2.1 Data cleaning – OLS regression for panel data
We initiate our data cleaning process by utilizing graphs, which serve as effective tools for visualizing the overall trends in our data According to Asterious & Hall (2007), a thorough graphical examination is essential for any robust empirical analysis Consequently, we commence our outlier detection for growth data by analyzing a variety of graph types.
We analyzed the individual time-series plots of real GDP for 58 provinces over five years to identify significant jumps in values at specific points in time This analysis revealed four within-variation outliers, specifically from the provinces of Dien Bien, Hai Duong, and Phu Tho in 2011, as well as Hoa Binh in 2010.
To identify potential outliers in the data, provincial real GDP per labor was plotted against the annual mean across five graphs for each year Can Tho and Vung Tau displayed significant data spikes compared to other provinces, leading to their exclusion from the sample.
In the final step of our data cleaning process, we applied the 3σ rule, eliminating any data points that exceeded their mean by more than three standard deviations As a result, we removed three additional data points from Dien Bien (2011), Hai Duong (2011), and Quang Ngai (2010), bringing the total number of excluded observations to 15.
Then, we employ panel data models of either Fixed Effect Model (FEM) or Random Effect Model (REM) and run OLS regression afterwards
The FEM is generally expressed as
The fixed-effects model (FEM), also known as the least-squares dummy variables (LSDV) model, permits each cross-section unit to possess a unique intercept value (β1i), which remains constant over time, hence the term "fixed-effects." Importantly, the estimators used in this model are consistently reliable.
Meanwhile, in the random-effects model, the intercept 1i is assumed to be a random variable with mean value of 1 and could be expressed as
The random error term, denoted as εi, has a mean of zero and a variance of σ², representing the individual differences in the intercept value for each entity This component is commonly known as the cross-section or individual-specific error term.
The model could be re-written as
The Random Effects Model (REM) assumes no correlation between the error component and the explanatory variables, which is crucial for consistent estimation of regression coefficients The choice between the Fixed Effects Model (FEM) and REM hinges on the correlation assumption; if the error components are uncorrelated with the explanatory variables, REM is considered more efficient To assess the appropriate model, the Hausman test is employed, indicating whether the error term is correlated with the explanatory variables in a specific regression analysis.
Null hypothesis: FEM and REM estimators do not differ significantly
The test statistic follows an asymptotic chi-square distribution, and if the calculated chi-square value surpasses the critical threshold, the null hypothesis is rejected, indicating a preference for the Fixed Effects Model (FEM) over the Random Effects Model (REM).
3.2.2.2 Modeling the overstatement of growth data – The Stochastic Frontier Analysis (SFA)
Where y it = provincial GDP that we aim to measure x it = controlling variables, including standard factors of production and institutional elements